Physics and Chemistry of the Earth 66 (2013) 89–100

Contents lists available at ScienceDirect

Physics and Chemistry of the Earth

journal homepage: www.elsevier.com/locate/pce

Key physical characteristics used to assess water harvesting suitability

L.M. Bulcock a,⇑, G.P.W. Jewitt a,b
a
Centre for Water Resources Research, School of Agricultural, Earth and Environmental Sciences, University of KwaZulu-Natal, Private Bag X01, Scottsville, Pietermaritzburg
3209, South Africa
b
Umgeni Water Chair of Water Resources Management, South Africa

a r t i c l e i n f o a b s t r a c t

Article history: Water harvesting (WH) techniques, which aim to increase water availability to crops, have long been
Available online 25 September 2013 used in arid and semi-arid areas to decrease the risk of reduced yields and crop failures due to dry spells.
The landscape conditions dictate the type of WH system that can be implemented as well as the quantity
Keywords: and quality of water that will be collected. The measurement and understanding of how these landscape
Water harvesting characteristics influence the hydrological function of WH systems is important and essential for further
Site characteristics studies which seek to understand and enhance efficiency, extend uptake and model the impacts of WH
Upscaling
within a catchment. However, commonly used guidelines often only prescribe optimal conditions for WH
Statistical analysis
Site guidelines
which results in many sites which may be suitable being over looked. Various statistical analyses was
performed on 28 WH sites gathered from the available literature to try and identify whether the land-
scape conditions under which WH is currently taking place differs to the recommended guidelines.
The results show that WH is taking place under a much broader range of conditions than those recom-
mended by the guidelines. The recommendations for minimum and maximum slope in particular are
too restrictive, with examples of successful WH taking place on slopes much steeper than the stipulated
guidelines. A new set of guidelines are suggested, which take into account not only optimal conditions
but also a range of suitable conditions on either side of the optimal range.
Ó 2013 Elsevier Ltd. All rights reserved.

1. Introduction Broadly, WH can be defined as the concentration, collection and

storage (in different structures or in the soil) of rainwater or runoff
1.1. The context of water harvesting for use either on-site or at a different location, immediately or at a later
time (Siegert, 1994). WH aims to improve the efficient use of rainfall by
Approximately 70% of the world’s poor live in rural areas where capturing it on the site where it falls or capturing the runoff it generates
they have little option but to rely on rainfed agriculture to sustain and storing it for later use to supplement plant water requirements
their livelihoods (CAWMA, 2007). The spatial and temporal distri- (Rockström, 2000; Agarwal, 2001; Ziadat et al., 2006). WH is limited
bution of rainfall in semi-arid regions varies greatly. When rainfall to in-field or small scales ex situ catchments and water is generally
is characterised by few, high intensity events, water may not be not available throughout the dry season. So differs from conventional
available for crops during critical growth periods (Rockström, irrigation in that water is not normally available throughout the dry
2000). Dry spells during critical crop growth periods and drought season from WH Structures (Mbilinyi et al., 2005). In the case of crop
further exacerbate the vulnerability of poor people relying on rain- production, WH aims to decrease the amount of rainfall ‘‘lost’’ through
fed agriculture (CAWMA, 2007). Dry spells and mid-season unproductive evaporation, namely soil evaporation, litter and canopy
droughts, rather than an overall reduced mean annual precipita- interception and through runoff and to increase the amount of water
tion (MAP) can, and often do, cause crop failures (Rockström, available to the plant for productive crop transpiration and as a result
2000). As a result, the adoption and development of water harvest- increase crop growth and production.
ing (WH) strategies to ensure the efficient use of water and sustain Whilst the Sharm el Sheik commitment’s highlight the intention
human livelihoods, both for domestic and agricultural purposes, to increase the portion of domestic water provided by WH systems
has evolved over centuries (Rockström, 2000; Vohland and Barry, in Africa to 15%, there are also continued efforts to expand the use
2009; Bossio et al., 2011). of these small scale systems, particularly in areas not considered
suitable for conventional irrigation development, but also an alter-
native thereto (AMCOW, 2008, 2012). Modern WH systems have
⇑ Corresponding author. Tel.: +27 33 260 5174. often evolved from traditional or indigenous systems and coupled
E-mail addresses: bulcockl@ukzn.ac.za (L.M. Bulcock), jewittg@ukzn.ac.za with improved agronomic practices have been shown to have
(G.P.W. Jewitt).

1474-7065/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.pce.2013.09.005
90 L.M. Bulcock, G.P.W. Jewitt / Physics and Chemistry of the Earth 66 (2013) 89–100

much potential in enhancing crop production (Falkenmark et al., generation area. The water may then be transported to the cultiva-
2001; Mbilinyi et al., 2005). There remains much to be learned tion area via channels or ducts (Gowing et al., 1999). Examples of
from the indigenous systems, however, such improvements often techniques used include terraces and dead level terraces (Fig. 1d,
include the use of more modern materials, different seed types e and g), runoff storage tanks (Fig. 1f) (Rockström, 2000; Agarwal,
and the application of fertiliser and herbicides. Such investment re- 2001; Desai and Ghose, 2001; Gandhi and Kirtane, 2001; Oweis
quires a sound site suitability assessment but this is an aspect et al., 2004; Nissen-Petersen, 2006). Other examples of macro-
where knowledge needs to be improved. Recommendations catchment WH include sand and sub-surface dams, small earthen
regarding, which WH system is best suited to a potential site, dams with in-flow channels which are often built in existing
how it could be implemented or where it should be sited requires depressions (Fig. 1c) and flood irrigation, stone lines (Fig. 1h).
through analysis of existing successful systems.
Water harvesting techniques can be divided into two main cat- 1.2. The key physical characteristics of water harvesting sites
egories, namely in situ and ex situ. In situ WH involves capturing
runoff generated in the field or cultivation sites where the crops 1.2.1. Criteria used to determine the suitability of water harvesting
are grown. Examples include pitting (Fig. 1a) and semi-circle or sites
contour bunds (Fig. 1b). Ex situ WH collects runoff from a larger Extensive literature exists on the social and economic value of
area and stores it in a storage area that is not adjacent to the runoff WH sites and how they benefit the communities that use them

Fig. 1. (Clockwise from top left) (a) Zai Pitting in Burkina Faso and Niger, (b) contour bunds/demi lunes in Kenya, Mali, Niger Burkina Faso, (c) micro runoff dames in Kenya,
(d and e) contours with borders of stone or grass in Kenya and Zambia, (f) surface runoff collection into tanks in South Africa, (g) dead level contours in Zimbabwe and
(h) stone lines in Burkina Faso and Niger.
 L.M. Bulcock, G.P.W. Jewitt / Physics and Chemistry of the Earth 66 (2013) 89–100 91

