Risk assessment and management of vulnerable areas to flash flood hazards in arid regions using remote sensing and GIS-based knowledge-driven techniques

Earth Observation from space has allowed characterizing, detecting, and managing natural hazards in spatiotemporal scale. Flash flood is the most frequent natural disaster that causes destruction to human lives, the economy, and infrastructure. Thus, developing a flash flood hazard zone (FFHZ) map is significant for comprehensive flash flood risk assessment and management to minimize its harmful effects, particularly in residential areas, because of climate change. Therefore, in this article, ten parameters derived from satellite images, including lithology, slope, topographic wetness index (TWI), Stream Power Index (SPI), Stream Transport Index (STI), Terrain Roughness Index (TRI), drainage density (Dd), distance to river, radar intensity map, and rainfall distribution map, were fused to predict the flood-vulnerable areas through GIS-based overlay analysis after normalization and assigning weight by applying Analytical Hierarchy Analysis (AHP). The findings allowed for the identification of the most vulnerable areas and provided an explanation for the flood's effects on New Qena City (NQC). The output FFHZs of the Wadi Qena Basin (WQB) were divided into six hazard zones, i.e., extreme hazard (6.86%), very strong (15.04%), strong (18.74%), moderate (22.58%), low (22.80%), and very low (13.98%) susceptibility. Furthermore, approximately 35% of the under-construction NQC is subject to the extreme to very serious hazards, as opposed to the extension area to NQC east of the Qena-Safaga Road. Interferometry Synthetic Aperture Radar (InSAR) change detection coherence (CCD) and spatiotemporal analysis of Landsat and Sentinel-2 data revealed steady changes in vegetation and infrastructure from 1984 to present. Based on GIS analyses about 10, and 14% of the NQC can be inundated if the flood extends 500, and 1000 m around the flood canal, respectively. Thus, several strategies were advised to safeguard the development projects, particularly the residential sections of the under construction NQC, including erecting four dams with a total capacity of 300 million m3, reinforce the dam at Wadi Shahadein, constructing concrete chevron bunds along the flood zone, and extending the depths of the flooding canal.


