Flood Risk Assessment Under Climate Change Scenarios in the Wadi Ibrahim Watershed

Abstract

Flooding poses a significant hazard to urban areas, particularly under the pressures of climate change and rapid urbanization. This study evaluates the flood risk in the Wadi Ibrahim watershed, located in Makkah Al-Mukarramah City, Kingdom of Saudi Arabia (KSA), by analyzing the impacts of climate change on flood hazards. The analysis incorporates projections from the Coordinated Regional Climate Downscaling Experiment (CORDEX) regional climate model (RCM) for three climate scenarios: representative concentration pathway (RCP) 2.6, RCP 4.5 and RCP 8.5. A novel aspect of this study is the integration of 2D HEC-RAS rain-on-grid (RoG) hydrodynamic modeling with climate change projection analysis, which has not been previously applied in this watershed. Flood risk maps are generated for each scenario at three return periods: 50, 100, and 200 years. The results indicate an increasing flood volume and depth under future climate scenarios. The flood risk mapping shows an expansion of medium- and high-risk zones compared to current conditions. Under the current climate, the low-risk areas (0–0.5 m) slightly decrease from 13.9 km² (50 years) to 13.8 km² (200 years), while the medium- (0.5–2 m) and high-risk areas (>2 m) increase from 6.5 km² to 7.0 km² and from 7.2 km² to 9.8 km², respectively. Under RCP 2.6, the low-risk zones decline from 13.6 km² to 13.0 km², the medium-risk zones grow from 14.5 km² to 16.2 km², and the high-risk zones rise from 4.3 km² to 6.5 km². The higher emissions scenarios show greater risk increases, with the high-risk areas expanding from 5.3 km² to 12.0 km² under RCP 4.5, and from 9.5 km² to 16.6 km² under RCP 8.5. These findings underscore the escalating flood risks due to climate change and highlight the need for mitigation in the Wadi Ibrahim watershed.

1. Introduction

Floods are among the most destructive disasters that cause loss of property and life. Since 2000, this disaster has risen by 134% compared to the previous decade [1], and from 2000 to 2019, it was estimated to affect 1.65 billion people [2]. These disasters result in approximately 20,000 deaths annually worldwide, with up to 20 million individuals displaced by flooding each year [3]. Moreover, global economic exposure to flooding is projected to triple by 2050, driven by population growth and expanding economies in regions that are susceptible to floods [4]. In the Kingdom of Saudi Arabia (KSA), flooding was the most frequent hazard during the period between 1982 and 2005, with an average annual return period of seven, resulting in an average annual economic cost of approximately USD 19 million [5]. In Makkah Al-Mukarramah City, flash floods are common during winter even though the amount of precipitation is low. On January 22, 2005, during the Hajj season, a severe rainstorm that was the most intense in over 20 years resulted in substantial human and infrastructural impacts. The storm led to 29 fatalities, 17 injuries, and extensive damage to infrastructure, including the sweeping away of vehicles and destruction of bridges, electrical towers, and communication networks [6].

Floods are becoming more severe and widespread due to climate change. Numerous studies have reported that climate change leads to an increase in extreme rainfall and rising sea levels, both of which play significant roles in the escalation of severe flooding [7,8,9,10,11]. Changing precipitation patterns, particularly in regions experiencing more intense monsoon rain or irregular seasonal rainfall, also increase the flood risk. As climate change profoundly impacts the hydrological cycle and increases the risk of flooding, it is crucial to predict the flood risks under future climate scenarios since future storms may bring significantly more rainfall than historical events [12,13,14].

Predicting future flood-prone areas requires forecasting rainfall patterns using global climate models (GCMs) and representative concentration pathways (RCPs). The ability of this method to project future climate conditions has been recognized by the Intergovernmental Panel on Climate Change (IPCC). The Coupled Model Intercomparison Project Phase 6 (CMIP6) data are currently the most widely available global climate models. CMIP6 includes several GCMs that simulate the 20th century climate to project 21st century conditions, and each GCM incorporates different RCPs (2.6, 4.5, 6.0, and 8.5) that represent varying levels of emissions and climate change projections. Downscaling is essential for applying GCMs to regional and local forecasts. Choosing the correct GCM is crucial for accurately predicting future events. A study explored the suitability of GCMs for impact analysis by downscaling the projected rainfall in Jakarta’s Ciliwung River Basin. The application of these data to simulate river discharge showed that flood-affected areas could expand by 6% to 31% under different GCMs and RCPs [15].

