El Nino Report

Preparing for the
2015-2016 El Nio:
Humanitarian Action in Zambia,
Somalia, Kenya, Ethiopia and Malawi
A Report for DFID By
Arielle Tozier de la Poterie
Julie Arrighi
Eddie Jjemba
Photo: Charlotte Floors/NLRC

With Support From
Stephen McDowell
Shaban Mawanda
Roop Singh
Red Cross Red Crescent Climate Centre
Final Report | October 2017
Contact the Climate Centre at: info@climatecentre.org

This is a Research Report produced by the Red Cross Red Crescent Climate Centre and partners.
This report was funded by the UK Department for International Development, through the Science
for Humanitarian Emergencies and Resilience (SHEAR) Programme. The views expressed do not
necessarily reflect the UK governments official policies or the official views of the Red Cross Red
Crescent Climate Centre, the National Societies in focus countries, or any other component of the
Red Cross Red Crescent Movement.
Preparing for the
2015-2016 El Nio:
Humanitarian Action in Zambia,
Somalia, Kenya, Ethiopia and Malawi
A Report for DFID By
Arielle Tozier de la Poterie, Julie Arrighi and Eddie Jjemba
With Support From
Stephen McDowell, Shaban Mawanda and Roop Singh
Red Cross Red Crescent Climate Centre
Final Report
October 2017
Contact the Climate Centre at: info@climatecentre.org

Table of Contents
Acronyms 8
Acknowledgements 10
Executive Summary 11
Chapter 1: Introduction 15
Purpose & Objectives 15
Research Team 16
Case Study Selection & Informant Sampling 16
Methodology 17
Limitations 19
Chapter 2: Zambia Case Study 21

Background 21
Zambian Forecast Production 21
El Nio Impacts in Zambia 22
Response to the 20152016 El Nio Forecasts in Zambia 23
Linking Social Protection to Climate Shocks 25
Conclusion 26
Chapter 3: Somalia Case Study 28

Background 28
Forecasts in Somalia 28
Impacts of El Nio in Somalia 29
Response to 20152016 El Nio Forecasts in Somalia 30
Early Warnings 30
Determining Likely Impacts 31
Contingency Plans 31
Funding for the Early Action 32
Actions by Organization 32
ICRC and SRCS 32
BRCiS 33
WFP 34
FAO 34
CARE International Somalia 34
Forecasts in the Context of Insecurity & Protracted Crises 35
Conclusion 35
4 Table of Contents
Chapter 4: Kenya Case Study 38
Background 38
Forecasts in Kenya 39
Impacts of El Nio in Kenya 40
Response to 20152016 El Nio Forecasts in Kenya 41
First warning 41
Contingency plans 41
Response 43
Social Protection in Kenya 44
Taking Advantage of Opportunities: Seed Distribution
win Kitui County 44
Conclusion 45
Chapter 5: Ethiopia Case Study 48

Background 48
Forecasts in Ethiopia 49
Impacts of El Nio in Ethiopia 49
Response to 20152016 El Nio Forecasts in Ethiopia 50
Obstacles to Acting on Forecasts 52
Conclusion 52
Chapter 6: Malawi Case Study 54

Background 54
Forecasts in Malawi 54
Impacts of El Nio in Malawi 55
Response to El Nio Forecasts in Malawi 56
Contingency Planning 56
Response 57
Forecast Limitations 59
Social Protection and El Nio Response in Malawi 60
Conclusion 61
Chapter 7: Measuring Impacts 64

Attributing Severe Weather Events to El Nio 64
Framework for Measuring Impacts: Options and Considerations 65
Framework for Measuring Benefits 65
Measuring Benefits in Kitui 66
Quantifying Value: Attributing Event Outcomes to Mitigating Harmful Impacts 68
Framework for Measuring Impacts Avoided 68
Comparing to Analogue Years 68
Comparing Against Other El Nio Events 68
Measuring Impacts Avoided in Somalia 69
Table of Contents 5
Chapter 8: Summary of Recommendations 72
Salient Recommendations 72
Chapter 9: Detailed Write-up of

Conclusions and Recommendations 74
Forecast Production, Dissemination, and Use 74
Which Sources Are Most Used? 75
There Can Be Too Much Information 75
Which Information is Most Useful? The More Localised and
Specific the Better 76
Limitations of Scale 77
Probabilities & Uncertainty 77
Confusion Over Terciles and Probabilities Was
Widespread Among the Informants 78
Interpretation Facilitates Action 79
Forecast Production, Dissemination, Use, and Content
Conclusions & Recommendations 81
Trust 82
Did the 2014 False Alarm Affect Early Actions in 2015? 83
Trust Conclusions & Recommendations 84
Gradual Action 84
Supplement Forecasts with Real-time Information 85
Gradual Action & Monitoring Conclusions & Recommendations 86
Funding 86
Slow Funding Slows Early Action in Response to Forecasts 86
Flexible Funding Facilitates Early Action 87
Funding in the Context of Competing Disasters 88
Impact Studies 88
Funding Conclusions & Recommendations 89
Triggers and Standard Operating Procedures 89
Risks of (De)Politicising Forecasts 89
Triggers and SOPs Conclusions & Recommendations 90
No-regrets Actions 90
No-regrets Conclusions & Recommendations 92
Social Protection Systems 92
Social Protection Systems Conclusions & Recommendations 93
Political Considerations 94
Political Considerations Conclusions & Recommendations 94
Other Influences and Findings 94
6 Table of Contents
Contingency Planning 94
Negative Experiences with El Nio Motivate Action 95
Realising Potential Benefits 95
Collaboration Facilitates Early Action and Response 96
Prioritising Action Based on Consecutive Events 96
Forecasts Can Narrow Humanitarian Focus Too Much 96
Flood Forecasts Prompt Faster Action than Drought Forecasts 97
Other Considerations Conclusions & Recommendations 98
References 99
Annex A: Overall Timeline 101
Annex B: Terms of Reference 104
Annex C: Interview Instruments 109
Table of Contents 7
Acronyms
APES Agricultural Production Estimate Survey
BRC British Red Cross
BRCiS Building Resilient Communities in Somalia
CARE Cooperative for Assistance and Relief Everywhere
CERF Central Emergency Response Fund
Cesvi Cooperazione e Sviluppo
CHF Common Humanitarian Fund
COF Climate Outlook Forums
DAG Development Assistance Group
DCCMS Department of Climate Change and Meteorological Services
DFID Department for International Development
DMMU Disaster Management and Mitigation Unit
DoDMA Department of Disaster Management Affairs
DRC Danish Refugee Council
DRM Disaster Risk Management
DRMTWG Disaster Risk Management Technical Working Group
ECHO European Civil Protection and Humanitarian Aid Operations
ECMWF European Centre for Medium-Range Weather Forecasts
ENSO El Nino Southern Oscillation
ERSC Ethiopia Red Cross Society
EU European Union
EWEA Early Warning Early Action
FAO Food and Agriculture Organization
FEWS NET Famine Early Warning Systems Network
FIRP Food Insecurity Response Plan
FISP Farm Input Subsidy Program
FSNAU Food Security and Nutrition Analysis Unit
FSNWG Food Security and Nutrition Working Group
GFCS Global Framework for Climate Services
GFCS-APA Global Framework for Climate Services Africa Program
GHACOF Greater Horn of Africa Climate Outlook Forum
HCT Humanitarian Country Team
HSNP Hunger Safety Net Program
HRD Humanitarian Requirements Document
IASC Inter-Agency Standing Committee
IAWG Inter-Agency Working Group
ICPAC IGAD Climate Prediction and Application Centre
ICRC International Committee of the Red Cross Red Crescent
IEC Information, Education and Communication
IFRC International Federation of Red Cross and Red Crescent Societies
IGAD Intergovernmental Authority on Development
IPC Integrated Food Security Phase Classification
IRC International Rescue Committee
IRI International Research Institute for Climate and Society
JFM January, February, March
KSh Kenya Shillings
KIRA Kenya Inter-Agency Rapid Assessment Tool
KMD Kenya Meteorological Department
KRSC Kenya Red Cross Society
8 Acronyms
LEAP Livelihoods Early Assessment Protection
Met Meteorological
MNSP Malawi National Social Programme
MoAIWD Malawi Ministry of Agriculture, Irrigation and Water Development
MRSC Malawi Red Cross Society
MVAC Malawi Vulnerability Assessment Committee
NDMU National Disaster Management Unit
NDOC National Disaster Operations Centre
NDVI Normalized Difference Vegetation Index
NGO Non-Governmental Organization
NMA National Meteorological Agency
NMS National Meteorological Services
NMHSs National Meteorological and Hydrological Services
NOAA National Oceanic and Atmospheric Administration
NRC Norwegian Refugee Council
NSNP National Safety Net Program
OFDA Office of US Foreign Disaster Assistance
OND October, November, December
PDNA Post-disaster Needs Assessment
PICSA Participatory Integrated Climate Services for Agriculture
PSNP Productive Safety Net Programme
RC / RC Red Cross Red Crescent
RCAT Red Cross Action Team
RUTF Ready to Use Therapeutic Food
SADC Southern Africa Development Community
SARCOF Southern Africa Regional Climate Outlook Forum
SCI Save the Children International
SCT Social Cash Transfer
SIDA Swedish International Development Cooperation Agency
SMS Short Message Service
SomRep Somalia Resilience Programme
SOPs Standard Operating Procedures
SRCS Somalia Red Crescent Society
SWALIM Somalia Water and Land Information Management System
TAMSAT Tropical Applications of Meteorology using Satellite data and ground-based
observations
ToR Terms of Reference
UKMO United Kingdom Meteorological Office
UN United Nations
UNDP United Nations Development Programme
UNICEF United Nations Childrens Fund
UNOCHA United Nations Office for the Coordination of Humanitarian Affairs
USAID United States Agency for International Development
VAM Vulnerability Assessment and Mapping
WASH Water, Sanitation and Hygiene
WFP World Food Programme
WHO World Health Organization
ZMD Zambian Meteorological Department
ZRCS Zambia Red Cross Society
ZVAC Zambia Vulnerability Assessment Committee
Acronyms 9
Acknowledgements
The lead author is grateful to the research team Eddie Jjemba, Shaban Mawanda,
Stephen McDowell, Roop Singh, and Julie Arrighi for their roles in helping to design the
research, conduct interviews, and provide feedback on the analysis.
The research team thanks Catalina Jaime, Erin Coughlan, Friederike Otto,
Elena Camilletti, Cecilia Costella, and Megan Bailey for their assistance with interview tools,
report drafts, and other feedback.
Finally, this report would not have been possible without the gracious participation of
each of the 74 informants who took time from their busy schedules to speak with us.
This is a Research Report produced by the Red Cross Red Crescent Climate Centre and
partners. This report was funded by the UK Department for International Development,
through the Science for Humanitarian Emergencies and Resilience (SHEAR) Programme.
The views expressed do not necessarily reflect the UK governments official policies or the
official views of the Red Cross Red Crescent Climate Centre, the National Societies in focus
countries, or any other component of the Red Cross Red Crescent Movement.
10 Acknowledgements
Executive Summary
The Department for International Development (DFID) commissioned the Red Cross Red
Crescent Climate Centre to conduct the following study of humanitarian preparedness
measures for the 20152016 El Nio event in five countries: Zambia, Somalia, Kenya,
Ethiopia, and Malawi. Chapters 26 of this study detail country-level measures to prepare
for potential El Nio related impacts. Chapter 7, addresses the challenges to and benefits of
early action as experienced by two organizations working in Somalia and Kenya. Based on
common themes trending across the case studies, Chapter 8 provides recommendations
to improve preparedness actions, before a potential disaster, following El Nio forecasts in
the future. The study concludes with Chapter 9, which provides detailed recommendations.
Summary of Case Study Findings
Organisations operating in all five countries found El Nio event messaging and associated
forecasts to be timely and credible. Not surprisingly, downscaled forecasts that include
detailed descriptions of potential impacts across sectors are considered more useful
than vague seasonal forecasts (above normal, normal, below normal precipitation).
Informants also noted that the gradual accumulation of information from many sources
is more important than any one warning or source. Stakeholders expressed a desire for
better information (see Chapter 9), but that did not hinder early action. What enabled or
impaired early action by organizations in each country was not information but the strength
of coordination mechanisms, funding mechanisms, the nature of the hazard, access to
forecast experts or intermediaries and the political climate.
Nearly every organization used forecasts to update its internal and national
contingency plans (often using data from analogue years to anticipate potential impacts). In
many cases, organizations had contingency plans but were not able to fully execute them
for lack of funds.
While organizations could access impact data from analogue years, they did not
have access to forecast experts or intermediaries. This was another oft-touted challenge.
Stakeholders desperately needed and requested help to interpret forecasts, anticipate
impacts, and plan for action. Unfortunately, these resources were scarce.
Initial El Nio forecasts triggered the first stages of response; they allowed
humanitarians to begin contingency planning. As seasonal forecasts became more relevant
(localised), organizations would gradually escalate their response accordingly. The most
common response to forecasts was pre-positioning staff and supplies.
Informants from every country credited access to flexible funding and programme
modification with their organizations ability to act sooner. In particular, crisis modifiers
empower organizations to modify existing programmes and take action more urgently.
Because uncertainty is inherent in forecast predictions, organizations and donors
are adopting a no-regrets approach to forecast-based early action. No-regret activities are
those that will build resilience or provide valuable benefits regardless of whether or not a
disaster occurs. No-regret actions based on forecasts can, therefore, reduce potential,
negative, political consequences of acting based on a forecast.
Of the countries in this study, only Kenya successfully used their existing safety net
system to deliver timely aid in response to forecasts. This is likely because, unlike Kenya,
Ethiopia, Malawi, and Zambia had not developed protocols for scaling their systems
in response to forecasts (or climate shocks) in advance. If social protection is to be a
mechanism for quickly delivering emergency aid, the work of registering beneficiaries,
verifying bank accounts, and coordinating donor funding is imperative before there is an
immediate need.
A summary of the recommendations most relevant to early action can be found in
Chapter 8, and the full analysis of lessons regarding the El Nio event is available in Chapter 9.
Chapter 2: Early Action in Zambia
International predictions of a possible El Nio event reached stakeholders in Zambia

between March and June in 2015. However, seasonal forecasts for normal to below-normal
rainfall in Zambia were not issued by international and national forecasting groups until
September and October 2015 when predictions could reliably be made. None of the NGOs
interviewed in Zambia reported mitigating the effects of El Nio through forecast-based
early action, and many felt they were unable to take early action. Although Zambias Social
Cash Transfer (SCT) programme was not able to provide early relief based on forecasts in
2015, government ministries and several NGOs are now testing options for scaling it up in
response to climate shocks in the future.
Chapter 3: Early Action in Somalia
In Somalia, SWALIM and FSNAU provided humanitarian organizations with precise,

timely information regarding potential El Nio-related flooding, enabling successful
early action and mitigation. Because organizations and communities have extensive
experience responding to floods, and because floods in Somalia are relatively localised
events, organizations used historical flood information, SWALIM and FSNAU reports, and
flexible humanitarian funding to pre-position supplies and reinforce riverbanks. Although
precipitation was not as intense as predicted, early action prevented significant flooding
in several areas. Concern Worldwide in Somalia estimated that for every dollar spent on
early action it saved $USD 28 in later aid and $USD 91 in farmer profits. Flexible, no-regrets
funding was essential to these responses. Response to drought forecasts in the North was
far slower, beginning only after the impacts were felt. This is likely because El Nio events
are typically associated with flooding in much of East Africa.
Chapter 4: Early Action in Kenya
Because of the devastation caused by the 19971998 El Nio, the Kenyan government was
quick to respond to El Nio forecasts. It set up a Multi-Sectoral Task Force and provided a
platform for KRCS and other stakeholders to coordinate their preparedness and response
efforts. Organizations updated contingency plans, pre-positioned supplies, led communities
in flood mitigation efforts, and disseminated flood warning messages via SMS. KRCS
and BRC were also able to provide farmers in Kitui with seeds to take advantage of the
additional rain. Because distribution systems were already in place, Kenyas Hunger Safety
Net programme successfully disbursed GBP 3 million in emergency preparedness funds
to 190,000 households to help them prepare for potential flooding. The rains were not as
damaging as they had been in 19971998, but actors believed they were better prepared
when the rains arrived.
12 Executive Summary
Chapter 5: Early Action in Ethiopia
As in Malawi, Ethiopia experienced two season failures, the second of which was due to
the El Nio event. Early action in Ethiopia was hindered by a variety of factors including
the timing of the rains in relation to the timing of the El Nio event, political constraints and
discrepancies over funding sources. Organizations could only reallocate funds to expedite
action once authorised by the government. Of the organizations interviewed, only one
was able to take early action, which it did by using crisis modifiers to adapt existing
programmes to emerging conditions. The PSNP, which was entering a new phase in
2015, was not able to facilitate early action but it was used to deliver benefits to some
households in March 2016.
Chapter 6: Early Action in Malawi
Malawi experiences regular climate stresses and is the beneficiary of ongoing efforts to
develop better climate services. Despite these efforts, respondents in Malawi often cited
the limitations of forecast information. Because the 20152016 El Nio followed a season of
severe flooding, a large disaster response effort was already underway. Some organizations
were able to modify or scale ongoing response efforts based on forecasts, but others were
too overwhelmed responding to pre-El Nio flooding to shift their focus to preparations for
potential El Nio impacts. Lack of funding was also a significant barrier to scaling activities
and implementing activities in contingency plans. Although Malawi has two social protection
programmes, neither was set up to deliver additional emergency aid. Efforts are under way
to adjust the programme for future events. As in other locations, flood preparations were
more readily adopted than early action to mitigate the impacts of drought.
The Benefits of Intervention, Impacts Avoided, and a

Framework for Assessing Early Action
Only two of the 58 organizations consulted for this study reported any efforts to track
benefits provided or impacts avoided because of early action. BRCiS in Somalia
demonstrated that early action to mitigate flooding provides significant value for money;
and, KRCS showed that forecasts could be used to help farmers.
The overarching theory of change underlying promotion of forecasts is that early
action and mitigation efforts based on forecasts will reduce the need for disaster response
and improve overall outcomes in the future. Because the indicators to be evaluated depend
upon the specifics of the project and a programmes anticipated benefits, it would be
impossible to outline a single method for evaluating early action. Taking care to collect
baseline data or to establish the basis of a counterfactual, as elaborated in Chapter 7 can,
however, help establish the benefits of early action in the future.
14
Chapter 1: Introduction
Interest in the use of forecasts for decision-making has grown in recent years due to
improved forecasting capacity and changing climate conditions. Advocates of forecast
based decision making posit that monitoring seasonal predictions in conjunction with
shorter-term forecasts can help disaster managers and responders to prepare for and
mitigate the impacts of climate shocks.
This study explores the use of forecasts for El Nio, a phenomenon that results
from the interaction between the atmosphere, ocean and air currents, and warmer-than-
average waters in the Eastern and Central Equatorial Pacific (ICPAC 2016). El Nio occurs
periodically, every 27 years.
During El Nio years, different regions of the world are likely to experience increases or
decreases in precipitationoften resulting in floods, drought, and associated humanitarian
impacts. In Kenya and the southern areas of Somalia, for example, El Nio is strongly
correlated with more precipitation, whereas, in southern and western Zambia, El Nio often
causes less precipitation. These correlations (known scientifically as teleconnections) allow
forecasters to issue seasonal forecasts in particular regions with greater certainty than is
possible in non-El Nio years.
The strongest and longest-lasting El Nio event since 19971998 (WFP 2016a), when
many countries around the world experienced its destructive force, occurred in 20152016.
In the two decades, scientists have increased their forecasting capacity significantly and
invested considerable resources in improving communication between those who produce
the forecasts and the humanitarian practitioners that prepare for and respond to those
severe climate shocks. The improved accuracy of these predictions has led donors, NGOs,
and governments across the world to incorporate El Nio forecasts into their contingency
planning exercises.
This study examines actions in five countries, in East and Southern Africa, to prepare
for potential impacts from the 20152016 El Nio forecasts; it assesses the challenges and
successes associated with those efforts; and, it posits improvements to future efforts to
reduce the impacts of El Nio events.
Purpose & Objectives
In early 2015, climate professionals predicted a formidable El Nio. As the year progressed,
international, regional and national agencies issued forecasts and advisories to warn
disaster management stakeholders around the world of the potential impacts. In theory,
these early warnings allowed governments, United Nations (UN) agencies, and non-
governmental organizations (NGOs), to plan, prepare, and implement early actions, thereby
reducing damage or harm. In practice, successful implementation of early action measures
varied across countries due to a number of reasons unrelated to specific forecasts.
To determine whether forecasts (a) prompted early actions and (b) reduced the
need for additional disaster response, the Department for International Development
(DFID) commissioned the Red Cross Red Crescent Climate Centre to review how climate
forecasts were used in preparing for and responding to the impacts of the 20152016 El
Nio (DFID 2016). Using interview data from five countries Zambia, Somalia, Kenya,
Ethiopia, and Malawi this study addresses the following primary research questions1:
1 Each question had a number of sub-questions. A full list of the research questions proposed in the ToR can be found
in Annex B.
Which information was used?
How did that information influence the decisions?
What was the role of communication, trust, and dissemination?
How have innovations in disaster preparedness and response since the
20112012 droughts led to earlier action and two has this mitigated the
impact of the El Nio?
Is there evidence that information led to earlier action and avoided
losses?
What is the perception of key actors on the value of information to the
response?
Answers to these questions will inform the readers understanding of how and
whenif at allinformation regarding El Nio is considered and incorporated into the
countrys planning and preparedness. Readers will also takeaway factors that enabled or
constrained early action. And lastly, this study includes a timeline of the decision-making
processes employed in each of the five countries.
Research Team
The research team consisted of four researchers Eddie Jjemba (interviews in Kenya,
Ethiopia, and Zambia), Shaban Mawanda (interviews in Malawi), Stephen McDowell
(interviews in Somalia), and Arielle Tozier de la Poterie (coordinated research design,
analysis, and synthesis of the results with support from the other partners). This team was
supervised by Julie Arrighi at the Red Cross Red Crescent Climate Centre.
Case Study Selection & Informant Sampling
Five countries were chosen based on the following criteria:

1. They received DFID funding;
2. They reported different responses to forecasts or varying degrees of
early action; and,
3. They experienced different weather patterns and impacts from El Nino
teleconnections.
Within each country, the research team selected organizations based on their
significant presence in the country, their involvement in El Nio response, and the availability
of staff for interviews.
Interviews with key humanitarian and disaster management personnel prepared
researchers to reconstruct overall response to forecasts in each of the five countries. A
common narrative emerged across all organizations within each country, which is why four
of the country case studies detail the response in a single narrative. However, in Somalia,
some respondents provided significant detail of their early action, warranting more explicit
separation of activities according to each organization.
16 Chapter 1: Introduction
Methodology
This study uses data from semi-structured, key-informant interviews to reconstruct

responses to El Nio forecasts in each of the five countries. The team interviewed
informants working in NGOs, UN agencies, government agencies, and national and
international forecasting centres to obtain the perspective of both information producers
and the users who were expected to respond to forecasts. Interviewers conducted a total
of 66 interviews with 69 informants. Interviews were conducted in-person or over Skype,
depending on informants availability. An additional five informants preferred to receive
the interview guide and fill it out in writing bringing the total number of informants to 74.
All interviews were recorded (with permissions from the informants), transcribed, and
qualitatively coded to identify case-specific details and cross-case themes. To respect
the privacy and anonymity of the informants, citations are attributed to broad categories
of actors (Humanitarian, Forecaster, Government see footnote for key) rather than
by organization or interview number.2 The semi-structured interview guideline used for
humanitarian actors can be found in Annex D of this report.
Given the large number of questions, researchers elected to ask informants to
reconstruct their organizations response to El Nio forecasts as they remembered it,
thereby focusing a discussion on the factors each informant found to be most important
to explain their reaction. Whenever possible, interviewers spoke to multiple contacts within
each organization to corroborate information and construct more accurate narratives and
timelines. However, because of respondent (un)availability and time and travel constraints,
the team was only able to interview one person from many organizations. A breakdown of
the informants for this study by country and organization is provided in Table 1 (on pages
18 to 19). To the extent possible, interview data was supplemented and corroborated by
documents and reports provided by informants or available on the organizations public
website. When this was not feasible, the details of specific organizational responses are
based on only one or two individuals within each organization. Fortunately, similar themes
emerged across organizations and often across countries increasing the research
teams confidence in their conclusions.
2 Interview informants are quoted according to four categories:

1 = Government officials
2 = Humanitarian (NGO, UN agency, consultant, or academic)
3 = Scientists, forecaster, or information providers (as in SWALIM or FEWS NET).
4 = Donors
Table 1: Breakdown of Informants by Country and Organization
Country / Informant Organization Number of Informants

Zambia UNICEF 2
Save the Children 1
World Vision 1
Zambia Red Cross Society 2
Oxfam Great Britain 1
FAO 1
Zambia Met 1
WFP 1
FEWS NET 2
CARE international 1
Concern Worldwide Zambia 1
DFID Zambia 1
Zambia Totals 12 Organizations 15 Informants
Somalia DFID Somalia 1
CARE 1
UNICEF 1
Somalia Red Crescent Society 1
BRCiS 2
Concern 1
WFP 2
FAO 2
ICRC 2
Somalia Totals 9 Organizations 13 Informants
Kenya DFID Consultant 1
Kenya Red Cross Society 3
UNOCHA 1
British Red Cross 1
KMD 1
WFP 1
UNICEF 1
Save the Children 1
Kenya Hunger Safety Net Programme 1
Kenya Totals 9 Organizations, 1 Consultant 11 Informants
Ethiopia Ethiopia Meteorological Service 1
Disaster Risk Reduction Department 1
WFP 1
University of Ethiopia 1
18 Chapter 1: Introduction
Country / Informant Organization Number of Informants
Disaster Management 1
Ethiopia Red Cross Society 1
UNOCHA 1
CARE 1
DFID 1
USAID 1
Ethiopia Totals 10 Organizations 10 Informants
Malawi DFID 1
WFP 3
DoDMA 1
FAO 1
Malawi Red Cross Society 3
MoAIWD 1
DCCMS 1
UNICEF 1
Malawi Totals 8 Organizations 12 Informants
Global Level Informants UNDP 1
WFP 1
UNOCHA 1
CARE International 1
Global Informant Totals 4 Organizations 4 Informants
International Forecasters NOAA 2
ICPAC 2
UKMO 1
IRI 3
Red Cross Red Crescent Climate Centre 1
Forecaster Totals 5 International Organizations* 9 Informants
Study Totals 58 Organizations, 1 Consultant** 74
* International forecast totals do not include National Hydro-meteorological Services, as those are included in country
totals.
** The organization totals treat informants from different levels within the same organization as separate organizations. For
example, because they received information at different times and responded at different scales, WFP headquarters,
WFP Regional headquarters, and WFP Malawi would be counted as separate organizations in this tally.
Limitations
Because of the studys emphasis on understanding peoples perceptions of the value of

information to early action, the majority of the analysis draws on interview data. Relying on
informant recall presents problems, as people may not remember details or sequences
of events properly. Assigning precise dates and reconstructing the exact details of each
organizations timeline presented a particular challenge, as people often did not remember
dates. As several informants explained, the El Nio warnings seem[ed] like a long time ago
now (3); therefore, many informants found it hard to pinpoint where [they] heard of El Nio
(2) or how exactly events unfolded. To the extent possible, dates were confirmed through
documents or additional informants, but because of the numerous organizations included in
this study (57 in total), it was not possible to corroborate every detail. When verification was
not possible, we reconstructed events based upon the memories of study participants. For
these reasons, some of the dates or the details of particular programmes may be slightly
subjective. The researchers have done their best to resolve any discrepancies or to note
them in the footnotes.
Finally, although government actors were shown to have a significant influence over
early actions in many countries, the majority of respondents for this study were from UN
agencies and NGOs, as this was the focus of the study. Only nine of the 74 informants
came from government agencies (one in Zambia, two in Kenya, three in Ethiopia, three in
Malawi, and none in Somalia). In Zambia and Kenya, informants were from the National
Hydro-meteorological services. As a result, the findings came from humanitarian actors
primarily and may overstate the humanitarian communitys willingness and ability to
respond to forecasts while overstating governmental or political obstacles.
20
Chapter 2: Zambia Case Study
Background
In Zambia, disaster management is supervised by the Disaster Management and

Mitigation Unit (DMMU) within the Office of the Vice-President. As in most countries,
donors and NGOs do not have a mandate to distribute forecast information to the public
or to act upon it independently. Zambia has an SCT programme that began around
2003 and has been scaled up since 2010 (2). As of 2014, the programme was prepared
to serve approximately 190,000 people in 50 of Zambias 105 districts. According to
informants for this study, by 2016 services had again been expanded, reaching people in
78 districts. The SCT programme focuses on assisting households with elderly, disabled,
and orphaned members (Ministry of Community Development, Mother and Child Health
2014). At the time of the 20152016 El Nio, the SCT system was not able to quickly scale
up in response to climate shocks.
Zambian Forecast Production
The Zambian Meteorological Department (ZMD) has the responsibility for producing
and distributing localized forecasts to the DMMU and other government ministries.
As with other national meteorological offices, the ZMD relies on forecasts from
international models which it then downscales to the Zambian context. ZMD issues
a seasonal rainfall forecast by the end of September each year, one or two months
ahead of the planting season.
Throughout the season, ZMD also issues ten-day Crop Weather Bulletins
summarizing rainfall conditions around the country after the fact, in addition to seven-
and three-day forecasts. These bulletins are available on its website and distributed
via email to institutional stakeholders. They reach the broader public through
newspapers, television, email, the internet, and radio. Radio is the most effective means
of reaching the general population, but programming does not reach the entire country.
In September 2015, ZMDs seasonal forecast indicated an increased chance of normal
to below-normal rainfall over most of the country for the October to December period.
This forecast also indicated that most of the country would return to normal rainfall
by the December to February period. In addition to the seasonal forecast ZMD also
created a crop calendar showing farmers planting and harvesting dates for typical crops
to help them select drought-resistant varieties. Because ZMD employs meteorologists
primarily, it coordinates with technical experts within other ministries or sectors to
determine the likely implications of the rainfall forecast based on these predictions.
ZMD meteorologists also participate in numerous forums to help disseminate
and interpret their forecasts. Every year, DMMU convenes government, UN, and NGO
stakeholders on the Zambian Vulnerability Assessment Committee (ZVAC) to conduct
an in-depth study of vulnerability in the country (see ZVAC 2015 for a complete list of
participants). At these meetings, participants share forecasts and early warnings as well.
During the 20152016 El Nio, ZVAC data was collected in April and May of 2016, and
the report published in June. ZMD also presents forecasts to the Disaster Management
Consultative Forum, a national platform chaired by the DMMU, which is called on as
needed to prepare for and respond to emerging threats. To help translate their forecasts
into action, meteorologists interact with sector representatives at the provincial level and
in districts where ZMD has employees.