Introduction
Flash floods are the most prevalent natural disasters in the world, and they occur when a large amount of rain falls in a short amount of time, resulting in a large amount of surface runoff (Abdelkareem 2017;Pham et al. 2020;Waqas et al. 2021). Due to climate change and other environmental conditions, flood disasters have become the most common natural phenomenon. Floods pose a serious threat to human lives around the world since most countries are vulnerable to them, and they cause various sorts of harm, including numerous environmental and socio-economic effects (Taylor et al. 2011;Oruonye et al. 2012;Zhang et al. 2015;Abdelkareem and El-Baz 2015;Abdelkareem andAl-Arifi 2021: Mansour andAbd El-Sadek 2021). Flash floods cause destruction in all places, particularly in agricultural areas and infrastructural sectors near rivers. Every year, over 5,000 people die as a result of flash floods around the world (Bui et al. 2019;Pham et al. 2020). This issue could be caused by a lack of effective mapping or preventative measures, as well as other factors such as drainage density and slope. To mitigate the impacts of flooding natural disasters and preserve healthy socio-ecological systems under changing catchment and climate conditions, spatially explicit and catchment-scale flood models will be required to assess landscape change and rainfall runoff scenarios (Tehrany et al. 2015).
Flood control design, particularly for flash flood hazards (FFHs) in residential areas, must take into account the effects of global warming to avoid future property destruction and loss of life (Szewraski et al. 2015;Recanatesi et al. 2017;El-Rawy et al. 2022). The FFHs modeling is an important tool for flood mitigation and management (Jin et al. 2015;Abdelkareem 2017;Vivekanandan et al, 2018;Abd-El Monsef 2018;Tehrany and Kumar 2015;Yariyanet al. 2020;Chen et al. 2020). Various methodologies have been used in many studies to map flood-prone areas. Several studies have used conventional morphometric analyses (e.g., Sujatha et al. 2013;Bhatt and Ahmed 2014;Abdel-Fattah et al. 2015;Farhan et al. 2016;Abdelkareem 2017;Zhu and Abdelkareem 2021). With a limited number of data points, these approaches do not produce credible quantitative flooding predictions. At present, geospatial analysis using RS and GIS methods provides a powerful and convenient tool for mapping prone areas to FFHs. In recent modeling studies, they have been afforded a meaningful role in flood risk management (Tehrany et al. 2015;Bui et al. 2018;Pham et al. 2020;Waqas et al. 2021). Many scholars have done flood hazard assessments utilizing RS and GIS methods, contributing significantly to hazard analysis. Integration of RS and GIS approaches, as well as the most up-to-date modeling tools, can help us mitigate and manage floods (Forte et al. 2006;Pradhan, 2014;Vojtek and Vojteková 2016;Tehrany et al. 2019).
Several hydrologic, geologic, ecological, and climatic factors that influence the flash flood in a catchment were utilized to initiate a FFHs (e.g., Zhao et al. 2019;Bui et al. 2018;Yariyan et al. 2020;Waqas et al. 2021). The most important of these factors include geology, slope, land use/cover, Dd, TWI, SPI, STI, TRI, distance to river, and rainfall distribution map (Vojtek and Vojteková 2016;Arabameri et al. 2020). Several modes were utilized to map the vulnerable areas to flood hazards, e.g., Frequency Ratio (e.g., Lee et al. 2012;Rahman et al. 2019;Yariyan 2020), Evidential belief function (GhorbaniNejad et al. 2017;Yariyan 2020), Support Vector Machine (e.g., Tavakkoli Piralilou et al. 2019), Logistic Regression (e.gTehrany et al. 2019; Rahmati et al. 2019), and the analytical hierarchy process (AHP). The AHP method (Saaty 1980;1990) is utilized by many studies to make reasonable decisions (Chen et al. 2011;Zou et al. 2013;Stefanidis et al. 2013;Yariyan et al. 2020;Waqas et al. 2021) to delineate flash flood prone areas. This technique allows strategists to utilize their analytical knowledge and experience to break down an issue into a hierarchical structure and solve it using the AHP technique. It is one of the multi-criteria decision-making methods that is a commonly used, efficient technique that is simple to use and understand, and it has gained popularity as a flood mapping technique using GIS (Abdelkareem and Abdalla 2021).
Flash floods are one of the most destructive environmental hazards in desert regions with little precipitation, and they are triggered by torrential rainfall that accumulates in the downstream area, where the environment, socioeconomic features, and infrastructure of new cities are located (Abdekareem 2017; Abdelkareem and Al-Arifi 2021). Egypt's New Qena City (NQC), which is located in the downstream area of Wadi Qena (WQ), is the true case, as the flood plain attracted attention for human settlements and development. Although WQ, generally receives little precipitation per year, it has witnessed severe flash flooding events. In the middle of the twentieth century, and after a destructive flash flood in 1954, the government established a torrent canal (El Sail canal) at the end of the wadi mouth to alleviate the effects of the flash floods. This channel did not guarantee complete protection as the massive flash flood events struck El-Ma'ana Village northeast of Qena City in November 1994 and 1996. Despite the high risk of the downstream area of WQ and the need for careful attention when proposing development in an urban flood plain, the government is constructing the NQC in areas that are sensitive to lateral erosion from flood waters and in areas with serious flash flood hazards. Human activities in the downstream flood plain, including habitation, commercial, agricultural, and industrial activities, have increased in recent years.
A few research studies have focused on evaluating and assessing the danger of a flash flood in the Wadi Qena basin (Fig. 1) using various morphometric characteristics (e.g., Abdekareem 2012; Abdel-Fattah et al. 2015 andTaha et al. 2017;Moawad et al. 2016;Feleb 2016;Elsadek et al. 2019). However, these studies did not evaluate and assess the risks of flash floods in the downstream areas, especially New Qena City (NQC). In addition, no detailed studies regarding the impacts of the FFHs on the WQB and NQC were addressed using high-resolution images of radar, and optical remote sensing to avoid the risk and to manage flood hazards. This is due to that the area has witnessed severe flash flooding events and the New Qena City is under construction which is highly required for future sustainable development to avoid the risk of FFHs. Thus, this study aims to (a) delineate the susceptible areas to the flash flood hazards of the WQB; (b) assess the effects of the flash flood on the under-construction urban areas and infrastructures of the NQC; (c) identify the safe and suitable land for future development of the downstream area, especially, NQC; and (d) provide strategies and solutions for managing and mitigating the effects of flash flood hazards to protect the Qena cities and the other development areas in the basin.

