Geospatial hazard modeling for the delineation of flash flood-prone zones in Wadi Dahab basin, Egypt

Geospatial prediction of flash flood hazard has become crucial in mountainous areas because of the socio-economic impact of flash floods and population growth (Abuzied et al. 2016c). Different studies around the world have supported the geospatial assessment of flash floods due to urbanization growth in mountainous regions (Youssef et al. 2011; Safaripour et al. 2012; Gioti et al. 2013). The Sinai Peninsula (Figure 1) has captured the Egyptian government's attention during the last decades in support of economic development. The Wadi Dahab area is considered one of the hazardous regions in the Sinai Peninsula due to its relatively high flash flooding range. Many researchers have considered Wadi Dahab as a strategic location for mitigation planning (Omran 2013; Selmi & Abdel-Raouf 2013). Therefore, the assessment of flash flood hazard in the Wadi Dahab area is extremely important to reduce catastrophic outcomes. This study reveals flash flood hazard zones through an integrated approach based on remote sensing and geographic information systems (GIS) techniques. Worldwide attention has focused on geospatial mapping of flash flood hazards using satellite images and geologic and topographic maps (Forte et al. 2006; Dawod et al. 2011; Clement 2013; Bajabaa et al. 2014; Abuzied et al. 2016c; Khosravi et al. 2018). However, few studies have conducted an integrated approach using remotely sensed, topographic, geologic, morphometric, and hydrologic data for spatial modeling of flash flood hazards (Elkhrachy 2015). Accordingly, this study adds an integrated multifactor analysis for a comprehensive assessment of flash flood hazard zones as the first comprehensive evaluation of flash flood hazards in Wadi Dahab basin, southeastern Sinai, Egypt.

The study included many remote sensing skills to examine the factors contributing to flash floods, such as Shuttle Radar Topography Mission (SRTM), Landsat 8 (OLI) satellite images, and geologic and topographic maps. Abuzied (2016) supported the use of Spot 5 and Landsat 7 to delineate lithology and land use in the mountainous region close to Wadi Dahab. SRTM has been used extensively in worldwide studies as a main source to extract several factors such as elevation, drainage network, slope, stream power index, and topographic wetness index (Pradhan & Lee 2010; Abuzied et al. 2016a, 2016d). Similar approaches were used in this study, utilizing OLI 8 to classify the lithological units, soil, and land use and SRTM to extract several factors such as elevation, drainage network, slope, stream power index, topographic wetness index, flow length, velocity, and flow time.

Drainage characterization plays an important role in evaluation of flash flood hazard (Dawod et al. 2012). Commonly, geomorphology has been essential to fluvial systems (Thorne 2002); hence, worldwide studies have characterized drainage networks in basins and sub-basins based on conventional geomorphologic methods (Horton 1945; Strahler 1964; Krishnamurthy et al. 1996). Watershed terrain analysis relies heavily on quantitative measures for drainage basins and their geometric characteristics (Abrahams 1984). Rationally, the morphometric variables of any basin can then reflect runoff routing, flood peaks, erosion estimates, sediment yields, and flood impact (Patton 1988; Gardiner 1990). Thus, this research adopted watershed terrain analysis to characterize the morphometric parameters of Wadi Dahab catchments and their drainage network.

The geospatial assessment of flash flood hazard in the Wadi Dahab area is challenging due to the rugged topography and lack of direct runoff data. Generally, flash floods are caused by complex interactions of topographical, geological, geomorphological, and hydrological effects (Abuzied et al. 2016c). For an integrated multifactor analysis, the Natural Resources Conservation Service (NRCS) rainfall-runoff model (SCS 1972) was selected in this study to evaluate hydrologic response at ungauged watersheds. The NRCS model predicts the runoff behavior in ungauged catchments better than more multifaceted models (Wagener & Wheater 2004), especially common in ephemeral catchments (Hedman 1970). Following NRCS approaches, several studies recognized relationships among basin geometric parameters and hydrological parameters (Fang et al. 2008; Masoud 2009). The spatially distributed unit hydrograph (SDUH) method was successfully used in several studies to estimate the hydrological parameters for highly hazardous sub-basins (Maidment 1993; Najjar 1999). The SDUH is generally used to evaluate the physical characteristics of the catchment and to parameterize a unit hydrograph in the absence of recorded rainfall and runoff (Cleveland et al. 2008). The geomorphological watershed characteristics of Dahab wadis are extremely variable but still definable, although synthetic hydrographs based on topographic characteristics have been adopted in this study to measure runoff velocity and flow time.