In an effort to further improve their services, ZMD is in the process of asking
stakeholders for feedback on their forecast products. One of their desired improvements
is employing sectoral experts to help translate forecasts into sector-relevant advisories
and recommendations. They would also like to develop new and stronger dissemination
channels to reach people at the village level who may not have access to the internet
or the radio.
El Nio Impacts in Zambia
Zambias climate is comprised of two seasons. The warmer wet season occurs between
October and April, and the colder dry season occurs between May and September.
El Nio events typically result in an increased number of dry spells during December,
January and February the peak of the rainy season.
The 20152016 El Nio came immediately after a dry spell during the 20142015
rainy season; it affected over 400,000 households in the Western, Southern, and Eastern
Provinces primarily (OCHA, Zambia VAC results, 2015). In other parts of the country,
El Nio meant higher than normal rainfall, though no one interviewed for this report cited
problems with flooding. Although seasonal forecasting skill increases during El Nino
years, rainfall forecasts over Zambia were inconsistent for the peak of the rainy season.
The September 2015 ZMD forecast for Zambia predicted that El Nio would lead to
normal to below-normal rainfall in the Western, Southern, and Eastern Provinces of
Zambia during the October to December period, returning to normal by the December
to February period. Regional and global forecasts generally concurred toward the
beginning of the season and differed toward the peak and end of the season, as detailed
in the next section. Forecasters around the world are unable to predict the onset of
the rainy seasons or how that rainfall will be distributed. As a result, humanitarian
stakeholders in Zambia knew overall there would be a problem of rainfall, but there
wasnt that level of sensitivity to tell them exactly how and where the problems would
manifest or allow them to make specific recommendations regarding planting times (2).
This common feeling of impending disaster may have been a result of perceived risk
as a result of global El Nino warnings and the increase in vulnerability due to the past
below-normal rainy season, rather than the seasonal forecasts issued in September
which did not indicate a significantly elevated risk at that time.
During the 20152016 El Nio, the rains started several months late in much of
southern Zambia, causing farmers there who planted according to the usual cropping
calendar to lose their crops. The lack of rain also caused problems with access to
clean water for household use and consumption. Nevertheless, farmers who planted late
or were able to replant after initial losses were able to get reasonable harvests.
Significant rainfall later in the season diminished the severity of feared impacts.
In February 2016, FEWS NET reported that despite high demand and above-average
prices in some areas, maize availability is adequate in Zambia; [government]
Assistance to food insecure populations in mostly western and southern regions
has continued.
22 Chapter 2: Zambia Case Study

The ZVAC report published in June 2016, which surveyed 42 districts in five
provinces, indicates that at a national level yields for most crops were better than
expected with sufficient yields to meet domestic needs. At a sub-national level however,
this report concludes that nearly one million people in the provinces of Lusaka, Central,
Western, Southern, and Eastern Zambia would require food support between August
2016 and March 2017 because of El Nios impacts, with southern districts being most
effected (ZVAC 2016).
On 15 June 2016, a Southern African Development Community press release that
reported more than 40 million people were in need to food assistance region wide in
the worst [El Nio-induced drought] in 35 years, said: Zambia is the only country
currently forecasting a cereal surplus (556,000 tons) during the 2016/17 marketing year.
Response to the 20152016 El Nio Forecasts in Zambia
By most accounts, El Nio forecasts reached stakeholders between March and

June 2015. These forecasts were most likely global El Nio warnings however, since
international and national forecasts for normal to below-normal rainfall in Zambia did not
occur until September and October 2015. This includes forecasts from FEWS NET, ZMD,
SADC, IRI and ECMWF. Organizations, in general, received information from their global
offices, international forecasters, and institutions like FEWS NET, SADC, and OCHA.
Communication channels ran relatively smoothly; most organizations first heard warnings
from their global or regional headquarters and then supplemented that information with
other sources of information. For partners like Concern Worldwide, monitoring forecasts
and being aware of potential risks is something that is embedded in the organizational
culture (2). Despite the global warnings of El Nio in 2014 which did not pan out, Zambian
stakeholders continued to monitor forecasts diligently.
Most partners in Zambia found the information from FEWS NET, OCHA, and other
international sources to be most reliable and useful. In contrast, several stakeholders
expressed concern that local forecasts are weak, and the information can be quite
questionable (2). Of the 11 organizations interviewed, only the Zambian Red Cross
Society (ZRCS) explicitly mentioned using ZMD forecasts; however, with no actionable
messages within the forecasts (2), it was up to ZRCS to determine the actions.
It is important to note that rainfall forecasts over Zambia were inconsistent for the
peak of the rainy season. The September 2015 ZMD forecast for Zambia predicted that
El Nio would lead to normal to below-normal rainfall in the Western, Southern, and
Eastern Provinces of Zambia during the October to December period, returning to
normal by the December to February period. The SADC regional forecast also indicated
a slightly elevated chance of normal to below-normal rainfall during the October to
December period, changing to normal to above-normal rainfall by December to February.
International forecasts, such as the seasonal forecast from the IRI, began indicating
normal to below-normal seasonal rainfall in October 2015, with increasing risk of below
normal rainfall during the peak of the rainy season, then returning to normal conditions
by the February to April period.

The warning of likely drought in the coming seasons was enough to raise concerns.
The 2014 and 2015 seasons were also quite dry and contributed to decreased electricity
generation, load shedding, bore holes running dry, and agricultural setbacks. In response
to forecasts, humanitarian partners and the government began talking about the likely
impacts in 20152016.
In August 2015, DFID met with the DMMU (the government department in charge of
emergency response) and in September staff began speaking to the Ministry of Community
Development about responding to the impacts of El Nio using the Social Cash Transfer
programme. Starting with the first warnings, FAO, ZRCS, CARE, and UNICEF incorporated
El Nio related drought predictions into their organizational contingency plans. As is
customary, those plans eventually informed the national contingency plan coordinated by
DMMU; it published in November of 2015.
Zambias 201516 National Contingency Plan did not explicitly mention El Nio,
outlining in general terms what is to be done before, during, and after anticipated dry
spells and flash floods in each of the following sectors: agriculture and food security;
health; WASH; nutrition; habitation and human shelter; early warning; alert systems
and communication; and logistics (Government of Zambia, 2015). According to the
ZVAC, assessments are to be coordinated by government and some NGO interviewees
expressed concerns with such requirements for government permission and coordination
of all assessments. They also expressed a need for stronger coordination and increased
access to information, specifically additional details in the governments annual crop and
vulnerability assessments to aid in fund-raising. Most NGOs are able to access funding
based on governments declaration of an emergency. As a result, some NGO interviewees
also felt that the potential political repercussions of requesting international assistance
delayed and diminished potential early actions. Early actions which did take place included
advising farmers to practice conservation agriculture and plant drought-resistant crops.
In January or February 2016, DFID and several NGOs with programmes in the
Southern and Western Provinces Concern Worldwide Zambia, WFP, Oxfam, Save
the Children, and World Vision came together to start thinking about how they
could possibly adapt their programmes to respond to El Nio effects (4). With DMMU
authorization, and funding from DFID through the Start Fund, these organizations
conducted four studies: (1) Concern Worldwide conducted a market assessment to
understand how markets might respond to cash transfers to mitigate El Nios impacts; (2)
Oxfam conducted a study on El Nios likely gender impacts; (3) World Vision evaluated
community-based disaster risk management; and (4) Save the Children conducted a
capacity assessment of District Committees and Infrastructure in Western Province. These
results were presented to the government during the Consultative Forum hosted by DMMU
in February or March.

As the rainy season ended in April 2016 and the impacts of drought became clear.
DFID allocated money through the Start Fund to NGOs for a trial of the emergency cash
transfers. The project began in earnest in mid-September 2016 and, as of October 2016,
was working with DMMU to determine the most food insecure districts according to the
most recent vulnerability assessment. Registration of households had not yet begun,
but the first cash transfer was anticipated to reach approximately 25,000 households in
December 2016.
Linking Social Protection to Climate Shocks
Stakeholders have now begun researching how Zambias Social Cash Transfer programme
might be expanded to help cope with climate shocks such as the last El Nio. Zambias
SCT programme targets the elderly and disabled but does not incorporate measures for
food insecurity. The Ministry of Social Welfare also distributes funds in cash rather than
electronically. Following assessments carried out by Oxfam, Concern Worldwide and
others, the government approved registering households in communities badly affected by
El Nio but not yet part of the social protection system.
DFID funded a pilot project to five districts suffering from El Nio-related food
insecurity, and the Ministry of Social Welfare, Concern Worldwide, and Save the Children
began distributing funds in September and October 2016. The pilot will continue through
April 2017. In the future, the government may also consider increasing the cash transfer
value in response to higher food prices.
Stakeholders in Zambia expressed a commitment to learning from more advanced
social protection systems elsewhere, and expressed the desire to use forecasts as a
trigger to register vulnerable households or deliver cash before impacts are felt. Most
important, the organization would like to ensure that scaling up social protection for El
Nio is not a one-off action, like only the humanitarian response but instead builds
capacity to make the whole cash transfers and social protection system more responsive
to shocks (2). These discussions demonstrate that while Zambias Social Cash Transfer
programme was not yet advanced enough to cope with the 20152016 El Nio, El Nios
impacts prompted dialogue about how such shocks might be addressed through social
protection programmes in the future.

Conclusion
The chief lesson from the 20152016 El Nio in the Zambian national context is that for
humanitarian stakeholders to be able to respond to forecasts in the absence of an officially
declared emergency, the system would have to change to allow them to adapt or initiate
programmes with a different form of governmental acquiescence. Such a system would also
benefit from a consultative and robust process to identify forecast thresholds that trigger
action well in advance of any particular forecast.

Timeline of Response to El Nio Forecasts in Zambia
Chapter 2: Zambia Case Study

27
Chapter 3: Somalia Case Study
Background
Humanitarian response to climate shocks in Somalia is complicated by 25 years of civil

war and the ensuing uncertainty about the governance and security situation. It has been
the site of ongoing humanitarian action since famine and conflict struck in 1991. When it
comes to disaster preparedness and response, the government of Somalia relies largely on
financial contributions from the international community of donors and services provided by
NGOs and UN agencies. Among these organizations, World Food Programme (WFP) and
International Committee of the Red Cross (ICRC) are the two humanitarian players; USAID
and DFID are the most flexible donors.
In addition to WFP and ICRC, representatives of two NGO consortia Building
Resilient Communities in Somalia (BRCiS) and Somalia Resilience Programme (SomRep)
were interviewed for this study. These consortia emerged to help coordinate funding and
action after the 2011 famine. BRCiS is a DFID funded consortium of five NGOs lead by the
Norwegian Refugee Council; they include Concern Worldwide, Cooperazione e Sviluppo,
International Rescue Committee, and Save the Children International. They coordinate
humanitarian intervention in response to shocks with the need for longer-term resilience
and development (BRCiS 2015). Through SomRep, an additional seven international
agencies (Oxfam, Adventist Development and Relief Agency, Action Contre la Faim, Danish
Refuge Council, CARE, Coopi and World Vision International) also coordinate common
efforts to build household and community resilience (Danish Refugee Council n.d.).
In addition to security challenges, the 201516 El Nio in Somalia followed two years
(four consecutive seasons) of below-average rainfall in many areas of the country, including
areas that have suffered from protracted political and economic instability (UNOCHA 2016).
The impact of El Nio, therefore, compounded already precarious livelihoods situations.
Historically slow reaction to early warnings also influenced the overall response to the
20152016 El Nio in Somalia. In 2011, Somalia experienced a severe drought, for which
a lot of early warning information was available, but the action to respond to the drought
came much later (2). This failure to act resulted in probably the biggest humanitarian
failure (2) in recent times and concern among NGO stakeholders in Somalia about
replicating the same mistake (2) in 20152016 (also expressed by 43).
Forecasts in Somalia
Although Somalias meteorological (Met) service is small (and was not mentioned by any
informant), Somalia hosts two well-developed monitoring partnerships with the Food
Security and Nutrition Analysis Unit (FSNAU) and Somalia Water and Land Information
Management System (SWALIM). NGO partners working in Somalia, therefore, largely
depend on the combination of international forecasts and the more detailed, localised
information provided by FSNAU and SWALIM. FSNAU is a monitoring effort funded
by SIDA, OFDA/USAID, DFID, the EU, and Common Humanitarian Fund (CHF). The
Met seeks to provide timely and relevant food security and livelihood information to
support evidence-based programmes (FSNAU.org) and is regarded as one of the
best famine monitoring systems in the world. It classifies food security according to the
Integrated Food Security Phase Classification (IPC) the same system used by FEWS
NET developed in the last decade to harmonise food security classifications across
organizations and contexts.
28
SWALIM is an information management group managed by FAO and funded by the EU,
UNICEF, and CHF. SWALIM acts as would a government ministry of water, providing advice
regarding irrigation, river management, monitoring, hydrological modelling, and other issues
related to land and water to emerging government institutions, NGOs, development actors,
and UN bodies. SWALIMs MayAugust 2015 update warned of the likelihood of increased
rains and associated flooding, particularly of infrastructural concerns (SWALIM, 2015).
By all accounts, the information provided by these two organizations was essential to
planning efforts. During the 20152016 El Nio, SWALIM provided FSNAU and the broader
humanitarian community with detailed information on areas that were likely to flood.
Informants for this study found SWALIMs maps of river locations along the Juba (21) and
Shebelle (47) particularly useful in their planning and recommendation to address these
weak points through river control activities. FSNAU then took the lead in estimating the
likely impacts of floods at these points, and humanitarian organizations then combine the
SWALIM and FSNAU information with their logistical expertise to determine how to act.
Impacts of El Nio in Somalia
The effects of El Nio in Somalia depend on the region. During El Nio, the northern
areas may realize lower than usual rainfall (or a continuation of dry conditions that typically
prevail), whereas the southern regions often experience increased rainfall and flooding
along the countrys two major rivers, the Juba and the Shabelle. Several organizations
faulted themselves for not adequately understanding the possibility of drought in the North
(UNICEF & WFP). They focused on the prominence of flood impacts during the 19971998
and 2006 El Nios as well as other similar years to determine effects. The flooding in
Somalia was likely not as bad as anticipated because it fell nicely in comparison to past El
Nio years when intense rains in a short time lead to higher rivers (2).
During the 20152016 El Nio, southern and central equatorial regions of Somalia
and the western region experienced flooding in October and December (IGAD & ICPAC
2016). Much of the flooding comes from increased rains in the Ethiopian highlands and
tributaries originating in Kenya which then drain to Somali rivers, affecting water levels and
potentially causing flooding (2). The impacts of flooding events include losses of crops,
livestock, and property.
Flooding was not as widespread or severe as forecast originally (WFP 2016); many
agreed that flooding was only slightly higher than what one would expect during the rainy
season. Nevertheless, international reports attribute many impacts in Somalia to the
20152016 El Nio. In January 2016 WHO estimated that the drought in the North had
affected up to 2.3 million people, the majority of whom were already internally displaced,
increasing food insecurity by 400,000 people. In response to warnings, partners agreed to
pre-position humanitarian relief items to assist at least 50,000 people for eight weeks in all
potential hotspot areas (IASC 2015: 4). In November of 2015, flooding impacted 145,200
people (OCHA 2016b). FAO was able to help with flood mitigation, including sandbag
diversions and elevated structures to protect grains from heavy rains or floods. They also
distributed fishing kits to allow people to take advantage of higher waters, but it is not clear
from the reports when precisely these efforts occurred (FAO 2016). Overall, early action
was hindered by a lack of funding (UNOCHA 2016b). As of March 2016, aid organizations
estimated that 385,000 people in Somalia were acutely food insecure, and another 1.3
million were at risk of becoming so approximately 37 per cent of people living in Puntland
and Somaliland (OCHA 2016b).

In an arid environment such as Somalia, increased rains can also bring benefits,
including increased water for domestic use, farming, and livestock (IGAD & ICPAC 2016;
FAO El Nio Update Jan 2016). Several actors Somalia Red Crescent Society (SRCS)
and members of the BRCiS Consortium reported that they were able to help people
plant crops to take advantage of potential bumper harvests. Because it is relatively
inexpensive to provide seeds and farming support, these initiatives were seen as positive
no-regrets actions; the potential benefits outweighed the risk of loss if rains were
insufficient (21; 28; 32; 43). The benefits, however, were never quantified.
Concern Worldwide Somalia is the only organization consulted for this study that
attempted to measure and quantify the benefits of their early action. Their estimates of
benefits and a discussion of their methods are provided in Chapter 7, which discusses
options and challenges to impact assessment at greater length.
Response to 20152016 El Nio Forecasts in Somalia
Response to El Nio forecasts in Somalia followed a similar pattern across organizations.

Upon first hearing of El Nio between March and October, stakeholders began researching
the consequences of El Nio in previous years. These likely impacts were then integrated
into contingency plans, the details of which were implemented as funding became
available. Activities were made possible by rapidly available DFID and OFDA funds, the
ability to reallocate OFDA funds from other initiatives, and capacity to coordinate with other
actors to move and store supplies. This section details how each organization proceeded
and concludes with some final observations.
Early Warnings
Organizations in Somalia first heard of the potential El Nio from international sources as
soon as March 2015. Some organizations, like CARE, were not informed until October
2015. SWALIM began providing Somali-specific information about river breakages in
July, which was received by FSNAU, WFP, FAO, SRCS, and BRCiS partners. Before that
formal advisory, WFP employees had heard vague references to El Nio via the news and
in informal discussions with donors. Partners working for Somali Red Crescent Society
(SRCS) and the ICRC first heard of the possibility of an El Nio through IFRC warnings
in April or May of 2015. Because of the drought that began several seasons earlier,
SRCS staff were already monitoring FEWS NET and Relief Web websites and reading
briefings from UN Cluster meetings. By late August and early September, NGOs within the
BRCiS consortium had substantial data from different sources addressing the increasing
probability of El Nio, which would result in flooding in the southern and central regions of
Somalia and pockets of drought in the northern part of Somalia (2). Overall, because of
the high level of coordination among the Humanitarian Country Team (to which UN Cluster
organizations report), UN Clusters themselves, and the two NGO consortia within Somalia,
most NGOs were made aware of the El Nio and its likely impact in time to coordinate their
early actions.
30 Chapter 3: Somalia Case Study

Determining Likely Impacts
Upon learning of a possible El Nio, several partners began investigating its potential
implications. WFP officers began conducting their research, also looking for analogue
years to get a better understanding of the potential impacts. They identified the 19971998
and the 20062007 El Nios as analogue years and noted that flooding had been the
primary effect in those years. FAO directed SWALIM to look through the historical archive
to determine what kinds of flooding had occurred in Somalia during previous six El Nio
events in Somalia, providing a sense of what was likely to happen. Once the analyses
were complete, FAO and SWALIM presented the possible implications of El Nio at the
Humanitarian Country Team (HCT) meeting in earlymid July. They put significant effort
into translating the forecasts into laymans terms and connecting it to potential actions for
all the members of the HCT. There was widespread agreement from all the stakeholders
that SWALIM did an amazing job. They really dug down into the archives; they started
to map and show us what the impacts were etc. I mean the power of SWALIM was really
harnessed in that case. I want to say, it took quite a lot of leadership from [sic] our part (2).
Following that meeting, ICPAC and SWALIM continued to provide regular updates on
the likely El Nio scenarios; by August, after the ICPAC meeting, partners had a relatively
accurate picture of what would later unfold (2). Because of these efforts to produce and
share information, by the time the rains started up in the Ethiopian highlands we were able
to monitor it people knew what to expect (2).
Contingency Plans
In response to the detailed SWALIM and FSNAU information, most organizations began
planning and coordinating early actions. When the initial forecasts were received, SRCS
was understaffed and did not have a disaster management coordinator because of staffing
transitions. Nevertheless, SRCS used these forecasts of heavy rain to update contingency
plans in August and identify what additional resources would be needed to respond to
potential El Nio-related flooding. WFP began planning their responses and coordinating
with ICRC, looking at the kinds of air and water transport they would need to move supplies
and the kinds of disaster response that would be required. During this planning process,
they identified triggers to initiate action, and by the end of July, WFP had reviewed past
events and developed an internal, unfunded, action plan. The CARE senior management
team met to decide how to address projections of floods and droughts. They sent teams
to do field visits and rapid assessments of the areas where SWALIM predicted riverbanks
would burst and then combined field information with lessons from flooding in 1997 and
2006 to devise action plans.
In contrast to other countries in this study, the Somali government combined these
plans to create a national contingency plan. When and where each organization would
intervene was determined through the HCT and NGO collaborations. These decisions
were determined by geographical considerations and who had secure access and the
connections to get things done in different areas (2).

Funding for the Early Action
Informants widely cited flexible DFID and OFDA funding as essential to early response to the
forecasts. CARE International also used the information to solicit funding from SIDA. DFID
first learned of the possible El Nio in March or April. Upon receiving the forecast, DFID
contacted partners like FAO and Humanitarian Country Team to find out what the likely
impacts would be. As DFID has 20 million pounds per year pre-approved for Somalia, they
were able to direct it toward early warning, preparedness, and early action. DFID agreed
with their implementing partners on several no-regrets actions that DFID could fund. DFID
is firmly committed to the no-regrets approach, believing it cannot be in a situation of if
an event doesnt materialize, but the expertise and evidence points to the fact that it should
have happened we shouldnt be in a position of questioning the fundamental principle of
no-regrets type finance (4). Without such funding and a no-regrets attitude, many of the
early actions outlined below would not have been possible.
Actions by Organization
Once plans were developed, organizations began taking early actions and implementing
their plans with the funding available. Because they took slightly different actions, each
organizations response to El Nio forecasts is outlined separately below.
ICRC and SRCS
Because SRCS was interested in working with ICRC to expand their activities in Somalia,
ICRC and SRCS worked together in their early action and response efforts. Because of
the lack of security in Puntland, on 10 September, 2015, IFRC reached an agreement with
ICRC to coordinate radio-based awareness campaigns there. In the beginning of October,
the two held a joint meeting to discuss the supplies and delivery systems already in place.
In October and November, SRCS began raising funds to support their contingency plan.
Unfortunately, the majority of the funds initially pledged for SRCSs contingency plan were
never received. Therefore, most implementation of the SRCS contingency plan stopped
in late December, and from January through March, the SRCS was unable to provide
assistance. Official appeals for funds were made in January and February, allowing disaster
response efforts to resume in March.
Because ICRC was already well established in Somalia, it responded to forecasts
primarily by pre-positioning additional supplies and continuing to provide support for health
and other programmes. Because floods occur every year and flooding in 20152016 was
not as severe as in past El Nios, the scale of the ICRC disaster response once the flooding
occurred was no larger than usual.

BRCiS
BRCiS partners were able to mobilise International Relief Response crisis modification
funds from DFID, as well as flexible funding from other programmes in Somalia, which can
be reallocated to emerging needs provided the overall objective is similar to the fundings
original intent. Whereas many actors in Somalia were unable to use their funding for
procurement and pre-positioning supplies, Concern Worldwide procured and positioned
supplies using flexible funding that could be used for awareness or preparedness (2). The
ability to quickly access funds, and to reallocate existing funds from existing projects to the
areas of greatest need were considered essential to these early actions as, according to
our informants, lack of access to such funds prevented other NGOs from pre-positioning
supplies in response to El Nio warnings.
Using these funds, BRCiS partners undertook numerous flood preparedness and
mitigation measures: repairing river embankments; pre-positioning sandbags and WASH
materials in various hubs around the country to expedite their use immediately before the
flooding; and, positioning inflatable boats for the transport of supplies after the flooding
had occurred. Concern Worldwide employees used the forecast to estimate which
villages in insecure areas were most likely to flood. They then used their community ties
to locate local, reliable merchants whose facilities could be utilised as hubs for storing
low-value flood response supplies like rope and sandbags. It was then up to Concerns
partners within each community to coordinate distribution of supplies in the event of an
actual disaster.
These supplies proved quite useful for a community-led response once the rains
began. According to community reports, people were able to access the supplies in 50
villages and coordinate amongst themselves to set up water blockades, and to improvise
solutions as the water patterns changed. Because the community has experienced this
kind of flooding before they know the waters coming, theyve been through this before, so
when you say hey I got list things here for you guys to set up, you dont have to ask them
twice (2). Communities ability to access the supplies and act proactively depended upon
who controlled the area, as in some places Al-Shabaab limited when and how people were
allowed to act. Despite these successes, based on satellite imagery, Concern estimates
that only half of the locations they prepared for flooded.
Concern also used its contacts to bring the El Nio forecasts to the attention of local
radio stations, who then largely of their own initiative developed and broadcast a
series of nearly 20 radio programs on Voice of America, one of the most listened to shows
in Somalia. The shows, which aired beginning in early October, just before the rains started,
featured experts who could discuss the various implications of El Nio. These broadcasts
reached an estimated 10.5 million people across Somalia, Djibouti, Kenya, and Ethiopia (2).