Site description and climate change
Wadi Qena represents a large, wide, long, and dry valley that slopes down from the high mountain terrains of the Red Sea Mountains to the city of Qena on the Nile River. It is considered a very promising area for land settlements and agricultural expansion (Abdekareem et al. 2012). It stretches for about 200 kms from upstream to downstream, between latitudes 26 o 10′ and 28 o 15′ N and longitudes 32 o 31′ and 32 o 45′ E, covering 16,000 km 2 (Fig. 1).
Wadi Qena drains the Red Sea Mountains and Ma'aza plateaus of the Eastern Desert of Egypt, debouching into the flood plain of the Nile River at Qena City. From the east, the wadis drain the basement rocks of the Red Sea Mountains, which are mostly impervious. The lower part is composed of gravel and gravelly sand soil of alluvial fans, outwash plains, Nile Terraces, and the Nile flood plain. The flora of the wadis is generally poor and represents a thin cover (Zahran& Willis 2009). Alluvial fans are widely reclaimed for agricultural activities and dispersed settlements of farmers. Geomorphologically, the WQB consists of major sub-basins, e.g., Wadi Jurdi, Wadi Shahadein, Wadi Naq'a El-Tier, Wadi Balad, Wadi Umm Omeiyed, and Wadi El-Qreiya (Fig. 1). These sub-basins carry a tremendous amount of water and sediment yields during heavy storm seasons (e.g., storms of December 2010, and March 2014) that flow through high-sloped gravel and sandy-gravel ravines with destructive speed and feed the outlet of the basin that covers the northern part of Qena City and the southern part of New Qena City. Based on rainfall intensity distribution derived from TRMM data, the WQB received about 150 million m 3 , with variation from 2 to 391 m 3 /cell during storm of March 2014, and about 80.7 million m 3 , with variation from 0.174 to 224 m 3 /cell with a mean of 111.21 m 3 /cell during storm of December 2010 (Fig. 2).Notably, the area is dominated by the Qena-Safaga shear zoneextends NE-SW, which overlapping the road of Qena-Safaga.
The closest weather station, which is located at South Valley University on the southern side of the downstream area, provides climate data for the Wadi Qena basin. The data shows the scarcity of rainfall with an average value of 0.55 mm a year, and in the summer, the highest recorded temperature reached 47° in July. For the winter season, the lowest temperature was reached in December at 0.0° at night.
In the context of hydrologic characteristics, Egypt has seen a marked decline in total annual precipitation amounts during the last 30 years, a decrease of about 22% (Climate Risk Country Profile: Egypt, 2020). While overall, annual mean precipitation is expected to decrease, the intensity of heavy rain events is expected to increase by the 2080s (GERICS 2019). However, many parts of Upper Egypt and the Red Sea areas were struck by severe flash floods (e.g., March 1976, April 1983, April 1985, October 1987, November 1996, January 2010, January 2013, March 2014, and October 2016. May 1976, October 1987, November 1996, and January 2010 are just a few examples of severe floods (Cools et al. 2012). There is no doubt that global warming has an impact on the frequency and magnitude of flash floods. In the Wadi Qena region, the frequency of rainfall is already very low, with recorded rainy storm events that usually only occur once every few years. These events can be considered alarming threats to the new settlements in the downstream area of WQ, especially the New Qena City.
Throughout history, flash floods have hit the WQ basin and the downstream residential area. Such natural events threaten the people, damage properties, destroy infrastructure in the area and are also responsible for a large number of deaths (Abdel-Fattah et al. 2015;Feleb 2016). In the context of climate change, the FFHs will increase, so the risk of settlements at the mouth of the wadi may cause disasters. Moreover, constructing infrastructure and cultivating large areas downstream of the wadi makes them more vulnerable to flash flood impacts. The government has created new urban areas and highways downstream of WQ as communal sustainability becomes a goal, especially given the significant population expansion. The city of Qena and its extension (new Qena city) are located downstream of the WQB, covering a total area of at least 35 km 2 (Fig. 1).

