A flood risk decision making approach for Mediterranean tree crops using GIS; climate change effects and flood-tolerant species

A flood risk decision making approach for Mediterranean tree crops using GIS; climate change effects and flood-tolerant species

Environmental Science & Policy 63 (2016) 132–142 Contents lists available at ScienceDirect Environmental Science & Policy journal homepage: www.else...

6MB Sizes 2 Downloads 108 Views

Environmental Science & Policy 63 (2016) 132–142

Contents lists available at ScienceDirect

Environmental Science & Policy journal homepage: www.elsevier.com/locate/envsci

A flood risk decision making approach for Mediterranean tree crops using GIS; climate change effects and flood-tolerant species Nektarios N. Kourgialasa,b,* , George P. Karatzasa a

School of Environmental Engineering—Technical University of Crete, Polytechneioupolis, 73100 Chania, Greece Hellenic Agricultural Organization—DIMITRA, National Agricultural Research Foundation (N.AG.RE.F.)—Institute for Olive Tree, Subtropical Crops and Viticulture, Agrokipio, 73100 Chania, Greece b

A R T I C L E I N F O

Article history: Received 16 March 2016 Received in revised form 27 May 2016 Accepted 27 May 2016 Available online xxx Keywords: Flood risk assessment Crete GIS Fruit tree areas Climate change Flood-tolerant fruit trees

A B S T R A C T

The aim of this study is to estimate flood risk in the Mediterranean island of Crete in Greece, using Geographic Information Systems (GIS). The island of Crete, covering an area of 8265 km2, is one of the most intensively Mediterranean agricultural areas dominated by fruit tree crops. In this study, the factors that are directly related to the creation of a flood are combined in a GIS environment in order to identify the most prone flooding areas. These factors are: (a) the Flow accumulation (F), (b) the Rainfall intensity (R), (c) the Elevation (E), (d) the Geology (G), (e) the Land use (L), and (f) the Slope (S). The initials of these factors gave the name to the proposed method: “FREGLS”. The above factors are presented in the form of grid maps and are used in order to determine the final flood risk map. Subsequently, the flood prone fruit tree areas of Crete can be estimated by applying a methodology based on weighting coefficients. The reliability of the final flood risk map is verified using historical flooded data. Additionally, the impact of global climate change scenarios A2 and B1, on flood risk in Crete is examined. Based on the above, this study highlights the flood prone fruit tree areas in the island of Crete under current and future climate conditions. Also, flood-tolerant fruit trees which appear to be economically important for Greece are recommended, especially for the high flood risk areas of the island. The proposed methodology can be applied as a decision making tool for flood risk mitigation to any river basin where tree crops are cultivated. ã 2016 Elsevier Ltd. All rights reserved.

1. Introduction Flooding has always been a real danger not only for lowland and semi-mountainous areas but also for mountainous plains (plateaus). Generally, most vulnerable to floods are low-lying areas and adjacent to rivers and streams. The risk of flooding in a particular area is mainly attributable to torrential nature of rainfall. Moreover, under the pressure of the population growth the continued expansion of building construction in the coastal zone, have engendered conditions of immediate cause of flooding (Thieken et al., 2007; Wang et al., 2011; Morelli et al., 2014). The main types of flooding that can occur in an area are: river floods, flash floods, urban floods, sewer flooding and coastal flooding. Especially, in small river basins with high slopes and low permeability geological formations intense rainfall can cause flash

* Corresponding author at: School of Environmental Engineering—Technical University of Crete, Polytechneioupolis, 73100 Chania, Greece. E-mail addresses: [email protected], [email protected] (N.N. Kourgialas). http://dx.doi.org/10.1016/j.envsci.2016.05.020 1462-9011/ã 2016 Elsevier Ltd. All rights reserved.

flood events, which due to their short-time flow concentration, are difficult to be predicted and treated (Camarasa Belmonte and Segura Beltran, 2001; Kourgialas et al., 2011). In addition, flash floods are very common in mountainous areas where rapid snowmelt or heavy rainfalls are quickly transformed into runoff. These events, due to the high capacity of transport, can be characterized as the most serious weather-related hazards in many mountain plateaus all around the world, causing considerable economic losses (Ballesteros et al., 2011). In the Mediterranean area, and especial in Crete, where small river basins and ephemeral streams are common, rainfall and runoff processes often lead to the generation of flash floods. These events happen suddenly with a minimal warning, causing significant economic damages. Specifically, flash flood waves move at very high speeds and could have a height from few centimeters to few meters (Archer, 1992). These characteristics of flood waves can destroy structures and tear out trees causing serious damages. These losses can be economically important especially in an intensive agricultural area such as the island of

N.N. Kourgialas, G.P. Karatzas / Environmental Science & Policy 63 (2016) 132–142