WFP
In accordance with its internal contingency plan, WFP began using DFID funds in October
to pre-position supplies, helicopters, and boats because they would likely need to move
supplies during potential floods as well as the logistical needs of other UN clusters
(WASH, nutrition, etc.). As with BRCiS, the availability of flexible funds and very good
contacts with the authorities and partners in the areas were essential to successful pre-
positioning of supplies (2).
FAO
Early support from DFID and OFDA allowed FAO to initiate several preparedness
and mitigation activities in response to the forecasts. In keeping with the donors
philosophies, FAOs approach to response planning was to look for no-regrets actions
and to act quickly so as not to replicate the mistakes of 2011. Among their early actions
were warning communities of the potential for flooding, distributing inputs, and helping
communities to construct platforms to protect seeds and other assets. FAO also
conducted animal health campaigns, vaccinating animals against diseases associated
with flooding and worked to protect and reinforce riverbanks, which are often damaged
by farmers seeking to use river water for irrigation to decrease the risk of flooding. Many
of these somewhat temporary fixes have survived into subsequent seasons, continuing to
protect nearby land from flooding.
CARE International Somalia
In response to the forecasts, CARE hired new personnel to help with coordination of the
EU Funded SomRep Consortium programme. Most of CAREs team in Somalia is made
up of Somalis, many of whom have experienced the impact of flooding and drought related
to El Nio in past years and are intimately familiar with the villages and local context. They
were able to mobilise villagers in response to the flooding quickly, and the very specific
river breakage points identified in SWALIM reports. In particular, communities evacuated
children to protect them from the flood waters and the dangerous animals they bring
(crocodiles and snakes).

Forecasts in the Context of Insecurity & Protracted Crises
Not surprisingly, security concerns factored into the ability to respond to El Nio forecasts
in Somalia. Response, for example, was limited to the four districts for which humanitarian
actors have safe access. In such security situations, all partners emphasized the
importance of good working relationships with local officials and communities. In some
instances, partners were only able to act through partnerships with ICRC.
Because of the security situation in Somalia, real-time monitoring can be a
challenge, as organizations do not always have access to affected areas. As one
informant explained, We were getting information from areas that were not accessible
so one of the things would be how do you verify or validate the information? So we are
stuck at that point. So you would work with the same secondary information in terms of
planning. You could have a good plan, but you cannot verify the information. How do you
verify the current situation in an area thats prone to conflict? This was a big challenge for
the Somalia context (2).
In the context of chronic insecurity and perpetual crises, some informants questioned
the value of forecasts for triggering action. As one informant put it, El Nio should not
matter to Somalia El Nio and La Nia forecasting because the things that happened
during La Nia or El Nio year are only 10 per cent less likely to happen in any other year
anyway. Therefore, we should be doing this stuff every year no matter what, and especially
when you look at the drought (2). Another informant echoed this sentiment, stating that
humanitarians should consider El Nio as a way of life and not as a kind of sudden thing,
basically meaning that two times a year, this is what we have ... droughts in the north
and floods in the south (2). In an ideal world, where there are sufficient resources to do
everything every year, this may be true. However, as indicated the anecdotes above, in the
context of inevitably limited resources, well-interpreted forecasts in conjunction with flexible
funding can help actors to initiate preparedness and mitigation measures based upon
where a disaster response is most likely to be needed.
Conclusion
These responses to El Nio forecasts demonstrate that in Somalia the information, the
funding, and the activities related to flooding were relatively straightforward. SWALIM and
FSNAU information clearly identified where to repair riverbank breakages and pre-position
supplies. Donors were willing and able to provide the necessary funds, and different actors
had the necessary access to provide these actions in different areas. UN agencies and
NGOs consulted with each other to coordinate the response. There was also agreement
among donors and practitioners that regardless of whether flooding eventually occurred in
2015, these interventions would benefit people in the near future, as floods happen even in
non-El Nio years.
In contrast, response to drought forecasts was almost an afterthought. Informants
believed that the lack of early action on droughts occurred because El Nio forecasts
drew attention to flooding, which is easier to address. During droughts, the risks are more
widespread, and the actions to be taken are less well defined.

36
Timeline of Response to El Nio Forecasts in Somalia
Chapter 3: Somalia Case Study

First responder, a Kenya Red Cross
Society volunteer from Malindi helps fellow
community members during a flood
Credit - Kenya Red Cross
Chapter 4: Kenya Case Study

Background
El Nio has strong teleconnections in Kenya, and its impacts are comparatively well known
by government officials and citizens alike. This is primarily because the 19971998 El
Nio had severe impacts and still looms large in the public consciousness. This El Nio
caused heavy rains damaging infrastructure, killing and displacing many people, cutting
communities off from disaster response (FAO 1998), and causing $USD 236 million in
damages to the agricultural sector (Karanja and Mutua 2000). Because of this experience,
most Kenyans understand El Nio to mean the potential for serious flooding.
When people heard there is such a strong El Nio [forecast in 2015],

everybody was ready for it because they didnt want to repeat
what happened in 1997, and even 2006. People actually took some
precautions because they knew. When people in Kenya hear there is
El Nio, they dont want to take things for granted. They think lets not
ignore like we ignored that other time and so many people died. (3).
The Kenyan Government takes a substantial role in coordination, though precise

procedures for disaster management are still in flux following devolution policies that
passed in 2010.1 The decentralisation process created county governments responsible
for basic service delivery. Following the 2006 El Nio which at least one respondent
1 Decentralisation began in 2010, but was not functional until 2013, when local elections were held.
See http://www.worldbank.org/en/country/kenya/brief/kenyas-devolution
38 Chapter 4: Kenya Case Study

said brought more intense rains than this 2015 (2) the government has changed its
structure, devolving powers to 47 counties, who now have greater responsibility for disaster
coordination. Because the transition to local control only began in 2013, there is still work to
be done on establishing connections and coordination at the county level (2); the roles and
responsibilities of national, county, and NGO officials remain somewhat unclear.
Actors expressed confusion about how much disaster preparedness should come
from local governments and how much from the national government. Current policy
states that disaster management is a shared function between the county and national
government, but rather than clarifying duties, the policy just ends at that (2). The national
government has the intention of passing legislation to harmonise actions and clarify disaster
preparation and response. Several policy proposals have called for creating a single agency
in charge of disaster response, but such legislation has not yet been passed. As a result,
there are several agencies involved in disaster preparation and management e.g.,
National Drought Management Authority, The National Disaster Operations Centre (NDOC),
and the National Disaster Management Unit (NDMU) but no official policy on how they
should coordinate.
Despite these outstanding issues, there was more coordination for early action in
Kenya than in the other countries selected for this study; most informants were pleased with
their response to forecasts. Memories of the devastating effects of the 19971998 El Nio
were a powerful motivation, and appear to have transcended any confusion over precise
responsibilities (2). Furthermore, Kenya experiences frequent flooding around the country on
a regular basis which has spurred agencies to invest in early action measures. (2)
Forecasts in Kenya
Kenya Meteorological Department (KMD) is the nations forecasting lead. The department
receives and downscales regional forecasts from the IGAD Climate Prediction and
Application Centre (ICPAC). After the international forecast has been downscaled to the
national level, county meteorological offices present in all 47 counties are responsible for
further downscaling predictions to the local context.
KMD first saw signs of changes in sea temperature in May and began issuing
preliminary advisories to key stakeholders. The KMD, however, was cautious in announcing
the El Nio to the public at this early stage. KMD and government officials were aware
that an El Nio forecast for 2014 had failed to materialise, and it did not want to alarm
Kenyan citizens unnecessarily. KMD issued their first El Nio advisory to the public on 24
August 2015, estimating a greater than 90 per cent chance that the evolving El Nio would
continue through to the short rains season (October through December) and around an
80 per cent chance it will extend into early 2016 (KMD 2015). KMD released its first formal
El Nio forecast on 2 September, 2015. Following the release of the seasonal forecast in
September, Kenya Meteorological Department (KMD) began providing weekly and monthly
email updates to government and NGO stakeholders. The government delayed formal
public announcements until early October 2015, after the predictions were more certain
and the Multi-sectoral Task Force released their El Nio Response Plan (on 29 September
2015). This plan outlines anticipated El Nio impacts and associated preparedness actions
to be taken across various line ministries, such as agriculture, health, water, education,
transport, etc. In October 2015, KMD issued another warning, stating that the heavy rains
were likely to last throughout the short rains season.
KMD is considered a highly credible source of information. Forecast information
reaches humanitarian stakeholders from many different sources, and it can be a challenge
for practitioners to filter through all the reports (2). KMD, however, provides the most
localised information, much of which is accompanied by general advisories, touching

on key areas that are expected to be affected, like they talk of infrastructure, they talk to
agriculture (2). The information and updates provided on the Kenya Red Cross Society
(KRCS) websites were quite regional and it was quite specific enough to be useful and
guiding enough on a lot of things (2). Because of the 2014 false alarm, the KRCS found it a
challenge to get some communities to believe something would happen in 2015 (2). Despite
this praise, there is always room for improvement. Some stakeholders wished the weekly
and fortnight updates, which responders use to monitor conditions as events unfold, were
delivered more consistently (2).
KMD has well-established dissemination mechanisms. At the national level, weekly
and monthly forecasts are available on their website; they are also disseminated to NGOs
and government partners via email. At the county level, KMD conducts dissemination,
education, and outreach programmes to ensure that no one may be dying of hunger
because they didnt have information (3). County Directors of Meteorology also even go
to the field to meet people, and give presentations to schools and universities (3). For
households, KMD disseminates forecast information in vernacular to individuals through
radio and the internet.
Outreach is driven by public interest. There was a turning point in 1997 for El Nio
awareness in Kenya; before that event people didnt understand what [El Nio] was and so
many people actually experienced heavy losses because of ignoring it (3). Now, the media
and the public take a keen interest in El Nio, so people are interested, they are hungry for
this information, and once you talk of the extreme events, in fact, they are the ones who
come to look for you (3).
Impacts of El Nio in Kenya
In Kenya, El Nio typically affects the short-rains season, which occur from October to
December. The 20152016 El Nio was forecast to cause drier than normal conditions
along the coast and in the northeast, and wetter than usual conditions during the short-
rains (KMD 2015). Although the impacts may be slightly different in different regions,
because of 19971998, Kenyans widely associate El Nio with flooding; most preparations
focus on mitigating the impacts of intense rains, including destruction of transport
infrastructure, landslides and mudslides, outbreak of both human and animal diseases, and
disruption of educational programmes like national examinations (IGAD & ICPAC 2016).
The short-rains, which occur from mid-October to mid-December, were stronger and
lasted longer than usual. In Western Kenya (Kisumu, Migori, and Bungoma) flooding eroded
soil leading, washed away soil nutrients, damaged infrastructure, and destroyed crops and
livestock (ICPAC 2016). Although the impacts were not as bad as anticipated, from October
to December 2015 flooding and landslides displaced over 100,000 people and led to 112
deaths and 73 injuries (IGAD & ICPAC 2016).2
In some instances, the additional rain brought by El Nio was a boon to local
livelihoods. In the arid and semi-arid regions of Garissa, Wajir, Isiolo, Baringo, Bomet,
Machakos and Kajiado El Nio increased rainfall to the benefit of agriculture and livestock
production (IGAD & ICPAC 2016). KRCS actively sought to capitalise on the increased
rainfall in Kitui. The details of this intervention are discussed below, and details of how
KRCS estimated their impacts are provided in Chapter 7.
2 This report made no mention of how these impacts were calculated.

Response to 20152016 El Nio Forecasts in Kenya
Of the five nations in this study, Kenya had by far the most coordinated response to
forecasts. Most actors felt that they were relatively prepared for the 20152016 El Nio
and that the impacts were less severe than had been anticipated based on the forecast
of a very strong El Nio. The generally positive outcome is likely the result of a confluence
of factors including political motivation because of dire impacts of 19971998 event, clear
teleconnections and understanding of the potential impacts, and good coordination. The
following sections outline how early actions in Kenya unfolded.
First warning
As in other countries, El Nio warnings first reached humanitarian stakeholders in Kenya in

early 2015. Many had also heard of the potential for an El Nio from international forecasting
centres in 2014. In contrast to other countries in this study, where international sources
of information were considered most credible, every informant interviewed mentioned
KMD forecasts. The OCHA Kenya office first heard of El Nio around March 2015, at
least six months in advance of the potential impacts (2). Because OCHA in Kenya plays a
coordinating role and assists the government with national contingency planning, it began
re-disseminating forecasts from KMD, as well as updates from FEWS NET, GHACOF,
and OCHAs weekly forecasts to government and humanitarian partners (2). UNICEF
heard about El Nio from KMD and FEWS NET in June or July 2015, and WFP was first
informed in July or August by KMD. DFID Kenya first received the forecasts from TAMSAT in
July 2015, and this prompted them to take action securing funds, reviewing programmes
that required funding, and selecting pre-existing projects through which to deliver the El
Nio aid (2); (Development Initiatives 2016). In August 2015, FSNWG also released an El
Nio alert to its members encouraging urgent preparedness actions to be initiated (FSNWG
2015). The IFRC also received updates from the IRI. It was noted that the consistency in
analysis from these multiple sources, as well as the increasing confidence, aided in the
certainty that preparation was necessary. (2)
Because of the way information flows through the KRCS, people learned of the El
Nio at different times. KRCS (and other organizations, like CARE) proactively seek out
forecast information to inform their programmes. Forecasts generally trickle down, arriving
first to operations and disaster response personnel and later to disaster risk reduction. As
a result, Red Cross staff working with partner national societies first heard of the possible
El Nio in May and June from international forecasting partners and the Red Cross Red
Crescent Climate Centre. The two informants from KRCS disaster operations, however,
only learned of the El Nio in early September (from IGAD, GHACOF, and KMD) one month
before the rains were anticipated to begin. Both agreed that more direct contact between
all departments and the KMD would facilitate planning on all levels (2).
Contingency plans
By all accounts, the National Disaster and Operation Centre began heeding the warnings
immediately, collaborating with sector partners on the development of a national
contingency plan. In September 2014, even before the 20152016 El Nio was announced
(and perhaps in response to false alarms in 2014), the Government of Kenya and
Humanitarian Partners published a five-year (20142018) El Nio Contingency Plan that
outlined the likely impacts of a weak, moderate, and strong El Nio on various aspects of
the economy (El Nio Contingency Plan 2014). Despite the existence of this plan, when

forecasts began to indicate that the 20152016 El Nio would be strong, the Ministry of
the Interior and Coordination of National Government set up a Multi-sectoral Task Force; it
was charged with creating a plan specific to October 2015 through January 2016 and with
mobilising funds for preparedness and response (IGAD & ICPAC 2016).
The first step of most organizations after receiving a forecast of a potential El Nio
was to update their internal contingency plans. For example, KRCS had been working
since 2015 to strengthen its Disaster Management programme. It had already completed
a contingency plan for flood and drought but decided to review and update it in July
and August 2015, after hearing of a possible El Nio. Likewise, after receiving the official
forecasts from KMD, UNICEF spent August and September draft[ing] their own internal
preparedness and response plan, based on their areas of expertise, such as nutrition,
WASH, health, and child protection (2). The Kenyan Ministry of Health partnered with the
WHO to develop health contingency plans for flooding; WHO allocated $USD 2.92 billion
to implement those plans (WHO 2016). During this time, county officials also updated their
local-level contingency plans.
Each organization had a process for developing internal contingency plans. For
example, throughout the contingency planning process, the Secretary General of the
KRCS disseminated information and conducted bi-weekly meetings with its partners
(private Sector and partner national societies) to track progress on developing and funding
contingency plans (2). WFP elected to focus on counties, in which it was already working, to
facilitate later response.
The impacts of El Nio in 19971998 and 2006 were well documented, helping
organizations to know what possible impacts to expect. Humanitarian responders were,
therefore, able to develop plans based on the previous experience of El Nio, and their
expertise and to advise on which specific counties are going to be of higher risk (2).
Because of this past experience, the country was well-guided in terms of the overall
picture and the overall possibilities. The scenario was set (2).
During the KRCS contingency planning processes, representatives of the KMD, the
Red Cross Red Crescent Climate Centre, and IGAD were brought in to speak with disaster
stakeholders and to help translate the forecasts into action. Together, stakeholders were
able to talk through the implications of El Nio as well as the uncertainties, and come with
simple, practical actions that could be incorporated into a revised contingency plan (2).
Participants in those meetings believed that it was incredibly helpful having a climate
practitioner not just a climate scientist, but a climate practitioner who was able to
do the translation job that is so often missing in forecasts (2). Without these experts to
help interpret the forecasts into applicable actions, most of us [humanitarian practitioners]
would look at a weather forecast map and actually not be able to interpret it very well (2).
Once the individual plans across line ministries were updated, the Multi-sectoral Task
Force, led by the National Disaster Operation Centre, held bi-weekly inter-sector planning
meetings. OCHA and the Kenyan government helped feed NGOs and sector-specific
responses into the national-level contingency planning. Despite devolution of authority to
the counties in Kenya, the contingency planning was still largely a top-down exercise (2).
The members of the Task Force then came together to consolidate the individual plans
with updated county-level contingency plans to develop the 2015 El Nio Contingency
Plan. Informants described the process as fostering good coordination and sufficiently
representing government officials, technocrats, ministry representatives, NGOs and the
Red Cross (2).
After the contingency plan was finalised (29 September 2015), humanitarian
organizations continued their planning and began to raise funds. The government put
forward a portion of the funding needed to implement the contingency plan (KSh 5 billion)
in September, but it also asked humanitarian organizations to raise funds to fill the gap
between projected needs and government allocations (KSh 10 billion). By late September,

Development Partners Group, the World Bank, the World Food Programme, and other
donors had pledged funds for the contingency plan (DPPS 2015).
Response3
In Kenya, county-level governments play a pivotal role in preparing for and responding to
impacts related to this forecasted El Nio. While a survey of these actions was beyond the
scope of this research study, a few of the examples of actions reportedly taken by county
governments to prepare for El Nio include: development of local county-action plans in at-
risk counties; dyke reinforcement; desilting flood-prone rivers; establishing loudspeakers for
warning messages in Kisumu; mapping flood-prone areas; alerting residents in those areas
of their risks in Homa Bay; mapping flood-prone roads and infrastructure in Nairobi; and,
drainage expansion in Narok (OCHA 2015; Onyango 2015).
WFP provided food assistance to 170,000 people in three affected counties and
seconded logistics staff to six flood-prone counties (Baringo, Garissa, Samburu, Marsabit,
Tana River, and Wajir) to provide logistical support for preparation and response (WFP
2015). When it first learned of El Nio in August and September, UNICEF re-programmed
$USD 250,000 as initial contingency funds to address emergency response needs during
the pending rainy season. This money was used to assist in purchasing and pre-positioning
non-food items. UNICEF also provided strategic direction to intensify ongoing cholera
response efforts (ongoing since 2014), including procurement of essential supplies to
prevent its spread to additional counties (UNICEF 2015). In December, UNICEF applied for
additional ECHO funding, which became available in January (2) and was used to respond
to the cholera epidemic in Dadaab.
In September and October, as the national El Nio Contingency Plan was being
finalised, KRCS began pre-positioning additional supplies to warehouses around the
country and provided response training to Red Cross Action Teams (RCAT) in counties
expected to be affected by El Nio. Community teams in some locations also took actions
to mitigate the potential impact of floods including digging trenches around farms,
clearing drainage canals for water to flow unobstructed, and moving irrigation pumps and
equipment away from riverbanks.
The forecasts and contingency plans also helped KRCS raise additional funds.
Through British Red Cross (BRC), DFID provided KRCS with a total of about
GBP200,000 extra funding to support and prepare for the anticipated flooding (2). In
October, KRCS began working with the government to disseminate information via an
SMS early warning system. The early warning system used Airtels mobile phone network
to distribute warning messages and tell people to move out of high-risk zones. The SMS
system also enabled two-way communication so that people could provide KRCS with
details on what was taking place in their communities (2). Additional funding was used to
help people who were moving out of floodplains with temporary shelter, medical support,
health, and sanitation (2). In other areas, KRCS helped people in Nairobis informal
settlements along the river move to higher ground.
3 Because 4 of the 10 people interviewed for this Kenya study a were from KRCS, the majority of our data focuses on the
RC response. This does not mean that those organisations did not play a more prominent role.

Social Protection in Kenya
In addition to the actions outlined above, Kenyas Hunger Safety Net Program 2 (HSNP)
was also able to distribute funds in response to drought from the 2015 season and in
preparation for El Nio-related floods. HSNP is one of five safety net programmes under the
National Safety Net Program (NSNP) in Kenya that aim to reduce hunger and vulnerability
by providing cash transfers to orphans and vulnerable children, the elderly, the disabled,
and the otherwise poor and vulnerable. The HSNP specifically is managed by the National
Drought Management Authority and provides regular cash transfers (via bank accounts) to
approximately 100,000 households in Turkana, Mandera, Marsabit, and Wajir counties; it
is also designed to be scaled up in response to floods and droughts (Ndoka n.d). Funding
comes from the Kenyan government and other humanitarian and development partners,
including DFID and the Australia Department for Foreign Affairs and Trade (DFAT). Under
normal conditions, the poorest quarter of households in these regions receive regular cash
transfers, but the number of households receiving assistance is scaled before and during
emergencies. Emergencies are funded entirely by humanitarian donors, including DFID.
Since the programme began in 2015, it has disbursed funds on ten occasions, but only
once in response to the 2015 forecasts.
Stakeholders in HSNP believed the response to the 20152016 El Nio was an
example of a successful no-regrets action. HSNP usually uses a Vegetative Condition
Index to determine pasture availability and the need for drought and flood response
(Fitzgibbon 2016) but decided to respond to El Nio forecasts and the anticipated flooding.
In September and October of 2015, based on information from the National Disaster
Operation Centre and KMDs El Nio Press release, DFID provided an emergency transfer
of GBP3 million to provide anticipatory assistance to 190,000 households who were likely
to be affected by floods. This payment was possible because El Nio arrived at the end of
2015, and DFID was also able to reallocate unused funds for emergency HSNP benefits.
These flood-preparedness payments (disbursed 29 October 2015) were in addition to
the approximately 80,000 households who received their regular payments, and another
5,000households who received drought mitigation payments (26 October 2015) during the
same time period (HSNP 2015). Flooding did not materialise as anticipated, but because
HSNP operates on a no-regret principle (Fitzgibbon 2016), our informant was pleased with
the response. Nevertheless, whether such funding will be disbursed in response to future El
Nio forecasts is uncertain.
Kenyas success in swiftly scaling up the HSNP to help likely flood victims in
advance of an extreme event demonstrates the value of pre-emergency investments
in registering households, opening bank accounts, and managing beneficiary data. In
contrast to other countries, where systems for scaling up were not in place, HSNPs
response went quite smoothly. Similarly, according to programme officials, now that the
programme infrastructure is in place, the cost of obtaining early warning signals and
satellite information is low.
Taking Advantage of Opportunities: Seed Distribution

win Kitui County
Although increased rainfall in El Nio years is more commonly associated with floods, it can
also be a boon to farmers and pastoralists in arid regions, like parts of Kenya. Recognising
this potential in a year when El Nio was forecast, KRCS and BRC took measures to
procure and distribute seeds in the hopes that farmers would realize bumper harvests.
The two organizations pooled nearly GBP 86,000 (some of which was from DFID) to
fund the opportunistic intervention. When most organizations looked to analogue years to

identify and plan for potential harms, Kenya was the only country that looked to the past to
glean possible benefits.
Aided by localised/downscaled forecasts, the partners selected Kitui County (an area
forecast to receive increased rainfall during El Nio); identified 21,000 farmers with two
acres of land available for planting; and purchased and distributed 2 kilogram bags of maize
seeds to each farm all within a three-week timeframe (KRCS 2016).
Even though the El Nio forecast was known months in advance, downscaled
forecasts took much longer and that meant decisions were arrived at hurriedly, because
by the time the forecast is localised enough, the time to get something in the ground is
really short (2).
What was sacrificed in the end was the interventions impact evaluation;
quantification of the yields and associated benefits proved difficult. Some farmers
reported a smaller harvest than usual, but most farmers reported harvesting between
one and four bags (between 90kg and 360kg) more than usual from the seeds they were
given (KRCS 2016: 2). Chapter 7 provides a thorough discussion of the methods used to
derive these numbers.
Securing funds for the potential benefit of an intervention is challenging because
people are used to investing in response and we are used to investing in preparedness,
but donors are not necessarily used to investing in [the upside] of forecasts (2).
Conclusion
In Kenya, previous experience with El Nio appears to have motivated both the
government and the broader humanitarian community to plan. Despite some confusion
regarding devolution policies, actors appear to have worked well, under government
supervision, to coordinate their response. Actors believed that good coordination and
collaboration was essential to planning a unified response. In the end, the flooding was not
as bad as anticipated (or, at least, not as bad as 19971998), but actors were prepared to
respond when the flooding arrived. KRCS was also one of the only organizations across
all five countries to use forecasts to mobilise funds to help farmers take advantage of
increased rains. Although evaluation of the project was a challenge, it appears to have
been successful.
An aerial photo of
Modogashe Garissa County
during El Nio season 2015
Credit - Kenya Red Cross Society

Timeline of Response in Kenya

Photo: Charlotte Floors/NLRC
Chapter 5: Ethiopia Case Study
Background
Ethiopia is one of the most populous countries in Sub-Saharan Africa. Although it is

considered to be within the Least Developed Countries, Ethiopias economy grew
impressively over the past decade, averaging 10.8 per cent per annum from 20032004
to 20142015 (World Bank 2016). Over a similar period, from 20002011, the number of
Ethiopians living in extreme poverty reduced significantly from 55.3 per cent to 33.5 per
cent and the Ethiopian government is currently implementing the second phase of its
Growth and Transformation Plan to continue these strong trends (ibid.).
In parallel to these positive trends, Ethiopia also faces regular food insecurity and
humanitarian situations. The government receives over $USD200 million (GBP 166.2) in
food assistance each year (USAID 2016). In collaboration with longstanding development
partners, the Ethiopian government has managed the Productive Safety Net Programme
(PSNP) since 2005, to reduce chronic food insecurity through a combination of food and
cash transfers to help households overcome food deficits. As of 2012, the programme
served 7.64 million food insecure people each year (PSNP n.d.).
The Disaster Risk Management and Food Security Sector (DRMFSS), under the
Ministry of Agriculture, coordinates humanitarian preparedness and response within the
Government of Ethiopia. Sectoral Task Forces manage sector-specific coordination. The
Development Assistance Group (DAG) coordinates support from 30 bilateral and multilateral
partners toward national development priorities. The humanitarian community in Ethiopia
has also established the Ethiopia Humanitarian Country Team (EHCT) to lead management
of the cluster system and provide overall coordination of non-governmental humanitarian
affairs. This country team is led by OCHA and includes representatives from WFP, FEWS
48 Chapter 5: Ethiopia Case Study