Assessing flood susceptibility by considering the future climate is vital for policymakers and hydrologists to sustainably manage flood risks. The accurate forecasting of flood-prone zones can significantly reduce fatalities and property damage. Researchers have modeled the flood risks at various levels, from global to local, using techniques such as artificial neural networks (ANNs), the analytic hierarchy process (AHP), the frequency ratio (FR), fuzzy logic (FL), logistic regression (LR), etc. This technique can predict future rainfall patterns up to 2100 under four RCP scenarios (2.6, 4.5, 6.0, and 8.5) using an ensemble of GCMs to produce flood susceptibility maps of the study area. This study offers valuable insights into early flood warnings, evacuation strategies, and flood management. This provides a clear understanding of how climate change affects the future flood risks in the area.

The novelty of this study lies in the integration of high-resolution 2D HEC-RAS rain-on-grid (RoG) hydrodynamic modeling with localized climate projection data from the Coordinated Regional Climate Downscaling Experiment (CORDEX), which has not been previously applied to the Wadi Ibrahim watershed. This approach enables a more detailed spatial analysis of the future flood risks, particularly under different climate change scenarios and return periods. The aim of this study is to assess the future flood risks in the Ibrahim watershed b considering climate change projections. Accurately evaluating the flood risks in the context of climate change and development is crucial as it forms the basis for effective water resource management and flood prevention, which is vital for safeguarding lives and property from flood-related hazards.

2. Study Area and Data Collection

The Wadi Ibrahim watershed is one of the most important watersheds in Makkah Al-Mukarramah City, primarily because of its proximity to the Al-Haram Mosque, located in the lower part of the watershed, as depicted in Figure 1. Makkah Al-Mukarramah City, the administrative capital of the Makkah Region, spans an area of approximately 1200 km². It lies between 39°53′ E and 40°02′ E longitude, and 21°09′ N to 21°37′ N latitude. The urbanization rate is high, with extensive residential and commercial development. By 2017, the city’s population had reached approximately 2,017,793, almost double the 2003 figure of 1,375,000 people [16]. The city is located at an elevation of 277 m above mean sea level (amsl) and 80 km inland from the Red Sea.

The Wadi Ibrahim watershed is classified as an arid catchment characterized by low precipitation, frequent droughts, and limited water resources. Geologically, the city is situated in the west–central section of the Proterozoic Arabian Shield, which includes three main rock types: igneous, metamorphic, and sedimentary. The city’s landscape is intersected by multiple structural valleys, faults, and fractures [17]. In particular, the Ibrahim watershed catchment is dominated by Precambrian rocks with quartz diorite and tonalite as the primary minerals [18].

Climatically, the winter season in Makkah Al-Mukarramah City is relatively short, lasting from November to February, with temperatures ranging from moderate to slightly warm, averaging between 18 and 28 °C. In contrast, the summer season is much longer and is characterized by high temperatures, often reaching a maximum average of 42 °C, contributing to intense daytime heat, while the average summer temperature is approximately 36 °C. The daytime temperatures typically peak at their daily maximum, whereas the nighttime temperatures are considerably lower. The World Meteorological Organization (WMO) reports that the long-term annual mean temperatures in Makkah Al-Mukarramah City from 1982 to 2011 were as follows: a maximum of 43.8 °C, an average of 31.4 °C, and a minimum of 18.8 °C. These data were obtained from the WMO (https://worldweather.wmo.int/en/city.html?cityId=700 (accessed on 10 September 2024)). Figure 2 illustrates the long-term mean monthly values for the maximum, average, and minimum temperatures.

3. Data Collection and Methodology

The flowchart in Figure 3 illustrates the general framework of this study.

3.1. Data Collection

The DEM of the Ibrahim watershed was sourced from the Copernicus DEM with a 30 m resolution, accessible via https://opentopography.org/ (accessed on 7 June 2024). The Copernicus DEM is a digital surface model (DSM) that represents the Earth’s surface, including features such as buildings, infrastructure, and vegetation. It is based on the refined WorldDEM, which incorporates adjustments, such as flattened water bodies and consistent river flow patterns. Additionally, it includes modifications to shoreline and coastline details, corrections for special features, such as airports, and adjustments to address unrealistic terrain structures. The delineation and calculation of the morphometric parameters of the Wadi Ibrahim watershed were performed using ArcHydro toolbox in GIS 10.8 version software. Some of the morphometric parameters are listed in

Table 1. Morphometric parameters of the Ibrahim watershed.