Crete, one of the most notable horticultural regions of the Mediterranean zone with significant areas of fruit trees. Floods in Crete are a relatively frequent phenomenon and it has become more intense over the last decades (Kourgialas et al., 2010). Various climate models for the region of Crete predict a gradual increase in rainfall intensity for the coming decades, which may increase the risk of flash floods (Nikolaidis et al., 2013; Kourgialas et al., 2015). Moreover, the strong orographic effect in Crete with a non-uniform spatial distribution of precipitation generates at specific regions of the island an intensive spatial variability of precipitation both in frequency and intensity. Taking into account the above, the generation of flood risk map for the island of Crete under current and future climate conditions certainly is a challenging task. In order to minimize the negative effects of floods, flood risk maps and mitigation plans must be developed, taking into account the geomorphological and hydrological conditions dominating in regions of potential flood risk (Liu et al., 2003; Chatterjee et al., 2008; Prime et al., 2016). Several models, presented in the past, can estimate potential flood risk areas for different scenarios using various numerical approaches to simulate flow and flood inundation (Brocca et al., 2011). However, for ungauged and/or extended areas where hydrological – hydraulic simulation is expensive or/ and time-consuming a multi-criteria analysis (MCA) in a GIS environment can be an essential tool. Floods are multidimensional phenomena with a spatial dimension, which makes GIS particularly useful in such applications (Zerger, 2002; Wang et al., 2011). Many MCA flood risk studies in GIS environment have been performed worldwide, especially in the last fifteen years (Liu et al., 2003; Van Der Veen and Logtmeijer, 2005; Kourgialas and Karatzas, 2011; Chau et al., 2013; Papaioannou et al., 2015; Kazakis et al., 2015) The combination of geoinformation techniques with multicriteria analysis (MCA) allows greater flexibility and accuracy in decisions with regard to the designation and the analysis of vulnerable to flooding areas (Makropoulos and Butler, 2006; Chenini et al., 2010). Specifically, in GIS databases analytical/ numerical tool and mathematical relationships among different layers of maps can be combined to a decision support system yielding a flood risk map. For an effective decision support system several criteria need to be evaluated which can be combined and create the final flood risk map. However, the main problem in MCA is that the factors as well as their weights are subject to the judgment of the decision-maker. In this study, in order to face this challenge we combined the literature with a Delphi approach and, historical flood events to validate the final results. Another aim of this study was the development of a flood risk map for the island of Crete taking into account all the particular hydrogeological characteristics of the island. Moreover, the flood prone fruit tree areas, based on current and future climate conditions, were determined. Specifically, the effects of the higher and lower emission climate change scenarios A2 and B1 in flood risk fruit tree areas were investigated. Also, this study highlights the tolerance to flooding of the existing fruit trees of Crete, suggesting new economically effective and flood tolerance fruit tree species that are recommended for high flood risk areas. The proposed methodology can be used as a flood decision making tool to any agricultural river basin. 2. Materials and methods 2.1. Study area Crete is the fifth largest island in the Mediterranean Sea and the largest in Greece, covering an area of about 8265 km2. The western part Crete includes Chania, and Rethymnon prefectures, while the

133

eastern includes the prefectures of Heraklion and Lassithi. The elevations in Crete range from zero to 2456 m MSL. In Crete lowlands (<200 m) cover an area of 2165 km2, 26% of the total area, semi-mountainous areas (200–800 m) cover an area of 4627 km2 which is 56% of the total area, while the mountainous area (>800 m) is approximately equally to 1473 km2 which is 18% of the total area. The study area is one of the most important horticultural regions of the Mediterranean zone; mainly, dominated by fruit tree cultivation such as olive, citrus, vineyard and avocado (Fig. 1a). Specifically, fruit tree cultivation in Crete reaches up to 32% (2608 km2) of the total area. Pasture, natural vegetation and forests cover 56% (4600 km2), annual crops cover about 11% (950 km2), while artificial structures (urban areas) are about 1% (107 km2). Also, Omalos and Lassithi plateaus are the largest mountainous plains of Crete, which are both cultivated with fruit trees and annual crops (Fig. 1a). In these areas as the agriculture has become more intensive a proper flood management plan is a challenge. The island of Crete is separated into 120 different river basins (Fig. 1b) and is characterized by a complex geological structure that consists of limestones, impermeable formations, and flysch formations (Water Resources Department of the Prefecture of Crete, 2014). Annual rainfall reaches up to 700 mm/year in low areas, 1000 mm/ year in semi-mountainous areas, and 2000 mm/year in mountainous areas. Due to the complex vertical and horizontal topography of the island significant climatologically differences occur. Thus, the western part of the island receives higher amounts of precipitation compared to the eastern part. Generally, about 65% of the total precipitation in the island is lost in form of evapotranspiration while 21% is converted to runoff and the rest of 14% recharges the groundwater. 2.2. Methodology In order to estimate spatial variability of flood risk in the island of Crete five different flood Risk Levels (RL) were considered (very high, high, moderate, low and very low). Specifically, the proposed RL are described as the probability of a flood event to occur within a hydrological year. Very high flood risk level indicates areas with 2% probability of a flood event to occur within a hydrological year (50year return period). High flood risk level indicates 1% probability (100-year return period). Moderate flood risk level indicates 0.5% probability (200-year return period). Low flood risk level indicates 0.2% probability (500-year return period) and very low flood risk level indicates lower than 0.2% probability (>500-year return period). This classification is performed by considering the factors that influence the generation of flood events. The final map of flood risk was created using the combination of six different thematic maps – factors, which are directly related to flooding events. This combination was achieved based on a linear algebraic function assigning relative weights to each factor. Analytically, estimating the final flood risk map, six different thematic map-factors such as Flow accumulation (F), Rainfall intensity (R), Elevation (E), Geology (G), Land use (L), and Slope (S) were created within ArcGIS environment. The (F), (S), (E), and (R) factors have numeric values, whereas the (G) and (L) factors are expressed in descriptive form. The effect of each factor is mapped as five different flood Risk Levels (RL) as following: very high, high, moderate, low and very low. For numeric-valued factors the Jenk’s Natural Breaks method was used for indicating the five different flood risk classes. In the case of nonnumeric-valued factors classification depends mainly on the influence of the factor on the recharging flood process. For example, impermeable geological areas as well as limited land cover indicates a very high flood risk for (G) and (L) factors, respectively. Next, for each of the proposed factors rating of the