NET, UNICEF, IFRC, ERCS, FAO, USAID and CRS (OCHA 2016d, EHCT 2015). Jointly, the
Government of Ethiopia and the EHCT publish a Humanitarian Requirements Document
semi-annually (at the beginning of the year in January and then mid-year in June or July)
(2). This document is the basis for humanitarian action in Ethiopia; during emergencies, it is
amended outside of the formal review periods.
Forecasts in Ethiopia
Ethiopias National Meteorological Agency (NMA or Met) produces three-day, weekly, 10-
day, monthly, and seasonal forecasts. They are distributed via its website, email, and media
outlets (with plans to distribute by SMS soon). NMA uses the regional consensus forecast
from the Greater Horn of Africa Climate Outlook Forum (GHACOF) to develop downscaled
forecasts for regions within Ethiopia. NMA then holds meetings in Ethiopia to allow officials
from government agencies to raise questions about the seasonal forecasts (1). During the
20152016 El Nio, the governments Disaster Risk Management Commission held regular
(every 24 weeks) update meetings, at which NMA presented on the evolving El Nio.
Government and NGO stakeholders in Ethiopia confirmed receiving updates from NMA as
well as several other sources.
The Government of Ethiopia also uses the Livelihoods, Early Assessment and
Protection (LEAP) system to manage acute food insecurity and integrate the PSNP into
the national disaster risk management framework. LEAP combines forecasts with over
20indicators (rainfall, potential evapotranspiration, crop soil, water balance, yield, moisture
index, NDVI, etc.); it is used to prepare briefings for the ministries of health, agriculture,
water and education (1). Regional products that include Ethiopia were also provided by
IRI, UK Met Office and ACMAD. Lastly, FEWS NET produces a combined climate and
agriculture product to project acute food insecurity throughout much of Africa, including
Ethiopia (FEWS NET 2016). This is a non-governmental system that is informally used by
non-government actors.
Impacts of El Nio in Ethiopia
Ethiopia is topographically and climatologically diverse. The three meteorological seasons

include: (1) the short, Belg, rains (FebruaryMay) that fall primarily in the eastern Oromia
and Hareri Regions as well as parts of SNNP, Amhara, and southern Tigray; (2) the
long, Kiremt, rains (June September) that fall primarily in Tigray, Amhara, Beneshangul-
Gumuz, Gambela, western Oromia and parts of SNNP; and, (3) Bega, the long dry season
(OctoberJanuary) (Singh 2016). According to Korecha and Sorteberg (2013), the Belg
rains are difficult to predict and are not susceptible to El Nio-Southern Oscillation (ENSO).
The Kiremt rains, however as documented by Korecha and Barnston (2007) are
governed by ENSO events. The result typically leads to below (above) normal rainfall during
El Nio (La Nia) events; local climate drivers have a secondary impact. There are two
rainy seasons in the Somali region of Ethiopia, the Gu from April to June and the Deyr from
October to November.
In early 2015, the Belg rains never fell. This was officially acknowledged by the NMA on
4 June 2015. Most global forecasting centres declared the presence of El Nio conditions
between March and May 2015.1 By mid-May, there was inconsistency in global and regional
predictions for Ethiopias seasonal forecast. Many of the seasonal forecasts produced by
1 Forecasting centres rely on slightly different parameters resulting in the varied timeframe.

global or regional forecasting centres showed some indication of unusually dry conditions
for the Kiremt rains. However, the area, confidence and timing differed depending on the
forecast (Singh 2016).
The short amount of time between the start of an El Nio event and the start of the
Kiremt rains, results in relatively short advanced warning that a below normal rainfall season
is likely. As a result, the ability of the government and humanitarian actors to confidently
take early action is limited. During interviews with various stakeholders the nuance
associated with forecasting El Nio impacts on Kiremt was not understood, leading to
somewhat biased assumptions.
North, central, and eastern parts of Ethiopia suffered higher than usual crop failures
and a shortage of pasture because the Belg and Kiremt rainy seasons failed to deliver the
usual precipitation. Food prices rose, greatly exacerbating food insecurity in the country
(IGAD & ICPAC 2016). As of late August 2016, 9.7 million people were in need of food
assistance (DRMTWG 2016); Ethiopia needed to rapidly import significantly more food and
supplies than usual. These needs were in addition to the almost 8 million people supported
through the PSNP. Much of the PSNP support is provided through food aid, compounding
the problem of food availability in the country during mid-2016. Ultimately, 10.2 million
people in northern Ethiopia needed food aid as a direct result of the failed Belg and effects
of the 20152016 El Nio (IGAD & ICPAC 2016).
Response to 20152016 El Nio Forecasts in Ethiopia
This section provides an account of how humanitarian response to the 20152016 El

Nio unfolded in Ethiopia as detailed by the ten informants from ten different government
ministries and humanitarian organizations consulted for this study.
Most stakeholders learned of the El Nio forecast in May and June of 2015 (2).
Because WFP is aligned closely with FEWS NET, they were warned of a potential El Nio
as early as February. OFDA, on the other hand, didnt learn of the El Nio from FEWS NET,
sources within USAID, and others2 until June. Given the overwhelming consensus for the
prediction, most informants believed the forecasts were credible and timely.
By all accounts, humanitarians understood the implications of the El Nio alerts and
related seasonal forecasts. In response to the forecasts, both international organizations
and the government began consulting with one another and monitoring the situation on the
ground. WFP monitored food prices and interviewed local people in vulnerable areas about
crop and weather conditions to identify emerging strain. The government of Ethiopia and the
EHCT released a revised Humanitarian Requirements Document in August 2015; it warned of
the potential impacts of El Nio on a failed Kiremt season and the increased number of people
in need of food relief (from 2.9 million to 3.5 million) due to the failed Belg (GoE 2015). At the
same time, OCHA began speaking to donors about the predicted impacts and what resources
were needed to respond. OCHA worked with DFID and other members of the Humanitarian
Country Team (HCT) to monitor El Nio conditions through July and August. On request
from the government, OCHA and other organizations used impact reports from analogue
El Nio events in 19971998, 20022003, and 20102011 to outline likely impact scenarios
and detail how the humanitarian community could work together to reduce impacts. (2). This
report was published in September by the government, OCHA, HCT and the DAG. In October
2015, the Government of Ethiopia announced an official increase in the number of people
requiring humanitarian assistance, raising it to 8.2 million. The response was formalised by an
addendum to the 2015 Humanitarian Requirements Document (OCHA 2015e).
2 National Meteorological Agency, the Agricultural Task Force, the UK Met office, FEWS NET, IFRC, and NOAA

El Nio forecasts enabled donors in Ethiopia to reprogramme their funds and make
resources more readily available. DFID Ethiopia acted very early because the UN identified
pretty early on that there was going to be a problem (4). OFDA and USAIDs Food for
Peace programme increased the amount of funding available for Ethiopia, in line with
HRD projections. But development partners were more hesitant to respond to forecasts.
Based on forecasts, USAID released funding to increase the number of districts receiving
assistance from 32 to 76. To make this possible, they shifted funds from the development
food programme to emergency response. DFID took similar actions and was, therefore,
able to quickly provide UNICEF and other partners with Ready to Use Therapeutic Food
(RUTF), alleviating suffering significantly.
Despite the shifts in funding, only one NGO interviewed reported that it acted
before the government officially declared a crisis, in response to deteriorating conditions.
This NGO used crisis modifiers to change existing programmes rather than developing
a response to a new problem. Using this approach, they began livestock vaccination,
rehabilitation of water points, and organized nutrition clinics as early as June and July 2015.
Other informants indicated that a lack of clarity between donors and the government over
who would fund various expenses for preparedness actions caused delays.
Reports in June estimated 2.9 million people would need assistance, but this number
increased rapidly. June is the most important planting season in Ethiopia; by August, it
was clear that the rains and the crops failed and that people needed assistance. When
the humanitarian community conducted a rapid assessment in late September, the needs
were already far higher than [the government] had thought (2). An assessment conducted
in October estimated that 8.2 million would need assistance. By the time the HRD was
published in early December 2015, the number of people in need was 10.2 million. After the
December release of the HRD, the government accepted and acknowledged the growing
need (4). It issued a humanitarian appeal that month (World Bank) and began contributing
funds toward disaster response.
Once the humanitarian response began, needs were beyond the capacity of
the national grain reserve and the PSNP, in part because the PSNP is historically a
developmental safety net programme and not a humanitarian programme. It can scale to
meet some needs but is not an instrument to manage large scale needs outside of PSNP
coverage areas. The sudden need to import vast quantities of food and other agricultural
and humanitarian supplies overwhelmed warehouses, distribution channels, and even the
port of Djibouti, causing backlogs and compounding delays. The PNSP was also entering
the first year of PSNP-4. This new phase of the PSNP focuses on integrating the PSNP with
disaster risk management systems, including early action and response to emergencies.3
Under this programme a new set of disaster risk management mechanisms were being
set-up but were not fully operational. This unfortunate timing challenged the governments
ability to register beneficiaries in the height of the crisis when the number of people in need
well exceeded the ordinary capacity of the programme. By February 2016, extraordinary
efforts by key humanitarian actors in Ethiopia resulted in the international communitys
recognition of the urgent need to address the crisis (2).
Most humanitarians felt that they anticipated the gravity of the situation accurately,
but were unable to take action beyond some minor pre-positioning of supplies
because of political constraints. (2). As a result, all main actions that should have been
taken between May and September were taken between December and February (2).
Most informants believed that a gradual scale up of the PSNP and more early action would
have alleviated suffering and reduced the overall cost of the crisis response. Nevertheless,
several stakeholders were proud of how quickly they scaled an additional payout to
3 http://projects.worldbank.org/P146883?lang=en

existing PSNP beneficiaries. PSNP also waived public works requirements and adjusted its
livelihood interventions to account for the drought.
Obstacles to Acting on Forecasts
Some informants noted early indications that development-focused stakeholders in

Ethiopia were reluctant to initiate action based on preliminary alerts that an El Nio event
was emerging. To a certain extent this is understandable. The intensity of ocean warming
that causes an El Nio event does not correlate to the intensity of its anticipated impacts
in the affected region, which is commonly misunderstood. Therefore, it is not possible to
predict the full extent of potential impacts of an El Nio event; such an event simply helps to
determine if a season is more or less likely to be wet, dry, or normal. While some informants
stated that they knew the severity of what was coming, there is no scientific basis for that
certainty. Furthermore, as detailed above, early seasonal forecasts showed little El Nio
influence and early international forecasts were inconsistent in their predictions.
However, meteorological drought and agricultural drought are not the same. Rapid
early action measures should have begun in June when the NMA notified stakeholders
of diminished Belg rains and a forming El Nio. Insufficient rainfall is common in the more
sparsely populated, pastoralists areas in southern Ethiopia, but during El Nio drought is
more likely to affect the eastern and northern regions of the Ethiopia, where a significant
number of people live. It is possible that, because development actors were accustomed
to effectively addressing droughts through existing systems, they genuinely believed the
government had the drought under control.
Humanitarians working in Ethiopia had high praise for the forecasts they received and
believed that the forecasts provided sufficient lead time to allow for early action and mitigation.
I think we scripted it almost perfectly. I think we were pretty spot on with all our forecasts.
This is one thing that is very, very clear: Ethiopias early warning system did not fail (2). As
another explained, Ethiopia does not lack tools, information, analysis, and data (2).
It was noted that various political constraints contributed to preventing early action (2)
and that good information is still subject to political influence. Stakeholders had good data
on time, but then when you interpret it into what it means in relation to how many people will
need assistance, thats where the difference and the debate come up (4). Forecasts only
provide projections; without certainty, some governments may be reluctant to act. When it
is politically challenging to take early action in response to forecasts, impact assessments
proved to be a powerful tool to convey urgency. The HRD was widely credited as the most
compelling impetus for action. In future events it will be critical to delineate between a forecast
of potential impacts and a humanitarian emergency; taking no-regret actions based on the
former, before the latter develops, is key to preventing humanitarian impacts.
Conclusion
In Ethiopia, communication constraints, a lack of clarity regarding funding for large scale
preparedness actions, and a possible lack of confidence in the gravity of the forecasts,
delayed action and prevented most actors from taking early action. Several donors began
freeing funds to enable a prompt disaster response once an emergency was declared,
but only one NGO interviewed acted in response to deteriorating conditions, before the
government needed to officially declare a crisis. These actions were made possible by crisis
modifiers that allowed funds to be allocated to scale up existing programmes. Humanitarian
action was eventually required, but it was only authorized after the HRD was released in
December 2015, providing concrete evidence of need.

Timeline of Response to 20152016 El Nio Forecasts in Ethiopia
Chapter 5: Ethiopia Case Study

53
Chapter 6: Malawi Case Study
Background
Malawi is a country that experiences annual shocks and needs consistent humanitarian
and development assistance (2). It ranks 173 of 188 countries according to the UNDPs
Human Development Index (Government of Malawi, 2016b). As a country with one of
the most erratic rainfall patterns in Africa, climate stressors have a major impact on
development (Government of Malawi 2016a: 5). For example, the country has experienced
eight major droughts since 1980 (Government of Malawi 2016a). High inflation, high rates
of HIV infection (ten per cent), undernutrition, and low education rates all affect Malawis
ability to cope with shocks and to secure higher levels of economic development (WFP
2016 Malawi Brief). According to reports published by Malawis Department of Disaster
Management Affairs, Malawis weather-related crises, coupled with a weak economic
profile, and weak land governance have combined to create a vicious cycle of food
insecurity and malnutrition, with devastating consequences on basic services and,
consequentially, on long-term development (Government of Malawi 2016b: 1).
The 20152016 El Nio followed a season of severe flooding. Because the
country already had exceptional flooding in early 2015, which very much shook up the
Humanitarian community in Malawi people were already geared up to expect some kind
of climate disaster in late 2015 and early 2016 (4). Rather than being an isolated shock, in
Malawi, the 20152016 El Nio manifested as one emergency connected to another (4).
Forecasts in Malawi
The Department of Climate Change and Meteorological Services (DCCMS) is responsible

for making localised predictions in Malawi. Malawis National Climate Forum for Producing
and Originating Forecasts convenes before the SADC forum to develop Malawis input to
the SADC process. After the SADC countries meet to develop a consensus forecast for
the region (Southern Africa Regional Climate Outlook Forum), regional consensus forecasts
are brought back to Malawi where DCCMS is then responsible for reintegrating local
information and downscaling the international forecasts to the national and sub-national
level. Malawi has only recently in 2014 and 2015 begun downscaling forecasts to
the district level.
Malawi has one rainfall season, which begins in October and extends through March
in the south of the country and through April in the North (DCCMS 2015a). DCCMS releases
two seasonal rainfall forecasts for this period. It releases the October to December (OND) in
September, in advance of the rainy season and a forecast for January to March (JFM) partway
through the season. Because El Nio signals typically emerge in May, June, and July, DCCMS
integrates El Nio information into the seasonal forecasts that are released in September.
The seasonal forecast released in September 2015, specifically advises farmers to consider
planting early maturing varieties to cope with below-average rainfall (DCCMS 2015b).
DCCMS has clear dissemination protocols, and all forecasts must be cleared for
dissemination by the Office of the President and Cabinet. Once approved, forecasts are
disseminated to Department of Disaster Management Affairs (DoDMA), who then works
with other stakeholders to develop contingency plans. When the seasonal forecasts are
released, DCCMS forecasters meet with DoDMA representatives in person to provide
briefings on what they should expect in the coming season (1). DCCMS also disseminates
five- and ten-day weather updates via email, SMS, television, and the DCCMS website.
Most informants interviewed for this study reported that they receive forecast information
via email lists as well.
54 Chapter 6: Malawi Case Study

Malawi is one of two pilot countries for the Global Framework for Climate Services
Africa Program (GFCS or GFCS-APA), an initiative that seeks to build capacity for forecast
production and dissemination. The project is premised on the idea that access to seasonal
forecasts and historical climate data can improve government and household decision
making. Because of this emphasis on influencing local decision making, DCCMS and other
GFCS partners in Malawi have invested considerable effort on disseminating forecasts to
the community level through radio, SMS, and WhatsApp to promote community-level and
individual action. However, they have found that due to complex concepts and terminology
many households are still unable to put it down to action level (2). Informants noted
that GFCS efforts to produce and disseminate forecasts and recent climate shocks have
generated increased interest in and demand for DCCMSs forecast information in Malawi.
Impacts of El Nio in Malawi
El Nio-related dry spells in the south and rains in the north exacerbated existing food
shortages. A DoDMA report released in June 2016, estimates that the 20142015 flooding
and the 20152016 El Nio impacts left 6.5 million Malawians in need of Humanitarian
aid 2.4 million of whom are farmers who lost their production (Government of Malawi
2016b). In the southern areas, dry spells are estimated to have affected over 51 per cent
of cropland. Production of maize, the countrys staple food, declined by 12.4 per cent
from April 2015 to April 2016, falling below national food requirements for the second
consecutive year (Government of Malawi 2016b). Shortages of water also contribute to food
insecurity and malnutrition.
Although government reports provide many statistics on the impacts of El Nio, they
do not explain how these numbers might be attributed specifically to El Nio rather than
development or governance challenges.
When asked about forecasts, Malawi informants spoke of recent responses to
weather variability in general, not just El Nio. There are at least three possible explanations
for this: (1) respondents did not understand what El Nio is, and what it means for Malawi;
(2) the 20152016 El Nio immediately followed four consecutive seasons of climate
shocks; and (3) many of the informants interviewed in Malawi are part of ongoing projects
seeking to use forecasts to improve decision making and humanitarian outcomes.
Whatever the reason, the fact that many early actions were extensions of efforts to respond
to climate shocks in 20142015, and that the 20152016 El Nio exacerbated such
problems, makes it difficult to untangle the precise impact of El Nio forecasts in prompting
action and mitigating impacts. As of our interviews, the humanitarian community in Malawi
was still overwhelmed responding to previous climate shocks, and had not conducted any
assessments of impacts avoided to credit early action from forecasts. While this chapter
tries to focus on responses to El Nio forecasts, it also gleans lessons from other recent
applications of forecasts for disaster mitigation.

Response to El Nio Forecasts in Malawi
By the time El Nio forecasts began reaching stakeholders in Malawi, the humanitarian
community there was already expecting a humanitarian disaster as a result of the severe
flooding that happened earlier in 2015. As outlined below, a few actors in Malawi were able
to shift funding and build upon existing efforts in response to El Nio forecasts. However,
early action was complicated by lack of funding, lack of understanding of El Nios potential
impacts in Malawi, and ongoing response to earlier floods. The sections below provide an
overview of early actions and disaster response efforts in Malawi, detailing these successes
and challenges.
Unlike in other countries, where informants reported learning of El Nio from March
to July, many stakeholders in Malawi appeared to have been informed later in the year.
This may have been because it took time to confirm the forecasts and understand the
implications for the country. Higher-level employees within WFP Malawi began to receive
warnings of a possible El Nio from WFP headquarters in Rome and other international
sources as early as February and March of 2015. Higher-level WFP staff eventually
conveyed the information to Malawi programme staff in July and August. Most other
organizations also reported hearing about a potential El Nio in July and August, but there
was no confirmation that we [were] going to have an El Nio in the 20152016 season.
Confirmation came later (2). UNICEF started receiving updates from their headquarters
in New York, as well as, their regional office in Nairobi, followed by updates from FAO and
FEWS NET in early August. The Department of Disaster Management Affairs (DoDMA) first
received warnings from DCCMS and SARCOF in June 2015. However, other important
government partners, such as Ministry of Agriculture, Irrigation, and Water (MoAIWD), did
not recall being informed until December 2015 or January 2016 through other stakeholders
at an Agricultural Production Estimate Survey (APES) meeting, and then through DCCMS
and FEWS NET emails predicting normal to below normal rainfall in the coming season.
FAO also reported learning of the El Nio as late as September 2015 when SARCOF
released the seasonal outlook. Official confirmation of the El Nio arrived in September and
October when DCCMS released their national seasonal forecast.
Contingency Planning
Most humanitarian organizations first responded to the forecasts by modifying their

contingency plans, a process that takes place every year in Malawi with an emphasis on
planning in districts that are particularly prone to disasters (1). WFP, for example, revised its
internal contingency planning from March to June, updating checklists for each department
within WFP to ensure everyone was prepared to respond to anticipated shocks. Similarly,
the Malawi Red Cross Society (MRCS), took stock of its resources and capacities and
began outlining operating procedures. UNICEF also updated their contingency plans and
pre-positioned stocks for managing and treating malnutrition.
Each year, once stakeholder contingency plans are complete, DoDMA leads an effort
to aggregate those plans into a national contingency plan in conjunction with stakeholders
from the UN Cluster groups. Sector-specific experts from each Cluster then define sector-
specific vulnerabilities and preparedness plans. Coordination among the partners is
considered imperative to effective disaster mitigation and response so that when it comes
time for it to be implemented we should all understand what we are doing and respond
in a coordinated manner (1). WFPs vulnerability assessment and mapping (VAM) team
conducted their food insecurity analysis and projections in June, for their response planning
and budget projections. Upon release of the assessment results, WFP supported the
government in coordinating response planning as the Food Security Cluster co-lead.

In 2015, the El Nio forecast prompted DoDMA, DFID, WFP, FAO, and other
stakeholders to study events in previous El Nio years to determine what the impacts were
likely to be and which areas were most likely to be hit. DoDMA specifically encouraged
partner organizations to compare the analogue years to the current year or the expected
season when we plan to develop the contingency plan (1). In October and November
2015, FAO used rainfall data and satellite imagery (NDVI indices) for the past three El
Nio years to prepare a summary of the likely impacts. In this way, the forecasts actually
helped [stakeholders] in coming up with the best national contingency plan with different
scenarios based on that weather forecast from our colleagues from the meteorological
department (1). The national-level contingency plan was published in November 2015,
and it states that El Nio has the potential of causing reduced rainfall during the main
December 2015 to March 2016 crop-growing season in southern Malawi. Conversely,
northern Malawi is more likely to receive normal to above-average rainfall from December
15 to March 16 and this may result in dry spells and flooding events respectively
(Government of Malawi, 2015: p. 9).
Response
Humanitarian stakeholders are only allowed to go into actual response, after the contingency
plan is completed, but the scale of the response depends on how much resources have
been mobilised at that point (2). As in many other situations, limited funding for contingency
plans creates bottlenecks. DoDMA began lobbying the Treasury for funding because if [they]
end up having these plans but then no resources, [they] will not be able to implement (1).
Unfortunately, the government was unable to finance the entire contingency plan.
Many donors are reluctant to release funds until disasters are under way. As one
respondent explained, I can say, I want money to pre-position these materials. They will
not give it to me. But when I say, I want money to rescue these people, they will give it to
me (2). In some instances, slow donor funding meant that partners were unable to take
early action. As late as September 2016, WFP was still in the process of responding with
the available resource because it had not gotten all the resources it needed (2).
By building on existing programmes, mobilising flexible funding, and shifting resources
from existing programmes in Malawi several organizations were able to take action in the
absence of large influxes of aid. MoAIWD enhanced programmes promoting irrigated
agriculture and climate smart agriculture (1) to better cope with the lack of water and
began negotiations with private sector food producers to import food for the strategic
granaries reserve and counter shortfalls in local production (1). Organizations without
existing funds found the forecasts a valuable resource mobilization tool, using them to
explain that, according to this years forecast, it is saying a, b, c, d and we need to have 1,
2, 3 things in place (2).
In October 2015, FAO made funds available from the Common Emergency
Response Fund and reprogrammed existing resources in a way that would help Malawi
cope with the anticipated lack of rain in the south. Based on forecast information and
remote sensing imagery from past El Nio seasons, FAO developed advice on when
farmers should begin planting. They also investigated drought-resistant seed varieties
(sorghum, millet, pigeon pea) and conducted market assessments to try to convince
farmers to change their crops. The programmes encountered other obstacles, as the
behaviour of people, the decisions, the choices of people are difficult to influence in the
middle of humanitarian response (2). In February, FAO conducted a rapid assessment of
the entire country and met with WFP, UNDP, MRCS, and other humanitarian partners to
coordinate a response to the impacts. Upon realizing the drought was widespread, they
decided to change their activities in the country, shifting from maize distribution, which is

very sensitive to drought, to more drought tolerant crops. It also shifted the timing of its
annual distribution of sweet potato, cassava and other drought-resistant crops to later in
the season, when the forecast indicated the probability of rain. Unfortunately, most crops
still failed because the rains never came (2).
DFIDs regional office was among the most proactive of those we interviewed
regarding seeking out and responding to El Nio forecasts. The DFID regional office
oversees the whole of Southern Africa, and in light of the climate shocks from the previous
few seasons, it was monitoring the forecasts to plan ahead. The regional office began
receiving information from DFID central in the Summer of 2015. DFID also contracts with
researchers at University of Reading directly to request additional weather and monitoring
information on a short-term basis. As a first step toward action, in July 2015 DFID Central
called a meeting to look at what the different UN agencies and the different donors were
doing in terms of response to the current food insecurity in the region (4). By November,
DFID hired additional humanitarian advisors and placed them in country offices in the areas
likely to be hit. By December 2015, DFID released GBP 4.5 million for early action. These
funds were given to WFP for pre-positioning items for floods and to increase nutritional
surveillance to detect impacts more quickly.
Forecasts also helped DoDMA and MRCS with flood preparations in the north. The
El Nio forecast indicated that typical drought and flood patterns would be reversed; there
would likely be dry spells in the southern region and flooding in the north. This information
was key in helping MRCS and DoDMA pre-position flood relief supplies and dispatch
search and rescue personnel in Karonga district in the north. Pre-positioning took place
in October and November so that partners had all the required resources on the ground
before the actual flooding (1). MRCS did not provide estimates quantifying resources
saved. However, without those forecasts, MRCS would [prepare] nationwide, which
would be a waste of resources. This information was giving us where, exactly, we should
concentrate (2). Because of the El Nio warnings, DoDMA claimed not to have lost any
lives to flooding in the north.
Existing disaster response efforts provided an avenue for action, but for some
organizations, they also diverted attention from preparatory action. While trying to prepare
for the El Nio, WFP and DFID were also responding to the weather shocks that affected
crop failure from the last season (2). When the El Nio forecasts came in, DoDMA was
coordinating the Malawi Floods Emergency Recovery Project in response to the 20142015
floods. Hearing about El Nio was, therefore, like a double tragedy in our case coming
from floods and then we are affected by dry spells. So immediately after getting the
information, I dont think there were any specific interventions to look at the El Nio. The
focus was on the recovery (1).

Forecast Limitations
In contrast to successes cited above, several problems with the information surfaced in
conversations with forecast users in Malawi. These included forecasts that came too late,
vague forecasts and advisories, confusion over likely impacts, and lack of trust in forecasts.
Although some organizations were able to make community-level recommendations
based on the forecasts, for others the forecasts came after critical decisions were made.
By the time Malawi Red Cross staff were told, we will have dry spells, there was nothing
[MRCS] could do because people had already planted (2). This indicates that while El Nio
warnings had surfaced as early as May 2015, it did not trickle down to NGO stakeholders
and farmers in time to influence key agricultural decisions.
Many forecasts users complained of the generally vague nature of the forecasts.
Although DoDMA and other stakeholders used the forecasts to develop contingency plans,
they also lamented that the forecast is mainly general. Its not area specific, and its not
even sector-specific (1). Sector-specific experts from ministries and NGOs are expected
to translate the forecasts into action, but many of them are not clear how to interpret them.
Some forecasts did not provide sufficient detail as to when events would occur, instead
indicat[ing] that there would be rains coming, but they did not say when (2).
The advisories accompanying the forecasts were often geared at individuals and were
overly simplistic Review your contingency plan and start making preparations (2) or,
Make sure you have an umbrella. Make sure you have a raincoat. Make sure children are
accompanied to school. Make sure you dont go out of the house (2). When it comes
to defining actions, current forecasts dont indicate the impact on any sectors. They just
say how the rains are going to perform (2). All those interviewed agreed that, in the future,
forecasts would be more useful if emphasis could [be] put on advisories because just
having information is not enough (2). Instead, more effort is needed to answer the question,
How do I use the information that I get? (2).
Understanding the forecasts in Malawi is complicated by the fact that El Nio has
opposite impacts in different parts of the country flooding in the north and droughts
in the south. Based on the forecasts alone, it was unclear to funders and humanitarian
organizations whether flooding or drought would be the primary impact. Some [donors]
were anticipating floods, but the information we were getting was not specific enough (4)
to act. It took stakeholders some time to untangle conflicting information, as everyone was
saying that normally, after El Nio, Malawi has an awesome harvest. When El Nio comes,
Malawi always does great, which was at odds with forecasts of below-normal rainfall (2). To
remedy misunderstanding of what El Nio means for Malawi in the future, there is a need
to systematise in a proper manner the different effects of El Nio in Malawi in the last year
(2) and to provide more nuanced analysis (4) of forecasts so that stakeholders understand
the implications from the beginning.
GFCSs efforts to disseminate forecasts to communities themselves led many
informants in Malawi to emphasize the need for better dissemination and interpretation of
the forecast information for use by individuals and households. Forecasts can be distributed
to individuals on social media, radio, and SMS, but they need to be translated into terms
people can understand. A forecast that talks about normal or above-normal rainfall is of
little help to most people. To somebody who is in the village, theres no information (2).
Tercile projections are difficult to translate into household planting decisions.