Basin Parameters	Ibrahim Watershed
Low Elevation (m)	210
High Elevation (m)	968
Area (km2)	110.8
Perimeter (km).	107.2
Longest flow path (km)	34.6
Basin Length (km)	28.6

. Figure 4 shows the watershed information, including the elevation, longest flow path, and slope. The elevation of a watershed plays a crucial role in its vulnerability to flooding; areas at lower elevations are generally more prone to flooding than those at higher elevations. Additionally, the slope of the basin significantly influences the flood risk, as it affects the energy, velocity, and flow direction of the runoff. Steeper slopes increase the surface runoff, which in turn reduces the infiltration rates. The relationship between the slope and the runoff intensity is a key factor in assessing and predicting the flood risk within a watershed [19].

The curve number (CN) in this study was used in the HEC-HMS software to estimate the runoff based on the land use, soil type, and antecedent moisture conditions. It plays a key role in frequency analysis by determining the runoff depth for various rainfall events, which is essential for flood risk assessment. In this study, the CN values were derived using the curve number generator plugin in QGIS (https://github.com/ar-siddiqui/curve_number_generator (accessed on 10 June 2024)) [20] as shown in Figure 5A). This tool employs various algorithms to create a curve number layer for any specified area using different datasets. The CN ranged from 40 to 100, and after calculating the composite CN, the resulting value was 74.

The land use and land cover (LULC) data for the Wadi Ibrahim watershed were also incorporated into the analysis. The LULC map for 2020 was obtained from the European Space Agency (ESA) WorldCover dataset, which provides freely accessible global land cover data at a 10 m resolution (https://viewer.esa-worldcover.org/worldcover/ (accessed on 20 June 2024)). The watershed was classified into seven land use types: tree cover, grassland, shrubland, cropland, bare land/sparse vegetation, built-up areas, and permanent water bodies. The analysis revealed that built-up areas dominate the region, covering approximately 29.9% of the total area. Urban and impervious surfaces are associated with higher CN values due to their reduced infiltration capacity, which increases the runoff potential. The hydrological soil group (HSG) is also a critical component of the U.S. Department of Agriculture’s Soil Conservation Service (USDA-CN) for estimating rainfall–runoff (Figure 5B). The four standard classes (A, B, C, and D) represent soils with different runoff potentials ranging from low to high. Class A has the lowest runoff potential, and Class D exhibits the highest runoff potential [21].

Rainfall data were collected from six stations located within the watershed: J114, MK139, Al Adel, Mena, Electricity, and Al Maesem (Figure 6). Station J114, an older station, has provided over 56 years of recorded maximum daily rainfall data, spanning from 1967 to 2022, with the recorded values ranging from a minimum of 0 mm to a maximum of 100 mm. In contrast, the more recent stations, Al Adel, Mena, Electricity, and Al Maesem, have recorded data from 2017 to 2024, while Mk139 has data from 2019 to 2022. The summary of the rainfall data, presented as new record data, is based on the average of overlapping records from these stations from 2017 to 2024, as illustrated in Figure 7.

Compared to Figure 2, which presents the monthly climate averages from 1982 to 2011 based on WMO data, it shows seasonal rainfall peaks in January, November, and December with dry summer months. Figure 2 displays the maximum daily rainfall records from 1967 to 2024 based on observed data from multiple local stations. It highlights greater rainfall intensities and more frequent extreme events. Station J114 offers long-term variability, while the inclusion of newer stations improves spatial coverage. The observed data in Figure 6 show a higher average daily rainfall (19.1 mm) compared to the WMO dataset (8 mm), indicating a shift in the precipitation patterns. These recent trends suggest more intense, shorter-duration, and frequent rainfall event characteristics aligned with anticipated climate change impacts.

Climate change scenario data were obtained from the Coordinated Regional Climate Downscaling Experiment (CORDEX) regional climate model (RCM) data at a single level (https://cds.climate.copernicus.eu/ (accessed on 3 October 2024)). The domain was the Middle East and North Africa (MENA), and the output variable was the mean precipitation flux, which is the deposition of water to the Earth’s surface in the form of rain, snow, ice, or hail. The precipitation flux is the mass of water per unit area and time (kgm⁻²s⁻¹). The data represent the average over the aggregation period, and the horizontal resolution was 0.44° × 0.44°, corresponding to a grid spacing of approximately 50 × 50 km. The global climate models (GCMs) were NOAA-GFDL-ESM2M and ICHEC-EC-EARTH, while the RCM was SMHI-RCA4. Detailed information on the data collection is shown in

Table 2. Data collection for the climate change scenarios (RCP 4.5 and RCP 8.5).

Domain	Middle East and North Africa
Experiment	RCP 2.6, RCP 4.5, and RCP 8.5
Horizontal resolution	0.44 degree × 0.44 degree
Temporal resolution	Daily mean
Variable	Mean precipitation flux
Global climate model	NOAA-GFDL-ESM2M (USA)
Regional climate model	SMHI-RCA4 (Sweden)
Ensemble member.	r1i1p1
Start year	2006
End year	2100

.