(Warren et al., 2003; Mbilinyi et al., 2005; Kessler, 2006, 2007; de teristics on a scale of 1–4 (with 4 being the most suitable) to develop
Graaff et al., 2008). Research has also been done to assess the suit- a multi-criteria decision support system. MAP was ranked with
ability of selected research catchments for the siting of WH struc- <100 mm/a being least suitable and >500 mm/a most suitable, slope
tures. Most of these studies (which are discussed in Section 1.2.1 <3% most suitable while slopes >10% least suitable, soils with clay
below) use one of three sets of criteria (or derivative thereof) content <10% as least suitable and clay content >35% most suitable.
developed by Critchley et al. (1991), Oweis et al. (1998) or the Inte- This allows marginal areas which may have only 2 of the 3 criteria
grated Mission for Sustainable Development (IMSD) for devised to regarded as suitable to be still be considered. For example, an area
determining the physical suitable of sites for WH technologies in with clay soils >35% and a slope of <3% but MAP of <100 mm/a can
India (in Singh et al., 2009). These guidelines are described in still be ranked as suitable, even though not all the selection criteria
Table 1. The guidelines of Oweis et al. (1998) are the most encom- have been meet. Al-Adamat et al. (2012) also altered the criteria so
passing through consideration of the often difficult terrain WH that MAP >50 mm/a, slope <5% and soil texture classified as silty
system can exist in and can operate under. They determine criteria loam, loam and silty clay loam were all considered suitable for
for different types of WH structures and set ideal and suitable lim- WH. However, the rankings in a multi-criteria decision support sys-
its for factors such as soil texture, soil depth, slope and vegetation tem such as that utilised by Al-Adamat et al. (2010), can be some-
and stoniness. Oweis et al. (1998) specify requirements specific to what subjective. Al-Adamat et al. (2010) acknowledge that in
different types of agriculture, i.e. the requirements for trees are dif- reality soil texture can be the deciding factor in determining site
ferent to field crops and to rangelands grazing. For example, trees suitability; however they only rank soil texture as the third most
require deeper soils (>100 cm) with a heavy texture, field crops important factor (3 out of 6). Rainfall has the highest ranking, even
require a medium depth soils (50–100 cm) with a medium texture though they describe it is the least determinate factor. Furthermore,
while rangeland grazing can have shallow soils (<50 cm) with a how the ranking of the criteria was determined is not well explained
medium texture. The ranges for slope are also more encompassing and can have a great influence on the results.
with contour benches being suitable to steep slopes (>12%) while The SCS-Curve Number (SCS-CN) approach has been used by
contour ridges and small basins are suitable for gentle (<4%) to many authors as an alternative approach to determining catchment
medium slopes (4–12%). These criteria are used by Ziadat et al. suitability for WH. Land use, slope, soil class and anticedal soil mois-
(2012) in assessing WH investment potential in Jordan using data ture conditions are used to assess how much runoff can be generated
collected through field survey and analysed in a GIS system. from a runoff area. This is often overlaid with drainage patterns
within the catchment and the areas with the highest runoff genera-
1.2.2. Methods used to determine the suitability of water harvesting tion capacity and closest to existing drainage lines are considered
sites most suitable location for a WH structure. This approach is used
Malesu et al. (2007) uses criteria, largely governed by the FAO by De Winnaar et al. (2007) in South Africa; Munyao (2010) in Zan-
standards, for creating index maps of Africa for the suitability for 4 zibar; Wang et al. (2012) in China and Gupta et al. (1997), Rama-
different types of WH systems, rooftop runoff tanks, ponds and pans, krishnan et al. (2009), Kadam et al. (2012) and Sharma and Singh
flood diversion and flow storage (i.e. sand dams) and in-field tech- (2012) in India. Senay and Verdin (2004) used the SCS-CN approach
niques such as zai pitting and terracing. They indicate that rainfall to produce low resolution index maps for the whole of Africa.
>200 mm/annum is sufficient for all methods, slopes of <2% were Jasrotia et al. (2009) used a water balance approach for deter-
best suited to most techniques however ponds could be situated mining the suitability of WH sites. Remote sensing data from the
on slopes up to 8%. Soil with low permeability are indicated, except Indian satellite IRS-1D (LISS + PAN) and GIS techniques were used
for sand dams where fluvial soil must be present. However, saline to determine the runoff potential (using the (TM) Model) of various
soils are not suitable for in field techniques. One of the major con- land use types in Jammu Himalaya in India. Runoff potential was
straints to the study of Malesu et al. (1997) is the lack of data. Some determined using rainfall and temperature data. THE IRS-1D data
assumptions, such as that all agricultural land is suited to infield was used to create a land use map which was intersected with a
WH, will result in an over estimation of land suitability for WH. Fur- soil texture map. A topographical map was used to create a slope
thermore, an exclusion of all other land currently not under agricul- map using a Digital Elevation Model (DEM). Runoff potential was
ture will also skew results as this land may be underutilised for other ranked on a scale from low to high and then overlaid with a drain-
reasons such as restrictions due to conservation areas, transport age channel map to confirm water availability. Lastly, settlements
infrastructure, political or social reasons. were given a 500 m exclusion buffer. This method determined that
Singh et al. (2009) use the water balance approach together with 11% of the Devak-Rui catchment in Jammu Himalaya, India was
the Integrated Mission for Sustainable Development (IMSD) guide- suitable for WH (Jasrotia et al., 2009).
lines for determining WH site suitability. These guidelines are spe- Elewa et al. (2012) mapped the potential of the Sinai peninsula in
cific to different types of WH systems and are more lenient in their Egypt for WH. A watershed modelling system was combined with a
site specifications than the Critchley et al. (1991) guidelines. Slopes multi-criteria decision support system using 9 thematic layers re-
up to 15% are considered suitable for some systems, while soil tex- lated to catchment hydrological characteristics such as the volume
ture characteristics are specific to the type of systems, e.g. soils in of annual floods, drainage frequency, maximum flow distance basin
the ponds should have low infiltration rate while soil around the slope and area. They found that between 5% and 12% of the catch-
percolation tanks should have a high infiltration rate. These guide- ment was considered highly suitable for WH while 64% was consid-
lines are more encompassing of the difficult climates under which ered moderately suitable. This is despite the annual MAP of the Sinia
WH systems are often founds. The land use guidelines present many peninsula being <100 mm/a in most areas, with only 5–15 rainfall
challenges. Singh et al. (2009) recommends land use classes such as days per year. This suggests that the criteria set out by many authors
shrub land, barren land or bare soils. These are often not used for is too limited and that using redefining the criteria used to assess the
agriculture, therefore suitable sites may be identified which are lo- suitability of an area for WH may identify much greater potential for
cated far from where the water is needed. Water harvesting struc- WH than most approaches predict.
tures are often small and can be fitted within the farm lands/ With so many guidelines and derivatives being used to deter-
cultivated areas therefore to exclude this land use is restrictive. mine suitability, it is difficult to assess which are the best methods
Some practitioners and authors have amended the available for site selections. Water harvesting often takes place in less than
guidelines to be more encompassing of local conditions. Al-Adamat optimal conditions as it is in these areas where it is most needed.
et al. (2010) researched WH suitability in Jordan and ranked charac- However it appears that by applying the criteria set by some
92 L.M. Bulcock, G.P.W. Jewitt / Physics and Chemistry of the Earth 66 (2013) 89–100

Table 1
Commonly used guidelines for determining water harvesting site suitability.