134

N.N. Kourgialas, G.P. Karatzas / Environmental Science & Policy 63 (2016) 132–142

Fig. 1. (a) Land uses, rainfall stations, and plateaus in Crete, (b) River basins in Crete.

flood Risk Levels (RL) a numerical value is assigned as follows: very high (RL) = 10, high (RL) = 8, moderate (RL) = 5, low (RL) = 2, very low (RL) = 1. Considering that all the studied factors do not have the same degree of influence on flood generation, a correlation analysis with different weights for each factor was applied. This analysis considers the effect of each factor on all other factors. Thus, two kinds of effects are employed: (a) major effect, that is, a change of the first factor bears a direct effect on the other [assigned (1) point] and (b) a minor effect, that is, the change of the first factor bears an indirect effect on the other factor [assigned (1/2) point]. The rate for each factor is calculated as the summation of the points of major and minor effects. These rates are presented in Table 1. The selection of the above mentioned factors as well as the flood Risk Levels (RL) and the Factors Rates (FR) were based on the literature and the Delphi approach (Eimers et al., 2000; Yahaya et al., 2010; Kourgialas and Karatzas, 2011; Kazakis et al., 2015). Specifically, according to literature surveys the aforementioned six factors can capture the necessary information improving the decision-making utility for flood risk modelling (Zerger, 2002). Additionally, in complex hydrogeological regions/systems, such as the Mediterranean area, hydro-environmental researches based on expert judgments can improve the accurate estimation of numerical parameters and/or logical functions which in turn

capture the physical responses of those systems in modelling approaches (Pshenichny et al., 2009). Taking into consideration the above, Delphi process was used to quantify and collect the opinions of hydrogeologists, hydro-agronomists, and environmentalist experts, from different organizations on the island of Crete (Ministry of Agriculture, Technical University and Technological Educational Institute, Institute for Olive Tree – Subtropical Crops and Viticulture, Water Resources Department of the Prefecture of Crete, and Hellenic Agricultural Insurance Organization). Specifically, in this study the Delphi method, which can be characterized “as a method for structuring a group communication process so that the process is effective in allowing a group of individuals, as a whole, to deal with a complex problem” (Linstone and Turoff, 2002), enabled: (a) the confirmation that the six selected factors have the most significant influence on flood risk assessment, (b) the identification of the weights and rates of factors, taking into consideration the interaction between factors. The final percentage of each factor regarding its effect on the flood risk occurrence is computed as the ratio of the total factor weight (calculated by multiplied FR and RL) to the overall total weight (i.e. the sum of all the six factors weights). Following, according to the weighted linear combination approach, each factor is multiplied by its percentage weight and all factors are summed to yield the final flood risk map (FREGLS) for the island of

Table 1 Interaction between factors that influence the flood risk [Factors Rates (FR)]. Factor Changing

Major Effect

Minor Effect

Factor Rate (FR)

Flow accumulation (F) Slope (S) Land use (L) Rainfall intensity (R) Geology (G) Elevation (E)

(L) (F), (L) (F), (R) (F) (L), (S), (F) (R), (L), (G), (F)

(S)

1.5 pts (1major + 1minor) 2.0 pts (2major + 0minor) 3.0 pts (2major + 2minor) 1.5 pts (1major + 1minor) 3.0 pts (3major + 0minor) 4.5 pts (4major + 1minor)

(G), (S) (L) (S)

N.N. Kourgialas, G.P. Karatzas / Environmental Science & Policy 63 (2016) 132–142

Crete using the following equation (Kourgialas and Karatzas, 2011): FREGLS ¼

X xi wi ¼ FwF þ SwS þ LwL þ RwR þ GwG þ EwE ;

ð1Þ;

where: xi = the map of each parameter i wi = the weight of each parameter i F = Flow accumulation factor, S = Slope factor, L = Land use factor R = Rainfall intensity factor, G = Geology factor, E = Elevation factor. All the above mentioned factors were georeferenced to the Greek Coordinate System EGSA’87 (EPSG Projection 2100). Geometric resolution of the produced raster maps equal to 20  20 m2. Geoinformatic and filed measurement techniques were used to determine and digitalize the aforementioned thematic maps/factors. Analytically, topographic maps (1:5000) were used in order to create the Digital Elevation Model (DEM) – (spatial resolution of 20  20 m, vertical accuracy equal to 5 m) of the study area. Using DEM, in GIS environment, slope map was produced applying a 3D Analyst tool. Topographic parameters such as slope and elevation are inversely proportional to the appearance of floods (Kourgialas and Karatzas, 2011), (Table 2). In addition, remote sensing methods (satellite images) and the traditional field explorative geological mapping were used to determine the land use map and the geological map for the Island of Crete, respectively. Land use factor is associated directly with the vegetation cover that controls both the amount of precipitation and the time that it takes to reach the soil surface contributing to surface runoff. Thus, dense vegetation cover can reduce the flood risk, while burned areas, without vegetation, can lead to increased

