Nat Hazards DOI 10.1007/s11069-012-0090-z ORIGINAL PAPER Floods in Greece, a statistical and spatial approach Michalis Diakakis • Spyridon Mavroulis • Giorgos Deligiannakis Received: 18 January 2011 / Accepted: 10 January 2012  Springer Science+Business Media B.V. 2012 Abstract Flooding is one of the most important types of disasters in southern Europe recording many victims and extended damages over the last century. The increased pressure for urban expansion together with the high population density has increased flood risk considerably in the region. Greece is not an exception in this regime, having a very rich flooding record since the ancient times. In this work, an extensive catalogue of flooding phenomena during the last 130 years in Greece has been compiled based on numerous sources. Based on this record the temporal and spatial distribution of flood events and victims was examined. In total, 545 events were identified, causing 686 human casualties and inflicting extensive damage across the country. Results showed seasonality patterns with more events clustering in November. They also showed that urban environments tend to present a higher flood recurrence rates than mountainous and rural areas. An increasing trend in reported flood event numbers during the last decades was discovered, even though the number of human casualties remains relatively stable during the same period. Moreover, spatial patterns were identified highlighting areas and administrational entities with higher flood recurrence rates across the country. Keywords Flood database
Historical data
Greece
Press archives
Flood history
Flood record 1 Introduction Flooding problem becomes an increasingly significant issue in the Mediterranean region as population expands to river deltas and coastal areas that are subject to inundation mostly from small rivers and ephemeral mountain torrents. Greece is not an exception in this regime having a very rich flooding record since the ancient times. During the last decade efforts to mitigate flood risk have been enhanced with new practices in Greece and across Europe in scientific as well as in civil protection terms. M. Diakakis (&)
S. Mavroulis
G. Deligiannakis Faculty of Geology and Geoenvironment, School of Sciences, National and Kapodistrian University of Athens, Panepistimioupoli, 15784 Athens, Greece e-mail: diakakism@geol.uoa.gr 123 Nat Hazards Moreover, the legal framework has been enriched with new legal binding instruments (e.g. European Commission Directive 2007/60). In Greece, regular recording of flood events by civil protection agencies started relatively recently, limiting the systematic official records to the last two decades. On the other hand, regional authorities, damage compensation organizations and the press documented disasters in official reports or in anecdotal form, maintaining archives with an extensive amount of data that were not systematically evaluated as a whole until now. Previous works suggested and demonstrated that examination of flood history is an important part in flood hazard assessment (Potter 1978; Bayliss and Reed 2001; Benito et al. 2004; Salvatti et al. 2009; Diakakis 2010; Diakakis et al. 2011a). In this context and given the scarcity of instrumental and systematic hydrological data in Greece, flood history examination was considered, in this work, a useful tool to assess preliminarily flood hazard spatial distribution. In recent years, studies that use qualitative information to recreate natural disaster history are becoming more frequent, especially in cases were scientific data and descriptions are not available. Especially in the Mediterranean region this approach has been applied in various types of disasters (e.g. floods, landslides) showing promising results. Successful examples include applications in the field of flood history reconstruction (Seidel et al. 2009), flood and landslide risk management (Guzzetti and Tonelli 2004) and study of disaster triggering phenomena (Cuesta et al. 1999). The use of press archives as a source of information has expanded even to international risk-mitigation organizations (Tschoegl et al. 2006) that extract data and subsequently develop databases to reconstruct and study disaster event time series. Goal of this paper is to exploit the possible benefits, in terms of civil protection strategy and operations, of the examination of a historical flood record in Greece. Therefore, in this work, an extensive database of flood events had to be developed by examining a broad number of reports, documents and articles stored in press databases, in civil protection agencies records and in disaster databases. Based on this catalogue we subsequently examined: • The spatial distribution of flood events and the associated casualties during the period between 1880 and 2010 • The temporal distribution of events and fatalities caused by flooding in the same period • The seasonality of flooding in Greece as a whole and possible differences in seasonality among different regions • The possible presence of any trends or patterns in the occurrence of flooding phenomena in time and space that can be useful to further understand flooding and protect lives and properties. This work was carried out within the context of enhancing the efforts towards the completion of preliminary flood risk assessment in Greece and all EU member states, under the European legal framework. It also aims in improving operational and strategic decision support in terms of civil protection from flooding across Greece. 