GIS is a standard technology to analyze, manipulate, and integrate the factors contributing to flash flood hazard with great accuracy and efficiency (Abuzied et al. 2016c). In this study, the causative factors for flash floods were constructed, analyzed, and integrated within the same georeferencing outline using GIS. Mapping of flash flood hazard requires ranking values corresponding to the importance of each factor and weights for its classes based on their contributions to flash flood hazard (Dawod et al. 2012). Several GIS models including logistic regression, bivariate statistical analysis, and analytical hierarchy process (AHP) have been used in different applications around the world to assign suitable ranks for any factor and suitable weights for its classes (Pradhan & Lee 2010; Moeinaddini et al. 2010; Regmi et al. 2014; Chapi et al. 2017). The AHP is an appropriate method to derive ranks for any factor corresponding to its relative importance based on a pairwise comparison analysis (Boroushaki & Malczewski 2008; Abuzied et al. 2016b). Therefore, the AHP was applied in this study to assign realistic ranks for the flash flood contributing factors. Our study goal is to explore a unique Flashflood Hazard Model (FHM) integrating multi-factor analysis. These factors are slope, stream power index, topographic wetness index, lithology, hydrological morphometrics, runoff depth, velocity, and flow time. The integrated approach adds to a more realistic flash flood hazard map and is a powerful tool for development actions in mountainous areas.

Wadi Dahab basin is located between the southwestern Gulf of Aqaba coast and the southeastern side of the Sinai rift zone (Figure 1). Dahab city is located on an alluvial fan extending from the Gulf of Aqaba to the main channel of Wadi Dahab, which represents one of the wadis with the greatest flash flood hazard on the Sinai Peninsula. The study area occupies 2,080 km² and covers an area between latitude 28° 22′ 43.4″ and 28° 52′ 18.5″ N and longitude 33° 55′ 46.9″ and 34° 31′ 28.8″ E (Figure 1). The topography of Wadi Dahab basin differs gradually from steep mountains to gentle plain, sloping towards the Gulf of Aqaba. The relief varies in the study area from low zones to high rugged mountains on both sides of Wadi Dahab that range in elevation from 100 to 2,527 m, respectively (Figure S1, available with the online version of this paper). Several wadis represent separated drainage systems running through the study area. Various cycles of sedimentation created the wadis during Quaternary times and rainy periods (Gilboa 1980). Numerous active wadis drain through the study area, causing a huge amount of flood flow, such as Wadi Zaghraa, Wadi Nasb, and Wadi Saal. The main tributaries of these wadis play an important role in the erosional processes that occur in the southwestern coastal zone because the wadis are fed by seasonal floods (Omran 2013).

Wadi Dahab basin is an extremely arid climate area in which winters are cold with intense rain, and summers are hot and dry. The maximum temperature at Wadi Dahab is recorded as 37 °C, while the lowest temperature may reach −3 °C in the highland areas such as Saint Katherine. The amount of precipitation increases in the western Wadi Dahab, where the average annual precipitation is about 64 mm (Figure 2). Occasionally, the precipitation exists as snow on the mountains peaks. In southern Sinai, the probability of active runoff can be produced from storms of more than 10 mm (Geriesh 1998). In the Middle East, literature on flash floods (Moore et al. 1991) recommended that a wet month could be characterized by one or two rainy days, alternating from 10 mm in the low coastal regions to 50 mm in the highland regions. The flash floods take place seasonally in the study area due to convective rains (Dayan & Abramski 1983). The hydrographical basins on the western side of Wadi Dahab are responsible for runoff and debris flow, because the western branches of their steep sloping channels drain from the highlands where high rates of rainfall prevail (Figures 2 and 3). Several flash floods have occurred along Wadi Dahab in the past years. The last destructive one occurred in 1994, in which the water level at the wadi exit increased up to more than 2.5 m, leading to huge damage to infrastructure and threatening life and property (WRRI 2006).

Geologically, Wadi Dahab represents the eastern extension of the Precambrian Arabian-Nubian shield that occupies the area extending from western Saudi Arabia to southern Sinai. Several lithological units exist with diverse hydrological properties (Figure 4). The lithological units consist of Precambrian basement rocks, sedimentary succession, and Quaternary wadi deposits (Said 1962). The Precambrian basement rocks are classified into metamorphic and igneous rock units. The Precambrian igneous rocks cover approximately 63% of the area, and the metamorphic rocks cover 7% of a small part of Wadi Dahab basin, especially in Wadi Nasb, Wadi Saal, and Wadi Zaghraa (Figure 4). The sedimentary rocks cover 13% at the northern part of the Wadi Dahab representing Cambrian to Upper Cretaceous rocks. The recent deposits occupy 17% of the area, including alluvial fans, terraces, and wadi deposits. Several wadis are responsible for shaping the Quaternary alluvial fans when the stream velocity reduces quickly (Figure 4). Commonly, the Precambrian basement rocks are extremely weathered, and exist on hillslopes as fragmented blocks vulnerable to transport when the shear strength decreases. The Precambrian basement rocks follow the steep hills along the main wadis, and the steep slopes accelerate weathering products into the wadis.