Data and methods
The data for this study came from a variety of sources, including (1) a digital elevation model (DEM) with a resolution of 30 m provided by the Shuttle Radar Topography Mission (SRTM); (2) a digital elevation model (DEM) provided by the Shuttle Radar Topography Mission (SRTM); and (3) a digital elevation model (DEM) provided by the Shuttle Radar Topography Mission (SRTM); (2) Landsat data obtained from the USGS website; (3) Sentinel-1 and 2 data; (4) TRMM data; and (5) ALOS/PALSAR data. Landsat and Sentinel-2 images were used for land-use mapping (Fig. 3). All data was processed using ArcGIS, SNAP, and ENVI ArcGIS software packages.

Optical remote sensing data
One scene of Landsat data (path175/row 042) was acquired on September 9, 1984, November 19, 2005, and November 1, 2019 to monitor changes in vegetation and urban activities. The obtained scenes were linked to the coordinate system of UTM Zone 36N. The visible and NIR bands and mid-SWIR were utilized and processed using ENVI software packages.
The ESA's Sentinel-2 spacecraft is an optical satellite platform. This mission consists of a pair of land-monitoring satellites that cover a larger area of the Earth's surface and generate large optical images with more consistency than previous Landsat operations. Sentinel-2 satellites have temporal resolutions of 10 and 5 days, making them extremely valuable in future investigations. The ENVI and SNAP software packages were used to process two scenes taken on August 20, 2015, and February 9, 2022, and January 3, 2019.

Fig. 3
Data and methods adopted in the present study 1 3

Radar remote sensing data
The SRTM DEM was generated by imaging two distinct radar frequencies: X-Band (λ = 3.1 cm) and C-Band (λ = 5.6 cm). Elevations, slopes, drainages, and low-elevation sites are all computed using SRTM (30 m) data. O'Callaghan and Mark (1984) employed the eight-deterministic (8-D) approach to operate the watershed's routine drainage. Drainage lines were transformed into a Dd map with the use of GIS analysis. In addition to drainage networks, the SRTM DEM can be used to delimit low-elevated areas that operate as sinks, allowing water to enter aquifers (Zhu and Abdelkareem 2021). In a quick and low-cost manner, the DEM allows qualitative and quantitative elucidation of a watershed's geomorphic and morphometric features (Abdelkareem and El-Baz 2016).
ALOS/PALSAR is a Japanese government-developed radar sensor that images the Earth's surface at any time. The L-band (1.27 GHz, = 24 cm) is used in this sensor, which can expose the roughness and texture of sediments and penetrate the sand sheet for many metres (Paillou et al. 2009). The location of fine vs. rough sediments is revealed using a piece of a mosaic (HH polarization) obtained from the Japan Aerospace Exploration Agency in 2017 (Abdelkareem et al. 2020).
Sentinel-1 is a satellite developed by the European Space Agency (ESA) that operates in the C-band microwave part of the electromagnetic spectrum at 5.405 GHz with a 12-day repeat cycle in a sun-synchronous orbit. The sensor's spatial resolution varies between 5 and 40 m. Sentinel-1 has single and dual-polarization modes as well as interferometric wide-swath, wave, strip map, and extra-wide swath modes. For single-polarization observations, vertical send and vertical receive (VV) or horizontal send and horizontal receive (HH) wave forms were utilized in data recording. On the other hand, the dual-polarization observations, on the other hand, were conducted utilizing VV + VH (vertical send and horizontally get) or HH + HV (horizontal send and vertical obtain) (Amitrano et al. 2014;Liu 2016;Attema et al. 2009). Two Sentinel-1 scenes were acquired on May 9th, 2018 and February 9 th , 2019. In addition, two scenes acquired on November 28th, 2017 and November 30 th , 2019 were collected, and VV polarization was used to derive an image depicting changes in landscape attributes throughout this time. The InSAR CCD image was made by extracting process and amplitude data from the single-look complex (SLC) SAR product.
The Tropical Rainfall Measuring Mission (TRMM) satellite, which offers useful daily rainfall for WQB, was used to collect rainfall data for this study. The collected data at 0.25° spatial resolution spanning the test area was estimated to contain the required daily rainfall from January 1, 1998, to November 30, 2012 (evaluated by the global TRMM-3B42 version 7). The tests were conducted to define the rainfall intensity and runoff across the study area. The GIS 10.3 software is then used to analyze and interpolate the precipitated data in order to interpret the spatial distribution of rainfall data.