Social Protection and El Nio Response in Malawi
Malawi has two existing social protection programmes, the Farm Input Subsidy Program
(FISP) and the Malawi National Social Support Programme (NSSP). The NSSP provides
cash transfers to the poorest ten per cent of households, most of which are labour
constrained. It also supports households that are not labour-constrained through work-for-
asset programmes in which individuals construct infrastructure in exchange for income.
In some regions, it also provides services such as, school feeding programmes, loan
and saving programmes, and microfinance (GIZ 2105), which provide social cash transfer
directly to beneficiaries. The Malawi Vulnerability Assessment Committee (MVAC) also
meets annually to assess the state of food security in Malawi every year. WFP is a major
partner of that committee. So the MVAC response is the humanitarian response that
happens every year (2).
World Food Programme and DFID attempted to link their humanitarian response
to the 20152016 El Nio to social protection programmes in Malawi, but these efforts
were complicated by the number of funding streams and the use of different distribution
channels. Although humanitarian social protection funds were intended to complement
existing social protection mechanisms, it was a challenge to integrate these additional
benefits with existing programmes. Humanitarian organizations were funding programme
expansion, which was coordinated through a parallel set of databases for registering
beneficiaries rather than through the existing systems.
These complications were largely seen as a symptom of the need for increased
coordination between humanitarian action and development in general. The plethora of
ongoing resilience, development, and market initiatives taking place in Malawi means that
when the humanitarian response comes, or the social protection programmes come, theres
a kind of crowd programmes at different levels and you always struggle to understand
who is doing what and what kind of activities and when you are programming (2).
As a result of these complications, stakeholders learned the importance of planning
for better integration with existing systems going forward, and there has been progress
on developing a single fund for the Malawi National Social Protection programmes. In the
future, rather than contributing haphazardly, in a fragmented fashion, donors will be able to
contribute to a single fund that would then be distributed through existing systems based
on specific climatic or weather-related triggers (2). Donors are also working to establish
a Unified Beneficiary Registry, which would include of the poorest 50 per cent of the
population so as to track who receives benefits in times of humanitarian crises. In order to
know who to target during disasters, more research needs to be conducted into whether
households receiving social protection benefits are particularly vulnerable to disasters and
which populations would benefit most from additional social support in times of crisis.
These steps may help to improve coordination of benefits during future events.

Non-information Considerations in Malawi
In response to a question about the challenges to using information, one informant

summarized the issues well:
The challenges are not necessary to do with weather forecasting. Its about how
robust and adequate the coordination mechanisms are; how good the real-time
monitoring is; learning across sectors; and how robust the beneficiary feedback systems
are. And then also, the overview and the leadership of the whole response, particularly
on the power of government. So, I think those are major, major challenges in Malawi,
which dont necessarily have anything to do with receiving good climate forecast
information. It is to do with leadership on the ground and how effective and efficient the
humanitarian system is (4).
Conclusion
El Nio forecasts in Malawi were delivered in the context of ongoing humanitarian response
and longer-term efforts to develop forecast production and distribution capacity. Because
of these pre-existing efforts, stakeholders in Malawi were conditioned to look to forecasts
and were already providing relevant services in many areas. Many organizations acted
on the forecast information by modifying existing programmes, but in some instances,
the need for response to previous disasters took precedence over planning for El Nio.
Delivery systems and beneficiary information were fragmented; humanitarian stakeholders
were unable to scale up existing social protection systems, but they were optimistic that
the lessons learned during the El Nio such as the need to pool funds and integrate
databases would encourage the use of these programmes in the future.

62
Timeline of Response to the 20152016 El Nio in Malawi
Chapter 6: Malawi Case Study

Photo: Charlotte Floors/NLRC 63
Chapter 7: Measuring Impacts
One of the studys important deliverables was collecting evidence showing that early action
in response to El Nio forecasts led to better humanitarian outcomes. At this time, only two
organizations Concern Worldwide (Concern) and the Kenya Red Cross Society (KRCS)
quantified the benefits of their early actions. The studys authors attribute this scarcity to
two factors: (1) organizations are still responding to the impacts of the 20152016 El Nio;
and, (2) conclusive measurement of programme impacts (or impacts avoided) is inherently
complicated as this section will demonstrate. However, these two reports serve as a
starting point for discussion of the opportunities for and challenges to quantifying benefits.
To quantify the benefits of early action in response to El Nio forecasts, organizations
must (1) attribute the impacts of a particular event to El Nio, and (2) determine how many
people and assets would have been affected, and in which ways, if early action were not
taken. This chapter is an overview of how each step could be accomplished given current
scientific capabilities.
Attributing Severe Weather Events to El Nio
During an El Nio year, stakeholders media, government officials, fund-raisers,

humanitarians are prone to blame any disastrous weather or climate event on the
phenomenon. However, when they do, these groups cite raw numbers of people
impacted. They do not, for example, deconstruct the data to specify the number of newly
malnourished, diseased, disenfranchised people, or those who suffered a myriad of other
harmful impacts attributable to and resulting from an event. They assume that the extreme
weather event would not have occurred in a neutral climate year. And they often fail to
report that the baseline needs of vulnerable people were only exacerbated by the disaster.
(see examples: IGAD & ICPAC 2016; UNOCHA 2016; WHO 2016).
The science of attributing weather events to climate change and El Nio is not only
new but difficult. As evident in the case studies (see Chapters 26), droughts, floods,
and other extreme weather events can and do occur in non-El Nio years. Therefore, to
definitively attribute impacts to El Nio, organizations must establish a more rigorously
conclusive connection between El Nio and the particular event in question.
A growing area of research is attempting to determine how climate change affects
the probability that a particular extreme weather event would occur (Allen, 2003; Stott et
al., 2013). The studys author interviewed a climate researcher working with World Weather
Attribution,1 a group trying to identify the role of climate change in extreme weather events.
The researcher believes it is possible to adapt the techniques used to attribute extreme
weather events to climate change to associate particular events to El Nio. In fact, one
study demonstrated that La Nia played a role in the drought that plagued East Africa (Lott
et al. 2013).
When attributing events to climate change, scientists ideally use at least two
independent methodologies statistical techniques and climate modelling to simulate
the event under pre-industrial conditions followed by current climate conditions. This
modelling (run hundreds of times) allows scientists to quantifiably asses the likelihood that a
particular event would occur under different climate scenarios.2
1 For more information on this group, see https://wwa.climatecentral.org

2 At least 2, ideally 3 or more methodologies are used for each scenario to cross-check the findings.
64 Chapter 7: Measuring Impacts

Climate modellers could use that method in the future to link particular weather events
to El Nio by simulating an event in neutral (non-El Nio/La Nia) years and comparing to El
Nio years. Adapting the model to this application would provide humanitarians with more
concrete scientific evidence that particular extreme events are linked to El Nio and that the
harmful impacts were less likely to occur in a non-El Nio year. The evidence would reveal the
extent to which El Nio drove both the predictability and magnitude of the events.
This method of attribution, however, is not without its challenges. In the case of
flooding for example, this approach would only attribute the extreme rainfall event. Further,
complex analysis would be needed to attribute the floods that result from the rainfall event,
including an understanding of the local hydrology. And even further analysis would be
needed to attribute the impacts from the flood, which would differ from location to location
due to the inherently local characteristics of vulnerability to flood risks. In both of the latter
cases, sound methodologies are not yet widely accepted. An additional complication is that
research scientists would also need rainfall data from affected areas as well as historical
data sets to conduct these methods. The data sets would need to be divided into neutral
years, El Nio years and La Nia years and as a result their smaller size could limit the
robustness of the findings. Another example pertinent to the humanitarian aid agenda:
linking severe weather events to El Nio may not supersede the benefit of forecast-based
action (regardless of the cause). This requires further investigation and discussion.
Framework for Measuring Impacts: Options and Considerations
The value of forecasts for humanitarian action is predicated on a particular theory of change
that connects forecasts to early action to outcomes. Advocates of forecast production and
dissemination believe that early information will lead to early action, which in turn will result
in better humanitarian outcomes.
At present, natural and social sciences are often incapable of definitively untangling
the myriad factors influencing how early action based on El Nio forecasts may have
altered the course of history, leading to a reduction in lost lives, property, and livelihoods.
Establishing a baseline to measure the benefits of an intervention or a counterfactual to
quantify the benefits of impacts avoided are two such models organizations could use.
The following section explores ways to quantify the value of early action based on
an El Nio forecast. We will refer to activities sponsored by the Kenya Red Cross Society
(KRCS) and British Red Cross (BRC) in Kitui.
Framework for Measuring Benefits
1. Establish a baseline against which to measure final impact

A baseline differs from a counterfactual in that it describes what the situation was
before the intervention. Changes in yields or other outcomes are then measured against
the baseline. Baseline data should, therefore, be collected before the intervention takes
place. Data can be collected through surveys, focus groups, interviews, or using existing
statistics. Unfortunately, in humanitarian situations, the need for prompt response may
further complicate the challenge of identifying indicators and collecting data before an
intervention begins.

2. Implement the Programme
3. Measure Results
To measure impact, return to the site and compare new outcomes to data collected
previously. In some instances, comparison with control groups or statistical analyses may
be necessary to attribute changes to the intervention being measured definitively. Less
rigorous interpretations will simply attribute any change in outcome to the intervention
without trying to account for other factors.
As seen in the Kitui case (below), when actions are taken quickly, it may be difficult
to collect baseline data. When that is the case, the researcher can reconstruct a
counterfactual in one of the following ways:
a. Ask people how their plans changed because of El Nio forecasts.

This method might ask people or organizations what they usually do, what they did
last season, or what they would have done had they not received specific information or
benefited from a specific intervention. Accounts of what would have been done differently
can translate to likely outcomes based on previous years or what occurred in other places.
This method can be done relatively easily in focus groups or, more precisely, using surveys.
Unfortunately, even when asking about recent events, recall data is often inaccurate and
difficult to verify.
b. Ask people to estimate the impact of the intervention.

This option is similar to the previous option but seeks qualitative impressions rather
than a direct comparison to previous years. In the Kitui intervention, people were asked to
estimate their past yields and how they compared to yields during the El Nio intervention.
These estimates are subjective and impossible to substantiate; in some instances,
researchers will have to be satisfied with perceptions of the forecast utility or the impact of
the intervention.
c. Compare communities that received intervention with those who did not.
This method assumes that severe weather events will impact neighbouring
communities similarly their culture, economy, geography, vulnerabilities, dynamics, etc.
While control groups are considered the gold standard for scientific research, this level of
rigour is not feasible for development or humanitarian researchers. Establishing control
communities would also be unethical intentionally depriving some communities of
assistance and relief to quantifying the benefits of an intervention. Therefore, identifying
control communities would need to be through natural occurrence such as limited funding/
resources resulting in not all communities in a particular area receiving an intervention.
Measuring Benefits in Kitui
The Kitui case study quantifies the value-added benefits of early action intervention (see
Chapter 4 for a more thorough description). Farmers in Kitui who received seeds from the
Kenya Red Cross Society (KRCS) and British Red Cross (BRC) reported year-over-year
increases in yield. However, researchers faced challenges trying to develop a rigorous
counterfactual.3
3 Although the reports executive summary concludes that farmers obtained higher than usual yields as a result of the
intervention, it is unclear from the Kitui report how the researchers were trying to establish benefits. The full report
mentions control groups (for which no baseline or harvest data was collected) as well as comparisons to previous
seasons. Rather than critiquing that report, the discussion here is meant to show the challenges with assessing impacts
more broadly.

First, there are numerous ways benefits could be measured. These include (1)
measuring yields in comparison to neighbouring communities; (2) measuring yields against
previous seasons;4 (3) determining whether farmers who received seeds had better
livelihood outcomes than those who did not get seeds; or (4) some combination thereof.
Any of these are acceptable metrics, but all require different baseline data. The first option
requires knowing what the yields were in neighbouring fields. Option two requires access
to historic yield data for previous harvests including those during an El Nio. The third,
and most challenging, requires knowing whether farmers would have procured seeds and
harvested crops had they not received assistance from KRCS. In other words, if these
farmers didnt get seeds from KRCS, would they ever have planted and harvested a crop in
the first place? The other options are not without their potential challenges, such as suitable
and reliable control communities or access to accurate yield data.
It is important to clearly define the metrics by which youll evaluate the programme well
in advance to ensure that baseline or counterfactual data is available when it comes time to
quantify the impacts. The complexities of impact evaluation are compounded by the short
turn-around times in forecast-based projects. KRCS had only a few months to plan and
execute their seed distribution. Time constraints likely explain why the baseline data in this
example compares two different seasons.5
Researchers in Kitui disclosed other challenges in their attempt to quantify the benefits
of seed distribution in response to an El Nio forecast. KRCS gathered impact data through
focus group discussions in which farmers estimated yields in comparison to previous
years. Given the potential for farmers to influence each others responses, focus groups
are not ideal for assessing and aggregating individual household impacts (KRCS 2016).
Surveys may not have been a practical alternative because they are resource intensive and
impractical in a short timeframe.
Additional measurement challenges resulted from the different ways in which farmers
used the seeds. Some farmers had mixed the seeds with seeds they already had. Others
measured yields in rows of maize rather than in bags of maize. Others did not plant the
entire 2kg provided. And some farmers neighbours (those who did not get seeds from
KRCS) also reported increased yields (KRCS 2016), which cast doubt on whether the
benefits were above and beyond what might have occurred without the donated seeds.
All told, both farmers and researchers had difficulty isolating yields from KRCS seeds from
those from other sources.
While most would agree that seed distribution in Kitui helped farmers, precise
quantification of those benefits despite considerable effort remains elusive. The
following section translates some of these challenges into more general terms, outlining key
considerations for assessing El Nio impacts in the future.
4 These first two options appear to be the methods used in the Kitui study.
5 KRCS attempted to capture a baseline, but asked for last seasons harvest, and therefore gathered yield data for the
March-April-May season rather than the October, November, December season.

Quantifying Value: Attributing Event Outcomes to Mitigating Harmful Impacts
This section explores ways to quantify the value of early action to mitigate the deleterious
effects of severe weather events during an El Nio year. We will refer to activities sponsored
by Concern Worldwide in Somalia.
Framework for Measuring Impacts Avoided
The baseline framework for measuring the benefits of an intervention (previous section)
cannot quantify the benefits of mitigating the impacts of a disaster entirely. Though the
techniques of data collection still apply (3a, 3b, and 3c), this framework employs a fourth
technique: the researcher must hypothetically reconstruct what would have happened in
the absence of intervention. In this framework, the first step begins after the intervention is
complete. This is referred to as developing a counterfactual.
Establishing a credible counterfactual is challenging in complex social-ecological
systems. Available data varies greatly from place to place as well as scale (e.g., a local
flood versus widespread drought), so recommending a single, universally applicable
methodology for measuring impacts is not feasible. That said, it is necessary to posit
a few possible methods for piecing together outcomes depending on the details of the
intervention and data available to analysts. Donors, practitioners, and researchers may
use a combination of these methods in the future.
Comparing to Analogue Years
To glean evidence that a severe weather event would have produced more damage had
responders not acted earlier when the forecast was released, organizations can look for
previous events of similar scope and scale. For example, where floods or droughts occur
regularly, look at data following a flood/drought from another year. Choose an event when
no attempts were made to mitigate impact with early action. The analogue approach has
strengths and weaknesses. In this example, the method is great for disaster preparedness,
because the impact has an experiential reference (it happened like this before and will likely
happen like this in the future). Unfortunately, communities are dynamic, and there is no
guarantee that the social, political, economic or environmental factors that contributed to
previous impacts have not changed. At best, anticipating and preparing for the impacts is
only an approximation.
Comparing Against Other El Nio Events
The 19971998 and the 20152016 El Nios were both powerful years according to
scientific measurements, but its interesting to note that the intensity of flooding in Kenya
and Somalia was far less in 2015. While some of the reduced impacts could be attributed
to better preparation, informants agreed that it was also because the rains fell differently in
2015 (they fell nicely as reported above) and were, therefore, less destructive. Comparing
El Nio events of similar magnitude is complicated by the fact that no two El Nio events
manifest in precisely the same way. As one meteorological expert put it no two years will
be exactly a photocopy of one another (3).

Measuring Impacts Avoided in Somalia6
This is a good example of how one organization quantified the benefits of early action to
mitigate the effects of a disaster. Concern Worldwide estimates that it spent $USD63,864
on pre-positioning supplies in Somalia. With that investment, it saved 4,779 households
from being displaced and prevented 12,111 hectares of land from being flooded and
damaged. Concern estimates that its actions obviated the need for $USD1.8 million in
post-disaster humanitarian aid and prevented $USD5.8 million in farmer losses. Because
of the low cost of flood control supplies, Concern estimates that for every $USD1 spent
on prevention, it saves $28.44 in post-disaster humanitarian aid and $USD91.03 in farmer
profits. As one informant put it, flood mitigation is so cost effective, and for that price,
why wouldnt you do it every season? (2).
How did they glean these totals? Concerns partners visited locations where flood
control structures were built to divert flood waters (in some areas people took action, but
flood waters did not reach the diversion structures). They cross-referenced data from past
floods with statistics about the people who lived in the area; they projected the extent of the
damage; and, their analysts calculated the number of houses and fields that would have
been inundated without the flood control barriers. Concern compared the cost of having to
assist people to rebuild ($USD380 per displaced household) and compensate farmers for
lost crops ($USD480 per hectare destroyed). A full cost-benefit analysis is in progress and
will be available in February or March of 2017.
As with most El Nio impact reports, disentangling the effect of El Nio from other
drivers of impact in the region is likely to be difficult. The problem of attributing impacts to
particular El Nio-related events is compounded by the complex humanitarian situation
in Somalia, and the fact that, according to informants for this study, shocks of some kind
occur every year in Somalia, independent of any ENSO influences (2).
Concerns analysis provides a good estimate of the impacts and value of their
programmes, but it may not be easy to replicate. Several factors simplified the analysis
described above:
Flooding in Somalia is recurrent and localised. It occurs at specific
points where human activity has weakened the river banks.
Recurrent flooding also provides researchers with data on past floods
(and their associated impacts) which can be used to construct a
counterfactual.
SWALIM and FSNAU have extensive, detailed knowledge of Somalian
hydrology, river characteristics (including weak points in the banks), and
livelihoods in flooded areas.
Reinforcing riverbanks stops damages entirely.
Experience and understanding of hydrological patterns, coupled with
a familiarity of where riverbanks broke and flooded communities and
where they did not.
In situations where such information is available, Concerns methods provide
a valuable guide to impact and value assessment. However, where damage is more
widespread or comprehensive data on comparable events is unavailable, or it is simply
harder to track results, other methods may be necessary.
6 Because the formal report is not yet out, this write-up is based on the informant interview and subsequent consultation
with those working on the report. As a result, final numbers and methods may deviate slightly from the final product.

It is hard to measure something that does not happen
In the case of most forecast-based interventions, the goals are to prevent damage,
not to provide additional benefit. In the case of flooding in Somalia, it was relatively easy
to guess what might have happened had riverbanks not been reinforced. The benefits
derived from other interventions are, unfortunately, less obvious and more complicated
to substantiate. What would have happened had we not pre-positioned supplies? How
many weeks delay would have resulted? If actions are taken, and no crisis arises, would
there have been a crisis without intervention?
I think its hard to monitor your impact when you dont have a disaster. The fact that
the disaster did not happen in some regions. Even though [other] regions we were able
to measure the impact because the rains were higher than usual so there would have
been more flooding. So in those areas, we were able to measure the impact. In some
other areas where it didnt happen, there is this question afterwards. Was it useful or was
it not useful? Its hard to measure, and people want you to measure things, and in this
case, sometimes its not possible (2).
Quantifying the benefit of early action be it intervening or mitigating impact has its
challenges. Experiences in Kenya and Somalia foretell the traps: memories of past events
can alter peoples perceptions and behaviours they may not act in the same way twice
which skews efforts to compare impacts with past events.
The solution is not an either/or as much as it will be a hybrid of methodologies that
can accommodate a range of estimates and other variables.

Chapter Julie
Photo: 7: Measuring
ArrighiImpacts
2010 71
Chapter 8: Summary of Recommendations
The five case studies presented in Chapters 26 provide an overview of actions taken,
or not, to prepare for potential impacts of the 20152016 El Nio in five countries.
The chapters provide insights into the contributing factors that catalysed or hindered
anticipatory actions, as well as recommendations to improve early action in the future.
This chapter summarises the most relevant data to improve the use of El Nio forecasts in
the future. A more detailed analysis of the recommendations emerging from this study is
available in Chapter 9.
Salient Recommendations
1. El Nio events are a periodic phenomenon and thus allow for research
and planning in advance of specific events. Humanitarian and
development actors need to account for this cyclical occurrence in their
planning. Identifying past El Nio and La Nia years, characterising
their impacts, and determining possible early actions can expedite the
planning and preparedness process before the next event is forecasted.
2. Informants called for better interpretation of forecast information.

Forecast producers and humanitarian practitioners require stronger
connections to intermediaries, ministries, and other sector-specific
experts who can help them to interpret forecast implications.
3. Governments may be reluctant to make official emergency declarations

before realised impacts. Humanitarians should work in partnership
with government on an agreement (or legislative remedy) that facilitates
forecast-based early action. Standard operating procedures or triggers
for humanitarian action are additional options for overcoming political
obstacles. Any procedures require joint cooperation with government to
ensure maximum reach.
4. Lack of funding is a common barrier to early action. Flexible funding

and crisis modifiers facilitate early action, but only a few donors provide
these options. Expanding funding for early action is essential to the
expansion of forecast-based early warning.
5. Few organizations attempt to capture the benefits or losses avoided as

a result of their early actions. Additional emphasis on impact evaluation
can help make a case for forecast-based early actions in the future.
6. Donors and practitioners dont think to record beneficial outcomes

resulting from a forecast-initiated action (as was done with seed
distribution in Kitui, Kenya). More can be done to document successful
interventions and identify potential financing mechanisms in advance of
El Nio events.
72 Chapter 8: Summary of Recommendations

7. A focus on no-regrets actions actions that contribute to
resiliency regardless of a specific climate event can be identified
or conceptualized in advance of forecasts to help overcome the
uncertainty associated with them. In advance of coordinating a no-
regrets action, donors and humanitarians must collaborate to define
acceptable levels of risk or investment based on various forecast
scenarios or probabilities. No-regrets actions may be particularly easy
to identify in areas prone to recurrent crises.
8. Social protection systems have potential to reduce the possible impacts

of an extreme weather event during an El Nio, especially among the
most vulnerable. To maximize this potential, mechanisms for scaling
social protection systems should be planned in advance. NGOs and
governments should collaborate to identify and register beneficiaries,
establish cash delivery systems, agree on forecast thresholds to trigger
additional payouts and ensure that a sustained financing mechanism
with built-in contingencies for rapid scaling is identified.
Chapter 8: Summary of Recommendations 73

Chapter 9: Detailed Write-up of Conclusions and
Recommendations
Drawing on the narratives reconstructed in Chapters 26, this chapter gathers evidence
from across all five countries to distil important themes and lessons regarding the use of
forecasts for anticipatory action. It divides the findings according to the following themes:
1. Forecast Production, Dissemination, and Use
2. Trust
3. Gradual Action
4. Funding
5. Triggers and Standard Operating Procedures
6. No-regrets Action
7. Social Protection Systems
8. Political Considerations
9. Other Influences and Findings
Each section concludes with recommendations based on these findings.
Forecast Production, Dissemination, and Use
Because El Nio influences seasonal weather forecasts, El Nio conditions are inherent in the
predictions. Forecasters and humanitarian informants in each of the five countries described
a similar process of seasonal forecast production, dissemination, and receipt. Forecasting
begins with the international agencies who have global monitoring capacity. As explained
by a forecaster at ICPAC, El Nio forecasts disseminated in Eastern and Southern Africa
rely on measurements from global forecasting centres with the capacity to take necessary
measurements of global sea-surface temperatures; therefore, National Hydro-meteorological
Services (NHMSs) have to rely entirely on other advanced centres to know whether El Nio is
coming or not (3). These forecasts are shared with national actors who develop preliminary
outlooks for their countries, which are distributed at regional Climate Outlook Forums (COFs).
At the COFs, forecasting experts meet to develop regional consensus forecasts, which are
then taken back to National Meteorological (Met) Services and downscaled into national or
sub-national forecasts depending on local forecasting capacity.
The process of confirming an El Nio event and of developing forecasts incorporating
El Nio conditions takes time; forecasts become more detailed, accurate, and hence
actionable, as the prediction times grow shorter. Weekly and daily weather forecasts
provide much more detailed information than seasonal forecasts, which cover large areas.
For example, whereas the daily weather forecast may predict with reasonable accuracy
(meteorological service capacity permitting) the hour and location of rains, seasonal
forecasts cannot provide information on exactly when and where rain will be distributed
throughout a season in a particular district or territory.
Early El Nio forecasting is complicated by differing definitions of El Nio around the
world, but because 20152016 was a strong El Nio, consensus was reached relatively
early, and potentially conflicting information did not appear to influence overall responses
to forecasts. For an El Nio to develop, there must be warming of the eastern equatorial
waters in the Pacific, which lead to a weakening or reversal of the trade winds over
the equator (3). Several different El Nio standards exist. During strong El Nio events,
organizations are likely to interpret the signals similarly, but when the signal is weaker,
there is greater potential for disagreement. NOAA and Australian Met services, for
example, may interpret early El Nio data differently because they monitor sea surface
temperatures in slightly different locations. Because of these discrepancies, in the early
74 Chapter 9: Detailed Write-up of Conclusions and Recommendations

stages of El Nio forecasting, organizations may receive conflicting information. However,
because this information only reaches global headquarters and higher-level actors and
because most organizations do not make significant investments in early action until an
El Nio is more certain, these discrepancies did not appear to influence the informants
confidence in the forecasts.
Stakeholders at the global, national, and sub-national level typically learn of El
Nio at different times. Most headquarter-level staff first learn of the possibility of El Nio
through their connections with international forecasting agencies such as NOAA, FEWS
Net, or the British and Australian Meteorological Offices more than six months in advance
of potential impacts, but this is still too early to have confidence in the forecasts. Given
the progressively more detailed nature of forecast information, the gradual dissemination
processes described above make sense. As the forecasts become more certain,
information is conveyed (usually via email briefings or in-person meetings) to national-
level actors or high-level programme officers within National UN or NGO offices. Often,
humanitarians at the national level first heard of the possibility of El Nio through
intra-organizational communication channels (global headquarters communicating to
the regional and national levels), and later received confirmation, additional details,
and downscaled forecasts from national forecasters. Using increasingly detailed
information, actors at progressively more localised levels of decision-making can begin
acting on the information.
Which Sources Are Most Used?
Overall, the gradual accumulation of information from numerous sources appeared more
important than any one warning or information source. The research team asked informants
about when they first heard about El Nio (and from whom), but rather than describe a
single warning, for most, news of El Nio gradually began coming up [their] consciousness
for so many months beforehand, through alerts within the [NGO] and also hearing
through the media (2).
Because most El Nio information is communicated via email lists or meetings,
dissemination to lower-level or operations personnel within each organization or country
depends upon ties to forecasting organizations, the frequency of intra- or inter-group
meetings, the frequency of email briefings, and who is on the email lists. While these
communication hierarchies make sense, in some instances, they cause delayed action within
organizations. In Kenya for example, those in charge of Kenya Red Cross Society (KRCS)
disaster operations learned of the potential for increased rains only one month before their
onset. They agreed that more direct contact between the Kenya Meteorological Department
(KMD) and all levels of KRCS operations would facilitate planning in the future (2).
There Can Be Too Much Information
Despite the value of repeated warnings from many sources, some stakeholders
complained that there were too many sources of, at times contradictory, forecast
information. Information from different listservs caused confusion or information fatigue,
as humanitarians struggled to keep up with their existing caseload. Over the course of
this study, informants mentioned information coming from at least 19 different sources,
including international forecasters (NOAA, IRI, SARCOF, GHACOF, ICPAC, FEWS Net),
UN agencies (FAO, OCHA, WFP), international NGOs (IFRC), donors (DFID), national Met
services or information providers (UK, KMD, SWALIM, FSNAU, etc.), and even private
forecasters. At the regional and national level, some partners found there [was] so much
and you could be spending your whole day looking at maps you dont quite understand
(2). For people unfamiliar with how to interpret forecasts it is very challenging to use
Chapter 9: Detailed Write-up of Conclusions and Recommendations 75

the different forecasts because sometimes they dont reconcile; they give you different
predictions (4). As a result, practitioners may limit themselves to looking at headlines
because they honestly dont have time to sit down and critically look at everything else (2).
The plethora of information sources means that practitioners dont have one report that is
comprehensive (2). Because many practitioners have limited time to really digest kind-of
long, technical, complicated report[s], the most useful information is short, simple and
sweet, easy to understand (4).
Given the challenge of information overload, attempts to better coordinate forecast
dissemination might be beneficial. One participant suggested a website consolidating
various sources of information and providing a concise interpretation of all the information
would be easier to understand (2). Individual organizations should consider their
dissemination protocols (from headquarters to local offices) and streamline processes
that interpret and distribute information so that clear, actionable intel is provided to key
personnel at predictable intervals. If individual practitioners could receive a regularly
scheduled digest of critical information, they can save time filtering through many
competing sources.
Which Information is Most Useful? The More Localised and Specific the Better
Unsurprisingly, stakeholders agreed that more localised information is the most relevant
and useful for planning and decision-making. Perhaps for this reason national-level
action (beyond contingency planning) generally did not begin until local Met departments
downscaled the forecasts to provide more localised interpretations of the likely events
and impacts. The three most important sources of forecast information in Somalia were
FEWS NET, SWALIM, and FSNAU because they provided precise information on the nature
and location of likely impacts. SWALIM simulated the impact of the last two El Nios and
disseminated information that was directly aimed at helping people programme their
response (2). In Malawi, national-level forecasts were more useful because they outline[d]
which part of the country has what chances of bad rains, good rains, etc.; [humanitarians]
are better able to see where the impact will be and where [they] can focus [their] energies in
terms of what preparedness or response actions [are need] (2).
More localised information also facilitates community action, whether independently
or facilitated by humanitarians. The risks of taking action are more pronounced when
engaging a community directly, because You cannot mobilise farmers in an area to
plough more fields if you are not fairly sure that the rains are gonna come there because,
actually, you could be helping people become poorer (2). Stakeholders in Kenya believed
that if [communities] can be able to interpret the information that is given by the Kenya
Met, with simple terms and simple actions that this means this and this means this
[communities can] do a lot to mitigate [impact] even before the big institutions come in to
help (2). In other instances, more localised information promotes humanitarian engagement
with the communities. In Malawi, NGOs took localised forecasts into the communities
directly so that they were able to see how the weather was transforming as per forecast
[and] so the communities were also more involved in [planning] (2). In countries like Malawi,
where programmes are targeting individual or community level-response to forecasts, there
is a need to tailor a lot of this national forecast to a local scale so that people can not only
understand it but also have access to it and actually know what to do with it (2). National-
level information remains too vague for local action.