The fault system present in the study area is greatly influenced by Red Sea rifting trending mainly in NW–SE (Gulf of Suez direction) and NNE–SSW (Gulf of Aqaba direction) (Figure 4). Wadi Nasb and Wadi Ghaieb faults originated with the formation of Gulf of Suez faults, while the majority of NE trending faults are in the direction of the Gulf of Aqaba and are younger, of Pleistocene age (Said 1962). Geologic structures, mainly faults and joints, have captured the interest of many hydrogeologists, as they act as good groundwater conduits, representing zones of high infiltration and groundwater potentiality. Most of these geologic structures are found in the basement complex with negligible abundance in softer rocks. Accordingly, there are two main types of aquifers present – alluvial and basement aquifers. The groundwater runs laterally and downwards via open fractures to recharge the alluvial deposits of the main wadis, i.e., Wadi Nasb, Wadi Saal, and Wadi El-Ghaib (El Rayes 1992). Therefore, these alluvial deposits receive large amounts of water recharge through the fractures in the basement rocks, producing good-quality reservoirs. Most of the groundwater wells in the study area are dug in these alluvial deposits, reflecting the high availability of groundwater. The alluvial aquifers have good hydraulic parameters with high storage capacity and good water quality with increasing thickness towards the downstream reaches (Shendi et al. 1997).

Wadi Dahab basin is extremely vulnerable to flash floods due to the cumulative effects of its physical environmental characteristics. Therefore, the holistic approach was considered to assess flash flood hazard based on topographical, geological, geomorphological, and hydrological factors. The assessment workflow is illustrated by Figure 5, which shows the data extracted from different sources.

This study suggests eight factors that contribute to the catastrophic impacts of flash floods. The following factors were selected carefully to develop a unique FHM: slope, stream power index, topographic wetness index, lithology, hydrological morphometrics, runoff depth, runoff velocity, and flow time. The preparation methods of these factors are briefly described in the following paragraphs.

SRTM data were adopted and processed to generate a Digital Elevation Model (DEM) at a spatial resolution of 29 m (Figure S1, available with the online version of this paper). The DEM is considered an essential data source to derive information on different topographic factors inducing flash flooding, such as slope, stream power index, and topographic wetness index. Slope is one of the key factors with respect to the possibility of existing flash flood hazards. We generated the slope map using SRTM DEM at a spatial resolution of 29 m, divided equally into six classes (Figure 3).

The TWI map of Wadi Dahab was also generated using Map Algebra in ArcMap 10.4 and classified to five classes with equal intervals (Figure 7).

The lithological units in Wadi Dahab were defined using remote sensing techniques and information from previous studies and geologic maps (Said 1962; Conoco 1987; EGSMA 1994). The United States Geological Survey (USGS) website was used to download two scenes of the OLI Landsat 8 satellite. The scenes were pre-processed to decrease the haze effects before mosaicking and sub-setting using ENVI 5.3. To obtain the highest contrast on lithological units, the combination of processing techniques was carried out to bands from each spectral zone, such as mid-infrared, short-wave infrared II, and visible. Decorrelation stretch and Intensity-Hue-Saturation transformation were used to enhance bands 6, 7, and 4 (Abuzied & Alrefaee 2017). Moreover, principal component analysis was used for all bands to evaluate the principal component (PC) containing the most information. The best information was depicted from bands 6, 7, and 4. A PC combination (3, 4, and 5) was created to enhance different rock units and compared with band ratio combinations (Abuzied & Alrefaee 2018). Algebra combinations and permutation were also considered to create band ratios. The contrast improved gradually with using bands from various spectral zones such as bands 6, 7, and 4. The combination of 4/3, 7/3, and 6/2 was considered as the suitable input to classify lithological units.

The processing techniques produced 17 classes, which were created by supervised image classification followed by post-classification smoothing and vectorization of the raster layer (Figure 4). The same especial rock units are validated and tested by the lithological map, which was added by EGSMA (Egyptian Geological Survey and Mining Authority 1994). The processing techniques classify Wadi Dahab basin into different rock units (Figure 4), including Precambrian igneous rocks (such as Granodiorite, Diorite, Alkaline granite, Monzogranite, Catherina volcanics, and ring dyke), Precambrian metamorphic rocks (such as Metagabbro, basic Meta-volcanic, acidic Meta-volcanic, Metasediments, and phyillite), and Phanerozoic succession (such as Firani group, Cambrian formation, lower Cretaceous formations, upper Cretaceous formation, and Quaternary deposits and recent wadi deposits). Most of the flash flood damage zones could be observed in association with the Precambrian rocks.