2 Setting The Balkan Peninsula has experienced a complicated geotectonic evolution during the alpine orogenic cycle, leading to the development of a variety of physiographic elements in Greece (Fig. 1). 123 Nat Hazards Fig. 1 Main physiographic elements of Greece A large part of the country is mountainous with the most urban development taking place in lowlands. Population is clustering mainly in coastal regions and on the islands, in areas in which during the summer, there is a noteworthy population increase contributing to augmented flood vulnerability. Population is concentrated in urban centers with Greater Athens (*4 million), Thessaloniki (*1 million), Patras (*230,000) and Herakleion (*140,000) being the host of a large portion of the total population (*11 million). A large part of the drainage network of the country consists of ephemeral mountain torrents and small to medium size drainage basins with limited amount of discharge for most of the year. In Mediterranean most of the floods are caused by intense rainfall in a short time frame (Martini and Loat 2007) making flash flooding the most common type of inundation. On the contrary to the central European rivers the lack of large river networks and regional rains make regional flooding virtually absent. Instrumental hydrological data are scarce as most of the catchments are ungauged making the historical records valuable in flood studies. 123 Nat Hazards 3 Data and Methods 3.1 Data A large number of reports and documents were collected from several sources, such as: • Flood event databases from state civil protection agencies • • • Official flood damage delineation declarations from the Earthquake Rehabilitation Center (2010) Flood event delimitation from the General Secretariat of Civil Protection (GSCP 2007) Database of technical reports accompanying flood damages from Prefectural Administration of East Attica (2007) • International research projects • • FLASH (Observations, Analysis and Modeling of Lightning Activity in Thunderstorms, for use in Short Term Forecasting of Flash Floods) HYDRATE (Hydrometeorological data resources and technologies for effective flash flood forecasting, Bain et al. 2008) • The press, with the following article-databases • • the Digital Newspapers Collection of the Greek National Library (Digital Newspapers Collection 2010) the Greek National Newspapers Archive (2010) of the Library of the Hellenic Parliament (microfilm database) In total 22 national and 4 local newspapers were researched extracting 457 articles on flood events • Scientific articles containing catalogues or information of flood events in Greece (Gaume et al. 2009; Llasat et al. 2010) or regions of it: Koutroulis et al. (2010) for Crete, Diakakis (2010) for eastern Attica, Diakakis et al. (2011a) for Peloponnese and Diakakis et al. (2011b) for Athens. • The EM-DAT International disasters database (CRED 2010). The types of records that were analyzed to extract information were: • Catalogues of flood-damaged structures and damage compensation catalogues with location and time information • News items (articles and headlines) describing the disaster containing specific information on casualties, location and time of the events. • Printed catalogues of state civil protection agencies with information on the location and time of events, evidence on the river networks that overflowed • Flood event databases in scientific articles, flood risk studies and international organizations with information on location, flooded river, casualties and time of the disaster. Although some of the sources contained information on some of the physical properties of the events, it was not possible to obtain systematic and accurate data on flood characteristics and therefore no particular analysis was carried out on this matter. 