135

surface runoff and greater risk to flooding (Kazakis et al., 2015), (Table 2). Geology formations through which the water flows can play a catalytic role in the runoff generation (Kourgialas et al., 2011). The larger the grain size and the more fractures (karst structure) the lower overland flow. On the other hand, impermanent geological formations can lead to high runoff rates increasing in turn the flood risk (Table 2). In addition, using the appropriate algorithm in GIS environment (flow accumulation – Arc Hydro) the flow accumulation map was generated (ESRI, 2008). This map indicates the number of cells that hydrologically contribute to each raster cell, where the higher the accumulation of runoff in a pixel-cell, the greater the flood risk (Table 2). In order to determine the Rainfall intensity (R) factor monthly rainfall values from 86 rainfall stations, located on the island of Crete and for the time period 1960–2014, were used (Water Resources Department of the Prefecture of Crete, 2014). The spatial distribution of these stations is shown in Fig. 1a. The map of Rainfall intensity (R) was created based on the Modified Fournier Index (Morgan, 2005): MFI ¼

12 2 X p ; P 1

ð2Þ

where: MFI: the Modified Fournier Index,

12 X

: the 12-month summa-

1

tion, p: the average monthly rainfall, and P: the average annual rainfall.

Table 2 Categorization—calibration and weight evaluation of the factors affecting flood risk areas on the island of Crete. Factors

Domain of effect

Descriptive level (flood risk Proposed weight of level) effect (RL)

Rate (FR)

Weighted rating (FR*RL)

Total weight

Percentage (%)

(F) Flow accumulation (pixels)

128.590–220.071 64.727–128.590 27.617–64.727 6.041–27.617 0–6.041 0–7.56 7.56–14.83 14.83–23.00 23.00–33.28 33.28–77.16 Urban & bare area Scrub, herbaceous, annual crops, Permanent crops, Fruit trees Agro-forestry areas, Pastures Mixed forest 554–838 377–554 244–377 151–244 0–151 Impermeable layer & flysch Clays Sands & fine grain content Gravels & conglomerates Karstic area & limestones 0–263 263–547 547–930 930–1.465 1.465–2.458

Very High High Moderate Low Very Low Very High High Moderate Low Very Low Very High High

10 8 5 2 1 10 8 5 2 1 10 8

1.5

15 12 7.5 3 1.5 20 16 10 4 2 30 24

39

9.68

52

12.90

78

19.35

Moderate

5

15

Low

2

6

39

9.68

78

19.35

117

29.03

403

100

(S) Slope (degree)

(L) Land use

(R) Rainfall intensity (MFI units)

(G) Geology

(E) Elevation (m)

SUM

Very Low Very High High Moderate Low Very Low Very High

1 10 8 5 2 1 10

High Moderate Low Very Low Very High High Moderate Low Very Low

8 5 2 1 10 8 5 2 1

2

3

1.5

3

4.5

3 15 12 7.5 3 1.5 30 24 15 6 3 45 36 22.5 9 4.5

136

N.N. Kourgialas, G.P. Karatzas / Environmental Science & Policy 63 (2016) 132–142

The MFI indicator expresses the sum of the average monthly rainfall intensity at a station. The MFI value of each station was interpolated at the GIS environment using the deterministic technique of spline method. This method was selected based on various studies which indicate that the spline method is the best method for representing smoothly varying surfaces of phenomena such as rainfall and the most appropriate method for cases with a small number of data points, as our case (Goovaerts, 2000; Lloyd, 2005). The monthly data used to estimate the rainfall intensity is not always applicable for flood risk analysis. Nevertheless, taking into consideration: (a) the large number of meteorological stations in the study area, as well as the extended time period of data- 55 years, (b) the seasonal character of the rainfall threshold level which can generate overbank flows in the Mediterranean region, where ephemeral streams and flash floods are common (Camarasa Belmonte and Segura Beltran, 2001), and (c) the fact that the proposed method does not require exact values for this factor but classifies it into five categories of risk, the monthly data could be

considered as an acceptable estimation method for rainfall intensity in the Mediterranean environment. In this study the climate change effect in flood prone areas for the island of Crete was investigated based on the global emission scenarios A2 and B1 for the year 2050. Choosing this short prediction time the uncertainty of the final results was minimized. Analytically, based on different Global Climate Models (Echam5, IPSL and CNRM) that were applied in the study area (Tsanis et al., 2011; Kourgialas et al., 2015), prediction of the average annual precipitation for the island of Crete indicate a small decrease of 2% for scenario A2 and 1% for scenario B1. In the present work, was chosen to simulate the higher and lower emission climate change scenarios A2 and B1 in order to capture their potential effects on flood risk areas. According to IPCC Third Assessment Report, these two extreme future climate change scenarios (A2 and B1), can be termed as following: A2 scenario indicates “a weakly globalized world with continuously increasing population. Economic development is primarily regionally oriented and per capita economic growth

Fig. 2. Flood risk maps for factors: (a) flow accumulation, (b) slope, and (c) land use.

N.N. Kourgialas, G.P. Karatzas / Environmental Science & Policy 63 (2016) 132–142

and technological change more fragmented and slower”. B1 scenario indicates “a confluent world that promotes sustainable development on a global scale, rapidly shifting towards a service and information economy based on clean and resource-efficient technologies”. These scenarios were considered to affect rainfall intensity uniformly over the island. Daily synthetic data for climate change scenarios A2 and B1 were generated using perturbed statistical characteristics of the observed data at each rainfall station. Next, in the GIS environment the maps of Rainfall Intensity (R) factor under climate change scenarios A2 and B1 for the year 2050 were generated. These maps, in turn, were combined with Eq. (1) in order to determine the flood risk areas under these future climatic conditions. Emphasis was given to the determination of the fruit tree areas located at very high and high flood risk level. Taking into consideration that these cultivations offer significant economic benefits to the local communities it is very crucial to investigate the tolerance of these species to flooding. Also it is very important to recommend flood-