To add a confident FHM, responses of surface hydrological processes are crucial tasks. These hydrological responses are essentially associated with spatial and temporal distributions of water flows and their accumulations, including runoff depth, velocity, and flow time. The surface runoff was evaluated from maximum daily rainfall data using the NRCS rainfall-runoff model (SCS 1972). Historical data from eight stations of the Egyptian Meteorological Authority (1990 to 2016) were contributed to create an Isohyet map (Figure 2) and to anticipate the runoff behavior. According to these data, the average daily rainfall in the Dahab area is small, but it might rise intensely within 1 day, resulting in a catastrophic runoff. The most disastrous storm might be recurrent several times through the next period of storms (Abuzied et al. 2016c). Accordingly, the maximum rainfall per day was suggested in this study to evaluate the runoff behavior and to assess flash flood hazard.

The synthetic hydrographs were adopted in the study using the raster GIS functions through ArcMap10.4. The climate data, topographic data, soil, and land use were prepared to estimate the runoff scenario. The highest storm values occurred in Saint Katherine station from 1934 to 2004 (76.2 mm/day) in November 1937. The Saint Katherine station is located at the west of Wadi Dahab basin and is characterized by having the highest elevation and slope gradient in the area. Therefore, the rainfall amount from this station represents a good record (WRRI 2006). Commonly, runoff happens when the rainfall intensity is more than the infiltration capacity at a location (Yair & Lavee 1985). Initial losses are affected by surface and subsurface storage, soil type, infiltration, land use, and evapotranspiration. Thus, initial losses take place before runoff approaches the drainage networks in the watersheds, whereas transmission losses take place as rainfall flows through the stream network. The initial abstractions considered in the study are the infiltration losses only as the result of arid conditions, sparse vegetation, and topography ruggedness of the Dahab area.

OLI Landsat 8 satellite images were processed to classify land use and soil types using unsupervised and maximum likelihood classification (Figures S2 and S3, available with the online version of this paper). Soil types were classified mainly based on hydrological and physical properties. Based on the infiltration capacity, the soil types were categorized into four main groups, including quaternary and recent deposits, soil containing clastic rocks and fragments, soil containing calcareous carbonate rocks, and soil containing hard rocks (Figure S2). Quaternary and recent deposits are deep, very well-drained gravel and sand producing high infiltration capacity and low runoff. These deposits were characterized as A-type group with infiltration capacity >0.76 cm/h. The clastic rocks were characterized as B-type group with infiltration capacity (0.38 to 0.76 cm/h) due to their fine to coarse texture and well-drained properties. Thus, low curve number (70.62 to 77.87) was assigned to Quaternary and recent deposits and the clastic rocks. The carbonate rocks were characterized as C-type group due to their infiltration characteristics. The Precambrian basement rocks were characterized as D-type group due to their texture and drained properties. The bedrock surface on the basement and the carbonate rocks is lacking soil cover. Hence, the infiltration rate is low in these areas, producing high runoff. A high curve number was assigned to both of the rock types (84.5 to 92.02). The different hydrologic soil groups were tabulated in each catchment to determine CN values. To estimate the rainfall excess, a single CN value was assigned at the outlet of all Wadi Dahab sub-basins with only one soil type, whereas area-weighted CN values were applied to sub-basins having different soil types. An area-weighted average method was applied to calculate CN values in the sub-basins having different soil types.

The multi-criteria evaluation represents a necessary analysis for modeling hazards due to flash floods. Three main steps were achieved to define the flash flood hazard index, including standardization of the criterion weights, estimation of each criterion rank, and summation of the criteria (Tables 3 and 4).

All factors increasing the severity impacts of flash flood hazard were derived in this study on different scales. Consequently, a linear scaling analysis is necessary to standardize the weight of each criterion, because it is the simplest procedure for standardization (Equations (3) and (4)). All classes of the effective criteria were given scores from 0 to 1, in which 0 represents less severity for flash flooding and 1 represents the highest severity for flash flooding (Table 4).

The geospatial hazard modeling of flash flood in Wadi Dahab basin was achieved by integrating the previously interpreted criteria through ranking and weighting processes. Based on the calculated weights and ranks, the influence levels of contributing factors and their alternatives were created. The hydrological factors, such as flow time and runoff velocity, and the SPI represent the most influential factors on the flash flood hazard ranking: 0.309, 0.271, and 0.155, respectively (Table 4). The different classes of all causative factors were assigned scores from 0 to 1 according to their relative contributions to occurring hazards. Hence, the shortest flow time, the highest runoff velocity, and the highest SPI were given scores of 1, representing the maximum severity for flash flooding (Table 4); while the longest flow time, the lowest runoff velocity, and the lowest SPI were given scores of 0, representing the least severity for flash flooding (Table 4). The runoff depth and morphometric index of Wadi Dahab catchments represent the following influence factors that cause flash flood hazard and have ranks of 0.092 and 0.071, respectively (Table 4). The highest runoff depth and the highest morphometric index were assigned scores of 1, representing the maximum contributing classes to occurring hazards; while the lowest runoff depth and the lowest morphometric index were assigned scores of 0, representing the least contributing classes to the occurring hazards. The slope, lithology, and TWI represent the least influential factors on the flash flood hazard, assigned 0.041, 0.037, and 0.023, respectively (Table 4). Steep slope, the Precambrian basement rocks, and the lowest TWI were assigned scores of 1, representing the maximum severity for flash flooding; while gentle slope, Quaternary and recent deposits, and the highest TWI were assigned scores of 0, representing the least severity for flash flooding.