Change detection
Change detection was conducted on Landsat and Sentinel-1 & 2 data along with those derived from InSAR. The InSAR CCD can be realized on this principle by evaluating interferometric similarity among sets of images with different acquisition data (Jung and Kim 2016;Ullmann et al. 2016;Clossan and Milisavlyjeevie 2017). The degree of matching between two SAR sceneries in terms of radar reflection is revealed by their coherence. Decorrelation in the phase of the suitable pixels is read as minor variations in the complicated reflectivity function of the scenery (Corr and Rodrigues 1998). On June 3, 2017 and January 12, 2019, two Sentinel-1 (2017) sceneries were collected, and VV polarization was used to derive an image depicting changes in landscape attributes throughout this time. The CCD image was made by extracting process and amplitude data from the single look complex (SLC) SAR product. The interferometric coherence (ɣ) of the two different scenes can be computed as below (Seymour and Cumming 1994;Derauw 1995): where Sc1 and Sc2 are the master and slave scenes' complicated signals, respectively. Kernels with L M pixels are used to determine a pixel's coherence.

AHP-overlay analysis technique
Each layer was given a weight in the AHP approach (Saaty 1980). In a pair-wise comparison matrix, the predictive layers were then compared to one another (Table 1). Every layer's sub-classes are assigned a weight based on how important they are in estimating mineral resources. Calculating the consistency ratio (CR) in this model may be done by determining the Principal Eigenvalue ( λ ), which was estimated using the eigenvector algorithm, and the Consistency Index (CI), which was derived using the equation: where's λ max represents the principal eigenvalue and n is number of factors.
Based on that the RCI is the Random Consistency Index value, whose values were quoted from the Saaty's standard (Saaty 1977;1980). Thus the CR (Table 2) can be computed as below: 1 3 Based on this equation, the estimated parameters' CR is 0 (CR = 0/1.49), and they are accepted as the consistency is acceptable if the CR is less than 0.1; else, the AHP is meaningless (Saaty 1980). The predictive layers are given a score, and the sub-features are given a rank according to their efficiency in controlling and holding FFHZ in this predicted methodology. The specified degree ranges from 1 to 8, with 8 being the most effective and 1 being the least hazards hazardous. To make the sub-feature rating, the score of each sub-feature is stanardized ( Sf ). The FFHZ (Eastman 1996) is calculated by multiplying the previous map's scores by the corresponding grade of a thematic map (Li), as shown follows (1):

Flash flood conditioning factors
Different thematic layers were employed as conditioning elements to prepare a flash flood susceptibility map. The lithology, slope, TWI, SPI, STI, TRI, Dd, distance to river (stream), radar intensity map, and rainfall distribution map are among the ten factors that can be used to create a flash flood conditioning factor dataset. The RS and geological data were used to gather these parameters (Figs. 4, 5 and 6).
The STI is a key criterion in the analysis of flash flood hazards because it depicts the eroding capacity of overland streams attributed to two variables: alluvium flow transport concentration and basin evolution (Devkota et al. 2013;Pradhan et al., 2014;Bui et al., 2019). It classified into four classes: 0-1, 1-10, 10-43, and 43-307 (Fig. 5a).
The aforementioned flash flood-related factors of TWI, SPI and STI were measured using following equations:

TWI = Ln
As tanβ  where As is the specific catchment area (m 2 m −1 ), and β (radian) is theslope gradient (in degrees) (Regmi et al. 2010). The TRI is used to create a flash flood hazard map. It's an essential aspect to consider while simulating and detecting flash floods. The TRI is among the morphological variables associated with flooding (Werner et al. 2005;Tehrany et al. 2019). The following equation can be used to determine this parameter: where the max and min are the largest and smallest values of the cells.
The Dd is the proportion of the basin's total area to the total length of the watershed channels. Drainage density is inversely related to flooding (Khosravi 2019). By dividing the total length of the complete drainage stream by the total area of the following, the drainage density (km/km 2 ) may be calculated: Dd = ∑ Li∕A Higher drainage density, which indicates significant surface runoff, is directly linked to a greater chance of floods (Paul et al. 2019). A natural break was used to divide the drainage density map into five categories (Fig. 5c): 0 to 0.24 (1), 0.24 to 0.42 (2), 0.42 to 0.61(5), and 0.61 to 1.10. (8).
The moisture content of soil and rock on the slope is affected by the distance to rivers (Miraki et al. 2019). In comparison to further distances from river networks, this element can alter the recharging process since the shorter the distance from the river, the higher the possibility of infiltration Bui et al. 2018). The Distance to River was created using five classes based on the DEM of the study area (Fig. 5d): 0 (7), 0 to 759.55 (4), 759.55 to 1210.14 (2), and 1210.14 to 3282.83. (1).
Satellite-based SAR can overcome fog, aerosol, gas, water molecules, and even sand cover at any time of day or night. It identifies and recognizes clastic sediment vicinities in opposition to bedrock. Except in seasonal conditions, precipitation is low in locations of high aridity. This encourages erosion and sand accumulation while also facilitating recharge. Due to scattering, the ALOS/PALSAR (L-band; = ƛ = 24 cm) has the ability to portray the properties of soil and expose fine sediments in a dark tone rather than bedrock in a brilliant tone (Jensen 2000). As a result, fine-textured stretches, such as sand and gravel, would be suitable recharging regions (Abdelkareem and Al-Arifi 2021). Arc GIS software systems (Fig. 6a) do a good job of categorizing data backscattering into five groups: very low, low, moderate, high, and very high zones (Fig. 6a), which occupy 0-20 (1), 20-80 (2), 80-100 (3), 100-209 (4), and 209-255 (5)% of the total range.
Rainfall is the primary cause of an increase in subsurface hydrostatic pressure and level. It is linked to river discharge, and heavy rain in a short period of time can cause flash floods  (Fig. 6b).

Flash flood susceptibility map
The multi-criteria decision-based AHP-overlay analysis fusion approaches were used to create flood risk vulnerability predictions. With the use of numerous classified criteria, these are capable of forecasting flood vulnerability. Through GIS-based overlay analysis and normalization using the AHP approach, a flash flood hazard map was created by integrating ten conditional components. Flooding vulnerability is indicated by a higher weight. On the other hand, lower weight levels indicate that flooding is less likely to occur. Using the natural break method, the outcome map was divided into six zones, viz., extreme hazard, very strong, strong, moderate, low, and very low susceptibility, covering 6. 86, 15.04, 18.74, 22.58, 22.80, and 13.98% of the entire WQB, respectively ( Fig. 7 and Table 4). The flood hazard grades indicate that these areas are more prone to severe flooding. The entire area (sectors I & II) under construction of the NQC is divided into six extreme hazard categories: very strong, strong, moderate, low, and very low susceptibility that cover 13. 23, 22.76, 25.68, 25.59, 12.62, and 0.13%, respectively (Fig. 7a, Table 4). According to the results of flash flood susceptibility maps (Fig. 7), the majority of sector one (I) of the New Qena city area, which is located in the flood plain of the WQ watershed, falls into the very strong to extreme susceptibility zone. This is due to the city's rapid growth and expansion, which has encroached on topographically sensitive areas susceptible to flash floods. Furthermore, the construction of buildings and roads in metropolitan areas increases the area of impervious infrastructure, resulting in less infiltration and more runoff, resulting in flash floods in the case of serious  The much more significant reasons for flood susceptibility in these places are human operations that generate substantial changes in watershed geometry, including terrain elevations, existing drainage alterations, and increased impervious surfaces in the metropolis (Fig. 8a-c, 9). This has increased the likelihood of flooding and flooding of basic infrastructure, posing a greater risk of death and economic damage (Pham et al. 2020). Despite the area in high risk, change detection maps allowed revealing the major changes in urban and vegetations since 1984 to the present