137

tolerate and economically important fruit trees suitable for cultivation in the very high flood risk areas of Crete. 3. Results and discussion Based on this final map the fruit tree areas at high flood risk can be determined. Table 2 presents the classification of the six factors into risk levels, the proposed rating for each risk level, expressed in points, as well as the rates for the factors. Additionally, the summation of all the factor weights yields the grand total weight and the contribution of each factor to the flood risk, expressed as a percentage (Table 2). The six maps that were developed with the classification method are presented in Figs. 2 and 3. Using Eq. (1), the final map of flood risk areas is generated by combining the above mentioned factors (Fig. 4). The combination of the six maps (flow accumulation, slope, land use, rainfall intensity, geology, and elevation) applied in the proposed methodology aims to identify the areas that are most prone to flooding on the island of Crete.

Fig. 3. Flood risk maps for factors: (d) rainfall intensity, (e) geology, and (f) elevation.

138

N.N. Kourgialas, G.P. Karatzas / Environmental Science & Policy 63 (2016) 132–142

Fig. 4. Final flood risk map for the island of Crete.

Consequently, Fig. 4 describes spatial distribution of flood risk in each of the 120 river basins of the island of Crete. Based on these figures, very high flood risk areas are distributed as much in the western part of the island as in the eastern part. Additionally, flood occurrence appears to be more intense in the lowlands of the island. A characteristic example is the southern part of the prefecture of Heraklion, where the final map describes an extended very high flood risk area (Fig. 4). Specifically, this area (valley of Messara) is an irrigated area that is cultivated intensively where groundwater over-exploitation has led to a land subsidence which in fact increases significantly the flood risk level (Xepapadeas, 1996; Paleologos et al., 2013). 3.1. Validation and final results In order to verify the final flood risk map, a review of historical flooding on the island of Crete (for the last 100 years) was introduced (Diakakis et al., 2012; Water Resources Department of the Prefecture of Crete, 2014). Specifically, in Fig. 5a and b the locations of historical flooded points, occurred within a 50-year return period, are marked with black symbols, while historical flooded points occurred within a 100-year return period are marked with green symbols. According to these figures, the majority of the historical flooded points are located in the western part of Crete in comparison with the eastern part. Additionally, accordingly of the frequency of the total 125 recorded historical flooded points 114 have 50-year return period, while the rest of them have a 100-year return period. To validate the final results it was necessary to convey the connection between historical flooded points and flood risk areas. To achieve this, the “Extract by Mask” tool in GIS environment was applied. Based on this tool, the cells of the final flood risk map (raster file) that correspond to the historical flooded points (mask data) were extracted. In this way, the extracted flooded points were identified based on their locations in the five different flood risk zones. Thus, the accurate number of the flooded points that were located in each flood risk area was defined. Based on the above, for the final flood risk map, almost all the recorded historical floods with a 50-year return period (111 points) took place at the very high flood risk areas with only three of them to be in the areas of high flood risk. On the other hand, all the recorded historical floods with a 100-year return period (11 points) were located within the identified areas of high flood risk. The good agreement between historical flood events and the final flood risk map appears to be not only at the low and semimountainous lands but also at the two main intensively cultivated mountain plateaus of Crete, Omalos and Lassithi (Fig. 5a and b). In this way, the uniqueness of the proposed parameterization was

ensured, minimizing the uncertainty inherent in the simulation results. All of the aforementioned information validate spatial reliability of the proposed methodology for the study area. According to the verified final flood risk map, an area of 1605 km2 (19.5%) in Crete is subject to very high flood risk and 2330 km2 (28.3%) is under high flood risk. In addition, 1915 km2 (23.2%), 1463 km2 (17.7%) and 931 km2 (11.3%) can be characterized as areas of moderate, low and very low flood risk, respectively. As mentioned above, the island of Crete is one of the most important tree cultivation regions in the Mediterranean zone. Based on the final flood risk map, 792 km2 (30.4%) of fruit tree areas are located in very high flood risk areas. Furthermore, fruit tree areas of about 1131 km2 (43.4%) are under high flood risk. Additionally, 270 km2 (10.3%), and 415 km2 (15.9%) of the fruit tree areas can be designated as areas of moderate, low or very low flood risk, respectively. Thus, a significant percentage of 73.8% of the fruit tree cultivation areas in Crete are under very high or high flood risk. The above classification results and the flood risk maps (including map coordinates) can be exported by local authorities using GIS. Thus, maps of high flood risk areas with summary information could be offered to the farmers to raise awareness. Moreover, using the proposed methodology, at very high flood risk areas of 792 Km2 economical effective and flood-tolerant fruit trees species can be investigated and suggested. This task is analyzed in a following section. 3.2. Climate change scenarios For predicting flood prone areas in 2050 under socioeconomic scenarios A2 and B1 all the factors of Eq. (1) except the rainfall intensity (R) were kept constant. Fig. 6 illustrates the rainfall intensity (R) results for the year 2050 under the scenarios A2 and B1. For scenario A2 the predicted flood risk results show that in 2050 the decrease of the expected precipitation gave a reduced area of very high and low flood risk, while high, moderate, and very low flood risk areas show a small increase, compared to the current conditions (Table 3). For scenario B1 the predicted flood risk results show that in 2050 the slight increase of the expected precipitation gave an increase in all flood risk levels except the high and moderate levels with comparison to the current conditions (Table 3), (Fig. 7). In addition, regarding scenario A2 (higher emission climate change scenario), fruit trees within the very high flood risk areas are expected to decrease for the year 2050 as compared to the current situation. Nevertheless, for 2050 and for scenario A2 fruit trees within high flood risk areas are expected to increase as compared to the present (Table 4). For scenario B1, very high flood risk fruit tree areas will increase in the year 2050, while