The flash flood hazard map classifies Wadi Dahab basin relatively into different hazard zones. The resulting hazard map consists of five main classes varying from very low to very high (Figure 14(a)). The interpreted hazard map refers to the most vulnerable zones to flash flooding. The most vulnerable zones are mainly distributed in the eastern side of the mapped area, where steep topography to downhill lands and Precambrian basement rocks exist, which cause high runoff velocity and short time for water accumulation. The very high and high hazard zones are located along several wadis and tributaries in the study area, such as El-Ghaib, Zaghraa, Nasb, Khasheib, Rimthy, and Saal (Figure 14(a)). Most of these wadis are characterized by rugged topography and widely spread fault actions controlling the drainage distribution.

The flash flood hazard map was tested with the flash flood markers that were recorded from field survey to validate the final hazard map (Figure 14(a) and 14(b)). Based on the correlation analysis, all flash flood markers are associated with very high and high hazard zones. The flash flood hazard map indicates that 1.59% of Wadi Dahab's total area is very high hazard zone (Figure 14(a) and 14(b)). The very high hazard zones occupy an area of 33.07 km², whereas the very low hazard zones occupy an area about 550.16 km² (26.45% of the total mapped area). The moderate flash flood hazard zones are distributed heterogeneously, covering 24.75% of the mapped area. The high hazard zones occupy 261.87 km², representing 12.59% of the Wadi Dahab area, while the low hazard zones occupy 720,096 km², representing 34.62% of the Wadi Dahab area (Figure 14(a) and 14(b)).

Geospatial mapping of flash flood hazards in Wadi Dahab was derived from multi-analyses including the SDUH method, NRCS rainfall-runoff model, morphometric analysis, topographic and geologic estimations, and field studies. The predicted hazard zones could be achieved from the findings of these analyses, which include multi-factors controlling hazard severity (Figure 14(a)). The factors were spatially integrated to provide comprehensive evaluation for flash flood hazards in Wadi Dahab basin. These factors include slope, stream power index, topographic wetness index, lithology, hydrological morphometrics, runoff depth, runoff velocity, and flow time. The influence of factors on occurring flash flood hazards varies spatially from one zone to another in Wadi Dahab. The ranks and weights describe more precisely and quantitatively the variation in the influence of factors on increasing the severity of flash floods (Table 4). The integration of hydrological data adds the holistic nature of the approach to reveal the flash flood-prone zones in mountainous regions (Figure 14(a) and 14(b)).

The SDUH method was applied successfully to estimate runoff velocity and flow time (Figures 12 and 13). As mentioned previously, the maximum storm hyetograph developed by WRRI (2006) of the Saint Katherine area was used to evaluate the scenario of the maximum storm in Wadi Dahab (Figure 15). The excess rainfall indicates direct runoff and represents the rain amount which falls at intensities exceeding the infiltration capacity. In this case, an infiltration capacity of 1.66 mm/h was calculated by the NRCS method. The flow velocity was calculated for the channel and overland flow. The complexity of the geometrical and hydraulic characteristics is too great for the Wadi Dahab channels. Furthermore, the overland flow velocity has more influence than channel flow velocity in Wadi Dahab basin. The velocity grid generally indicates the time needed for water passing through each cell of the grid. The flow velocity in Wadi Dahab basin ranges from 0.223 to 21.199 m/s during the maximum storm (Figure 12). The high velocities occur in Wadi Saal, Wadi Nasb, Wadi Rimthy, Wadi Khasheib, and Wadi El Ghaieb, ranging between 12.808 and 21.199 m/s (Figure 12). The high velocities in these wadis are affected by steep slope, high topographic elevations, and Precambrian basement cover (Figure S1; Figures 3 and 4). The runoff velocity decreases at the northern side of Wadi Dahab because of gentle slope, low elevations, and sedimentary rock cover (Figure S1; Figures 3 and 4).