Limitations of Scale
As previously identified in a number of studies (Lemos et al. 2002; Patt et al. 2007;
Patt&Gwata 2002; Ingram et al. 2002; Hansen 2002), the desire for the most local,
detailed forecast information possibly stands in contrast to the information scientists are
able to provide. In contrast to localised predictions, international and national seasonal
forecasts are not granular enough to predict when and where, in time and space, rainfall will
occur. SADC forecasts, for example, predict 45 per cent chance of bad rains for Southern
Africa as a whole, without specifying which areas will be hardest hit. In most countries,
El Nio will have different impacts in different regions, and these need to be understood
in order for stakeholders to act. Without more local detail, the regional forecasts are
informative, but if you dont have specific information about the areas within the country, its
not that useful; it just makes the noise that something is happening (2). The most useful
information is, therefore, not from seasonal forecasts, but rainfall estimates, crop estimates,
and other analyses that connect the forecasts to impacts in specific locations.
Forecasters are currently unable to provide mid- to long-range forecasts that specify
when and where rain will or will not occur. Unfortunately, the most common petition
aside from the request for more interpretation is that the department of metrological
services provides area-specific information (2). One Met officer lamented, People always
want [information] at higher spatial or temporal resolution than [the forecasters] can provide.
[Users] really want to be told what the forecast is, when its going to start raining, how
much its going to be, which is not really the way that the longer rains forecasting works
(3). Climate impacts materialise and are felt locally. Within a single community, rains can
be localised such that some farmers do very well and his neighbours do poorly because
he didnt get the same rains as his neighbour (4). Informants for this study complained
regularly that the information was too generic, and the country is vast, so we dont know
where it is applied (2). Even those who understood the scientific limitations of a detailed
forecast could acknowledge that more nuanced information would be preferable. From
the perspective of preparedness experts, in order to be really effective, we really need to
know the amounts of rainfall we are likely to receive and then which specific areas will be
receiving a lot of rains and which areas will likely be affected (1).
Probabilities & Uncertainty
In all countries, the ability to work with experts (both climate scientists and sector experts)
to interpret the forecasts and identify targeted actions was essential to early action.
Most people working within humanitarian organizations are not climate experts. They
do not necessarily understand probabilistic forecasts, nor do they have time to learn the
nuances of probabilities and what they mean. Many informants, especially those closest
to operations expressed that the forecasts were quite scientific and very difficult for
someone who is not in that field to interpret (2). As a result, most organizations need more
support in the interpretation of forecast, and the answer to How does it translate and how
can it assist the organization to programme and give appropriate support? (2).
Seasonal forecasts incorporating El Nio conditions were particularly difficult to
interpret as presented. Many informants found the classification systems typically used in
seasonal forecast communications normal, above normal, and below normal rainfall
to be particularly opaque. One informant asked, What does below normal mean in terms
of agricultural productivity? What percentage of loss will be there as a result of that below
normal rainfall? Some more interpretation, in terms of agricultural productivity, should be
added into the forecast because that is where all this debate and dialogue happens (4).

Confusion Over Terciles and Probabilities Was Widespread Among the Informants
Although there are still limitations to what the science can provide, forecasters are aware of
the need to map out in detail the magnitude of the events, and where, and how, and what
timing the event will hit (2). Even Met officials recognised that forecast recipients never
understand what it means even if we say there will be above normal rainfall or below
normal rainfall so they plan life as usual even if theres a potential danger that is coming in
(2). In Malawi, forecast recipients approached DCCMS to say that our information is not
actionable at all (2) because of the tercile format. As a result of such feedback, theres an
increasing body of work to translate forecasts into action, but in most places, more needs
to be done (2).
There is also a tension between decision-makers desire for certainty and the
probabilistic nature of forecasts. There is always some degree of uncertainty in the
forecasts and the climate scientists are required to really convey the uncertainty in the
forecasts (3). Forecasters are aware that users would like more precise information about
when and where the rain will fall, with sufficient lead time to allow for action, but there is
no science in the world that can tell you, in this period, prepare for 200mms of rain. (3).
These constraints present a problem when dealing with a user community who doesnt
want to deal with probabilities, but instead, want to have a determined forecast (3).
From a practitioners perspective, action that is based on probabilities and large
geographic areas, does not help him or her to decide when and where to act. It only
increases the risk of wasted resources. Organizations often wait for more detailed forecasts
because until the forecasts are localised enough, you cant do very much at all, because
its still a whole country, and you dont know where its going to dump (2). Taking action
too early may mean acting in vain, which leads to practical problems (wasted resources)
and/or political problems (issues with donors).
Beyond calling for investments in science to produce localised proof-positive data at
longer timescales, more needs to be done to help users interpret the information that is
available so that they understand how to incorporate it into their decision-making.
Probabilities
Informants expressed widespread confusion over how to act upon probabilities.

For some, the probabilities are such that they do not warrant any action. I hate to be down on the
forecasting stuff, but for 98 per cent of [possible early action measures] the short-term [forecasts] are too
short-term, and the long-term forecasts are not even close to specific enough What we see is that when
these early warnings come they say, okay there Is 40 per cent chance that youre gonna have heavy rains. In
the normal season, youre gonna have [a] 30 per cent chance. So is that 10 per cent of [the] difference really
worth changing our whole decision-making process? No. We should still do the same thing we always do (2).
If you say that there is an 80 per cent or 90 per cent chance of rain, [people] are going to bring an
umbrella. If you tell them theres a 10 per cent chance, they are not. It is those middle probabilities, and those
are the ones that climate forecasts typically have. You know, about 40, 50, 60 per cent. People probably carry
an umbrella, but I dont know perception-wise whether they think it is going to always rain if there is a 50 per
cent chance of rain, because it also means there is a 50 per cent chance that it is not going to rain (3).
Unless [forecasters] help interpret those maps, then it is just a big blob of blue or its a big blob of green.
And what does a 40 per cent probability actually mean? (2).
There are big questions for organizations. How much and [how] far ahead can you invest? How much
certainty? What does that probability bit look like in order to make sense? It is a difficult one (2).
There are other phenomena that can really interact or interfere with the expected impact [of El Nio] and
thats why I would be very uncomfortable going to a group and saying its going to be wetter than average here
or drier than average. I wouldnt do it for the US, I certainly wouldnt do it for other parts of the world (3).

Interpretation Facilitates Action
In contrast to the confusion caused by vague forecasts, well-interpreted information

contributed positively to early action. In Somalia, FAO took the lead in directing SWALIM
to investigate the impacts of past El Nios and then used their role as a technical
agency to help communicate that to [their] partners in the humanitarian and development
communities (2). Understanding that it is not as easy as we think as techies for people
to understand what the information means, FAO explained what the forecasts predicted
rather than simply distributing email warnings and newsletters. As a result, their partners in
Somalia used forecast information in conjunction with analysis of previous flood events from
FEWS NET, FSNAU, and SWALIM to identify areas that were likely to flood. They then chose
the best locations to reinforce riverbanks, pre-position supplies, and support preparedness.
Support for interpretation does not necessarily need to be developed within each
organization. National Met services do not have the in-house expertise to interpret
the forecasts and translate them into action, so developing this capacity is unrealistic.
Discussion forums for national stakeholders and climate experts to come together,
analyse the forecast, and provide relevant advisories for different players could be more
manageable (2). Contingency planning processes or other regular meetings where
organizations already come together to work on preparedness, mitigation, and response
could serve as appropriate venues for collective interpretation. The contingency planning
process facilitated such discussions in Kenya. During at least one planning meeting, KRCS
and other disaster stakeholders met with intermediaries people with training in both
climate science and humanitarian action. These external experts helped stakeholders
understand what the forecasts meant and translated them into action. Stakeholders in
Kenya found it incredibly helpful having a climate practitioner not just a climate scientist,
but a climate practitioner able to do the translation job that is so often missing in these
reports, because most of us would look at a weather forecast map and actually not be able
to interpret it very well (2).
The nearly universal decision to search for analogue years, to better understand
the impacts of El Nio, reinforces the importance of interpretation. In the absence of
intermediaries, stakeholders sought past events so that they could understand how El
Nio might impact their country and how they should prepare. The interpretation was done
through analogue years, despite the fact that its not gonna be identical, but at least [it]
starts to give you a sense, Is this something that should be on my radar? Is this something
we need to actually invest our resources into to understand what are the potential impacts?
(2). In many situations, it was not the forecasts themselves that were useful, but the
scenario development or interpretation made possible by analysis of past El Nio events.
Efforts to contextualise normal, above normal, and below normal tercile values by
incorporating historical data, may yield more concrete predictions. Although more geared
toward farmers and households than institutional decision-makers. Participatory Integrated
Climate Services for Agriculture (PICSA), out of University of Reading, is one attempt to
use historical data to help users interpret seasonal predictions and make decisions. Similar
efforts, though applicable to organizational decision-makers, needs further exploration.

It is not the forecasts that lead to action, it is the interpretation
The consensus among informants is not that there is a lack of forecasts, but that there is a lack of
interpretation of said forecasts. Better understood outcomes strengthen contingency planning and action
well in advance of the extreme weather event. The quotes in this box illustrate the importance of such
interpretation in the informants words.
Forecasts are not area-specific and not even sector-specific. When it comes to agriculture, they would
just say: Ministry of Agriculture, prepare. We expecting first rains lets say in November. But they wouldnt
be specific to say Central region will receive rains on such and such a date. So the forecast is not sector-
specific, and its not area-specific (1).
We are prone to using the words there is likely to be above normal, normal to above normal. For
farmers or any other users, what is above normal? They dont want to use that (3).
So when I look to FEWS Net, FEWS Net will be specific to say that we think that so many people are likely to
be affected and the other reports will just say there are these chances of bad rains in this area, in that area. Then
when I go to FEWS Net they will say even so many people will be affected from this period to that period (2).
Organizations also need to get the support to interpret the forecast and plan appropriately. The thing
that can also be done by the Met department that issues the forecasts, I want emphasis more on advisories.
They are already doing the advisories but they can do more (2).
As a practitioner, I dont have time to learn every bit of information that comes across. I need people who
know that stuff to be interpreting and saying, It would be a good time to start doing A, B, C. If, in few weeks
time, this situation evolves, it would then be a good time to do X, Y, Z. If this situation continues to evolve,
its likely that these sorts of measures may be needed. Were not saying they are needed now but you may
want to have them on your radar. So, interpretation is key but from people who have a practice perspective:
What does that mean in terms of practical application? What can I do with that information? And thats why I
was impressed with some stuff Im hearing from some of the regional Kenya Met teams. Some of them were
taking that really into local action (2).
In order to respond financially we need detailed information on likely humanitarian impact (4).
Its important, as [my colleague] stressed, that you have people who are looking at this information, that
know how to look at it and interpret it in the right way (3).
Practitioners want language that can easily be understood by somebody who is not an expert in that
field. Or put simple indicators of What does this mean? For example, when you say above normal rainfall,
above normal what? Where? What are some of those thresholds that a layman like myself would sit here and
say, This above normal. The normal is around here. so if anything goes up to this per cent and this per
cent, this is how you measure in simple terms and then action can be done quickly (2).
There is a lot of meteorological data; there are a lot of forecasts; there is a lot of rainfall, weather, what-
what. The question is, How actionable they are made? Is that information actionable? (2).
The other thing I want to say is that, there are people in the early warning system that genuinely believe
that if they provide the information, they have done their job. But its like when youve got child crossing
the road and you say, Cross the road now and the child doesnt move. Youve got to move now because
theres a bus coming down the road and if you dont go now, its possible that [the bus will] arrive [first].
You cant tell the kid, Lets go now and then set off [while] the kid stays. You have to ensure that the kid
is moving. Some people in early warning, they get the feeling that, Yes, we gave them the information.
No, that is not enough. If people have not gotten the message, then you have to find a new way to give the
people the message (2).
How do I translate this 90 per cent chance of a moderate El Nio vis--vis a 50 per cent chance of a
very strong El Nio? You see? How do I translate that into my preparedness? Its two scenarios. On one
side, the high probability is decreased by the strength so I know the impact will not be too high. So
there I can kind of prepare. But the other [forecast], the probability is really like 50 per cent chance. But, if it
happens, the impact will be really high (2).
If you know what to do, then the warning is infinitely useful. If youre gonna run around and distribute
Puritabs, or something like that, then you dont know what to do. Its a fund raising scheme. The utility of the
information itself is relative to your understanding of why its important to people (2).

Forecast Production, Dissemination, Use, and Content Conclusions & Recommendations
1. Organizations assimilate information from many sources. The progression

of information that accumulates from numerous sources appears more
important than any one warning or single source of information (except in
Somalia, where detailed SWALIM information was key).
2. Initial warnings filter down from headquarters (who have direct links
to international forecasters) to national-level actors through intra-
organizational channels; clear dissemination pathways are needed to
ensure that information reaches people in key decision-making positions.
3. International El Nio forecasts are widely seen as credible.
4. Localised information (downscaled forecasts) is (are) more useful for
planning, but the quality of that information is more variable; it depends
upon National Hydro-meteorological Services capacity.
5. Because localised information is most valuable, international
stakeholders should build national Met service capacity to develop and
tailor local information to user needs. Capacity building efforts might
also work to connect National Meteorological and Hydrological Services
(NMHSs) to each other.
6. Multiple information sources can lead to information overload;
organizations should develop internal dissemination protocols that distil,
interpret, and disseminate critical information at predictable intervals.
7. Although forecasts are useful, localised and interpreted information
is more valuable. More work needs to be done help humanitarians
interpret forecasts. This can be done by connecting humanitarians with
expert practitioners who interpret forecast probabilities (data that can
inform planning and action) or with ministries or sector-specific experts.
8. Leveraging local connections and networks (such as providing
information to radio hosts) can help forecasts reach a broader audience.
9. Practitioners universally request more detailed information than is
scientifically possible. While no one anticipates that science will soon
provide more granular information at longer timescales, more can be
done to help practitioners interpret, and therefore use, the information
that is available.
10. Analogue years can serve as a valuable means of interpreting forecasts
and deriving possible scenarios. Because El Nio forecasts allow for
easy identification of analogue years, they may be more useful than
other seasonal forecasts.
11. Analysis of analogue El Nio and La Nia years including identification
of analogues, characterisation of impacts, and determination of possible
early actions can be done before a specific El Nio or La Nia forecast
emerges, expediting subsequent planning processes.
12. Because El Nio and La Nia are periodic events, planning for
their occurrence should be made part of the regular contingency
planning process.

Trust
At the organizational level, there appears to be widespread trust in El Nio information.1

Only a few respondents reported any scepticism regarding the quality of the information,
and this was usually reserved for downscaled forecasts provided by the National
Meteorological Services when they were perceived to have limited capacity.
Trust at the local level, however, appears to be more tenuous and is easily eroded
by what are perceived to be erroneous forecasts. In Kenya, farmers who had planted in
response to predicted rains that had never arrived preferred to wait and see the rains (2)
than to act based upon KRCS warnings. Employees in two Red Cross offices (Zambia and
Malawi) expressed concern that they would relay information about a possible extreme
event and later face charges that aah the Red Cross, theyre the same with the Met. They
are not telling the truth (2). This rapid disillusionment likely results from a misunderstanding
of the probabilistic nature of forecasts. People mistake forecasts for the real thing
and believing that if they say this, then it is going to happen (2). The real or perceived
dissemination of inaccurate forecasts erodes trust in the forecasts, because nothing
happened last time (2). Such deterministic interpretations of forecasts can undermine their
credibility over time, reducing local-level responses to forecasts in the future.
The potential for mistrust suggests that more needs to be done at the community
level to communicate uncertainty and help people to understand what the science can
and cannot provide. Such efforts have been successful in Malawi where, as a result of
ongoing programmes, trust in the forecasts is increasing. Previously, most of the climate
information which was shared could not be believed by many, but because of outreach
efforts, MRCS has seen people believing in this type of forecasts, this type of information,
and they are able to make decisions in terms of what they should do and what they
should not do (2). In the past, since [forecasts were] not localised, people would not take
it seriously, but this time [El Nio 2015] they did (2). These efforts help bridge the gap
between information as it is presented scientifically and what people understand (2) and
preserve trust in the forecasts.
Uptake by communities is further complicated by a lack of trust and a failure to
reconcile forecast with local knowledge. When climate predictions are at odds with
historical patterns and personal experience, people trust their past experience. In Malawi,
for example, forecasts indicating unseasonable rain went unheeded by community
members because they predicted rain during times that would normally be dry. MRCS
warned that by January it was going to flood, but people couldnt believe it, because
flooding in January is unusual (2). Consequently, many people died (2). Citizens there are
accustomed to flooding at specific times of the year, and without a greater understanding
of the possibilities for anomalies, particularly in El Nio years, they are less likely to heed
warnings. This suggests the need for more education and outreach, but also for additional
efforts to integrate climate forecasts with local knowledge.
In areas where the impacts of El Nio are less clear, or trust in forecasts is lacking,
indigenous knowledge and signs around climate are actually really important sometimes
experienced people living in rural areas have an understanding of climate change which is
really important and is more granular than some of the global forecast might (4). Greater
efforts to use forecasts in conjunction with local knowledge at the local level may also
encourage individuals to take forecasts as one of many sources of information rather than a
deterministic expectation of what is to come.
1 The one exception to this was the government and humanitarian organizations in Ethiopia who, according to
humanitarians, were reluctant to take forecasts seriously. Additional follow-up would be required to understand why, from
their perspective, the forecasts were not taken seriously.

These experiences with community-level reactions to weather predictions
demonstrate that providing forecasts at the community level, without explaining their
limitations or providing additional support, has the potential to hinder the use of forecasts
in the long-term. While donors, NGOs, and government officials trust the El Nio forecast,
more needs to be done to help household users of climate information understand the
strengths and limitations of such information.
Did the 2014 False Alarm Affect Early Actions in 2015?
Although most informants had heard of the potential for an El Nio in 2014, few reported
it having a deleterious affect on responses to 2015 forecasts at the organizational level.
The El Nio was never officially declared in downscaled forecasts, and was therefore
not disseminated as part of a downscaled seasonal forecast. As demonstrated in the
case study chapters (Chapters 26), the first course of action in most countries is for
governments and organizations to prepare contingency plans, which are relatively
inexpensive and generally account for several possible scenarios. Therefore, at the early
stages of forecasting, the risks of preliminary action are minimal.
False alarms do, however, appear to impact trust in forecasts at the community level
(2). Several informants relayed stories suggesting that the 2014 forecasts eroded trust at
the community level. In Kenya for instance, belief that the 2014 El Nio forecast had been
inaccurate, led some villagers to dispute [the 2015 El Nio forecast] quoting the earlier
forecast] of the previous year when we were told El Nio is coming, and then it never
came (2). For individuals, the implications of acting in vain are potentially much greater than
for organizations, and as the result of predictions that do not materialise, individuals appear
quick to conclude that because [El Nio] didnt happen after the last two [forecasts], were
not going to rely on this information anymore (2).
2015 Mogodashe floods in

Garrisa County, Kenya -
Credit Kenya Red Cross Society

Trust Conclusions & Recommendations
1. Organizations largely believe that international forecasts are credible,

though trust in National Hydro-meteorological Services vary from country
to country.
2. Individuals are more likely to interpret forecasts deterministically, which
can contribute to frustration and lack of trust in future predictions. More
education and outreach regarding how to interpret forecasts has been
successful in overcoming mistrust in some areas.
3. Efforts to integrate climate knowledge with indigenous knowledge can
enhance both the quality and credibility of the information and increase
local understanding of forecasts.
4. The 2014 false alarm did not prevent organizations from acting in 2015,
but it may have eroded credibility with individuals.
Gradual Action
Another common theme among the respondents is the need to gradually adjust the level
of action as the accuracy and specificity of the available forecast information increases.
Long-term seasonal projections are necessarily more vague than shorter-term weather
predictions and feedback about actual conditions on the ground. The initial forecasts are
useful in prompting organizations to identify the likely impacts, develop contingency plans,
and begin securing funds and mobilising resources. However, organizations agreed on the
need for continual monitoring of conditions on the ground to track how the situation was
unfolding and to know when to initiate actual disaster response. As particular climate events
become more certain and the impacts begin to be felt, additional actions can be taken
more quickly because of the advanced preparation. The ability to monitor conditions on the
ground, to compare emerging conditions to original projections, allows actors to adjust and
make more sound decisions.
Evidence for Graduated Action
The following quotes demonstrate how informants themselves described gradual response based on
forecasts and subsequent monitoring and assessments.
At the beginning of your planning process you dont have those short-rain forecasts. All you have is the
whole season; what you need to do is to be sure your adjustments in your planning scenarios are done
using your short-rains forecast because those are the ones that will give you distributions at a given space
and time. So your research proposal should actually keep on emphasising that in as much as you have your
plan generated after the forecast has been made, you need to adjust your planning scenarios using your
short-range forecast. 7 days etc. (2).
That far ahead[forecasts] are vague enough that actually you cant do very much other than be
conscious about it and continue to monitor. Because its too geographically wide to be able to narrow down
for preparedness other than very, very generic preparedness (2).
One of the big challenges, as I said, with the forecasts, is being able to adjust as the forecasts are
getting closer. If a forecast [says] this is the area, you cant prepare for the whole area thats a whole
continent, so you cant prepare for that, its too big. So you prepare in a very generic way, and you can
only adapt and zoom-in as the forecasts come closer as the forecasts become shorter. So I think thats
inherently challenging to keep adjusting because if you make a decision and pre-position something
here but your event happens over there and then a week or five days before you realize its going to happen
over there, youve got an expense to move it over there, and you may or may not get there in time (2).

The second point that I wanted to highlight too was that its important to continually come back to these
forecasts and reassess what new information in the season might tell us about how the rest of the season is
going to go. [My colleague] mentioned earlier that at FEWS NET at least we do a monthly seasonal forecast
call. This seems to be a good period or good timing for us and how we do our analysis. And doing the
seasonal forecast call every month was particularly helpful this last year with forecasts for Afghanistan and
how the season in Afghanistan was progressing. We had made some assumptions earlier on that given the
likelihood of El Nio and then the eventual declaration of El Nio, that precipitation in Afghanistan was going
to be above average. And then there had been quite a few things that were going on with the global climate
last year that meant that that wasnt the story for the whole season. So the fact that we were revising and
looking at these forecasts assumptions every month helped us get a better grasp of what was going on in
Afghanistan and allowed us to update our analysis and provide the best available information to decision
makers for food security outcomes that was available (3).2
I think making sure that at the end of the day you are keeping your actions relevant to what is
materialising, what was forecasted to materialise in the next seven days is very good. There is a lot of
information. You know Somalia does have a lot of relatively active flood information or rainfall information in
the Ethiopian highlands basically I think it helps to provide a very good picture. And certainly the VAM
and programme during that period were closely monitoring the flood levels being reported [upstream].
And we were also working closely with our partner in our area offices and local authorities to understand
what we are seeing in terms water moving and then the frequent and updated satellite imagery helped to
understand the scope of flooding in rain breakages (2).
I mean the thing is you have these forecasts. You start the response. The forecasts become clearer and
the difference with us is we have a presence there so you can actually start to triangulate because you are
seeing what is happening (2).
2 This global level informant was speaking to the use of forecasts beyond the five countries in this study; his response demonstrates that the
experience is similar elsewhere.
Supplement Forecasts with Real-time Information
An important part of being able to respond gradually and methodically, as described by the
informants, is having access to real-time information about conditions across the country.
Because of the difficulty interpreting forecasts, political considerations, lack of funding, and
other impediments to acting on the forecasts, informants were quick to emphasize the need
for on-the-ground monitoring and assessments. Contingency planning begins based on
forecasts, and when the time comes, you adjust to what the reality is, and that is confirmed
by the vulnerability assessments (2). For many informants, the climate science that is really
useful is not the forecasting, it is the monitoring The monitoring is revolutionary (2).
Disaster response efforts in Malawi are an apt example because stakeholders
emphasized the importance of real-time assessments of the impacts as events unfold.
Such evaluations provide real-time monitoring information in Malawi, which is more
granular, and which is coming from the field and give a more detailed picture of what
the situation is at [the] district level in real-time (4). Until vulnerability assessments provide
absolute numbers, stakeholders are not really able to completely say these are our
requirements (2). It was not until April 2016, when the impacts of El Nio were more
certain, that the President of Malawi declared a State of National Disaster. The declaration
triggered the Food Insecurity Response Plan (FIRP), targeting food security, nutrition,
agriculture, health, education, and water and sanitation (WASH) as critical priorities for
immediate assistance (Government of Malawi 2016b: ii). DFID one of the more proactive
donors released only GBP4.5 million in December 2015 for early action, followed by
another GBP31 million in April when the disaster was declared, and impacts were evident.