Author
Critchley et al. (1991) Oweis et al. IMSD (in Singh et al., 2009)
(1998)
Mean annual <150–750 50–300 Not defined
precipitation (mm/a)
Slope (%) Runoff generation: short, steep slopes <4% <15%
Runoff collection: <5%
Soil texture (class) Medium textured loamy soils are best, otherwise all which Not defined System dependant: percolation tanks suited to sandy soils,
are suitable for agriculture. ponds suited to clay soils
Soil depth <1 m is unsuitable, 2 m is ideal but rarely found <50 cm Not defined
Other Not defined Not defined Land use should be barren, scrubland or bare soil

authors, especially the FAO guidelines, many sites currently under sites we cannot begin to understand how and why WH systems
WH would not be classified as suitable. function and studies into the potential for up-scaling, understand-
ing catchment and environmental responses to WH, modelling the
local and regional impacts or any other physically based studies
1.2.3. Data sources are futile.
Several studies have used GIS and/or remote sensing to locate
sites suitable for both macro- and micro-catchment rainwater
harvesting schemes based on these criteria (Al-Ahmed et al., 2. Methodology
2008; Sekar and Randhir, 2006, 2007; De Pauw et al., 2006;
Bodhankar, 2004; Durga Rao and Bhaumik, 2003; Patrick, 1997; 2.1. Data collection
Mwenge Kahinda et al., 2009; Mbilinyi et al., 2005). The FAO guide-
lines for the siting of rainwater harvesting ponds as well as other Twenty-eight site reports were analysed, of which sixteen
FAO documents on WH are also based on the Critchley et al. were from the World Overview of Conservation Approaches and
(1991) criteria (FAO, 2003). Technologies (WOCAT) database (https://www.wocat.net/en/
One of the major constraints in the curve number approach is knowledge-base.html), eleven from peer reviewed scientific pa-
the resolution and quality of available input data. Ramakrishnan pers and two from research project reports (Table 2). Additional
et al. (2009) used data from the Indian Remote Sensing Satellite papers and reports would have been useful, however many lack
(IRS-LISS-III) to prepare land use maps, a digital elevation model proper descriptions of the study sites or are not readily available.
(DEM) derived from the Shuttle Radar Topographic Mission Data regarding the physical characteristics of the site which may
(SRTM) to derive the slope map and drainage lines and a affect the hydrological functioning of the WH system were ex-
1:50,000 hydrological soil group (HSG) map generated by the In- tracted. These include slope, soil texture and depth and mean an-
dian National Bureau of Soil Survey and Land Use Planning (NBSS- nual precipitation. Other data extracted were latitude, longitude
LUP) to provide a WH suitability map. Kadam et al. (2012) used and altitude. Data regarding drainage patterns types, land cover
LANDSAT Thematic Mapper to identify landuse. These high reso- and mean annual evaporation may have also been valuable but
lution data resulted in more accurate input data and more accu- were generally not available. Even scientific papers which focus
rate results. However, Senay and Verdin (2004) used SCS-CN to on runoff related characteristics lacked a good description of
create index maps for Africa for pond suitability using 10 km res- key hydrological influencing characteristics and these needed to
olution data. This coupled with inaccuracies in the model meant be in-filled. The most difficult data to gather were the percent-
that the absolute values for runoff amounts could not be used ages of clay, sand and loam respectively. When this data was
and only a relative comparison between the suitability of areas not available from the paper/report the authors of the papers
for ponds could be done. Arguably, a 10 km data resolution for were emailed and asked to verify the characteristics of the site.
identifying areas for relatively small scale structures such as Very few responded but those that did were able to provide up-
WH structure is not useful. Furthermore, some assumptions made dated soil data or alternative references. Often a general descrip-
in the SCS-CN studies are bold and unsubstantiated. For example tion of the soil texture was available (e.g. sandy loam) or a
Gupta et al. (1997) assume that all areas with slopes <5% are un- ranking of the textures classes was available, (i.e. more clay than
der agricultural production and therefore can be defined as suit- sand and more sand than loam). In these cases the percentages
able to WH, this may result in an over estimation of suitability were estimated using the soil texture triangle (Saxton et al.,
for WH. Conversely, Kadam et al. (2012) identify agricultural land 1986). Table 2 show the 28 sites used and their characteristics.
as unsuitable for a WH structure, which contradicts the design When no information was available, the site was discarded. Sites
and purpose of a WH structure. This results in the sites suitable were classified as either in situ or ex situ according to the defini-
for WH being located far from the fields and areas where the tion in Section 1.1 above.
water is needed. Lastly, assumption in the antecedent soil mois-
ture calculations often result in an over assumption in the runoff 2.2. Statistical Analysis
generation potential of the site.
A study was done to examine under what conditions WH is cur- In order to assess the current status of WH knowledge available
rently taking place, to see if there is indeed a strong correlation be- through published (including informal publication) literature and
tween the actual sites and the guidelines. Without a proper data, a principal components analysis (PCA) was undertaken to
understanding or inventory of the physical characteristics of WH evaluate the relationships between the characteristics of the
 L.M. Bulcock, G.P.W. Jewitt / Physics and Chemistry of the Earth 66 (2013) 89–100 93

Table 2
Database of WH sites used for the PCA analysis.

Site Characteristics WH Technology Slope (%) Texture (%)