In addition, the time of concentration was determined by the SDUH method, in which the time defines the maximum flow time to the catchments' outlet. Generally, time of concentration is the time required by the entire drainage area to cause the runoff. It can be estimated as time from the start of excess rainfall to the inflection point on the recession limb (Figure 15). Hence, the shorter time of concentration represents the higher peak discharge and greater flash flood hazard in the same rainfall condition. The shortest flow time was calculated in Wadi Zaghraa, Wadi Khasheib, and Wadi El Ghaib, ranging from 2 to 4 hours during the maximum storm (Figure 13). The longest flow time was calculated in Wadi Saal, Wadi El-Genah, and western parts of Wadi Nasb and Wadi Rimthy, ranging from 12 to 20 hours during the maximum storm (Figure 13). Many factors essentially control flow time through Wadi Dahab catchments, such as the basin length, runoff velocity, and slope gradients. The lengths of Wadi Khasheib sub-basin and Wadi Rimthy sub-basin are short, whereas the lengths of Wadi Nasb sub-basin and Wadi El Ghaieb sub-basin are long. In addition, the runoff velocity is high in Wadi Rimthy and in Wadi Khasheib. The values of flow time are relatively decreasing with increasing amounts of storm and rainfall excess in all Wadi Dahab sub-basins. In short, there is an inverse relation between rainfall excess and time of concentration, which is the longest travel time for runoff (Saghafian et al. 2008). Therefore, the flow time and runoff velocity deserved the highest ranks in FHM, and their classes were assigned weights corresponding to the influence of the respective classes on occurring flash flood hazards (Table 4).

From the obtained Dahab hydrograph for the maximum storm (Figure 15), the peak discharge (Q_p) was 743.5 m³/s and was achieved after 42 h (time to peak T_p). The time of concentration (T_c), which is the maximum time of flow to the basin outlet, was about 52 h. The time between maximum rainfall and peak discharge (lag time T_l) was 33 h. The runoff volume coming through the outlet was calculated to be 63.4 million m³. The base flow hydrograph was separated from the direct runoff hydrograph by joining the first sign of the hydrograph rise and the point of inflection on the falling limb (McCuen 1989). The shape of the maximum storm hydrograph indicated that Dahab basin has a very high hazard from flash floods, with high peak discharge and runoff volume, steep curve flanks and short time to peak caused by the steep slope, the sparse vegetation, impermeable rock types and high drainage density of the basin. Another indication of its hazardousness is the shorter lag time compared with time to peak and the nearly low constant base flow.

The analysis of runoff characteristics in Wadi Dahab drainage network was applied using NRCS empirical equations. The runoff depth was calculated in each Wadi Dahab sub-basin to assess streams' likelihood of flash floods (Table 4). Precipitation and infiltration play basic roles in runoff production. Commonly, a smaller infiltration amount causes a higher runoff behavior and a greater flash flood hazard. In Wadi Dahab, most lithological units are Precambrian rocks, which are characterized by low infiltration rates (Figure 4). In addition, urban area and highway pavement are characterized by smaller infiltration and thus produce greater runoff. The highest values of runoff depths are distributed at the northwestern and central parts of Wadi Dahab, especially close to Saint Katherine, ranging from 47.09 to 70.47 mm (Figure 10); while the low values of runoff depth are distributed at the downstream of Wadi Dahab, ranging from 12.02 to 35.4 mm (Figure 10). However, the downstream of Wadi Dahab represents the most hazardous zone because of physiographic characteristics such as basin geometry, slope, and topographic elevation. These characteristics support the travel and accumulation of surface runoff from Saint Katherine to Dahab city (Figure S1 and Figure 3). The temporal and spatial distributions of runoff depths control the flooding characteristics and thus produce hazards in some zones in Wadi Dahab. Hence, the runoff depth and its classes deserved the given rank and weights in FHM corresponding to its actual influence on occurring flash flood hazards in the study area (Table 4).

The morphometric analysis of Wadi Dahab catchments and their stream network resulted in the MHI (Figure 9(l)). The morphometric parameters were estimated in all 43 catchments, and their stream orders (Figure 9(a)–9(k)). The parameters of catchment shape including elongation ratio, circulation ratio, and shape factor mainly affect the stream efficiency and thus the resulting flash flood hazard (Figure 9(l)). Surface runoff travels the same distances in a circular catchment and probably arrives at the catchment outlet at the same time, resulting in a high flood peak. However, a surface runoff spreads out over time in an elliptical catchment because of its outlet at one end of the major axis, resulting in low flood peak (Schumn 1956). The values of elongation ratios increase in proportion to decreasing elongation shape of the catchments, and thus high values indicate high flash flood hazard. The high values of elongation ratio range from 0.459 to 0.643 distributed in several zones at Wadi Khasheib and Wadi Rimthy (Figure 9(a)). These catchments are characterized by steep slope and high relief, resulting in a high tendency for flash flood hazards. The high values of circulation ratio range from 0.24 to 0.64 and are distributed also in several zones at Wadi Khasheib and Wadi Rimthy (Figure 9(b)). Surface runoff flows through the same distances in these catchments and accordingly reaches the outlet at the same time, causing a high flash flood hazard. The low values of shape factor range from 0.0032 to 0.011, distributed likely at Wadi Khasheib and Wadi Rimthy (Figure 9(c)). The shape factor is directly proportional to the catchment length, and thus, the low values of shape factor indicate the rapid travel of runoff through the catchment, resulting in a high tendency for runoff hazards.