N.N. Kourgialas, G.P. Karatzas / Environmental Science & Policy 63 (2016) 132–142

139

Fig. 5. (a) Validation of the final flood risk map with historical flooded points (Western part of Crete) (b) Validation of the final flood risk map with historical flooded points (Eastern part of Crete).

high flood risk fruit tree areas will significantly decrease, compared to the baseline conditions (Table 4). 3.3. Flood-tolerant fruit trees, current species and prospects for the island of Crete In the present subsection some recommendations are presented for selecting flood-tolerant fruit tree corps suitable to very

high flood risk areas. During the past 30 years research has been conducted on the effects of flooding on various fruit tree species (Crane and Davies, 1989; Larson et al., 1993; Schaffer, 1998; Schaffer et al., 2006). According to those studies, the primary effect of flooding on tree crops physiology and growth is related to the reduction of soil oxygen. The earliest detectable physiological symptoms of flooding stress include decreased net CO2 assimilation, stomatal conductance, and transpiration. Prolonged flooding

140

N.N. Kourgialas, G.P. Karatzas / Environmental Science & Policy 63 (2016) 132–142

Fig. 6. Rainfall intensity maps under scenarios A2 and B1.

usually results in a stoppage of root operation and tree growth, in a decrease of nutrient uptake, and often causes tree death. In addition, flood leads to a decreased flowering, yield and fruit quality for several species of fruit trees. Different fruit tree species have different flood tolerance depending on the growing season, the duration and the magnitude of flooding as well as on the age of the plants (Schaffer et al., 1992). Specifically, flooding during the dormant season is much less harmful than during growing season. In addition, some fruit trees are killed by less than one week of flooding while some others can survive continuous flooding. On the other hand, flash flooding with moving water may increase growth of some flood-tolerant species. Furthermore, mature trees tolerate flooding better than seedling or over-mature trees (Ismail and Noor, 1996). Generally, flood-tolerant species could be characterized as the fruit trees that could survive under excessively wet (high water table) and flooded conditions that last from several days to a few weeks. Moderately flood-tolerant trees could be characterized as the species that could survive for several days of excessively wet or flooded soil conditions. However, the stress of wet conditions may reduce tree growth and fruit production, while root disease may result in tree damage or death. Finally, not flood-tolerant trees

could be characterized those trees that are not tolerant to wet or flooded soil conditions. These trees may bear heavy damage after few days of wet soil conditions (Schaffer et al., 1992). A very important task of this study was to review the existing literate related to flood-tolerant fruit species. Additionally, in this study taking into consideration the above literature evidences, the flood-tolerant fruit trees under Mediterranean conditions were investigated and ordered. This was achieved considering that flash floods are common in the study area taking place mainly in rainy period (winter, spring and autumn) coinciding or not with the growing season of different fruit trees. Moreover, in order to have trustworthy results, it was assumed that all the investigated Mediterranean fruit trees were at the same age (mature trees). Based on the above, in the case of the most common Mediterranean cultivated fruit trees the order of flood tolerance is: quince, pear, apple, plum (flood-tolerant species), citrus, cherry, apricot, peach, and almond (moderately flood-tolerant species), followed by olive, vineyard and avocado (not flood-tolerant trees) (Rowe and Beardsell, 1973; Schaffer et al., 2006). As mentioned in previous sections, the island of Crete is mainly dominated by olive, citrus, vineyard and avocado cultivations. These cultivations are sensitive to flooding. Thus, for an adopting agricultural approach,

Table 3 Flooded areas under current and future scenarios. Flood risk level

Area (Km2) Baseline

Area (Km2) scenario A2

Percentage change Scenario A2 vs. Baseline

Area (Km2) scenario B1

Percentage change Scenario B1 vs. Baseline

Very High High Moderate Low Very Low

1605 2330 1915 1463 931

1459 2392 2000 1437 956

1.77 0.75 1.03 0.32 0.30

1710 2160 1894 1488 992

1.27 2.06 0.25 0.30 0.74

N.N. Kourgialas, G.P. Karatzas / Environmental Science & Policy 63 (2016) 132–142

141

Fig. 7. Final flood risk maps under scenarios A2 and B1.

this study highlights the fruit tree species that show tolerance to flooding and are economically profitable which could be suggested for cultivation in very high flood risk areas. Based on the above, fruit trees which could be effectively cultivated at the specific climatological conditions of the study area, and could ensure at the same time adequate income to local farmers could be the quince, pear and apple trees. Specifically, the cultivation of quince trees, that its domestic production covers only 25–30% of the total demand in Greece, could ensure a good income to the farmers (indicative price approximately equal to 15000 Euros/ha/year), (Ministry of Agriculture of Greece, 2010). Thus, this cultivation could be a prospective horticulture for low and semi-mountainous areas of Crete where flood risk can be characterized as very high. Moreover, considering the flood tolerance of pear and apple trees they also could be suitable for cultivation at very high flood risk areas of semi-mountainous and mountainous regions of Crete, ensuring a very good income to the farmers (indicative price approximately equal to 40000 Euros/ha/year).