Updated information is made possible through close connections to communities
and by conducting vulnerability assessments (both ad hoc and regularly scheduled).
Organizations with close attachments to potentially affected communities attributed their
success in planning appropriate actions to those connections.
It is important to note however that while this information helps to spur response to a
disaster, it may not help to spur early action to prevent a disaster from happening in the first
place which is the true value of an El Nio forecast.
Gradual Action & Monitoring Conclusions & Recommendations

Forecasts prompt actors to identify impacts, revise contingency plans,
secure or shift funding, and mobilise resources, but shorter-term
forecasts and on-the-ground monitoring are essential to moving from
planning and early action to disaster response.
Forecasts are most useful if considered part of a gradual process of
planning, monitoring, and refining actions.
Connections to local communities facilitate monitoring and collection or
real-time information.
Impact assessments are more useful than forecasts in overcoming
political obstacles.
Impact assessments are easier for stakeholders to interpret and
translate into action than forecasts.
Funding
Stakeholders across organizations in all five countries emphasized the importance of

timely access to flexible funds to facilitating early action. When funding is not available,
contingency plans cannot be implemented, negating the potential benefits of forecast-
based planning for early action measures. In addition to significant evidence that the lack
of funding slowed, or prevented, early action in several cases, three themes emerged:
(1) flexible funding and crisis modifiers improve response times; (2) when governments
are reluctant to act, competing disasters in neighbouring areas will eclipse the potential
urgency generated by forecasts; and (3) few organizations attempt to capture the impacts
of their interventions.
Slow Funding Slows Early Action in Response to Forecasts
The availability and timeliness of funding were critical factors to determining early action
efforts. While contingency-planning can take place as part of everyday operations, funds
are needed in order to mobilise any specific actions. Unfortunately, many humanitarian and
development partners dont have money that is flexible to assist in emerging disasters
so money was available to make contingency plans but not to implement actions (2). El
Nio warnings arrived as early as March 2015, but warnings were still not met with early
action and funding at scale (Baudot and Hillier 2016: 6). Organizations consulted for this
study frequently cited slow funding as a barrier to action.
Pressure to wait until there is confirmation of a disaster appears to play a significant
role in delaying funds. Some partners were able to use the forecasts to develop
contingency plans and cost estimates, which allowed them to begin raising funds in
advance of pending disasters (2). However, many donors are still reluctant to provide
money in advance, and responses often remain underfunded until the impacts are more
certain. Some organizations found it difficult to mobilise, to receive resources, based on
contingency plans (2). Others believed the donors sometimes do not release funds until

[a disaster] happens (2). Humanitarian informants regularly expressed a sense that both
donors and governments are waiting to see the real people dying, then they release
the money (3).
Feelings aside, resources are always limited, and funding can be a challenge both
before and after disasters. Organizations around the world experienced significant
funding gaps between humanitarian appeals. In Kenya the contingency plan was only
partially funded (2) and in Ethiopia, only 51 per cent of the funds needed according
to the Humanitarian Requirements Document were available as late as March 2016,
after impacts were being reported (Venton 2016). As late as June 2016, Oxfam estimated
the global funding gap to be $USD2.5 billion (Baudot and Hillier 2016) or approximately
GBP1.73billion.3
Flexible Funding Facilitates Early Action
Projects or donors with flexible funding, on the other hand, were lauded as critical to early
action efforts. Most organizations who were able to take early action did so using flexible
funds, or through scaling up or modifying flexible projects. In Somalia, stakeholders go
through great lengths to make sure that we have flexible funding because it is the
difference between what [they are] able to do and what most other NGOs in Somalia
are able to do (2). WFP in Somalia took advantage of flexible internal-funding policies
to pre-position supplies before it received funds pledged by DFID, overcoming potential
administrative delays (2).4 In Kenya, El Nio arrived at the end of the year. As a result, DFID
reallocated unused funds through the Hunger Safety Net Programme (HNSP) and other
existing channels to scale up cash benefits to those already receiving assistance and to
expand the programme to people who do not usually get assistance, but could use it
for disaster preparedness (2). Flexible funding in Somalia also allowed NGOs to quickly
shift focus from flooding to drought when it became apparent drought in the North was
becoming a problem (2).
In countries where governments were hesitant to declare an emergency based on
a forecast, flexible funding and crisis modifiers allowed preliminary actions to be taken.
In Ethiopia, DFID scaled up funding to existing programmes and partners, shifting funds
from development programmes to emergency response to meet emerging needs (4). On
the basis of forecasts, USAID in Ethiopia shifted its funding to increase the number of
districts receiving assistance through existing programmes from 32 to 76. One NGO in
Ethiopia acted earlier than others because of an early action fund and crisis modifiers
(2). As a result of that success, it is expanding an early action fund to make early action
disbursements easier. Some donors in Ethiopia also re-programmed some funding so that
it could be immediately available once the government declared a need. DFID took similar
actions and was therefore able to quickly provide UNICEF and other partners with as much
Ready to Use Therapeutic Food (RUTF), alleviating significant suffering.
DFID (and to a lesser degree OFDA, SIDA, ECHO) stood out to informants as those
donors with the most advanced funding mechanisms. To my knowledge the only donor
that actually put some money on the table for preparedness was DFID. And all the other
donors specifically took a wait and see position (Somalia). Whereas DFID scored very
highly because of the Risk Facility certain other large donors found themselves wrapped
up in the internal red tape where they werent able to make disbursements before it
happened (4). Informants credited the relative success of DFIDs Internal Risk Facility to
the fact that it channels money through existing programmes and allows those decisions
3 In June 2016, the average monthly exchange rate was GBP 1.4451 to $USD1
4 According to informants, WFP is able to advance internal funds to pre-position supplies, but it must wait for pledged
donor funds to arrive before it can actually respond.

to be made by actors on the ground who know the context and can make the decisions as
events take place (4).
Funding in the Context of Competing Disasters
In a world with numerous humanitarian crises, funding will always be limited. Informants in
Kenya, Zambia, and Ethiopia all believed that ongoing emergencies in South Sudan, Syria,
Somalia and other neighbouring countries stretched humanitarian capacity, and diverted
attention from El Nio-related problems (2). This was, unsurprisingly, particularly true when
governments were reluctant to declare a crisis or accept external help.
Impact Studies
One challenge to advanced funding may be the lack of concrete impact studies. As we
found in this review, few organizations have or take the time to assess their activities and
rigorously demonstrate benefits (qualitatively or quantitatively). Although some informants
believed the evidence for early action was there, donors remain reluctant to act based
on forecasts (2). More attention to demonstrating the benefits of early action may help to
overcome this reluctance.
The Importance of Flexible Funding
Timely and flexible funding were widely cited as essential to fast action. The comments below reinforce the
value and importance of flexible funding mechanisms in the eyes of humanitarians.
The primary challenge I think is funding. As you know, most of the emergency preparatory activities
were not funded, so it becomes difficult (2).
Donors always wait for something worse to happen then they give their funding for support. We
shouldnt wait for a disaster to happen for donors to open up their doors, that is a waste of resources.
Because we could save more by doing more and intensive preparedness rather than waiting for something
to happen and yet we have the forecast (2).
I think where most of the work needs to be done is maybe on the bridge between donors and decision
makers make them realize that they cannot expect perfect information. That doesnt exist. They will have
to act based upon imperfect information; that will fail some of the time (2).
When the forecast is going on, there has to be some form of donor commitments around financing or
government commitment around financing this. Like Ive said, this early action piece because the forecast
is good but without resources, it is not so good. I think DFID are really the leaders in terms of the donors (2).
The biggest challenge that we can say also applied to you guys from the Red Cross is funding. To
have a contingency fund in place which is sufficiently funded to take those early actions. Now we, from our
side, we face the biggest challenge in mobilising the financial means to have a small or a midsize contingency
fund which can be triggered in those countries which will be most affected by climatic events (2).
To be clear there was money that was available that helped that pushed that encouraged
everyone even the sceptics to move forward, because when you get the dollars pushing, that moves
things as well. And I think thats part of the game (2).

Funding Conclusions & Recommendations
Timely, flexible funding is essential to early action.

Crisis modifiers can help organizations to take early action before
emergency declarations are made.
Because resources are limited, declared disasters in neighbouring areas
will draw resources away from forecasted problems if governments are
slow to request assistance.
Little funding is available for organizations seeking to help households
benefit from El Nio conditions.
Few organizations attempt to measure or quantify the impact of their
early action. Additional studies may help convince donors of the
potential value of forecasts.
Triggers and Standard Operating Procedures
Discouraged by slow response through normal funding channels but heartened by the
perceived successes of crisis modifiers and more flexible funding mechanisms, many
stakeholders suggested the need to develop triggers for action and standard operating
procedures (SOPs). Groups, such as the East African Interagency Early Warning Early
Action Working Group, are currently developing a set of indicators and thresholds that
would trigger pre-defined actions based on forecasts or other assessments.
Humanitarian organizations support the development of such triggers because they
determine the actions to be taken in advance (as in scaling up social protection),
expediting response processes.
Despite the enthusiasm for developing SOPs, none of the organizations interviewed
for this study were prepared to use thresholds, triggers, or SOPs during the 2015
2016 El Nio. These mechanisms may hold promise, but more needs to be done to
identify indicators, thresholds, and early actions. Because these mechanisms are new,
organizations would do well to monitor and evaluate new trigger and SOP programmes, in
order to demonstrate their value to potentially sceptical donors.
Risks of (De)Politicising Forecasts
Although triggers and SOPs are appealing as seemingly objective ways to trigger action,
forecasts can be used to generate a crisis or raise and distribute funds, thereby
becoming a political tool rather than an objective trigger for action. Several informants
for this study questioned how humanitarian organizations use climate events to raise
funds. Some believed, that the 20152016 El Nio itself was looked at as a fund-raising
tool by organizations and fund-raising tool for response (2). And both development and
government actors in Ethiopia may have in part disregarded the warnings because they
believed humanitarians want to declare emergencies because they get money (2). In
some instances, funds may expedite developmental or no-regrets activities that will have
obvious benefit but may otherwise lack funding. However, good intentions do not erase
the potential for politicisation of information and an eventual loss of credibility. At least
one of the governments in this study reportedly may have attempted to use forecasts to
funnel El Nio support to its constituents rather than those who would be most affected
by El Nio. It illustrates how a scientific tool can easily be co-opted and used for political
gains. Arguments over what is the right science have emerged in other contentious policy
debates (Sarewitz 2004). If forecasts are used to justify actions that are clearly politically

motivated, or if the benefits are not demonstrated, meteorological departments and the
forecasts they produce could quickly lose their credibility, becoming just another alarm- and
fund-raising tool.
Proponents of forecasts are quick to dismiss the risks of acting in vain, but the
reluctance among many humanitarian donors and governments to release funds (and act)
on the basis of forecasts, is understandable. Advocates of early action believe donors have
to understand that the forecasts you can make about these events are always affected
by a level of uncertainty so that should not become an excuse for not taking action early
enough (2). However, hesitancy to act reflects the inherently political nature of resource
allocation decisions. There may be political risks for both governments and donors both
of whom will always have limited funds to disbursing funds in anticipation of crises that
may never materialise. For this reason, the concept of no-regrets, discussed in the next
section, is particularly appealing.
Triggers and SOPs Conclusions & Recommendations
1. Humanitarians see triggers and standard operating procedures as a

way of streamlining response to forecasts.
2. Although these mechanisms have promise, more needs to be done to
identify indicators, thresholds and associated actions, and to test such
early actions in the field.
3. Actions based on triggers should be monitored and evaluated to
demonstrate desired benefits.
4. Although automated response may seem more scientific, using
forecasts to determine resource allocations may lead to the politicisation
of science and loss of forecast credibility in the long-term.
5. Proponents of forecasts and early action must take into account the
potential political implications of responding to disasters.
No-regrets Actions
The value of no-regrets actions was a common theme across interviews in Kenya and
Somalia. Informants used the term no-regrets to refer to actions that will not result in
wasted resources or the feeling of having acted in vain. The East Africa Interagency Early
Warning Early Action Working Group suggests defining no-regrets as action undertaken
preliminary to a potential crisis (not confirmed at the time of the action) and activated by
early warning signs, characterised by a cost-effective use of the funds that will, in any case,
benefit the beneficiaries (without creating distortion) whether the crisis happens or not
(EWEA Working Group n.d.).5
5 The East African Interagency EWEA Working Group concepts paper contains two definitions with similar meanings.
Definition 1
The Inter Agency Working Group has defined No Regret actions as ones which are implemented early in the crisis,
throughout the crisis and at scale. They are relevant during crisis and seek to support long-term, strategic priorities or
responses which manage the risk, not the crisis. The idea of no-regrets action is based on probabilities and argues that
the benefit of acting timely overweight the cost of acting in vain. While perhaps simple in concept, it has been difficult to
define in practice.
Definition 2 (suggested definition, based on the understanding that many members of the group had, and maybe easier to
define in practicesubject to any adjustment by the group)
A No Regret action is an action undertaken preliminary to a potential crisis (not confirmed at the time of the action) and
activated by early warning signs, characterised by a cost effective use of the funds that will, in any case, benefit to the
beneficiaries (without creating distortion) whether the crisis happens or not.

No-regrets actions include purchasing and pre-positioning supplies or implementing
projects that will increase resilience to future threats. In Somalia, BRCiS partners attempted
to identify actions that would contribute to future disaster response, development, or
well-being even if those actions were not used to respond to the specific disaster event for
which they were intended. Because Somali rivers flood nearly every year, identifying likely
breakage points in river banks and fortifying them (4) was likely to provide benefit at some
point in the future. In some areas, BRCiS partners implemented all the [flood] preparations
for El Nio, and nothing happened. But, six months later, there was a flood which would
have been much bigger if there was no preparation for El Nio six months before (2). So
although the preparations did not have an immediate impact, they were still considered
beneficial. In Kenya, cleaning road drainage systems was likewise considered a no-regrets
action. Even if El Nio floods did not materialise, it would protect the infrastructure from
future heavy rains (2). In essence, no-regrets actions bridge longer-term development or
resilience and humanitarian response by using forecast triggers to justify projects that
benefit communities and manage risk.
Interview data suggest that no-regrets actions may be easier to identify and justify
in countries or contexts where humanitarian assistance is needed on a regular basis.
Ongoing security concerns and frequent flooding in Somalia means that, in most cases,
organizations will either use [supplies] in another region because you can move them
around, or you can use them in the same place in the next season. It is very unlikely that
you will regret it (2). Likewise, in Kenya, there are certain areas where you are going
to have hotspots of floods and droughts. So if you prepare with that context, youre not
really gonna go wrong because El Nio only makes it a little bit worse (2). Because no-
regrets actions were most commonly mentioned in the context of recurrent hazards, their
applicability to less-chronic hazards needs to be explored.
A no-regrets approach requires flexibility and understanding on the part of the donors.
This flexibility is essential so that (1) organizations are not penalised for taking actions that
may not be immediately necessary; and (2) funds or items can be repurposed down the line
in response to different events. Partners in Somalia applauded DFID for championing the
no-regrets approach challenging us and pushing us to react and look for no-regrets
options (2).
The concept of no-regrets actions is clearly useful in some contexts, but it is still highly
subjective, and it may not be universally appealing. Proponents of no-regrets emphasized
that actions should be commensurate with the probability and availability of resources
(2), but political considerations may also be important. In some countries there may be
significant political consequences for taking action. If governments consistently act on
forecasts for which events do not materialise, distributions based on forecasts could
eventually be perceived as favouritism, eroding trust in the science. Therefore, in some
instances, it may be better to maintain the separation preparedness and development
aligned to actions that are triggered by forecasts. It is up to those providing the money and
implementing the programmes to decide what level of investment is appropriate and at
what point they will regret having invested in certain programmes.
Most informants were convinced that the cost of preventive action far outweighed the
cost of disaster response, but only one (Concern Worldwide in Somalia) had attempted to
prove this through impact assessment or cost benefits analysis. Given donors reluctance
to fund forecast-based activities, humanitarians would do well to document their no-regrets
actions more rigorously.

No-regrets Conclusions & Recommendations
1. No-regrets actions refer to interventions that can be initiated based on

forecasts but will provide a benefit regardless of whether an extreme
event occurs.
2. No-regrets actions may be easiest to identify in areas that suffer from
recurrent crises.
3. No-regrets actions require buy-in from donors and flexible funding.
4. Organizations can identify no-regrets actions in advance of forecasts.
5. The concept of no-regrets is subjective. Donor and governments
may have different levels of tolerance to interventions that are not
immediately needed.
6. More work needs to be done to define acceptable levels of risk
or investment in no-regrets actions based upon various forecast
probabilities.
7. There is a need to document the outcomes of no-regret actions more
rigorously.
No-regrets
Many informants spoke of the value of no-regrets programmes:

The approach was that if the flooding had been massive and destroyed all these
preparations for farming, too bad, but at least there was a chance that it could be solved.
This no-regret approach works a little bit on two sides (2).
If we take it from a much more no-regrets framework of thinking so, then quite
a lot of our investments can be done in a way. So actually we can prepare and then,
if the event doesnt happen, its not wasted anyway. Investments in training some of
the communities in areas that are routinely affected by those disasters, is not wasted
anyway (2).
It ended up that they were not used for their initial purpose; however, those assets
were still used and they ended up on other components of the overall humanitarian
operations, so its not a total loss. But again, those commitments had to be made, and
thats why management was brought in very early in this whole process (2).
Social Protection Systems
Evidence from Kenya suggests that pre-established beneficiary and delivery systems can
be used to quickly and successfully provide pre-emergency benefits based on forecasts.
The Kenyan experience also highlights, however, that such pre-disaster disbursements may
be taken in vain, as events may not affect all of those who receive benefits. Therefore,
governments and donors need to be willing to accept a no-regrets approach when it
comes to pre-emergency disbursements.

In countries where safety net systems were not fully set-up for disaster response,
stakeholders were unable to effectively use social protection programmes to respond early
to the 20152016 El Nio (see individual cases for more details). In Zambia DFID and other
NGOs were piloting efforts in several districts to provide emergency cash transfers to those
affected by the 20152016 El Nio and to learn from this experience. In Malawi, multiple
funding streams and parallel databases of beneficiaries prevented stakeholders from
scaling-up.
Perhaps because of these challenges, the 20152016 El Nio prompted dialogue
in Zambia and Malawi as to how to better use safety net systems in future crises. These
experiences taught stakeholders that a social protection programme started in the middle
of an emergency is not as successful as an existing social protection system or mechanism
that is scaled up (2); the mechanisms for service delivery and for scaling-up need to be
determined in advance. More research is also needed to understand whether people who
qualify for normal social protection programmes are more vulnerable to disasters.
Social Protection Systems Conclusions & Recommendations
Beneficiary identification and registration, and cash delivery systems

(bank account information, etc.) need to be set up in advance of crises
to ensure prompt action.
NGOs and governments should work together to determine how to
scale such systems in advance of actual crises.
In order to know who to target during disasters, more research needs
to be conducted into whether households receiving social protection
benefits are particularly vulnerable to disasters and into which
populations would benefit most from additional social support in
times of crisis.
Raimond Duijsens/NLRC

Political Considerations
As the narratives in earlier chapters clearly demonstrate, political considerations can

easily overshadow forecasts. For politicians (and many donors), given the risk of acting in
vain, it may be politically safer to wait until a disaster has occurred than to take proactive
measures. If a forecast event doesnt occur, its very easy for people to start asking, and
challenging, and saying that was a waste of money, the event didnt happen (2). Both
politicians and donors may then be in the position of explaining why they spent money on
interventions that were not needed. In the case of politicians specifically, preventive action
could be perceived as wasteful, or as an excuse to benefit their constituents. Declaring a
disaster early may make governments look weak or unprepared, whereas responding to
disasters may not. It makes [politicians] look good when they distribute blankets when
they distribute mosquito nets. It makes them look better to their voters rather than when
they start building dykes (2). This leaves humanitarians, who would prefer to depoliticise
resource allocation, in a difficult place. In countries with strong governments, unless the
government declares that theres a need, [NGOs] should not show up with your resources
to help (2). Decisions about how to allocate resources are inherently political, and therefore
delay humanitarian response to forecasts.
Political Considerations Conclusions & Recommendations
For many governments the political implications of acting on forecasts

may outweigh the potential for crisis, delaying early action regardless of
forecast lead-times.
Because governments may be reluctant to declare the need for
emergency response, agreements between governments and
humanitarians (or national legislation) must specify that forecast-based
early action or appeals are distinct and separate from official emergency
declarations; this may alleviate some political obstacles to early action.
Other Influences and Findings
Contingency Planning
Every organization interviewed for this study used forecasts to modify or develop
contingency plans in response to early forecasts. These plans outline scenarios and
possible responses to hazard events likely to be associated with El Nio. Contingency
plans allow actors to plan based on available information and, if the situation does not
materialise as forecasted, then there is no response (2). Updating contingency plans are
the most logical response to the earliest, least certain, forecasts because it is a low-cost
activity that takes place every year (or every few years) irrespective of specific forecasts.
Contingency planning is the logical first step in the gradual planning described above.
Before stakeholders can act, they have to develop plans and coordinate who will be
responsible for various actions.
Another lesson regarding contingency plans is that having a contingency plan is only
the first step; both funding and motivation are required to move from plans to early action.
At the GHACOF 41 meeting in late August 2015, the governments of Kenya, Ethiopia, and
Somalia shared their contingency plans and budget requirements. However, only Kenya
and Ethiopia had committed some funds, whereas Somalia anticipated using its plan to
appeal for funding (IGAD & ICPAC 2016). As of GHACOF 42 in February 2016, Kenya
was still approximately $USD35 million (KSh3.52 billion, GBP 24.41 million) short, and

Ethiopia still needed $USD174 million (GBP 121.35) to fully fund those contingency plans
(IGAD & ICPAC 2016). Without funding, contingency planning is a relatively fruitless form
of early action.
Negative Experiences with El Nio Motivate Action
In both Somalia and Kenya, past negative experiences proved a strong motivator for early
action. Because they remember the catastrophic events of the 19971998 El Nio, even the
general population in Kenya understands the potential gravity of the impacts. Government
and humanitarian stakeholders alike were committed to preventing similar levels of damage.
In the end, the precipitation in Kenya was not as intense as it had been in 1997, leading
informants to remark that somehow maybe [responders] were too well prepared (2) or a
little over-hyped (2), though both were still committed to early action.
Humanitarians in Somalia were eager to take early action because of a number of
past experiences. Several informants cited the collective failure to respond to early drought
warnings in 2011 as an impetus for early action. Many in Somalia also remembered the
impacts of the 19971998 and 2006 El Nios, so there was and still is a lot of
pressure from the donor community and from humanitarian actors, in general, to be more
responsive (2). These experiences suggest that trauma from failure to respond to past
forecasts likely encourages organizations to respond more promptly down the line.
Realising Potential Benefits
Of the 57 organizations interviewed for this study, only BRCiS, SRCS in Somalia, and KRCS
in Kenya reported attempting to help people reap benefits on the basis of the forecasts.
Although increased rainfall in El Nio years is more commonly associated with floods, it can
also be a boon to farmers and pastoralists in arid regions. Recognising this potential, these
organizations took measures to procure and distribute seeds in the hopes that farmers
would obtain bumper harvests. The risk of losing money purchasing and distributing
seeds that might still wash away or not receive enough rain was considered worth the
potential benefits to farmers.
Most organizations immediately looked to analogue years to identify and plan for
potential harms, but none mentioned looking to the past to learn of possible benefits.
The documented success of seed distribution in Kitui, Kenya, and scantly documented
successes in Somalia, suggest that increased efforts in identifying possible benefits might
be a valuable response to future El Nio forecasts.
Despite this promise, lack of funding for positive action may hinder efforts to
explore benefits. As the KRCS remarked, donors are not used to funding on the basis of
forecasts in general and funding positive projects, perhaps even less. To make a case for
such funding in the future, there likely needs to be more dialogue between donors and
humanitarian organizations as well as additional scoping of low-cost or no-regrets projects
that might make sense in the future.

Collaboration Facilitates Early Action and Response
Successful coordination between NGOs, UN Agencies, and government officials was

widely cited as essential to El Nio early action and response efforts. Contingency
planning often facilitated such collaboration, reducing overlap and ensuring that different
sectors and clusters were well considered. Kenya most clearly demonstrated government
coordination of national response efforts, but informants believed coordination was
important in other countries as well. In Zambia, the ability of NGO stakeholders to work
together to collect information and move the response forward was cited as essential to
the to El Nio campaign that got underway eventually. The group felt very comfortable
working at that level together sharing efforts and sharing information (2). These
stakeholders were only able to get permission to conduct their assessments because of
their joint influence. Partners felt it would have been much more difficult to influence the
government had they been working independently. In Ethiopia, when the government and
humanitarian community responded to the eventual impacts of the El Nio was through
strong coordination between the Government of Ethiopia, NGOs and donors working in
the country. We needed to see action by everybody. Because there wasnt gonna be
one agency or NGO saving the day, so it requires massive collective effort. So in order to
publish [the HRD], we needed to have WFP on board, UNICEF on board, we needed to
have the NGOs on board, we needed to have the expertise of the government, we needed
to have everybody agree. All this in a weeks time, so that is what we did (2). Organizations
in Somalia worked as a team really well, enabling them to interpret the forecasts together
and have fruitful discussions about the smartest and best ways of responding to this
early information (2). Conversation among different organizations drove out of the box
thinking and innovation regarding what activities could have the most impact in light of the
information available.
Prioritising Action Based on Consecutive Events
Stakeholders in Somalia described similar scenarios and advocated for monitoring

conditions and prioritising areas affected by consecutive shocks. Because of limited
resources, and the fact that the impacts of climate events are not felt uniformly across most
of the country, humanitarian informants suggested monitoring weather and climate patterns
and their impacts locally to identify areas affected in two consecutive seasons. In any given
year, the rains will likely be insufficient in some or many areas of Somalia. Using the Africa
Flood & Drought Monitor, CWW can identify areas in which there are two consecutive bad
seasons. Those suffering from back-to-back crises are likely to be more severely affected
and can be prioritised for humanitarian assistance. The Africa Flood & Drought Monitor was
cited as having the most useful forecasting information available because it allows partners
to track the likelihood of flood or droughts in specific areas of Somalia.
Forecasts Can Narrow Humanitarian Focus Too Much
International attention to forecasts and their likely impacts also have the potential to focus
attention away from other potential drivers of risk or unexpected scenarios. For example,
aid representatives in Somalia looked to past years to understand the potential impacts,
and because trend analysis of past events did not indicate drought, the humanitarian
community focused on potential flooding in the south rather than the drought in the north.
The El Nio forecasts combine with past analysis caused humanitarians in Somalia to be
taken by an event that never happened and, in some ways, [we] almost put on blind[fold]s
as humanitarian community because the word El Nio was there (2). As a result, they
found themselves in a situation where there were no floods, [at least] not at the level

that we expected, but there was a drought in the north (2). It then took a bit of time for
everybody to realize or accept that we have an event that is taking place in a totally different
location that was not part of the plan and to begin raising funds for the drought and
advocating for intervention in the North. Forecasters also emphasized that El Nio impacts
may not manifest as they have in the past. As a result, focusing too closely on past events
may lead to overly narrow planning.
Flood Forecasts Prompt Faster Action than Drought Forecasts
Organizations in Malawi, Kenya, and Somalia described relative success in taking early
action to mitigate floods. In all three countries, disaster managers were able to pre-
position necessary supplies in preparation for potential floods. Flooding affects specific
areas and communities, and the actions to be taken (pre-positioning transport, sandbags,
disseminating warnings, and acquiring other supplies for quick disaster response) are
relatively low-cost and low-regret. Humanitarians were accustomed to floods, and knew
what engineering was required to mitigate the impacts (2). As one informant explained,
if you say theres a probability theres going to be flooding in Uganda, there are specific
spots that there is going to be flooding not the whole part of Uganda. The forecasts may
be giving a bigger picture, but the actual [event] may be localised (2). Flooding is often
localised; by combining the forecasts with more extensive knowledge of hydrology and
livelihood patterns, organizations like SWALIM and FSNAU, can provide relatively simple,
actionable information. As a result, informants were far more confident and pleased with
their responses to flood forecasts than to drought forecasts.
In contrast to the successes organizations had in mobilising for floods, response to
drought forecasts was generally slower and considered far more complicated. Because
the effects of drought are more diffuse and slower to materialise, drought is just a lot
more complicated, and we [the humanitarian community] havent really articulated clearly
what the preparedness actions are (2). In the case of drought, it just wasnt so possible
to trigger early action because the information available was more general and the
connection between the data and the actions to be taken was less straightforward (2).
Perhaps because responses are more complicated, consistent monitoring and on-
the-ground impact assessments were more effective than forecasts in triggering response
to droughts. Although humanitarians were quick to respond to flood forecasts in Somalia,
drought forecasts were either overlooked or ignored. Partners there needed to wait until
FSNAU post-Deyr [assessment] was released to be able to talk about the severity of the
drought and to actually respond (2). The tendency to wait for impact assessments and
overt suffering (whether political, intentional or something else) to trigger disaster response,
prevents humanitarians from taking early action no matter what the forecast.
Why is flooding easier than drought?
One informant summarized the contrast between floods and droughts particularly nicely:
Basically Im more concerned with drought than I am with flooding. Flooding is
pretty localised. Thats important because we know the places that flood, more or less.
It makes it much, much easier to respond flooding because, okay, now we narrowed it
down it 70 villages instead of 1,000 villages. That makes your job a lot easier. We also
have some pretty tangible things that we can do to mitigate flooding preparing banks,
pre-positioning sandbags [the] community is gonna respond with those right away or
while [the] flooding is taking place (2).