123 Nat Hazards 3.2 Approach In order to create a systematic record, a database was developed with detailed data on each flood event. Information was standardized to match the database’s requirements. Then data were stored with information on: a. b. c. d. e. f. g. location time and date flooded river flooded villages, towns or settlements, casualties damage effects of each event. sources of information In total 545 entries were found and stored in this inventory. Subsequently, simple checks such as cross-referencing between different sources of information were carried out to assure quality control of the record and standardization of information. The development of this database improved the analysis capabilities of the data, and made them easier to retrieve and process. It also highlighted possible lack of information and presented the record in a more systematic form in a way that it could be evaluated for the country as a whole and for specific regions in the same time. The database allowed also easy quantitative analysis of the information. Subsequently, simple numerical calculations were used for quantitative analysis of the data (i.e. calculation of percentages, sums and subsets). On the basis of this process, several conclusions were drawn, concerning the spatial and temporal distribution of events and casualties. In addition, flood locations were plotted on map in GIS environment by connecting the developed catalogue with GIS databases. Density calculations were carried out using GIS tools to assess the spatial density of events across the country. Furthermore, seasonality analysis was performed for the country as a whole and for separate geographic entities, in order to examine possible differences between the various regions. 3.3 Data accuracy and uncertainties Accuracy of the data in terms of location, timing and casualties was considered satisfactory as multiple checks between different sources were carried out to ensure consistency of the data. Uncertainties inherent in historical information were treated using cross-referencing between different sources. Approximately 89% of the events and the details associated (mentioned above) were cross-checked with two or more sources. 30% of the events were cross-referenced in more than two sources. Only 11% of the events were based on a single source. As far as the definition of flood is concerned, flood events were considered the destructive incidents associated with overflowing of specific parts of the drainage network presenting liquid flow and covering temporarily land outside the limits of the river banks, which is not normally covered. Additionally, minimum prerequisite for a flood event to be included in the database were accounts of damages of above ground structures, or agricultural land or reports of casualties or injuries. For the purposes of this work, flash flood and riverine flood events were examined as no cases of regional floods were found. Debris 123 Nat Hazards flows, floods connected with urban drainage systems failures were not included in this catalogue. Point features were used to determine and illustrate flood locations. This method already used successfully by Guzzetti and Tonelli (2004) and Salvatti et al. (2009) was considered adequate for representing the spatial distribution of events in a country scale, considering the absence of regional floods and large river networks and the lack of information to accurately delineate the exact flood limits of each inundation incident in the catalogue. In terms of completeness of the record, in the first decades of the period under study and during the time of the First and the Second World Wars, only the most important events are reported. This is attributed to poor recording capabilities on behalf of authorities and the press and higher tolerance thresholds with respect to disastrous phenomena on behalf of the society. However, casualties are reported sufficiently enough to provide a reliable representation of their fluctuation over time. In addition, there is uncertainty associated with the evolution of population. Areas that were not inhabited at the beginning of the period under study were developed into villages, small towns or even urban areas at the end of this period. This phenomenon lead official descriptions and accounts to report flood events mainly when properties were damaged, people or animals were in peril and generally when population was established near the flooded areas. Very intense or noteworthy events are an exception to this rule. These two issues lead to possible lack of data on flood occurrence during the first decades at the beginning of the considered period. Another source of uncertainty is the subjective nature of historical information depending on the different perceptions of the observers. This problem concerned mainly the magnitude and the effects of an event, factors that were not studied in this paper. When the issue was extending to the time, the location of the incident then it was treated by cross referencing documentary information with scientific sources. Finally, there is a flood event in Trikala (Thessaly, central Greece) that occurred in 1907 for which it was not possible to determine with accuracy the number of casualties. In this work, the view of official reports that agreed to the minimum of 300 casualties was adopted. However, there is anecdotal evidence that there may be more fatalities associated with this event. Undoubtedly, due to the large amount of deaths, this event affected the statistical analysis and created difficulties in the evaluation of the results mainly concerning seasonality and spatial distribution of fatalities. 