The CORDEX climate projection experiments used representative concentration pathway (RCP) forcing scenarios, typically covering the period from 2006 to 2100. In this study, three scenarios, shared socioeconomic pathway (SSP) 2.6, SSP 4.5 and SSP 8.5, were selected to represent the intermediate- and high-emission pathways, respectively. Figure 8 presents a comparison between the three climate change scenarios and the observed data (2006–2022). Both RCPs generally show higher values than the observations. Interestingly, there is some variability between the two RCPs in this period, where the RCP 4.5 values exceed those of RCP 8.5 in certain years and vice versa. However, when considering the long-term projection up to 2100, RCP 8.5 consistently presents higher values than RCP 4.5, as expected given the higher emission assumptions.

GCMs often contain systematic errors or biases such as overestimation of the number of rainy days, underestimation of extreme rainfall, or misrepresentation of the temperature and seasonal rainfall. Therefore, the RCM or downscaling process plays a crucial role in enhancing climate change data at the regional and local spatial scales. They resolve physical processes at a higher resolution than GCMs. This downscaling approach enables a more detailed representation of regional climate features that are not captured by GCMs, such as complex terrain atmospheric dynamics, some land–atmosphere interactions, and coastal influences.

This study employed the CMhyd tool for bias correction to refine the simulated climate data to match the observed values at the gauge locations within a watershed model. Bias correction adjusts the climate model outputs using transformation algorithms via several methods, such as delta change (DC), which applies historical differences to future projections but assumes constant biases; multiple linear regression (MLR), which uses linear relationships to correct averages but struggles with extremes; and quantile mapping (QM), which aligns the entire data distribution to correct both the average and extreme conditions [22,23]. The results showed that the total rainfall from 2006 to 2022 increased from 1288.9 mm (average: 17.2 mm/year) before correction to 2224 mm (average: 29.7 mm/year) after bias correction.

3.2. Methodology

3.2.1. Rainfall Analysis

In this study, the Gumbel distribution was selected to ensure consistency with previous research in arid regions, where the Gumbel analysis is widely used for rainfall frequency modeling. For instance, In a study of the Makkah Al Mukarramah region, six different distributions were compared, and the Gumbel distribution was found to provide the best fit for the four stations analyzed [24]. Given the available data and the regional context, it remains a reliable and practical choice for this study. The rainfall depths for the various return periods were then estimated using the Gumbel distribution equation as follows [25]:

$$x=\beta -{\displaystyle \frac{1}{\alpha }}\ln \left[\mathrm{l}\mathrm{n}Tr-\ln \left(Tr-1\right)\right]$$

(1)

$$\alpha ={\displaystyle \frac{1.2825}{\sigma }}$$

(2)

$$\beta =\mu -0.45\sigma$$

(3)

where $x$ is the rainfall depth, $\alpha$ and $\beta$ are the distribution parameters, $Tr$ is the return period, $\mu$ is the mean of the rainfall data, and $\sigma$ is the standard deviation of the rainfall data. This formula was also successfully applied in the development of intensity–duration-frequency (IDF) curves for SA [26]. The HEC-HMS model was utilized to assess and simulate the rainfall–runoff dynamics within the watershed.

Table 3. Average rainfall values across the stations for the various return periods and climate scenarios.

		Coordinates		Rainfall (mm) at Different Return Periods (Years)
Scenarios	Stations	Long (E)	Lat (N)	50	100	200
	Al Adel	39.85	21.44	84.70	94.80	105.00
	Mena	39.87	21.43	77.90	87.50	97.00
	Al Maesem	39.92	21.46	59.71	66.47	73.19
	Electricity	39.88	21.46	88.36	99.42	110.45
	J114	39.83	21.44	97.19	110.56	123.89
	M139	39.82	21.41	98.64	113.55	128.41
Current	Average			94.51	107.09	119.62
RCP 2.6				141.72	164.58	187.35
RCP 4.5				150.50	175.90	201.10
RCP 8.5				184.30	215.80	247.30