4. Conclusion The estimation of the spatial flood risk in a GIS environment is essential for an effective flood risk management. In this paper, the flood risk in the Mediterranean island of Crete is identified using GIS. Specifically, the fruit tree areas which are most prone to flood risk are indicated through the application of the proper methodology. The determination of flood risk areas is based on a classification of the flood risk in five categories ranging from very high to very low. Thus, the fruit tree areas that are most susceptible to flooding were determined. The identification of these areas resulted from the combination of the six thematic flood risk maps for the region of Crete. The thematic map – factors used for the present study were: (a) flow accumulation, (b) slope, (c) land use, (d) rainfall intensity, (e) geology, and (f) elevation. The results of this assessment were verified by historical flood events within the study area (including low, semi-mountainous and mountain plateau areas). Taking into account that Crete is one of the most

Table 4 Flooded fruit tree areas under current and future scenarios. Flood risk level

Fruit Tree Area (Km2) Baseline

Fruit Tree Area (Km2) scenario A2

Percentage change Scenario A2 vs. Baseline

Fruit Tree Area (Km2) scenario B1

Percentage change Scenario B1 vs. Baseline

Very High High Moderate Low/Very Low

792 1131 270 415

695 1212 285 416

4.42 3.69 0.68 0.05

828 1049 313 418

1.38 3.14 1.65 0.12

142

N.N. Kourgialas, G.P. Karatzas / Environmental Science & Policy 63 (2016) 132–142

intensively fruit tree cultivated areas in Mediterranean, where significant flash flood events occur from time to time, the proposed methodology can contribute to the determination of horticultural areas that are most prone to flooding. GIS results show that 47.8% of the total area of Crete is under very high and high flood risk, while a very significant percentage of fruit tree areas, about 73.8%, are under very high or high flood risk. Another aim of this study was to predict flood risk areas for the year 2050, under climatic scenarios A2 and B1. Based on this, for scenario A2, a decrease of very high flood risk areas was observed in the whole area of Crete and especially in fruit tree cultivations. On the contrary, a significant increase of high flood risk areas was observed for fruit tree cultivations in comparison to the present situation. Regarding scenario B1, the increase of rainfall intensity generates a slight increase in very high flood risk areas where fruit tree cultivations exist. The opposite behavior is observed in high flood risk areas. Also, this study investigates and suggests floodtolerant and economically effective fruit tree species, suitable for cultivation at very high flood risk areas of Crete. Such species could be quince, pear and apple trees. All these approaches can be incorporated into a flood decision support system and applied to any river basin where tree crops are cultivated, ensuring the sustainability of agriculture. References Archer, D., 1992. Walls of water. Circulation-British Hydrological Society Newsletter Society 44, 1–3. Ballesteros, C.J.A., Eguibar, M., Bodoque, M.J., Diez-Herrero, A., Stoffel, M., GutierrezPerez, I., 2011. Estimating flash flood discharge in an ungauged mountain catchment with 2D hydraulic models and dendrogeomorphic palaeostage indicators. Hydrol. Process. 25 (6), 970–979. Brocca, L., Melone, F., Moramarco, T., 2011. Distributed rainfall-runoff modelling for flood frequency estimation and flood forecasting. Hydrol. Process. 25, 2801– 2813. Camarasa Belmonte, A.M., Segura Beltran, F., 2001. Flood events in Mediterranean ephemeral streams (ramblas) in Valencia region. Spain Catena 45 (3), 229–249. Chatterjee, C., Förster, S., Bronstert, A., 2008. Comparison of hydrodynamic models of different complexities to model floods with emergency storage areas. Hydrol. Process. 22 (24), 4695–4709. Chau, Vu Ngoc, Holland, J., Cassells, S., Tuohy, M., 2013. Using GIS to map impacts upon agriculture from extreme floods in Vietnam. Appl. Geogr. 41, 65–74. Chenini, I., Mammou, A.B., May, M.E.L., 2010. Groundwater recharge zone mapping using GIS-based multi-criteria analysis: a case study in central Tunisia (Maknassy Basin). Water Resour. Manage. 24, 921–939. Crane, J.H., Davies, F.S., 1989. Flooding responses of Vaccinium species. HortScience 24, 203–210. Diakakis, M., Mavroulis, S., Deligiannakis, G., 2012. Floods in Greece, a statistical and spatial approach. Nat. Hazards 62, 485–500. ESRI, 2008. ArcView 9.2 User Manuals. Environmental System Research Institute, Redlands, CA. Eimers, J.L., Weaver, J.C., Terziotti, S., Midgette, R.W., 2000. Methods of Rating Unsaturated Zone and Watershed Characteristics of Public Water Supplies in North Carolina. Water-Resources Investigations Reports, Raleigh, NC, pp. 99– 4283. Goovaerts, P., 2000. Geostatistical approaches for incorporating elevation into the spatial interpolation of rainfall. J. Hydrol. 228 (1–2), 113–129. Ismail, M.R., Noor, K.M., 1996. Growth and physiological processes of young starfruit (Averrhoa carambola L.) plants under soil flooding. Sci. Hortic. 65, 229–238. Kazakis, N., Kougias, I., Patsialis, T., 2015. Assessment of flood hazard areas at a regional scale using an index-based approach and analytical hierarchy process: application in Rhodope–Evros region Greece. Sci. Total Environ. 538, 555–563. Kourgialas, N.N., Karatzas, G.P., 2011. Flood management and a GIS modelling method to assess flood-hazard areas: a case study. Hydrol. Sci. J. 56 (2), 212–225.