4 Results Statistical processing of the data showed 686 casualties resulting from 545 flood events in total during the period 1880–2010 including 3 events in 1715, 1805 and 1833. The most destructive event in terms of human casualties was undoubtedly the 1907 incident in Trikala (Thessaly, northern Greece) that caused at least 300 fatalities. Analysis of the database showed that most of the flood events took place during autumn (Fig. 2). As far as monthly distribution is concerned the analysis showed that November is the month with the richest flood record (Fig. 3). Flood casualties occurrence does not follow the events’ monthly distribution (Fig 4). It is important to note that substantial differences appear in the casualties/events ratio amongst January and November, the months with the richest flood record. November 123 Nat Hazards Fig. 2 Seasonal distribution of events. SON stands for September, October, November), DJF stands for December, January, February, MAM stands for March, April, May, JJA stands for June, July, August Fig. 3 Distribution of flood events in months expressed as percentage of total 545 events. November and January present the highest proportion with 23.8% and 17.9% of the total events. October, September and December follow with 17.3%, 9% and 8.8% respectively records 112 events and 234 casualties (approximately 2.1 deaths per event), whereas January records 97 events and 21 casualties (approximately 0.2 fatalities per event). Seasonal distribution of events was also examined in various administrational entities in respect with their geographical location. The only pattern noted was observed between eastern and western parts of the country. Studying seasonality on third level administrational units (called ‘‘regions’’), the analysis showed that there is a small difference in the season floods are occurring between eastern and western regions (Fig. 5). In the western and the central part of the country (brown in Fig. 5) maximum frequency of floods occur in November (27.4% of the total 409 events). On the other hand, in easternmost part of the country, maximum frequency occurs in January (31.6% of the total 133 events). 123 Nat Hazards Fig. 4 Distribution of flood casualties in months expressed as percentage of total (686). The highest percentage appears in June due to the 1907 flood in the city of Trikala which caused 300 deaths accounting about 43.7% of the total. June is followed by November (34.1%), October (10.2%) and January (3.1%) As far as the temporal distribution of flooding is concerned, events and casualties were grouped in five-year periods between 1881 and 2010, in order to examine their 5-year occurrence variations (Figs. 6, 7). By examining the period between 1951 and 2010 where the record is considered more complete in terms of events and reported casualties, and therefore more reliable, one can identify a significant increase in reported flood event reports. On the other hand casualties remain roughly at the same level, and do not present a respective increase (Fig. 8). The spatial distribution of flood events was examined primarily on the basis of second level administrational entities (‘‘Prefectures’’) and expressed as the number of events per unit (Fig. 9). The number has been subsequently normalized by incorporating the area of each prefecture in the calculations, expressing the distribution as the number of events per 100 km2 for each unit (lower left corner in Fig. 9). Results in this case show an increased clustering of events in entities with extended urban environments like Attiki (city of Athens), Thessaloniki (city of Thessaloniki) and Larissa prefectures. However, increased flood occurrence is also recorded in Laconia (south Greece) where there are no large cities. Flood casualties follow approximately the same pattern, clustering mainly in urban environments (Fig. 10). Their numbers have been normalized by incorporating the population of each prefecture in the calculations, expressing the distribution as the number of casualties per 100.000 individuals for each unit (lower left corner in Fig. 10). Population data were based on 2001 census (ELSTAT 2001). Finally, flood events were plotted on the map to illustrate spatial patterns within the administrational units and the country as a whole (Fig. 11). Based on this projection spatial density was calculated across Greece (Fig. 12). Results showed that increased frequency of flood events is noted mainly along coastal areas and close to cities, with Athens, Thessaloniki, Patras and Larissa presenting a rich flooding inventory. Moreover, there is high density in east Thrace near Evros River and the borders with Turkey, and across Pineios River in Thessaly. In addition, it is evident, that there is a relatively reduced occurrence of flooding phenomena in the mountainous central 123 Nat Hazards Fig. 5 Seasonality difference between the easternmost and the central and western regions of Greece part of the country and the central part of Peloponnese. It is also noteworthy that the south parts of the island of Euboea and Crete present substantially lower density of events than the central and the north part respectively. 