Characteristics
Site Country Catchment Name/Site Lat Long Altitude Type of WH system aWH MAP Min Max Soil depth Sand Loam Clay
No. location (Reference) (masl av) type (mm/a) (cm)
*
1 Ethiopia Bilate watershed (Pretorius, 2011) 7.07 38.07 1950 Micro catchments/ 2 900 2 8 50 60 32 8
ponds
*
2 Chad Eastern chad (Zähringer, 2012) 11.58 15.46 550 Water-spreading 2 275 0 5 85 8 9 83
weirs
*
3 China Gansu Province, Zhuanglang County 38.98 110.94 1250 Terraces 1 500 16 30 150 41 40 19
(Wang, 2006)
*
4 Ethiopia Bilate watershed (Danano, 2011) 7.07 38.07 1750 Earth checks 1 875 5 8 100 41 40 19
*
5 Ethiopia Alaba special woreda (Pretorius, 2011a,b) 7.49 38.19 1950 Grass strip, bunds, 1 900 2 8 100 69 26 5
contours
*
6 India Madhya Pradesh, Ratlam, Mohanpada 23.31 75.03 750 Sand wells in 2 800 0 8 25 80 4 16
(Gandhi, 2002) stream bed
*
7 India Bijapur district, Hadalsang village, 16.81 75.71 594 Farm Ponds 2 550 2 5 30 42 17 41
Karnataka (Metri, 2004)
*
8 Kenya KiMuiki, Kitise, Mburo, Kwa Kauisi 2.23 37.85 1000 Retention/ 1 350 6.5 12 50 59 28 13
(Pretorius, 2003) infiltration ditch
*
9 Nepal Kavrepalanchowk district/Lamdihi, 27.68 85.68 1000 Plastic lined ponds 2 1070 5 16 100 7 51 42
Patalekhet, Chiuribot (Dhakal, 2006)
*
10 Philippines Nueva Ecija (Pretorius et al., 2000) 15.79 120.82 250 Micro Dams 2 1980 5 30 115 1 43 56
*
11 Philippines Pangasinan, Nueva Ecija, Tarlac, Isabela, 14.79 120.88 250 Micro Dams 2 1867 8 16 35 29 47 24
Bulacan, Ilocos Norte (Labios et al., 2000)
12 South Botshabelo, De Wetsdorp, Bloemfontein, 29.21 26.83 1508 Micro Basins 1 600 0 2 100 28 12 60
Africa Thaba Nchu (Botha, 2001)
*
13 Senegal Tabanding, Tambacounda Ndiaye et al. 13.91 13.41 50 Concrete barrier 1 750 2 8 35 20 9 71
(2010)
*
14 Syrian Hannaser Valley Turkelboom, 2004) 36.28 37.66 300 Zai mixed with 1 250 5 16 65 40 37 23
Arab drainage ditch
Republic
*
15 Tajikistan Kabodion/Khudoikulov, Khatlon (Firdavs 37.21 68.21 300 Flood irrigation/ 2 100 0 2 35 94 3 3
et al., 2011) water spreading
*
16 Zambia Southern Province (Zähringer and Malesu, 16.72 26.53 750 Small earthen 2 700 2 40 65 66 23 11
1970) dams
17 South Thukela, KwaZulu-Natal (de Winnaar et al., 28.81 29.35 1351 JoJo Tanks 1 1800 4 16 250 17 69 12
Africa 2007)
18 Burkina Gampela 12.33 1.33 275 Stone Lines 1 600 0 5 100 78 11 11
Faso
19 Mali Kaniko, Koutlala Region (Bodnár et al., 12.32 5.36 351 Stone rows/water 2 1200 0 5 100 37 42 21
2007) spreading, grass
strips
20 Jordan North-East Jordan - S1 (Abu-Zreig and 32.56 36.02 520 Sand ditch 1 211 10 12 750 22 48 34
Tamimi, 2011)
21 Jordan North-East Jordan – S2 (Abu-Zreig and 32.56 36.02 520 Sand ditch 1 211 10 12 200 14 48 38
Tamimi, 2011)
22 China Gaolan County, Lanzhou, Gansu Province 36.22 103.78 1780 Micro compacted 1 263 7 8 100 12.3 66.9 20.8
(Li and Gong, 2002) plots
23 South Thohoyandou, Limpopo (Mzezewa et al., 22.97 30.43 596 Micro-catchments 1 450 2 8 180 27 24 49
Africa 2011)
24 India Udhagamandalam, Tamil Nadu (Madhu 11.40 76.68 2217 Furrow contours 2 1204 25 25 100 45 20 35
et al., 2011)
25 India Aravalli hills, Banswara (Singh, 2009) 23.42 74.41 280 Trenches 1 990 3-53 53 100 13 5 82
26 Ethiopia Kushet – Gobo Deguat (Hengsdijk et al., 13.66 39.42 2575 Stone bunds 1 538 0-50 50 78 78 14 8
2005)
27 Kenya Kalalu, Diaga, Laikipia (Ngigi et al., 2006) 0.10 37.14 1900 Micro-dams 1 1024 4 4 125 20 30 50
28 South Baynesfield, KwaZulu-Natal (Everson et al., 29.75 30.34 1150 Micro catchments 1 900 5 16 150 19 35 46
Africa 2011)
a
Code for type of Water Harvesting system 1 = in situ, 2 = ex situ.
*
Indicates soil texture inferred as opposed to reported.

reported WH sites. A PCA is a type of cluster analysis which groups able for the majority of variance, the second principle component
variables according to the variance within the dataset (Morell et al., for the second largest and so forth (Selle et al., 2013). Together
1996). The PCA converts a set of possibly correlated variables into a the variables should account for most of the variance within the
set of values of linearly uncorrelated variables i.e. principal compo- dataset. The PCA aims to expose trends and relationships between
nents so that each is homogenous (Sârbu and Pop, 2005; Chaplot variables in the data, which represent a common process/impact
et al., 2010, 2011). The principle components describe the variance and may not be apparent otherwise (Selle et al., 2013). It identifies
in the data set with the first principle component being account- patterns in data, and expressing the data in such a way as to
94 L.M. Bulcock, G.P.W. Jewitt / Physics and Chemistry of the Earth 66 (2013) 89–100

highlight the similarities and differences. The base equation under- WH sites and should be prioritised for measurement, and which
lying the PCA method is: variable is less important. PCA groups variables, along the princi-
ple component axes, thereby classifying measured variables
X
m
according to their importance to WH. The PCA also illustrates
Z ij ¼ bji PC il
i¼1
how the characteristics of the site are similar or where they differ
and reveals relationships not always apparent by just looking at
where Z is the observed data, which was standardized by subtract- the data. Descriptive statistics (Minimum, Maximum, Mean,
ing the mean and dividing by the standard deviation for each of the Mode, Median, Standard deviations, skewness testing and kurto-
measured water quality variables; b denote ‘loadings’ representing sis) were computed for MAP, Altitude, minimum slope (%), maxi-
coefficients of correlation between observed variables and principal mum slope (%), soil depth (cm) and the percentages of sand, loam
components; PC are the ‘scores’ reflecting the relationship between and clay in the soil. These results were displayed and are pre-
samples and the principal components; i represents an index of sented in the next section.
samples indexing both locations and time of sampling; j is an index
of m variables that were measured for each sample; and l is an in- 3. Results and discussion
dex of m principal components.
A correlation matrix PCA was applied to the data using the 3.1. Determining the most important factors in positioning an in situ
Ade-4 software to evaluate the relationship between the environ- WH site
mental variables (Chessel et al., 2004). PCA is used to identify
whether two variables measure the same characteristics, thereby The correlation circle in Fig. 2 illustrates the results of the PCA.
showing which variable have a greater impact on the location of There is a clustering of soil characteristics around the first principle

Fig. 2. Correlation circle PCA which expresses the relationship between variables of in situ WH sites.

Table 3
Statistical analysis of the site characteristics of in situ WH sites.

WH type 1 – In situ, n = 17
MAP (mm/a) Altitude (masl) Min Slope (%) Max Slope (%) Soil Depth (cm) % Sand % Loam % Clay
Mean 659.53 1103.2 4.7 15.8 155 35.2 31.9 32.9
Min 211 50 0 2 35 12 5 5
Max 1800 2575 16 53 750 78 69 82
Median 600 1150 4 12 100 27 30 23
Mode 900 520 2 8 100 41 40 19
Std Dev 405.2 737.8 4.2 14.9 162.9 22.7 19.1 23.3
Skew 1.3 0.3 1.3 1.9 3.4 0.9 0.5 0.7
Kurt 2.7 0.9 1.9 2.7 12.6 0.4 0.3 0.5
 L.M. Bulcock, G.P.W. Jewitt / Physics and Chemistry of the Earth 66 (2013) 89–100 95