Risk management, field observations and recommendations
The computed FFHZs map, based on a basic multi-criteria decision analysis model combining geologic, hydrologic, and topographic parameter calculations, showed the vulnerable areas to FFHZs, including the NQC (Figs.7, 8 and 9). This map is very necessary for the management of water resources and the sustainable development of the area, in addition to the protection from flood hazard. This is because the low-elevation area is surrounded by elevated hills and serves as a water storage location. As the NQC in a flood channel, if water levels are raised to 90 m, 100 m, and 110 m, an area of roughly 16, 53, and 95.95% will be inundated by surface water (Fig. 12 b) (Fig. 12c). In addition, in case of the water surplus the flood channel for a distance of 500, and 1000 m an area of about 10 and 13.72% will be inundated as the area in low topography (Fig. 12d,e). Noteworthy, the area of elevation > 98 m that covering the NQC is the lowest hazards safe and suitable land for future development. The Wadi Qena basin and its sub-basins collect precipitated water during rain storms. The collected water and sediments drain downstream into the Nile through a flood channel (Fig. 12c,d,e). Several obstacles, artificial barriers, and infrastructures (e.g., roads and highways, farms, and bridges along the torrent canal) are subjected to high risk and most of them are destroyed, causing high devastation in the area from the Red Sea -Sohag highway at W. Jurdi to the area surrounding the flood route at the NQC. Since the segments of the Red Sea-Sohag highway cross the wadi at an elevation higher than the wadi floor's level, water blocking cannot be avoided, and hence erosion impacts the highway sides when flooded with water. For example, during the flood of 1996, two of the three bridges on the torrent canal were clogged with remnants of electric rollers that were thrown into the stream outlet. The water floods on both sides of the canal, destroying the village of El Ma'ana on the northern side and the highway of Qena-Safaga, and several houses on the southern side of the canal.
Despite this, some protection measures were considered by the government to reduce the flow of water and prevent water from blocking the culverts in some segments of the roads and highways, dams to change water routes, and raising the bridges on the torrent canal (Figs. 13,14). Due to the clogging of the culverts by trees and plant remains, torrent waters are blocking and forming lakes behind the highways. This allowed the excess water to overflow onto the roads, causing destruction and high risk to the surrounding areas. In addition, the reclaimed lands during the wadi are subjected to complete inundation by water and transported sediments (e.g., sand, gravel, and rock blocks). Moreover, many roads are still sensitive to lateral erosion from flood waters and may still be at risk from hazards associated with flash flood events. Generally, the meandering in the torrent path (Fig. 13b, c, 14) particularly in the area of the NQC into the Nile River, leads to the slaughter of livestock in places facing the rushing waters that cut bridges and destroy property in these places. These places range from low risk at the wide stream outlet to high-risk areas at the narrow stream outlet. These risky places should be protected by an appropriate means (Fig. 13).
Nonetheless, city planners devised a strategy to move the flood problem from the NQC to an artificial flood channel (Feleb 2016). The flood conduit captures water and diverts it east of the city, where it flows for about 12 km (Fig. 14) into the main Nile Valley flood channel. The purpose of this waterway was to handle rainwater runoff and overflow. The newly created tunnels in the floodway channel were meant to quickly drain rainwater during rainstorms (Fig. 13a). Flash floods occur when heavy rains surpass the capacity of the flood channel, destroying dams and canals and causing overflow into the urban areas of the NQC. This operation is insufficient to ensure complete flood safety as > 35% of the entire city is susceptible to FFHs and the area experienced major changes based on InSAR CCD (Fig. 14a, b).
Many procedures should be considered in the WQ basin to provide safety and sustainable development to the NQC (Fig. 15), based on the aforementioned information generated from the FFHZs map and field observations. This assessment, which is required to prevent potential damage, was aided by the creation of a methodology for predicting flood hazard areas' vulnerability.
• Construct concrete bunds of zigzag geometry, water-spreading dams with earth-fill embankments, and retention dams in the area north of the NQC to the highway of the Red Sea -Sohag (Fig. 15b, c).
The management strategy of Wadi Qena floodwater includes different types of embankments and dams especially contour stone bunds, water-spreading dams with earth fill embankments, and retention dams. Local materials can be used for constructing the contour stone bunds and water-spreading dams with earth-filled embankments. Simple stone lines in a zigzag pattern, on the other hand, can be partially effective (Critchley et al., 1992). Contour stone bunds are built in the upstream parts of Wadi Qena sub-basins due to the availability of large stone blocks and to reduce bed erosion and run-off velocity, followed by water-spreading dam's further downstream. Moreover, sediment yield will be reduced downstream, and run-off water will have sufficient time to recharge the groundwater aquifer.
Sub-basins with high peak discharges and sediment yields such as Wadi Jurdi, W. Shahadein, and W. Naq'a E-Tier must be protected with a series of contour stone bunds and water-spreading dam systems. Sub-basins with moderate peak discharge and sediment yield, such as WQ, may be protected by fewer water-spreading dams. Sub-basins with very high sediment yield, peak discharge, and run-off volume, such as WQ, should have a series of water-spreading dams in their upstream region, in addition to downstream retention dams to retain and harvest the run-off water. These dams also trap most of the sediment yields, mainly gravel and stone blocks, transported from the neighboring mountains by these flood sand consequently, preventing the reclaimed areas from re-desertification.
• Construct artificial water storage lakes north the highway of Red Sea -Sohag and north of the NQC. Retention dams are used to store runoff water either to protect Qena cities or to use water for planting reclaimed areas and recharge the groundwater aquifer. Due to the high evaporation rate in the area, it is advisable to reduce the surface extent and increase the depth of the lakes to decrease the water losses through evaporation. To safely discharge the water beyond the dam's storage capacity, a suitable design of the water outlet of the dam is highly recommended.
• Increase the depth of the flood canal at the NQC as the canal is shallow (Figs. 12  and 13).
Retention dams are used to store runoff water either to protect Qena cities or to use water for planting reclaimed areas and recharge the groundwater aquifer. Due to the high evaporation rate in the area, it is advisable to reduce the surface extent and increase the depth of the lakes to decrease the water losses through evaporation. To safely discharge the water beyond the dam's storage capacity, a suitable design of the water outlet of the dam is highly recommended.
• A concrete wall should surround the northeastern side of New Qena city to protect it from inundation and destruction of the flood. • The roads that cross the wadi are built higher than its floor with culverts usually clogged by trees and plant remains and obstructing their work.
Based on that, the highways are elevated above the wadi bed. The water overflows onto these roads, destroying them and forming a high, very destructive wave of water. However, segments of the roads and highway crossing the wadi bed should be lowered to their floor level in order to avoid water blocking and prevent erosion along the highway sides during flash flood events. Warning residents of potential floods enables a timely evacuation. Building the capacities of people living in the downstream area through awareness and training to deal perfectly with flash flood events reduces the scale of the disaster.
• Reinforce the dam (e.g., W. Shahadein) and ensure that the torrent canal is clear of obstructions.