5 Discussion and conclusions Analysis of the database shows that there are seasonality patterns in flooding with autumn being the season with the most events while November and January are presenting the 123 Nat Hazards Fig. 6 Temporal distribution of flood events between 1881 and 2010 Fig. 7 Temporal distribution of flood casualties between 1881 and 2010 richest flood record. A conclusion, that is not fully in accordance with the conclusions of Stathis (2004) who suggested that November and December are the most flood prone months of the year in Greece. However, it is noteworthy that there is a significant difference in casualties per event between January and November. This difference can be attributed to a number of reasons such as differences in storm characteristics or in magnitude of flooding phenomena. There is also a possibility to be related to the level of preparedness and the risk appreciation of communities and individuals during autumn which could be reduced comparing to the winter. Furthermore, analysis showed that there is substantial difference in seasonality between the eastern and western regions of the country, indicating a possible pattern in storm development. Distribution of the events in months shows a primary indication of seasonality of flooding in the area and certainly can provide a preliminary guidance for disaster 123 Nat Hazards Fig. 8 Evolution of flood events against flood casualties during the period 1951–2010. Even though the number of reported flood events per decade increases, there is no significant change in casualties preparedness and operational decision making for civil protection agencies. The analysis shows autumn as the season with more frequent flood occurrence, a result which comes in accordance with the findings of Maheras et al. (2004), Barnolas and Llasat (2007) and Llasat et al. (2010) for the region. Especially the difference observed between eastern and western regions (Fig. 5) is important and has to be taken into consideration in regional preparedness planning. Concerning the temporal distribution of floods (illustrated in Figs. 6, 8) the record presents a significant increase during the last decades. This cannot be considered by any means an accurate representation of a natural trend or a trend in natural processes (i.e. climate change). The increase is attributed to the following factors: • The increase of population, leading to augmented pressure for urban expansion, sometimes in unacceptable locations increasing in turn the number of individuals and properties at risk. • The enhancement of means of reporting and recording disasters through the years (advances in IT technology and media). It is also important that during specific periods such as 1941–1945 (Second World War), poor reporting capabilities and lack of means prevented the community from recording sufficiently flood events. • The increased social and media interest in climate related catastrophes in the last decades and the lower tolerance threshold of the society with respect to natural hazards which lead to reporting of events of smaller significance. A factor acknowledged from other researchers too (e.g. Llasat et al. 2009). • The increased human interference in hydrological processes, through the expansion of public works, road networks and impervious surfaces especially near the cities. However, due to the fact that reporting of floods is related with the damages inflicted, the increase of events is a measure of increase in damages and properties at risk, indicating an increase of flooding interference with human activities. This fact suggests that there is a deteriorating trend in flooding problem in Greece and a need for improvement of the current land use planning. 