component, which accounts for 28.51% of the variability in the data slopes with a minimum and maximum range of 67–20%. How-
for micro-catchment WH (Fig. 2). Therefore soil characteristics ac- ever examples are present of in situ WH taking place on slopes
counts for most of the variance in the data. This is further sup- of 18–53%, so these potential sites with steep slopes should not
ported by the large standard deviation seen in the soil be excluded. This suggests that the recommendation by all
characteristics in Table 3. This implies that in situ WH is taking three guidelines are too restrictive as the steepest slope recom-
place under a wide variety of soil conditions. Fig. 3a shows that mendation, by IMSD, is <15%. Altitude and MAP are all highly
in situ WH is taking place in soils with clay content across the spec- variable and so it appear that in situ WH can take place at
trum therefore clay content does not limits WH. No in situ WH is any altitude and across many MAP regions (Table 3).
taking place in very sandy soils (>80%) or soils with a high loam
content (>70%). Most WH takes place when loam content is 3.2. Determining the most important factors in positioning ex situ WH
between 10% and 50% and sand content is between 20–30% site
and 40–49%. According to this analysis, the % of sand and loam
has an influence on the location of in situ WH sites. Minimum The results of the PCA show that the first principle component
slope (%) and Longitude are closely aligned with the 2nd prin- accounts for 36.33% of the variance in the data while the second
ciple component which accounts for 20.25% of the variability principle component accounts for 19.41% of the variance in the
in the data. Climate zones are closely linked to their position data. In ex situ WH there is a strong relationship between the soil
relative to the Equator, i.e. tropical, sub-tropical, arid and depth, MAP and the % of loam in the soil (Fig. 4). Deeper soils
semi-arid regions. The distribution of the location of in situ (P1 m) were found in areas with a higher MAP and had a higher
WH sites, across the longitudinal zones indicated that there is percentage of loam. These represent the most ideal conditions for
no direct relationship between in situ WH and climate zones WH as suggested by all three sets of guidelines. The standard
and therefore in situ WH can be utilised under a wide range deviation of the macro-catchment mean soil depth is much lower
of climatic zones. The variability in the minimum and maxi- than the in situ WH (34 cm as opposed to 163 cm) indicating that
mum slope angle (in%) varies greatly, evident in a standard ex situ WH is taking place in mostly shallow soils as compared to
deviation of 4 and 14% respectively (Table 3) however analysis in situ WH which is taking place in a wide range of soil depths
of skewness of the data (Fig. 3b and c) illustrates that mini- (Table 4). This is again confirmed by the skewness test which
mum and maximum slope is skewed towards the left, around shows little variation from the mean of the ex situ WH (skew-
the more gentle slopes. This shows that more gentle slopes ness = 0.063) (Table 4). There is a strong inverse relationship
are better suited to in situ WH with most occurring on gentle

Fig. 3. (a) Frequency distribution of soil texture classes of in situ WH sites, (b and c) frequency distribution of minimum and maximum (respectively) slope classes for in situ
WH sites.
96 L.M. Bulcock, G.P.W. Jewitt / Physics and Chemistry of the Earth 66 (2013) 89–100

Fig. 4. Correlation circle PCA which expresses the relationship between variables of ex situ WH sites.

Table 4
Statistical analysis of the site characteristics of ex situ WH sites.

WH type 2 – Ex situ, n = 11
MAP (mm/a) Altitude (masl) Min Slope (%) Max Slope (%) Soil Depth (cm) % Sand % Loam % Clay
Mean 967.8 814.7 4 14.5 67.3 42.6 26.5 30.9
Min 100 250 0 2 25 1 3 3
Max 1980 2217 20 40 115 94 51 83
Median 900 594 2 8 65 42 23 24
Mode N/A 750 0 5 100 N/A N/A N/A
Std Dev 588.7 673.4 5.9 12.3 33.7 30.5 17.5 23.8
Skew 0.4 1.4 2.3 1.1 0.1 0.2 0.1 1.1
Kurt 0.3 1.1 5.8 0.2 1.9 0.9 1.6 0.9

between the amounts of clay and sand in the soil, which is to be slopes indicates that the guidelines are restrictive and perhaps ex
expected (Fig. 4). The soil texture frequency distribution (Fig. 5a) situ WH can take place on more steep slopes.
shows that no ex situ WH is taking place in soils with a loam con-
tent >60% however ex situ WH is taking place across the full range
of sand and clay contents, indicating that the clay and sand are 3.3. Consequences for the guidelines
not limiting factors in location of ex situ WH. The minimum
and maximum slope lines are not closely aligned with either prin- The results of the statistical analysis have shown that while
ciple component indicating that there is not much variability most of the WH sites are situated in landscape characteristics sim-
within the data. The lines are also positioned very close to each ilar to those recommended by the three sets of guidelines, some
other indicating that there is not much variability between the sites have characteristics quite different to those specified in the
two characteristics (Fig. 4). Examination of the frequency distri- recommendations. The guidelines may be described as the optimal
bution of the slope shows a strong clustering around the gentle conditions for WH, but a second set of recommendations should be
slopes for the minimum slope (except for one outlier) (Fig. 5b) made to allow for conditions that may still be suitable, even if they
and most of the sites having a maximum slope of <20%, except are not optimal. Furthermore, separate recommendations for in situ
for 3 sites (Fig. 5c). This again confirms the recommendations and ex situ WH are necessary. Table 5 presents a new set of guide-
by the three sets of guidelines that more gentle slopes are best lines for the positioning of both optimal and suitable conditions for
suited to WH, but evidence of ex situ WH taking place on steeper in situ and ex situ WH sites based on the results of this study.
 L.M. Bulcock, G.P.W. Jewitt / Physics and Chemistry of the Earth 66 (2013) 89–100 97

Fig. 5. (a) Frequency distribution of soil texture classes of ex situ WH sites, (b and c) frequency distribution of minimum and maximum (respectively) slope classes for ex situ
WH sites.

Table 5
New guidelines for WH site requirements based on a statistical analysis of existing WH sites.

Criteria Low limit suitability Optimal Upper limit suitability

In situ WH site requirements
Altitude No restrictions
Latitude No restrictions
Longitude No restrictions
MAP (mm/a) No restrictions
Slope No lower limit <20% Up to 53% proven to work
Soil texture Loam: No lower limit Loam: 0–50% Loam: <70%
Sand: 10% Sand: 10–30% Sand: <80%
Clay: No lower limit Clay: 0–50% Clay: No upper limit
Soil depth 35 100 No upper limit
Ex situ WH site requirements
Altitude No restrictions
Latitude No restrictions
Longitude No restrictions
MAP (mm/a) 100 900 No upper limit
Slope No lower limit <10% Up to 40% proven to work
Soil texture Loam: No lower limit Loam: 0–50% Loam: 60%
Sand: No lower limit Sand: 20–50% Sand:95%
Clay: No lower limit Clay: 0–50% Clay: 90%
Soil depth (cm) 25 100 115

4. Conclusion within the catchment. Localised soil, slope and land cover condition
will have an influence on how runoff is generated, how much is gen-
Detailed information on the physical characteristics of WH sites erated and how it naturally moves through the landscape. These
is imperative to understand the hydrological functions of these sites conditions will also dictate the extent and what type of WH can be
98 L.M. Bulcock, G.P.W. Jewitt / Physics and Chemistry of the Earth 66 (2013) 89–100