The parameters of catchment surface comprise relief ratio, relative relief ratio, ruggedness, and hypsometric index control hydrological responses in sub-basins (Schumn 1956). Generally, higher relief and steeper slope support the fast runoff through catchments, producing shorter flow time, higher flow velocity, less time for infiltration and thus higher possibility for flash flooding. The high values of relief ratio range from 5.38 to 9.29, distributed in several zones at Wadi Khasheib and Wadi Rimthy (Figure 9(d)). The high values of relative relief ratio range from 0.024 to 0.045, located in several zones upstream of Wadi Khasheib and Wadi Rimthy (Figure 9(e)). The high values of ruggedness ratio range from 0.71 to 1.32, distributed in several zones at Wadi Zaghraa, Wadi Khasheib, Wadi Nasb, and Wadi Rimthy (Figure 9(f)). The low values of hypsometric index range from 0.24 to 0.41, existing mostly at Wadi Zaghraa, Wadi Khasheib, Wadi Nasb, and Wadi Rimthy (Figure 9(g)).

The parameters of drainage network, comprising drainage density, stream frequency, texture ratio, and bifurcation ratio, indicate runoff behavior in each catchment. The values of these parameters reflect the topographic and geologic characteristics of any terrain, including lithology, infiltration capacity, slope, and relief. For example, the high values of drainage density, stream frequency, and texture ratio generate a high flash flood hazard. Generally, the physical characteristics of underlying terrain, such as impermeable sub-surface rocks and sparse vegetation, control drainage density, stream frequency, and texture ratio. A high drainage density indicates probably a low infiltration rate and high runoff potential. The high drainage density range is mostly from 0.9 to 1.34 and exists at Wadi Khasheib and Wadi Rimthy (Figure 9(h)). The high stream frequency range from 0.52 to 0.78 occurs at Wadi Khasheib and upstream of Wadi Rimthy (Figure 9(i)). The high texture ratio range from 0.48 to 0.69 is distributed at Wadi Khasheib and Wadi Rimthy (Figure 9(j)). Additionally, the low bifurcation ratio results in high flash flood hazard because the precipitation probably accumulates in one channel rather than dispersing freely. The low bifurcation ratio range from 2 to 3.7 is located at Wadi Zaghraa, Wadi Khasheib, and Wadi Rimthy (Figure 9(k)).

The MHI for Wadi Dahab catchments was estimated based on the total normalized values of the morphometric parameters in each catchment (Figure 9(l)). The summation of the normalized values adds a suitable layer to evaluate relatively the high or low flood-prone catchments in Wadi Dahab. The morphometric analysis suggests that the upstream of Wadi Khasheib and Wadi Rimthy are the most hazardous wadis, especially at the catchments numbered 3, 4, 10, 12, 13, 14, 17, 29, and 38 (Figure 9(l)). These catchments are associated with high relief topography, steep slope hills, Precambrian basement rocks, and draining of higher order wadis. These catchments' characteristics indicate the high potentiality for flash flooding and low change for groundwater recharge. Therefore, the maximum weight was assigned to the highest value of MHI, while the minimum weight was assigned to the lowest value of MHI (Table 4).

The lithological units represent an essential indicator to evaluate flash flood-prone zones. Most of the Precambrian basement rocks in Wadi Dahab have been exposed to an extended period of weathering, producing very vulnerable rocks for fracturing and sliding. In addition, the structural features in Wadi Dahab cause different degrees of stress, producing some weak zones in the weathered rocks. The weathered disintegrated rocks in the vicinity of faults/structural lineaments cause excessive water flow along the drainage channels, mainly on the steep slopes, and during the rains. Hence, any water moving along fault planes promotes erosional processes. Generally, the Precambrian basement rocks are distinguished by low infiltration capacity, steep slope, and high relief, resulting in a high possibility of runoff hazards. Therefore, the Precambrian basement rocks, especially Dokhan volcanics and old granitoids, have the highest weight in FHM, while the sedimentary succession has the lowest weight in the model. Consequently, the zones close to Precambrian rocks are the most vulnerable to flash flooding.

For slope angles, the highest weight was assigned to the slopes ranging from 40° to 50°, while the lowest weight was assigned to very gentle slopes ranging from 0° to 20° (Table 4). Generally, the flash flood hazards increase with the increase of slope gradient up to a particular extent, and then decrease (Abuzied et al. 2016c). In the case of SPI, the maximum weight was assigned to the highest value of SPI (11.6–25.01) indicating a high probability of flash flooding; while the minimum weight was assigned to the lowest value for SPI (−13.8–6.7) indicating a low probability of runoff hazards (Table 4). The erosive power of water flow depends mainly on the assumption that discharge increases with the increase of catchment area and slope gradient (Moore et al. 1991). Therefore, the high values of SPI cause a high probability of runoff hazards. The erosional and torrential activities are usually associated with drainage, and thus, SPI deserves its rank in FHM. In addition, the maximum weight of TWI was assigned to the lowest value class (−6.3–0.10). However, the minimum weight of TWI was assigned to the highest value class (22.69–27.78). TWI refers to the topography impact on the location of saturated areas by runoff (Moore et al. 1991). Commonly, the sediment transportation increases with the increase of catchment area and slope gradient. Hence, the higher slope gradient relatively gives lower TWI and higher runoff probabilities.