Conclusions
Qena city and its inhabitants were subjected to several flash flood events, some of which were highly destructive and caused serious damage to Qena city and its inhabitants. Now, the new Qena city was established in the downstream area of Wadi Qena and could potentially be damaged by floodwaters and sediments. Therefore, generating a flash flood hazard map is important for assessing and managing flash flood risks. Through GIS-based overlay analysis, several parameters primarily derived from remote sensing data, such as slope, TWI, SPI, STI, TRI, Dd, distance to a river, radar intensity map, and rainfall distribution map, were merged to map flood-vulnerable areas after normalization and weighting by implementing the AHP method. The data enabled the identification of the most vulnerable areas, as well as an evaluation of the flood's impact on NQC. Extreme hazard (6.86 %), very strong (15.04 %), strong (18.74 %), moderate (22.58 %), low (22.80 %), and very low (13.98 %) vulnerability were all found in the WQB output FFHZs. Additionally, approximately 35% of the NQC under development is exposed to extremely significant to incredibly severe hazards.
Regarding the risk assessment of maximum runoff and sediment yield, some sites throughout the wadi were designated moderate to high flood risk categories. Construction of a series of contour stone bunds in the upstream part of the wadi Qena sub-basins, followed by water-spreading dams further downstream, is part of a flood mitigation strategy to reduce bed erosion and run-off velocity. Some retention dams with storage lakes were recommended downstream of some sub-basins for flood protection, water demand control, and aquifer feeding.
Funding Open access funding provided by The Science, Technology & Innovation Funding Authority (STDF) in cooperation with The Egyptian Knowledge Bank (EKB). The authors have not disclosed any funding.

Conflict of interest
The authors declare no competing interests. The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.