123 Nat Hazards Fig. 9 Distribution of floods across Greece expressed as the number of events per administrational unit for the period 1880–2010. In the lower left corner distribution of events is expressed as the number of events per 100 km2 in each of these units Casualties’ temporal distribution, taking into consideration the number of events, shows that there is a slight decreasing trend in victims per event, indicating that events with large number of casualties are reduced. However, the relatively stable number of victims especially during the last decades supports the conclusion that the status of threat to human life is not improving although many actions were carried out in this direction. Thus, it is imperative to enhance preparedness and protection from floods. Concerning the spatial distribution of events and casualties as expressed in numbers per administrational unit (Figs. 9, 10) and in spatial density (Figs. 11, 12), it is shown that certain areas present increased flooding problems, such as: Attiki and Greater Athens area, Thessaloniki, Patras, south Peloponnese (Messinia and Laconia), the east part of Evros and central Greece (mainly the prefectures of Larissa, Trikala, Magnesia and Karditsa). There is a certain amount of clustering of flooding and casualties in urban environments. In addition, sparsely populated areas, such as, the central mountainous part, (Western 123 Nat Hazards Fig. 10 Distribution of floods across Greece expressed as the number of casualties per administrational unit for the period 1880–2010. In the lower left corner distribution of fatalities is expressed as the number of casualties per 100.000 individuals based on the 2001 population census (ELSTAT 2001) Macedonia, Epirus and Central Greece and central Peloponnese) and the islands record lower measures of victims and flooding phenomena. The variety of climatic and physiographic characteristics of the above flood-rich urban areas indicates that the causes of this clustering of flood events are not environmental. On the contrary the higher frequency of reported flood events should be attributed to the increased vulnerability of an urban environment in comparison with a rural one, the increased capabilities of local media to report localized flood events and possibly a lower disaster tolerance threshold on behalf of the society. The use of documentary data in conjunction with scientific flood analyses and technical reports is an approach that has demonstrated significant potential (Benito et al. 2004, Guzzetti and Tonelli 2004, Llasat et al. 2009). However there are several inherent limitations that arise from the use of documentary, non-scientific evidence. One of these is the discontinuous nature of the records formed by such data. For example, a single event that 123 Nat Hazards Fig. 11 Locations of 545 flood events across Greece in the period 1880–2010, including 3 events in 1715, 1805 and 1833 occurred in the past in a remote community could have failed to be reported due to lack of means or individuals to record it. In addition, there is the risk that the community could lose the interest or the means to report disastrous phenomena during certain periods (e.g. during Second World War as shown in Fig. 6). In addition, the subjective nature of historical information and the fluctuation of social tolerance in disastrous phenomena introduce an uncertainty connected with the number and the magnitude of events reported. In this work this factor is clearly seen in the increasing trend in the temporal distribution of reported events through the decades. Overall, examination of flooding history through a database of flood events proved to be a useful tool in understanding the occurrence of floods in Greece, in terms of seasonality, temporal trends and spatial distribution. Although flood risk assessment is a basin-specific issue, flood history information in a country-scale can be used to highlight priorities and identify higher risk areas. In this way, it can support decision making in the field of civil protection and operational preparedness (e.g. allocation of emergency response units), 123 Nat Hazards Fig. 12 Spatial density of floods in events per square kilometer across Greece between 1880–2010 improve information available to the general public and enhance the efforts towards detailed flood risk mapping under the European legal framework. References Bain V, Gaume E, Bernardara P (2008) Hydrate Project: Collation of Primary Flash Flood Data. HYDRATE WP 1, Available at: http://www.hydrate.tesaf.unipd.it/WareHouse/ProjectDeliverables/ DELIVERABLE_HYDRATE_WP1.pdf. Accessed 10 October 2011 Barnolas M, Llasat MC (2007) A flood geodatabase and its climatological applications: the case of Catalonia for the last century. Nat Hazards Earth Syst Sci 7:271–281 Bayliss AC, Reed DW (2001) The use of historical data in flood frequency estimation. Report to MAFF, Centre of Ecology and Hydrology, Wallinford, 87p. Available at: http://nora.nerc.ac.uk/8060/1/ BaylissRepN008060CR.pdf. Accessed 9 January 2011 Benito G, Lang M, Barriendos M, Llasat MC, Francés F, Ouarda TBMJ, Thorndycraft VR, Enzel Y, Bárdossy A, Coeur D, Bobée B (2004) Use of systematic, palaeoflood and historical data for the improvement of flood risk estimation, review of scientific methods. Nat Hazards 31(3):623–643 123 Nat Hazards CRED (2010) Summarized Table of Natural Disasters in Greece from 1900 to 2011. Internation Disaster Database. Centre for Research on the Epidemiology of Disasters (CRED). Available at: http:// www.emdat.be/result-country-profile. Université catholique de Louvain—Brussels—Belgium Cuesta MJD, Sanchez MJ, Garcia AR (1999) Press archives as temporal records of landslides in the North of Spain: relationships between rainfall and instability slope events. Geomorphology 30:125–132 Diakakis (2010) Flood history analysis and its contribution to flood hazard assessment: the case of Marathonas, Greece. Bull Geol Soc Greece 43:1323–1334 Available at: http://www.geosociety.gr/ GSG_XLIII_3.pdf. Accessed 9 January 2011 Diakakis M, Deligiannakis G, Mavroulis S (2011a) Flooding in Peloponnese, Greece: A contribution to flood hazard assessment. In: Lambrakis N, Stournaras G, Katsanou K (eds) Advances in the Research of Aquatic Environment. 