The FHM was holistically developed using spatial integration of several analyses including SDUH, NRCS rainfall-runoff model, morphometric analysis, topographic and geologic estimations, and field studies. Flash flood markers with GPS locations were recorded in the field study for mapping flash flood hazard zones. These hazard markers include highway undercutting, damage to traffic signs and electrical lines, and movement of large boulders and debris onto asphaltic roads (Figure 14(b)). The flash flood signs and damage locations were added to test and validate the flash flood hazard map (Figure 14(a) and 14(b)). The recorded damage locations match high and very high hazard zones on the flash flood hazard map. The flash flood hazard map refers to several hazard zones at Wadi El-Ghaib, Wadi Zaghraa, Wadi Nasb, Wadi Khasheib, Wadi Rimthy, and Wadi Saal (Figure 14(a) and 14(b)). These wadis have rugged topography, steep slope terrain, and Precambrian basement rocks which create high runoff velocity and short flow time for water accumulation. In short, the hard rocks in Wadi Dahab cover the steep hills along the main wadis, accelerating runoff accumulation into the wadis and creating several vulnerable zones for flash flood hazards (Figure 14(a) and 14(b)).

According to the final hazard map, some trails were made to decrease the destructive power of Dahab flash floods on the main infrastructure connecting Dahab city with Saint Katherine city. Eight culvert crossings made of circular pipe sections reinforced with concrete were suggested to be put under this main road to allow the flow of flash flood waters, preserving the road. The culvert locations were proposed based on the intersection between the highway and the flash flood hazard (Figure 16). Another trial to create a benefit from the flash flood water was by proposing new groundwater well locations. The locations were chosen based on the intersection of the flash flood hazard with the areas of high TWI and fault buffer zones of 300 m (Figure 16). As mentioned before, most of the study area is covered by fractured and faulted Precambrian rocks, which are considered to be good conduits and storage for groundwater in the area. The high TWI also gives indications of high groundwater potentiality in terms of high flow accumulation and gentle slope which allows water to percolate underground. The TWI map was classified into ten classes, with the highest class including all wells and dams present in the area (Figure 16), as they represent locations of high groundwater availability. The locations of the 26 new proposed wells were suggested in Wadi Nasb and Wadi Rimthy, as these high hazard zones allow the water to stay much longer on the gentle land surface, increasing the groundwater recharge. The areas with low flash flood hazards, especially the northern areas of Wadi Dahab with flat land surface, are considered suitable for new land use plans. However, flash flood hazards should be considered during planning, as these areas have high fault densities.

The development of a predictive model for flash floods in Wadi Dahab is the main outcome of this study. The FHM was built based on eight causative factors indicating flash flood hazards. The factors include slope, SPI, TWI, lithology, hydrological morphometrics, runoff depth, runoff velocity, and flow time. The ranks and the weights for the contributing factors and their classes were assigned based on AHP. Therefore, the holistic approach based on topographical, geological, geomorphological, and hydrological factors was used to delineate hazard zones. The selected factors were spatially integrated to calculate the FHI for each 29 × 29 m cell.

A flash flood hazard map was created in which 1.59% of Wadi Dahab total area represents very high hazard zones for flash flooding. The high hazard zones occupy 12.59% of the Wadi Dahab area. All the highly hazardous zones are located in structurally controlled channels, rugged topography, and Precambrian rocks. The very high and high hazard zones are distributed along several wadis in the study area, such as El-Ghaib, Zaghraa, Nasb, Khasheib, Rimthy, and Saal. The moderate, low, and very low hazard zones occupy 24.75%, 34.62%, and 26.45% of the total Wadi Dahab area, respectively. Several flash flood markers were described from field observations to test and validate the flash flood hazard map. Most damage locations were recorded in association with rugged topography, steep slope terrain, and Precambrian rocks.

In short, most of the highly hazardous zones are distributed in different locations around built-up communities. Hence, the development actions and management plans should consider these zones in order to reduce and compensate for greater hazard potential. The results of this study provide new locations which are susceptible to flash flooding. Briefly, this study recommends that decision-makers apply mitigation efforts in Wadi Dahab basin to avoid the socio-economic impacts of flash flood hazards.

The authors wish to express their appreciation to the Editor of Journal of Hydroinformatics and three anonymous reviewers, for constructive and fruitful criticism on an earlier draft of the manuscript.