sustained within a catchment and the quantity and quality of water niques such as GIS, remote sensing, geo-information and a more
that can be harvested. It will also have an effect on the environment, common acceptance of the role of indigenous knowledge can allow
downstream and upstream users, at a local scale, at catchment sale for more wide spread data collection. Remote sensing has proven
and potentially even a regional scale. Without a good understanding to be successful in identifying potential WH sites by various
of the physical characteristics of existing and potential WH sites, we authors as discussed above. Further use of this technology to col-
cannot begin to model or understand the possible impacts of exist- lect data about existing WH sites and analysed through statistical
ing WH sites nor can we explore the potential for expanding and analysis as presented in this paper (but with a larger sample size)
upscaling of WH within a catchment. will further strengthen the guidelines for WH site suitability.
The results of the statistical study show that the existing guidelines
which are most commonly used to determine suitable locations for Acknowledgements
WH only represent the optimal conditions for WH. WH can, and cur-
rently does, take place under an much wider range of conditions. This The research leading to these results has received funding from
implies that many more sites may be suitable for WH than currently the European Community’s Seventh Framework Programme [FP7/
acknowledged. Therefore the strict application of existing guidelines 2007–2013], under the WHaTeR project (Water Harvesting Tech-
is too limiting. Evidence from existing WH sites shows that a new set nologies Revisited) Grant Agreement No. 266360.
of guidelines, taking into account not only optimal conditions but also
upper and lower limits of suitability need to be developed. The ranking References
of conditions may allow for many more potential WH sites to be iden-
tified and allow WH to make a greater impact in supplying water for Abu-Zreig, M., Tamimi, A., 2011. Field evaluation of sand-ditch water harvesting
technique in Jordan. Agricultural Water Management 98, 1291–1296.
agricultural, livestock and domestic uses. Agarwal, A., 2001. Water harvesting in a new age. In: Khurana, I. (Ed.), Making
Water Everybody’s Business. Centre for Science and Environment, New Delhi,
pp. 1–13.
African Ministers’ Council on Water, 2008. Sharm El-Sheikh Commitments for
5. Recommendations Accelerating the Achievement of Water and Sanitation Goals in Africa:
Decisions, Declarations, Tribute and Resolution. <http://www.amcow-
One of the limits to this study was the availability of data. Site online.org/images/Resources/sharm_el-sheikh_decisions_english.pdf>
(05.08.13).
descriptions often left out important details such as land use, African Ministers’ Council on Water, 2012. The Pan African Water and Sanitation
drainage patterns and mean annual evaporation which have all Monitoring, Evaluation and Reporting Format: The Guidelines. <http://
been identified in previous studies as being important in determin- www.cedare.int/namcow/attachments/article/122/07%20%28ENG-
4%29%20Pan%20African%20M%20E%20Guidelines%20.pdf> (05.08.13).
ing WH site suitability. Several other authors such as Malesu et al.
Al-Adamat, R., Diabat, A., Shatnawi, G., 2010. Combining GIS with multicriteria
(2007) highlighted a lack of data as a constraint on their mapping decision making for siting water harvesting ponds in Northern Jordan. Journal
exercises. Malesu et al. (2007) also highlighted that their study on of Arid Environments 74, 1471–1477.
sand dams was further compromised by poor slope data and so Al-Adamat, R., AlAyyash, S., Al-Amoush, H., Al-Meshan, O., Rawajfih, Z., Shdeifat, A.,
Al-Harahsheh, A., Al-Farajat, M., 2012. The combination of indigenous
slope had to be discarded, weakening the study. Prinz et al. knowledge and geo-informatics for water harvesting siting in the Jordanian
(1998) highlighted that large scale planning and implementation Badia. Journal of Geographic Information System 4, 366–376.
of WH structures, requires quantitative information of the spatial Arrindell, W.A., Van der Ende, J., 1985. An empirical test of the utility of the
observations-to-variables ratio in factor and components analysis. Applied
distribution of physical land characteristics which is often not Psychological Measurement 9, 165–178.
available in arid or semi-arid areas at the resolution required for Bodhankar, N., 2004. Application of vectors for suitability of landforms in siting
accurate land suitability assessments for small scale structures surface water harvesting structures. Journal Environmental Geology 44, 176–
179.
such as WH structures. Traditional soil and land surveys (1:10 Bodnár, F., Spaan, W., Hulshof, J., 2007. Ex-post evaluation of erosion control
000) do give sufficiently accurate information however these are measures in southern Mali. Soil & Tillage Research 95, 27–37.
often only available over limited areas, are expensive and time Bossio, D., Jewitt, G., van der Zaag, P., 2011. Smallholder system innovation for
integrated watershed management in Sub-Saharan Africa. Agricultural Water
consuming to produce (Prinz et al., 1998). Management 98, 1683–1686.
Unfortunately there are few guidelines for the describing the Botha, C., 2001. WOCAT Database Reference: RSA045en. <https://www.wocat.net/
minimum accepted sample size for PCA (Osborne and Costello, en/knowledge-base.html> (08.08.13).
Cattell, R.B., 1978. The Scientific use of Factor Analysis in Behavioral and Life
2004). Some authors recommend an absolute minimum sample
Sciences. Plenum, New York.
number (N) of 50, while others suggest a ratio between the number Chaplot, V., Lorentz, S.A.L., Podwojewski, P., Jewitt, G.P.W., 2010. Digital mapping of
of cases per variable (N/p) with recommendations range from 3:1 A-horizon thickness using the correlation between various soil properties and
to 6:1 (Cattell, 1978). Later studies by Arrindell and Van der Ende soil apparent electrical resistivity. Geoderma 157, 154–164.
Chaplot, V., Jewitt, G.P.W., Lorentz, S.A.L., 2011. Predicting plot-scale water
(1985), Jackson (2001) and MacCallum et al. (1999) show that rec- infiltration using the correlation between soil apparent electrical resistivity
ommendations on absolute N and the N/p ratio can be misconceiv- and various soil properties. Physics and Chemistry of the Earth 36, 1033–1042.
ing. MacCallum et al. (1999) developed a theoretical framework for Chessel, D., Dufour, A.B., Thioulouse, J., 2004. The ADE4 package. One-table
methods. R News 4, 5–10.
testing the effects of sample size on factor recovery and concluded Critchley, W., Siegert, K., Chapman, C., 1991. A manual for the Design and
that there are no absolute thresholds for a minimum sample size Construction of Water Harvesting Schemes for Plant Production. Food and
however it is accepted that the smaller the sample size, the less Agriculture Organization of the United Nations, Rome, Italy.
Comprehensive Assessment of Water Management in Agriculture, 2007. Water for
confidence can be placed in a result (de Winter et al., 2009). Food, Water for Life: A Comprehensive Assessment of Water Management in
In this study the sample size of N = 28, means that caution Agriculture. London: Earthscan, and Colombo: International Water
should be applied when examining the results. Rather than con- Management Institute.
Dhakal, M., 2006. WOCAT Database Reference: NEP022en. <https://www.wocat.net/
cluding that the ranges and recommendations for new guidelines en/knowledge-base.html> (08.08.13).
be regarded as absolute truth, the trends in the data should be con- Danano, D., 2011. Zena Estifanos Ministry of Agriculture and Rural Development.
sidered. It is apparent that WH is taking place under a much wider WOCAT database reference: ETH039en. <https://www.wocat.net/en/
knowledge-base.html> (08.08.13).
range of conditions than those currently considered suitable. This
de Graaff, J., Amsalu, A., Bodnár, F., Kessler, A., Posthumus, H., Tenge, A., 2008.
should be the basis or motivation for future work to improve the Factors influencing adoption and continued use of long-term soil and water
availability of WH data, to allow for more robust analysis of exist- conservation measures in five developing countries. Applied Geography 28,
ing data and also to expand WH research and more rigorously anal- 271–280.
de Pauw, E., Oweis, T., Youssef, J., Bashar, N., 2006. Assessing biophysical potential
ysis existing WH sites so that researchers can learn under what full for water harvesting at national and global scales. In: 18th World Congress of
range of conditions WH can take place. Developments in tech- Soil Science July 9–15, 2006. Philadelphia, Pennsylvania, USA.
 L.M. Bulcock, G.P.W. Jewitt / Physics and Chemistry of the Earth 66 (2013) 89–100 99