1:199–206, Springer Diakakis M, Foumelis M, Gouliotis G, Lekkas E (2011b) Preliminary flood hazard and risk assessment in Western Athens metropolitan area. In: Lambrakis N, Stournaras G, Katsanou K (eds) Advances in the Research of Aquatic Environment. 1:147–154, Springer Digital Newspapers Collection (2010) E-efimeris: digital newspapers collection of the Greek national library. Available at: http://www.nlg.gr/digitalnewspapers/ns/main.html. Accessed 10 January 2011 Earthquake Rehabilitation Center (2010) Disaster delimitations, Earthquake Rehabilitation Center (YAS), Ministry of Public Works Available at: http://www.yas.gr. Accessed 10 October 2010 ELSTAT (2001) Population Census 2001: De facto population, Area and population by range, density and elevation zones. Hellenic Statistical Authority. Available at: http://www.statistics.gr/portal/page/portal/ ESYE. Accessed 10 January 2011. http://flood_risk/flood_atlas. Accessed on 13 April 2010 Gaume E, Bain V, Bernardara P, Newinger O, Barbuc M, Bateman A, Blaskovicova L, Bloschl G, Borga M, Dumitrescu A, Daliakopoulos I, Garcia J, Irimescu A, Kohnova S, Koutroulis A, Marchi L, Matreata S, Medina V, Preciso E, Sempere-Torres D, Stancalie G, Szolgay J, Tsanis I, Velasco D, Viglione A (2009) A compilation of data on European flash floods. J Hydrol 367(1–2):70–78 Greek National Newspapers Archive (2010) National Newspapers Microfilm Database of the Library of the Hellenic Parliament, Hellenic Parliament, Greece GSCP (2007) Flood-prone areas of Attica. General Secretariat of Civil Protection. Ministry of Interior, Greece Guzzetti F, Tonelli G (2004) Information system on hydrological and geomorphological catastrophes in Italy (SICI): a tool for managing landslide and flood hazards. Nat Hazards Earth Syst Sci 4:213–232 Koutroulis AG, Tsanis IK, Daliakopoulos IN (2010) Seasonality of floods and their hydrometeorologic characteristics in the island of Crete. J Hydrol. doi:10.1016/j.jhydrol.2010.04.025 Llasat MC, Llasat-Botija M, Lopez L (2009) A press database on natural risks and its application in the study of floods in Northeastern Spain. Nat Hazards Earth Syst Sci 9:2049–2061 Llasat MC, Llasat-Botija M, Prat MA, Porcu F, Price C, Mugnai A, Lagouvardos K, Kotroni V, Katsanos D, Michaelides S, Yair Y, Savvidou K, Nicolaides K (2010) High-impact floods and flash floods in Mediterranean countries: the FLASH preliminary database. Adv Geosci 23:47–55 Maheras P, Tolika K, Anagnostopoulou C, Vafiadis M, Patrikas I, Flocas H (2004) On the relationships between circulation types and changes in rainfall variability in Greece. Int J Climatol 24:1695–1712 Martini F, Loat R (2007) Handbook on good practices for flood mapping in Europe. European exchange circle on flood mapping (EXCIMAP), Paris. http://ec.europa.eu/environment/water/ Potter HR (1978) The use of historic records for the augmentation of hydrological data, Institute of Hydrology Report No.46 Wallingford 59p. Available at: http://nora.nerc.ac.uk/5773/1/IH_046.pdf. Accessed 5 January 2011 Prefectural Administration of East Attica (2007) Technical Services Archive, Natural Disaster Damages (1987–2005). Technical Services Department, Prefectural Administration of East Attica Salvatti P, Bianchi C, Rossi M, Guzzetti F (2009) Societal landslide and flood risk in Italy. Nat Hazards Earth Syst Sci 10:465–483 Seidel J, Imbery F, Dostal P, Sudhaus D, Burger K (2009) Potential of historical meteorological and hydrological data for the reconstruction of historical flood events – the example of the 1882 flood in southwest Germany. Nat Hazards Earth Syst Sci 9:175–183 Stathis D (2004) Extreme rainfall events and flood genesis in Greece. Proceedings of the 7th Panhellenic Geografical Conference of the Hellenic Geographical Society (7PGC/HGS), Mytilene, Greece, 1–8, 2004. Available at: http://www.srcosmos.gr/srcosmos/showpub.aspx?aa=5991. Accessed 10 January 2011 Tschoegl L, Below R, Guha-Sapir D (2006) An analytical review of selected data sets on natural disasters and impacts. UNDP/CRED Workshop on Improving Compilation of Reliable Data on Disaster Occurrence and Impact, Centre for Research on the Epidemiology of Disasters, Université catholique de Louvain, Brussels, Belgium, March 2006 123