Other Considerations Conclusions & Recommendations
El Nio can help organizations focus on specific response strategies,

but humanitarians should also remain vigilant as events unfold.
Countries in which El Nio may have different impacts in different
regions should take care to understand geographical differences.
Failure to act early to past extreme events can be a powerful political
motivator to both NGOs and governments.
El Nio can have positive, as well as negative, impacts.
Few organizations consider early actions that might help people take
advantage of beneficial El Nio conditions.
There is limited funding for interventions that can help people take
advantage of benefits.
Donors and organizations should give greater thought to potential
benefits.
Collaboration among organizations and government agencies was
essential to coordinating early action in all countries.
Mechanisms for coordination should be clearly defined in advance to
prevent duplication of efforts.
Response to flood forecasts is more straightforward than response to
drought forecasts.
Humanitarian stakeholders need to work to identify early responses to
drought forecasts in advance of specific forecasts.

References
Act Alliance. Malawi Severe Drought Response. An alert Iss.d 23 August 2016.
Allen, M. R. Liability for Climate Change in Nature, pp. 421, 892, 1180; 2003
BRCiS. Building Resilient Communities in Somalia: Areas of Intervention. 2015.
Cabot-Venton, C. The Economic Case for Early Humanitarian Response to the Ethiopia
2015/2016 Drought for DFID, 3 November 2016.
Danish Refugee Council. Somalia Resilience Program Information Sheet. N.d.
DCCMS. Seasonal Outlook for 2015-2016. Lilongwe, 2015.
DCCMS. Weekly Forecast for Balaka. Lilongwe, 2015.
Development Initiatives. DFIDs preparedness to El-Nino and wider humanitarian
emergencies in Kenya. Report to DFID, 2016
DFID. Terms of Reference: Understanding the use of forecasts in preparing for and
delivering the humanitarian response to the El Nino, Annex C. 2016
DPPS. More Funds to Tackle El Nio Rains Pour In in Daily Nation, 25 September 2015.
DRMTWG. Ethiopia: Drought Response Situation Report, No. 04, 31 August 2016.
EHCT. Ethiopia: Slow onset natural disaster. September 2015.
EWEA. Early Warning Early Action: Working on 3 Concepts. Internal Document of its
Working Group, provided by interview informant. N.d.
FAO. Special Report: Heavy Rains Attributed to El Nio Cause Extensive Crop Damage in
Parts of Eastern Africa. 1998.
FAO. Update on Somalia El Nio 2015/2016 Status, Successes and Needs. January 2016.
FEWS NET. Famine Early Warning System Network: Our Work. 2016.
Fitzgibbon, C. Shock-responsive social protection in practice: Kenyas experience in
scaling up cash transfers, in Humanitarian Practice Network. 2016.
FSNWG. The Potential Deterioration of Food and Nutrition Security Condition in the Great
Horn of Africa (GHA). An alert Issued August 2015.
GIZ. Social protection for people in extreme poverty in Malawi: Supporting the
establishment of a sustainable and inclusive social protection system in Malawi.
Lilongwe, 2015.
Government of Ethiopia (GoE). Humanitarian Requirements Mid-Year Review 2015. August 2015.
Government of Malawi. National Contingency Plan 2015-2016. Lilongwe 2015.
Government of Malawi. Malawi Drought 2015-2016 Post-disaster Needs Assessment. 2016.
Government of Malawi. 2016-2017 Food Insecurity Response Plan. The Republic of Malawi
Office of the Vice President Department of Disaster Management Affairs, 2016.
Government of the Republic of Zambia. Disaster Management Act No. 13 of 2010. 2010.
Hansen, J. W. Realizing the potential benefits of climate prediction to agriculture: Iss.s,
approaches, challenges in Agricultural Systems, Vol. 74, Iss. 3, pp. 309330,
December 2002.
HSNP. On 29th October 2015 HSNP paid drought and floods emergency payments. N.d.
IASC. Early Warning and Readiness Report. Outlook Period November 2015April 2016.
IGAD & ICPAC. Report on the 2015/16 El Nio Effects and Lessons Learned.
Intergovernmental Authority on Development Climate Prediction and Applications
Centre. N.d.
Ingram, K. T., Kirshen, P. H., and Roncoli, M. C. Opportunities and constraints for farmers
of west Africa to use seasonal precipitation forecasts with Burkina Faso as a case
study in Agricultural Systems, Vol. 74, Iss. 3, pp. 331349, December 2002.
Karanja, F.K. and Mutua, F.M. Reducing the impact of environmental emergencies through
early warning and preparedness - the case of el nio-southern oscillation (ENSO),
May 2000.
KRCS. Forecast Based Action Kitui: a case study. Prepared by the Kenyan and British Red
Cross, 2016.
References 99
KIRA. Kenya Inter-agency Rapid Assessment: Mandera County IDPs Situation Report.
2015.
KMD. El Nio Forecast. 2015
KMD. Short Rains Season October to December 2015. Press release, 2015.
Korecha, D. and Barnston, A. Predictability of June September Rainfall in Ethiopia in
Monthly Weather Review, Vol. 135, Iss. 2, pp 628650, 1 February 2007.
Lemos, M. C., Finan, T. J., Fox, R. W., Nelson, D. R., & Tucker, J. The use of seasonal
climate forecasting in policymaking: lessons from Northeast Brazil in Climatic
Change, Vol. 55, Iss. 4, pp 479507, December 2002.
Lott, F. C., Christidis, N. and Stott, P. A. Can the 2011 East African drought be attributed
to human-induced climate change? in Geophyical. Reearch. Letters, Vol. 40, pp.
11771181, 27 March 2013.
Ministry of Community Development Mother and Child Health. Social Cash Transfer (SCT)
Programme Overview. September 2014.
Ndoka, C. Hunger Safety Net Programme Phase 2 (Presentation), n.d.
OCHA. El Nino in Kenya: Yesterdays tragedies, tomorrows preparedness. December 2015.
OCHA. Zambia Vulnerabilty Assessment Committee Results. 2016.
OCHA. Somalia: Call for Aid - Drought and El Nio. March 2016.
OCHA. El Nio: Overview of Impact and Projected Humanitarian Needs and Response. 29
January 2016.
OCHA. Ethiopia Country Overview. 2016
OCHA. Ethiopia: Government and Humanitarian partners scale up ot meet additional
immediate relief needs of El Nino driven crisis. UN Press release, 13 October 2015.
Onyango, D. Counties in Kenya prepare for safety as El Nino approaches in The Standard
Media, 20 September 2015.
Oxfam. A Preventable Crisis: El Nio and La Nia Events Need Earlier Responses and A
Renewed Focus On Prevention. Baudot, C. and Hillier, D. 2016.
Patt, A. G., and Gwata, C. Effective seasonal climate forecast applications: examining
constraints for subsistence farmers in Zimbabwe in Global Environmental Change,
Vol. 12, Iss. 3, pp. 185195, October 2002.
Patt, A. G., Ogallo, L., and Hellmuth, M. Learning from 10 Years of Climate Outlook Forums
in Africa in Science, Vol. 318, Iss. 5847, pp. 4950, 5 October 2007.
PSNP. Productive Safety Net Programme Fact Sheet. 12 September 2012.
Sarewitz, D. How science makes environmental controversies worse in Environmental
Science & Policy, Vol. 7, Iss. 5, pp. 385-403, October 2004.
Sayagie, G. Naroks rush to improve drainage in Daily Nation, 7 October 2015.
Singh, R., et al. Reality of Resilience: perspectives of the 2015-16 drought in Ethiopia in
BRACED, 14 December 2016.
Stott, P. A., et al. Attribution of weather and climate-related extreme events in Monograph:
Climate Science for Serving Society: Research, Modelling and Prediction Priorities,
publication pending.
SWALIM. SWALIM Update 9th Edition May-August 2015. 31 August 2015.
UNICEF. Kenya Country Office Humanitarian Situation Report July to September 2015. 2015.
USAID. Food Assistance Fact Sheet. N.d.
WHO. El Nio & Health Global Overview. January 2016.
WFP. Kenya El Nino Situation Report #1. 02 December 2015.
WFP. WFP Malawi Country Brief. November 2016.
WFP. ENSO: Humanitarian Implications and Scenarios The El Nio Aftermath and
Perspectives for 2016-2017. 4 August 2016.
World Bank. World Bank Country Overview: Ethiopia. October 2016.
World Bank. World Bank Provides Funds to Help Ethiopia Mitigate the Impact of Drought.
30 June 2016.
100 References
Annex A: Overall Timeline
In addition to the image below, timeline files have been provided to DFID as Aeon and PDF
files. Events in the timeline are color coded by country as follows.
Timeline Color Key:
Red International forecasts or events

Orange Somalia
Yellow Malawi
Green Zambia
Purple Kenya
Blue Ethiopia

102 Annex A: Overall Timeline
Annex B: Terms of Reference
Understanding the use of forecasts in preparing for and delivering the humanitarian
response to the El Nio
Draft v2.0 (22nd April 2016, Nicola Ranger)
This ToR is for research assistance/partnerships in reviewing how climate forecasts were
used in preparing for and responding to the impacts of the 201516 El Nio. It also explores
the role of recent innovations to shock-responsive systems, such as social protection, in
enabling earlier action based upon forecasts. We envisage a close partnership with DFID
advisors during the project. The products will support DFID and counterparts in developing
best practice in the use of forecasts in future disasters.
1. Objectives
The overall objective of this project is to identify and share best practice in using
climate forecasts to deliver timely, efficient and effective humanitarian response
to climate-driven disasters. It will also provide evidence on the role of shock-
responsive systems, such as adaptive social protection, in enabling earlier action
based on forecasts.
This project will review how climate forecasts were used by key humanitarian and
development organisations in preparing for and responding to the impacts of the 201516 El
Nio and assess the outcomes (i.e. actions on the ground) and impacts (e.g. lives saved) that
resulted. It will assess the pathway through which information led to actions and will identify
key actors, approaches, and events that ensured successful uptake and use of forecasts to
inform action. From this, it will draw out learning and recommendations on best practice.
2. Key Research Questions
The project will address the following key interconnected research questions:
Which information was used?

a. What sources of information where important initially in driving action? (e.g.
media, online sources, internal communications, national communications,
international organisations, met services forecasts).
b. What (if any) climate-related information was accessed and used by key
humanitarian and development actors in preparing for and responding to the
impacts of the 2015/16 El Nio (e.g. the types and sources of information)?
c. Which information was most pertinent in driving action and why was this type/
source particularly important?
How did information influence the decisions and actions?

a. What was the course of events that led from information to action; what
were the key decision points? What role did climate information play?
b. What other factors where important in driving the timeline of preparation and
response (e.g. logistical considerations, approvals of funding, requests for
support from government, agreements by coordinating bodies)?
104 Annex B: Terms of Reference

c. How did uncertainty in forecasts influence action? e.g. was uncertainty a
barrier to action? did the actors wait for a level of certainty to be reached
before acting, or did activity ramp up gradually as certainty improved?
d. Where any other barriers important in limiting the uptake and use of
information and delivery early action based upon this?
Role of communication, trust and dissemination?

a. How well did the humanitarian actors understand El Nio prior to the event?
Was knowledge improved? Did this create constraints?
b. How well did humanitarian actors understand the forecasts that were
available? Did they find any source of information particularly helpful or
particularly unhelpful and why? What challenges were encountered in
interpreting the forecasts and using them to guide action?
c. What is the perception of the quality of communication of the forecasts to
the humanitarian community (e.g. the format of forecasts and dissemination
channels)? How could this be improved?
d. What role did national or regional information (e.g. RCOFs) play vs. globally-
derived forecasts (e.g. from Met Office, ECMWF and NOAA?)
e. Which information was most trusted and why?
Have innovations in disaster preparedness and response since the 2011/12 droughts led to
earlier action and how has this mitigated the impact of the El Nio? It has been suggested
that countries/regions with existing shock-responsive systems, were able to respond more
quickly based on forecasts. What is the evidence for this and what can we learn for future
systems? E.g.
a. What advances and innovations have we observed in disaster preparedness
and response at an international, national or sub-national level since the 2011/12
drought and what role did these play in facilitating earlier action in 2015/16?
b. What role did shock-responsive systems, e.g. social protection systems,
play in driving earlier action in response to the El Nio? Is there evidence that
countries/regions with shock-responsive systems were more responsive than
those without them? What types of shock-responsive systems proved most
effective? Did they prove more effective than more traditional humanitarian
response? Are they any counter examples?
c. What was the process through which systems, social protection or otherwise,
made use of climate information and forecasts, what were the steps between
information and action, and what was the impact?
d. Did any barriers limit the effectiveness of the information in driving the
response of shock-responsive systems?
e. How did these systems respond to the false alarm El Nio in 2014? Did
this impact the response in 2015/16?
Is there evidence that information led to earlier action and avoided losses?
a. What is the objective, quantitative and qualitative, evidence that information
drove earlier action and avoided impacts? What factors were important here?
b. What was the (approx.) scale of the avoided impacts?
What is the perception of key actors on the value of information to the response?
a. Is there a perception from the humanitarian and development actors
that climate-information was important in driving earlier action and
avoided losses?

3. Methodology
The project will focus on four countries: Ethiopia, Kenya, Somalia and Zambia (further
countries could be added). For each country, key humanitarian and development actors to
be included will be identified. This includes, but is not limited to, DFID country offices, DFID
HQ and other key donors (e.g. USAID), UNOCHA, various Social Protection programmes,
World Food Programme and Red Cross. Key actors in these organisations will be
interviewed our preference is for structured interviews rather than surveys. In some cases,
DFID advisors will join these interviews, particularly those with DFID country office staff.
Results should be based on objective data collected through field work, though subjective
information, such as perceptions of key actors, is also valuable.
The research team will provide a summary of recommended humanitarian and

development actors and other stakeholders to include in the study and these will
be agreed with DFID prior to proceeding with the analysis.
The research team will propose a method and analytical framework to address
the research questions. This will be agreed with DFID. We envisage that this will
include the following:
Phase 1:
Analysis of Global/regional/national forecasts

Assembly of a timeline with data on the main global/regional/national
forecasts of the El Nio (and its impacts) that were (a) available to and
(b) used by humanitarian and development actors, with associated
data including regarding uncertainties and communications channels.
Data sources will include (not limited to), for example, IRI, NOAA and
Met office forecasts, FEWSNET and RCOF advisories and national met
service information.
In addition, assemble data on other sources of information identified as
important, e.g. agencies communications, media reporting.
Identifying a timeline of decisions and actions

Assembly of a timeline of data on decisions and actions taken by
the key humanitarian and development actors and other relevant
stakeholders. This should include data on allocation of resources, e.g.
dates and amount of funding committed, as well as key meetings,
actions, statements etc.
Identification of when key information, including forecasts and climate
data, were received and collection of information on how this influenced
the decisions and actions, including qualitative information from
interviews, as well as objective evidence demonstrating how information
was used (e.g. briefings, reports, etc.).
Drawing preliminary conclusions

Address the research questions using the data gathered above
and information obtained through interviews with key humanitarian
and development actors and other stakeholders.
Be transparent about the limitations of the analysis, e.g. the
assumptions made and the balance between factual data
and perceptions.

Phase 2:
Estimating impacts avoided

A full formal impact evaluation is beyond the scope of this study.
New data collection on the impacts of the El Nio and impact avoided
through the use of forecasts is not expected. However, we expect
the team to make use of existing data sources to provide estimates,
and where possible, to leverage other research and evaluation
initiatives to provide more robust quantification. This might include, for
example, ongoing humanitarian evaluations in the region and research/
evaluations on shock-responsive social protection.
Gather and analyse available information on the disaggregated impacts
of the El Nio in the country.
Develop a framework for analysing (qualitatively and quantitatively) the
link between humanitarian/development actions and losses avoided,
using well-reasoned and transparent assumptions. We expect this to
make use of a Theory of Change to support the framework.
Provide estimates of the losses avoided due to humanitarian actions
and any developmental shock-responsive actions.
Drawing final conclusions

Finalise conclusions, drawing in additional analysis where necessary
Draw out best practice and test this with DFID advisors and others.
Draw initial recommendations on how the process of information into
action could be improved
4. The team
We envisage that the research team will include the following characteristics and expertise:
Excellent awareness of the humanitarian landscape in the focus
countries and existing links with key humanitarian actors to facilitate
the interviews.
Excellent awareness of shock-responsive systems.
Recognised ability to provide an objective view in this area
Technical expertise in seasonal and sub-seasonal weather forecasting
Experience/expertise in humanitarian preparedness and response
Track record of conducting similar research

5. Deliverables
Initial kick-off meeting with DFID London.................................................... Start June 2016
Scoping report: methods and framework.................................................... End June 2016
Meeting to agree methods and framework and
agree humanitarian/development actors to interview................................... End June 2016
Interim report on Phase 1 findings................................................................. End July 2016
Interim report on Phase 2 findings, including
initial conclusions on best practice and recommendations....................... End August 2016
Workshop on best practice....................................................................... September 2016
Final report.................................................................................... End of September 2016
Communications on findings.............................................................. End of October 2016
Contacts:
The main contacts will be Nicola Ranger and Yves Horent of the Department for
International Development (DFID)

Annex C: Interview Instruments
This is the tool used for humanitarian actors who represent the majority of informants
interviewed for this study. The tools used for Forecasters and Social protection express are
available upon request to the Red Cross Red Crescent Climate Centre.
Humanitarian Organization Interview Guide:
Understanding the use of forecasts in preparing for and

delivering the humanitarian response to the El Nio
Interviewee Name and Title: [Type here]
Organization: [Type here]
Date, Time, & Location: [Type here]
Interviewer Name:
Instructions to interviewers:
Keep in mind our general objective:
To obtain documentation of the forecasting products that were made available

at national, regional and global levels, communication channels for sharing these
products, the decisions and actions that were, or were not, made based on these
products or other sources (ToR).
The following is meant as semi-structured interview guide (rather than a structured more
survey-like tool). The purpose is therefore to guide discussion so that each of the questions
is answered, but not necessarily in the order presented here. You may not have to ask
every question in order to get an answer. It is best to start the discussion with general
questions, and follow-up with more specific questions as necessary to get sufficient detail.
The numbered questions are more general and the lettered sub-questions may be asked if
additional detail is needed in order to answer all the questions. Prompts are also provided
should you need to clarify or provide examples. In general, it is best only to use prompts
if it becomes clear that additional clarification is needed (the respondent stalls, answers a
different question, or asks for clarification).
The language below was selected so as not to lead the respondents to particular answers.
In general, it is best to ask if something happened (for example if the respondent received
forecasts, or if they were able to respond to forecast and follow up based on their response
rather than launching immediately into: how did you respond to X forecast? The latter
assumes there was a response and may lead the informant to attempt to provide the right
response rather than recounting what actually happened.

Research Permission:
Do you have any questions regarding the purpose of this research?
Do I have your permission to record this interview?

Yes
No
Semi-structured Interview Questions
Interview Questions
A. Background
1. To begin, please tell me more about your role in your organization (and
in receiving and responding to forecasts).
2. In your understanding, briefly, what is El Nio? How/when did you first

learn about EL Nio as a phenomenon?
3. What are the anticipated impacts of El Nio in your country?
B. Forecasts & Action
4. Did you or anyone in your organization receive information, forecasts or

warnings of the 2015-16 Nio?
If not, skip to Section D below.
5. Would you be willing/able to share the forecasts with us (especially if

via email)? If yes, I will get these documents from you at the end of the
interview or via email.
6. When was the earliest you remember hearing of El Nio? From what
source?
7. Did your organization undertake any new programs or activities in

response to the 2015/16 El Nio forecasts?
If not, skip to Section D below.
8. If yes, please describe your organizations process of planning for and

preparing for potential El Nio impacts
a. What sources did you get information from? (Prompt: Media,
forecast from Met Service, international forecasts, other
organizations etc.)
b. What was the course of events that led from information to action?
c. What were the key decision points?
d. What role did climate information play?
e. Were there multiple sources of forecast information?
110 Annex C: Interview Instruments

f. How did this information reach you (email, telephone, word of
mouth, you requested it etc.).? (If multiple methods, which one
was most preferred/effective?
g. Were the forecasts accompanied by any advice, suggested
actions, sector-specific warnings?
i. If yes, what were they? Were they useful?
h. Did the warning include information on potential impacts of
El Nio in your area?
i. If yes, what are the anticipated impacts of El Nio in
country X?
C. Usefulness of Information (if not covered in the discussion above)
9. Which information did you use in preparing for and responding to the
impacts of the 2015/16 El Nio (Prompt: media, online sources, internal
communications, national communications, international organisations,
met services forecasts)?
a. Which information was most useful? Why?
b. Which information prompted action?
c. Which information was least useful? Why?
d. What other information would you have liked to have had?
e. How might the information have been improved?
f. Did uncertainty in forecasts influence action/in-action? (Prompt:
was uncertainty a barrier to action? did the actors wait for a level
of certainty to be reached before acting, or did activity ramp up
gradually as certainty improved?)
D. Successes & Challenges (if not covered in the discussion above)
10. In your opinion, what were the primary successes, achievements

or benefits of your early response to El Nio forecasts? (Prompt:
Institutional Readiness to Respond? Reducing losses or damages?
scale of impacts avoided?)
11. What were the primary challenges of responding to the El Nio

forecasts?
a. What other factors influenced your organizations ability to
prepare and respond to the El Nio forecasts?
b. Were there any barriers to the use of the information for early
action? (Prompts: logistical considerations, approvals of
funding, requests for support from government, agreements by
coordinating bodies)
c. Did you encounter any problems in trying to interpret the forecast
and associated warnings? (Prompt: Were you able to understand
the format of the El Nio forecast without outside help/expertise?)
d. In your opinion, how accurate or credible are the El Nio
forecasts produced for your region/country?
12. Based on your experience, what suggestions would you make to

improve El Nio forecasts or their dissemination in the future?
a. What other information would you have liked to have had?
b. How might the information you received have been improved?

E. Previous El Nio Events, Social Protection & Lessons
13. Were you involved in any El Nio preparedness in 2011/2012 or during

earlier El Nios? If yes, was the 20152016 response different in
anyway? How?
14. Did you receive any information regarding an El Nio in 2014? If yes, did
this impact the response in 201516?
F. Closing
15. Anything else related to preparedness and the use of El Nio forecasts
we did not cover that you think is important for us to understand?
16. Anyone else in your organization or other organizations working in your
country that you recommend I speak to?
17. Do you know of any studies of El Nio impacts, or impacts avoided,
from your country?
G. Get Documents if possible
Copies of any emails etc. of forecasts or warnings received

Any impact studies that might be available
112 Annex C: Interview Instruments

El Nino Report

Uploaded by

Copyright:

Available Formats

El Nino Report

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

El Nino Report

Uploaded by

Copyright:

Available Formats

Preparing for the

A Report for DFID By

Arielle Tozier de la Poterie

Photo: Charlotte Floors/NLRC

Red Cross Red Crescent Climate Centre

Final Report | October 2017

Contact the Climate Centre at: info@climatecentre.org

A Report for DFID By

Arielle Tozier de la Poterie, Julie Arrighi and Eddie Jjemba

With Support From

Stephen McDowell, Shaban Mawanda and Roop Singh

Red Cross Red Crescent Climate Centre

Contact the Climate Centre at: info@climatecentre.org

Chapter 2: Zambia Case Study 21

Chapter 3: Somalia Case Study 28

Chapter 5: Ethiopia Case Study 48

Chapter 6: Malawi Case Study 54

Chapter 7: Measuring Impacts 64

Chapter 9: Detailed Write-up of

Summary of Case Study Findings

Chapter 2: Early Action in Zambia

International predictions of a possible El Nio event reached stakeholders in Zambia

Chapter 3: Early Action in Somalia

In Somalia, SWALIM and FSNAU provided humanitarian organizations with precise,

Chapter 4: Early Action in Kenya

Chapter 6: Early Action in Malawi

The Benefits of Intervention, Impacts Avoided, and a

Purpose & Objectives

Case Study Selection & Informant Sampling

Five countries were chosen based on the following criteria:

This study uses data from semi-structured, key-informant interviews to reconstruct

2 Interview informants are quoted according to four categories:

Country / Informant Organization Number of Informants

Because of the studys emphasis on understanding peoples perceptions of the value of

In Zambia, disaster management is supervised by the Disaster Management and

Zambian Forecast Production

Chapter 2: Zambia Case Study 21

El Nio Impacts in Zambia

22 Chapter 2: Zambia Case Study

Response to the 20152016 El Nio Forecasts in Zambia

By most accounts, El Nio forecasts reached stakeholders between March and

Chapter 2: Zambia Case Study 23

24 Chapter 2: Zambia Case Study

Linking Social Protection to Climate Shocks

Chapter 2: Zambia Case Study 25

26 Chapter 2: Zambia Case Study

Chapter 2: Zambia Case Study

Humanitarian response to climate shocks in Somalia is complicated by 25 years of civil

Impacts of El Nio in Somalia

Chapter 3: Somalia Case Study 29

Response to 20152016 El Nio Forecasts in Somalia

Response to El Nio forecasts in Somalia followed a similar pattern across organizations.

30 Chapter 3: Somalia Case Study

Chapter 3: Somalia Case Study 31

ICRC and SRCS

32 Chapter 3: Somalia Case Study

Chapter 3: Somalia Case Study 33

CARE International Somalia

34 Chapter 3: Somalia Case Study