Kourgialas, N.N., Karatzas, G.P., Nikolaidis, N.P., 2010. An integrated framework for the hydrologic simulation of a complex geomorphological river basin. J. Hydrol. 381 (3–4), 308–321. Kourgialas, N.N., Karatzas, G.P., Nikolaidis, N.P., 2011. Development of a thresholds approach for real-time flash flood prediction in complex geomorphological river basins. Hydrol. Process. doi:http://dx.doi.org/10.1002/hyp.8272. Kourgialas, N.N., Dokou, Z., Karatzas, G.P., 2015. Statistical analysis and ANN modeling for predicting hydrological extremes under climate change scenarios. the example of a small Mediterranean Agro-Watershed. J. Environ. Manage. 154, 86–101. Larson, K.D., Schaffer, B., Davies, F.S., 1993. Physiological, morphological, and growth responses of mango trees to flooding. Acta Hort. 341, 152–159. Linstone, H., Turoff, M., 2002. The Delphi Method: Techniques and Applications. Information Systems Department, New Jersey Institute of Technology. Liu, Y.B., Gebremeskel, S., De Smedt, F., Hoffmann, L., Pfister, L., 2003. A diffusive transport approach for flow routing in GIS-based flood modeling. J. Hydrol. 283 (1–4), 91–106. Lloyd, C.D., 2005. Assessing the effect of integrating elevation data into the estimation of monthly precipitation in Great Britain. J. Hydrol. 308 (1–4), 128– 150. Makropoulos, C.K., Butler, D., 2006. Spatial ordered weighted averaging: incorporating spatially variable attitude towards risk in spatial multi-criteria decision-making. Environ. Modell. Software 21 (1), 69–84. Ministry of Agriculture of Greece, 2010. The Greek Agriculture in Numbers: Basic Characteristics. (http://www.minagric.gr/.). Morelli, S., Battistini, A., Catani, F., 2014. Rapid assessment of flood susceptibility in urbanized rivers using digital terrain data: application to the Arno river case study (Firenze, northern Italy). Appl. Geogr. 54, 35–53. Morgan, R.P.C., 2005. Soil Erosion and Conservation. Blackwell Publishing Ltd., Oxford. Nikolaidis, N.P., Bouraoui, F., Bidoglio, G., 2013. Hydrologic and geochemical modeling of a karstic Mediterranean watershed. J. Hydrol. 477, 129–138. Paleologos, E.K., Mertikas, S., Papadaki, E.S., Kourgialas, N.N., 2013. GroundwaterInduced land subsidence in crete, Greece. Porto, Portugal, 26th–29th June8th International Conference of EWRA, 013. , pp. 1325–1333. Papaioannou, G., Vasiliades, L., Loukas, A., 2015. Multi-criteria analysis framework for potential flood prone areas mapping. Water Res. Manage. 29 (2), 399–418. Prime, T., Brown, J.M., Plater, A.J., 2016. Flood inundation uncertainty: the case of a 0.5% annual probability flood event. Environ. Sci. Policy 59, 1–9. Pshenichny, C.A., et al., 2009. The event bush as a semantic-based numerical approach to natural hazard assessment (exemplified by volcanology). Comput. Geosci. 35 (5), 1017–1034. Rowe, R.N., Beardsell, D.V., 1973. Waterlogging of fruit trees. Hortic. Abstr. 43, 534– 548. Schaffer, B., Andersen, P.C., Ploetz, R.C., 1992. Responses of fruit crops to flooding. In: Janick, J. (Ed.), Horticultural Reviews, vol. 13. Wiley, New York, pp. 257–313. Schaffer, B., Davies, F.S., Crane, J.H., 2006. Responses of subtropical and tropical fruit trees to flooding in calcareous soil. HortScience 41, 549–555. Schaffer, B., 1998. Flooding responses and water-use efficiency of subtropical and tropical fruit trees in an environmentally-sensitive wetland. Ann. Bot. 81, 475– 481. Thieken, A.H., Kreibich, H., Müller, M., Merz, B., 2007. Coping with floods: preparedness, response and recovery of flood-affected residents in Germany in 2002. Hydrol. Sci. J. 52, 1016–1037. Tsanis, I.K., Koutroulis, A.G., Daliakopoulos, I.N., Jacob, D., 2011. Severe climateinduced water shortage and extremes in Crete. Clim. Change 106 (4), 667–677. Van Der Veen, A., Logtmeijer, C., 2005. Economic hotspots: visualizing vulnerability to flooding. Nat. Hazards 36 (1–2), 65–80. Wang, Y., Zhongwu, L., Zhenghong, T., Guangming, Z., 2011. A GIS-based spatial multi-criteria approach for flood risk assessment in the Dongting Lake Region, Hunan, Central China. Water Resour. Manag. 25 (13), 3465–3484. Water Resources Department of the Prefecture of Crete, 2014. Sustainable Management of Water Resources in Crete – Hydrometeorological Data (in Greek). Region of Crete Information Bull. (Available from: http://www.regioncrete.gr.). Xepapadeas, A., 1996. Quantity and quality management of groundwater: an application to irrigated agriculture in Iraklion, Crete. Environ. Model. Assess. 1, 25–35. Yahaya, S., Ahmad, N., Abdalla, R.F., 2010. Multicriteria analysis for flood vulnerable areas in Hadejia-Jama’are river basin Nigeria. Eur. J. Sci. Res. 42 (1), 71–83. Zerger, A., 2002. Examining GIS decision utility for natural hazard risk modelling. Environ. Model. Softw. 17 (3), 287–294.