Desai, A., Ghose, S., 2001. Traditions to the rescue. In: Khurana, I. (Ed.), Making Mzezewa, J., Gwata, E.T., van Rensburg, L.D., 2011. Yield and seasonal water
Water Everybody’s Business. Centre for Science and Environment, New Delhi, productivity of sunflower as affected by tillage and cropping systems under
pp. 19–21. dryland conditions in the Limpopo Province of South Africa. Agricultural Water
De Winnaar, G., Jewitt, G.P.W., Horan, M., 2007. A GIS-based approach for Management 98, 1641–1648.
identifying potential runoff harvesting sites in the Thukela River basin, South Ndiaye, D.S., Salif, N., Baïdy, S., Aldiouma, S., 2010. WOCAT Database Reference:
Africa. Physics and Chemistry of the Earth 32, 1058–1067. SEN314fr. <https://www.wocat.net/en/knowledge-base.html> (08.08.13).
de Winter, J.C.F., Dodou, D., Wieringa, P.A., 2009. Exploratory factor analysis with Ngigi, S.N., Rockström, J., Savenije, H.H.G., 2006. Assessment of rainwater retention
small sample sizes. Multivariate Behavioral Research 44, 147–181. in agricultural land and crop yield increase due to conservation tillage in Ewaso
Durga Rao, K.H.V., Bhaumik, M.K., 2003. Spatial expert support system in selecting Ng’iro river basin, Kenya. Physics and Chemistry of the Earth 31, 910–918.
suitable sites for water harvesting structures – a case study of Song watershed, Nissen-Petersen, E., 2006. Water from Small Dams. A Handbook for Technicians,
Uttaranchal, India. Geocarto International 18, 43–50. Farmers and Other on site Investigations, Designs, Cost Estimates, Construction
Elewa, H.H., Qaddah, A.A., El-Feel, A.A., 2012. Determining potential sites for runoff and Maintenance of Small Earth Dams. ASAL Consultants Ltd. and Danish
water harvesting using remote sensing and geographic information systems- International Development Assistance (Danida), Kenya.
based modeling in Sinai. American Journal of Environmental Sciences 8, 42–55. Osborne, Jason W., Costello, Anna B., 2004. Sample size and subject to item ratio in
Everson, C.S., Ghezehei, S.B., Everson, T.M., Annandale, J., 2011. Agroforestry principal components analysis. Practical Assessment, Research & Evaluation,
Systems for Improved Productivity Through the Efficient use of Water. WRC 9(11) (Electronic Journal). <http://PAREonline.net/getvn.asp?v=9&n=11>
Report No: 1480/1/12. Water Research Commission, Pretoria, South Africa. (retrieved 08.08.13).
Firdavs, F., Zähringer, J., Ibragimov, F., 2011. WOCAT Database Reference: TAJ112en. Oweis, T., Oberle, A., Prinz, D., 1998. Determination of potential sites and methods
<https://www.wocat.net/en/knowledge-base.html> (08.08.13). for water harvesting in Central Syria. In: Blum, H-P., Eger, H., Fleischauer, E.,
Falkenmark, M., Fox, P., Persson, G., Rockstroem, J., 2001. Water harvesting for Hebel, A., Reij, C., Steijner, K.G. (Eds.), Towards Sustainable Land Use, Vol. 1.
upgrading of rainfed agriculture. Problem Analysis and Research Needs. SIWI Advances in Geoecology 31. Catena Verlag, Reiskirchen, Germany, pp. 83–88.
Report II. Stockholm International Water Institute, Sweden. Oweis, T., Hachum, A., Bruggeman, A., 2004. Indigenous Water-Harvesting Systems
FAO, 2003. Land and Water Digital Media Series, 26. Training Course on RWH in West Asia and North Africa. International Centre for Agriculture Research in
(CDROM). Planning of Water Harvesting Schemes, Unit 22. Food and Agriculture the Dry Areas. Aleppo, Syria.
Organization of the United Nations, Rome, FAO. Patrick, E., 1997. Assessing water harvesting suitability. In: Proceedings of the
Gandhi, K., Kirtane, S., 2001. Bounty of Bundelkhand. In: Agarwal, A., Narain, S., International Workshop Remote Sensing and Water Resources, November–
Khurana, I. (Eds.), Making Water Everybody’s Business. Centre for Science and December 1995. Food and Agriculture Organization of the United Nations.
Environment, New Delhi, pp. 22–27. Montpellier, France.
Gandhi, D., 2002. Comprehensive Watershed Development WOCAT database Prinz, D., Oweis, T., Oberle, A., 1998. Rainwater harvesting for dry land agriculture –
reference: IND003en. <https://www.wocat.net/en/knowledge-base.html> developing a methodology based on remote sensing and GIS. In: Proceedings,
(08.08.13). XIII International Congress Agricultural Engineering, 2–6 February 1998,
Gowing, J.W., Mahoo, H.F., Mzirai, O.M., Hatibu, N., 1999. Review of rainwater ANAFID, Rabat, Morocco.
harvesting techniques and evidence for their use in semi-arid Tanzania. Pretorius, C., 2003. WOCAT Database Reference: KEN032en. <https://www.wocat.
Tanzania Journal of Agricultural Sciences 2, 171–180. net/en/knowledge-base.html> (08.08.13).
Gupta, K.K., Deelstra, J., Sharma, K.D., 1997. Estimation of water harvesting potential Pretorius, C., 2011a. Melese Teshome Alaba special woreda, SNNPR, WOCAT
for a semi-arid area using GIS and remote sensing. In: Baumgartner, M.F., Database Reference: ETH045en. <https://www.wocat.net/en/knowledge-
Schultz, G.A., Johnson, A.I. (Eds.), Remote Sensing and Geographic Information base.html> (08.08.13).
Systems for Design and Operation of Water Resources Systems. International Pretorius, C., 2011b. WOCAT Database Reference: ETH040en. <https://www.wocat.
Association of Hydrological Sciences, London, United Kingdom. net/en/knowledge-base.html> (08.08.13).
Hengsdijk, H., Meijerink, G.W., Mosugu, M.E., 2005. Modeling the effect of three soil Pretorius, C., Samuel, C., Teresita, S., Rodolfo, L., 2000. WOCAT Database Reference:
and water conservation practices in Tigray, Ethiopia. Agriculture, Ecosystems PHI004en. <https://www.wocat.net/en/knowledge-base.html> (08.08.13).
and Environment 105, 29–40. Ramakrishnan, D., Bandyopadhyay, A., Kusuma, K.N., 2009. SCS-CN and GIS-based
Jackson, D.L., 2001. Sample size and number of parameter estimates in maximum approach for identifying potential water harvesting sites in the Kali Watershed,
likelihood confirmatory factor analysis: a Monte Carlo investigation. Structural Mahi River Basin, India. Journal of Earth System Science 118, 355–368.
Equation Modeling 8, 205–223. Rockström, J., 2000. Water resources management in smallholder farms in Eastern
Jasrotia, A.S., Majhi, A., Singh, S., 2009. Water balance approach for rainwater and Southern Africa: an overview. Physics and Chemistry of the Earth 25, 275–
harvesting using remote sensing and GIS techniques, Jammu Himalaya, India. 283.
Water Resources Management 23, 3035–3055. Saxton, K.E., Rawls, W.J., Romberger, J.S., Papendick, R.I., 1986. Estimating
Kadam, A.K., Kale, S.S., Pawar, N.J., Sankhua, R.N., Pawar, N.J., 2012. Identifying generalized soil-water characteristics from texture. Soil Science Society of
potential rainwater harvesting sites of a semi-arid, basaltic region of Western America Journal 50 (4), 1031–1036.
India, Using SCS-CN method. Water Resource Management 26, 2537–2554. Sârbu, C., Pop, H.F., 2005. Principal component analysis versus fuzzy principal
Kessler, C.A., 2006. Decisive key-factors influencing farm households’ soil and water component analysis A case study: the quality of danube water (1985–1996).
conservation investments. Applied Geography 26, 40–60. Talanta 65, 1215–1220.
Kessler, C.A., 2007. Motivating farmers for soil and water conservation: a promising Sekar, I., Randhir, T.O., 2006. Spatial assessment of conjunctive water harvesting
strategy from the Bolivian mountain valleys. Land Use Policy 24, 118–128. potential in watershed systems. Journal of Hydrology. http://dx.doi.org/
Labios, R.V, Teresita, S., Dolores, M.G., 2000. WOCAT Database Reference: PHI005en. 10.1016/j.jhydrol.2006.09.024:1-1.
<https://www.wocat.net/en/knowledge-base.html> (08.08.13). Sekar, I., Randhir, T.O., 2007. Policies for sustaining groundwater resources. Water
Li, X.Y., Gong, J.D., 2002. Compacted Micro-Catchments with local earth materials International (Journal of International Water Resources Association) 32, 697–
for rainwater harvesting in the semiarid region of China. Journal of Hydrology 709.
257, 134–144. Selle, B., Schwientek, M., Lischeid, G., 2013. Understanding processes governing
MacCallum, R.C., Wildaman, K.F., Zhang, S., Hong, S., 1999. Sample size in factor water quality in catchments using principal component scores. Journal of
analysis. Psychological Methods 4, 84–99. Hydrology 486, 31–38.
Malesu, M., Khaka, E., Odour, A., De Bock, T., Nyabenge, M., Odour, V., 2007. Mapping Senay, G.B., Verdin, J.P., 2004. Developing index maps of water-harvest potential in
the Potential of Rainwater Harvesting Technologies in Africa: A GIS overview Africa. Applied Engineering in Agriculture 20, 789–799.
and atlas of development domains for the continent and none selected Sharma, A and Singh, VV. Identification of potential runoff harvesting sites in a
countries. World Agroforestry Centre (ICRAF) ICRAF Technical Manual No. 7. water scarce rural watershed using GIS approach. India Water Week 2012 -
Nairobi, Kenya. Water, Energy and Food Security: Call for Solutions, 10-14 April 2012, New Delhi.
Metri, C., 2004. WOCAT Database Reference: IND008en. <https://www.wocat.net/ Siegert, K., 1994. Introduction to water harvesting: some basic principles for
en/knowledge-base.html> (08.08.13). planning, design and monitoring. In: Water Harvesting for Improved
Madhu, M., Sahoo, D.C., Sharda, V.N., Sikka, A.K., 2011. Rainwater-use efficiency of Agricultural Production. Proceedings of the FAO Expert Consultations, Cairo,
tea (Camellia sinensis (L.)) under different conservation. Applied Geography 31, Egypt, 21–25 November, 1993. FAO, Rome, pp. 9–23.
450–455. Singh, G., 2009. Soil water dynamics, growth of Dendrocalamus strictus and herbage
Munyao, J.N., 2010. Use of Satellite Products to Assess Water Harvesting Potential in productivity. Forest Ecology and Management 258, 2519–2528.
Remote Areas of Africa: A Case Study of Unguja Island, Zanizbar. MSc Thesis. Singh, J.P., Singh, D., Litoria, P.K., 2009. Selection of suitable sites for water
International Institute for Geo-Information Science and Earth Observation, harvesting structures in Soankhad watershed, Punjab using remote sensing and
Enschede, The Netherlands. geographical information system (RS&GIS) approach-A case study. Journal of
Mbilinyi, B.P., Tumbo, S.D., Mahoo, H.F., Senkondo, E.M., Hatibu, N., 2005. the Indian Society of Remote Sensing 37, 21–35.
Indigenous knowledge as decision support tool in rainwater harvesting. Turkelboom, F., 2004. WOCAT Database Reference: SYR003. <https://
Physics and Chemistry of the Earth 30, 792–798. www.wocat.net/en/knowledge-base.html> (08.08.13).
Morell, I., Gimhez, E., Esteller, M.V., 1996. Application of principal components Vohland, K., Barry, B., 2009. A review of in situ rainwater harvesting (RWH)
analysis to the study of salinization on the Castellon Plain (Spain). The Science practices modifying landscape functions in African drylands. Agriculture,
of the Total Environment 177, 161–171. Ecosystems and Environment 131, 119–127.
Mwenge Kahinda, J., Taigbenu, A.E., Sejamoholo, B.B.P., ELillie, S.B., Boroto, R.J., Wang, Y., 2006. Meili Wen Department of Resources and Environmental Sciences,
2009. A GIS-based decision support system for rainwater harvesting Beijing Normal University, PR China. WOCAT database. <https://
(RHADESS). Physics and Chemistry of the Earth Parts A=B=C 34, 767–775. www.wocat.net/en/knowledge-base.html> (08.08.13).
100 L.M. Bulcock, G.P.W. Jewitt / Physics and Chemistry of the Earth 66 (2013) 89–100

Wang, H., Wang, X., Wang, X., Jiang, L., 2012. Identification of potential rainwater Ziadat, F., Oweis, T., Mazahreh, S., Bruggeman, A., Haddad, N., Karablieh, E., Benli, B.,
harvesting sites using SCS-CN and GIS. Nongye Gongcheng Xuebao/Transactions Abu Zanat, M., Al-Bakri, J., Ali, A., 2006. Selection and Characterization of Badia
of the Chinese Society of Agricultural Engineering 28, 108–114. Watershed Research Sites. ICARDA, Aleppo, Syria.
Warren, A., Osbahra, H., Batterbury, S., Chappell, A., 2003. Indigenous views of soil Ziadat, F., Bruggeman, A., Oweis, T., Haddad, N., Mazahreh, S., Sartawi, W., Syuof, M.,
erosion at Fandou Béri, south-western Niger. Geoderma 11, 439–456. 2012. A participatory GIS approach for assessing land suitability for rainwater
Zähringer, J., Malesu, M., 1970. WOCAT Database Reference: ZAM001en. <https:// harvesting in an arid rangeland environment. Arid Land Research and
www.wocat.net/en/knowledge-base.html> (08.08.13). Management 26, 297–310.
Zähringer, J., 2012. WOCAT Database Reference: CHA001en. <https://
www.wocat.net/en/knowledge-base.html> (08.08.13).