presents the rainfall depths under the current climate, as well as projections under RCP 2.6, RCP 4.5 and RCP 8.5 scenarios, for return periods of 50, 100, and 200 years. These return periods are commonly used in hydrological and flood risk assessments to represent varying levels of risk, corresponding to annual exceedance probabilities of 2%, 1%, and 0.5%, respectively. The 50-year return period is typically applied in urban stormwater system design, the 100-year is the standard for floodplain management, and the 200-year is used for critical infrastructure that must account for rare, high-impact events. As shown in Table 3 and Figure 8, the RCP projections exhibit variability compared to the observed data during the historical period (2006–2022). In some years, the RCP models underestimate the rainfall events. For example, in 2008 and 2016, RCP 4.5 projects much higher rainfall (204.6 mm and 203.9 mm) compared to the observed 86 mm and 17 mm, indicating a tendency of overestimation for extreme events. Conversely, in other years, such as 2011 or 2012, the RCPs underestimate the rainfall compared to the observations. The differences between the observed and projected rainfall are expected due to the inherent uncertainty in climate models at local scales. RCP models are designed primarily for future climate projections rather than exact replication of past events; however, they still capture the general trends of increased rainfall under higher emission scenarios (as seen from the increasing averages across RCP 2.6, 4.5, and 8.5 in Table 3).

Furthermore, the analysis shows that the average rainfall is projected to increase across all the return periods under the future climate scenarios. For the 50-year return period, the rainfall increases by 50% (RCP 2.6), 59.3% (RCP 4.5), and 95% (RCP 8.5). For the 100-year return period, the increases are 53.7%, 64.3%, and 101.6%, respectively. For the 200-year return period, the rainfall rises by 56.6%, 68.1%, and 106.8%. These findings highlight substantial increases under all scenarios, with RCP 8.5 nearly doubling the rainfall at higher return periods.

Figure 9 presents the rainfall distribution in the Wadi Ibrahim watershed under the current climate scenario across the different return periods. For the 50-year return period, the rainfall ranges from 91 to 99 mm, with higher concentrations in the northern part of the watershed and lower values in the south. As the return period increases to 100 years, the rainfall intensifies to 99–114 mm, maintaining a similar spatial pattern, with the peak rainfall still concentrated in the north. For the 200-year return period, the rainfall further increases to 106–128 mm, indicating a notable rise compared to the shorter return periods and reaffirming the northern region as the most impacted area. Under the future climate scenarios, the projected rainfall increases significantly, reaching up to 201 mm under RCP 2.6, 216 mm under RCP 4.5, and 265 mm under RCP 8.5. The isohyet lines illustrate variations in the rainfall across the region, with denser lines in the northern part, indicating a sharp gradient of rainfall change over a short distance.

The frequency analysis in Figure 10A shows the relationship between the rainfall depth and the return periods under different climate scenarios based on the NOAA model projections. The observed data (green) represent the historical rainfall trends, while the RCP 2.6 (blue), RCP 4.5 (purple) and RCP 8.5 (red) scenarios project the future rainfall depths under low, moderate and severe climate change conditions, respectively. The Gumbel distribution for RCP 8.5, represented by the red line, shows a steeper slope, indicating that extreme rainfall events will become more frequent and intense under severe climate change. This trend suggests a heightened future flood risk. Although the data show some deviation from the Gumbel distribution, particularly for higher return periods, it was selected due to its widespread use and acceptance in hydrological frequency analysis. The Gumbel distribution provides a simple and reliable method for estimating extreme events and demonstrates the best goodness of fit based on the Kolmogorov–Smirnov (K-S) test, with a lower error compared to six other distributions [24].

Figure 10B presents the simulated flood hydrographs that illustrate the peak discharge and flow duration for the 50-, 100-, and 200-year return periods. The results indicate a significant increase in the peak discharge under future climate conditions, with RCP 8.5 producing the highest peaks, followed by RCP 4.5, and then RCP 2.6, which remains closer to current conditions. Under RCP 8.5, the 200-year return period exhibits the most extreme flood response, characterized by a pronounced increase in the peak flow and a prolonged recession period.

3.2.2. Hydraulic Modeling Using Rain-on-Grid (RoG)

Rain-on-grid (RoG), also known as the direct rainfall method (DRM), is a two-dimensional (2D) hydrodynamic model that integrates both hydrological and hydrodynamic processes in flood simulation. This method simulates the catchment runoff by directly applying rainfall to a 2D modeling grid. Some routing parameters, such as the topography, roughness, and loss mechanisms, should be considered. This technique has gained popularity in storm risk management because it has some advantages, such as the ability to produce satisfactory stage hydrograph responses with proper calibration and data representation [27], success for effective rainfall generation using SWAT [28], and sensitivity to topographic data quality, and its potential as an alternative to traditional hydrological modeling methods [29]. The application in this study was performed by some researchers, such as succeeding to determine the still flooded areas around the Pluit Polder Jakarta, with a flood depth ranging from 0.25 m to 2.75 m [30]. It has also been used to replicate flash flood events in various locations, such as the East Branch Du Page River in Illinois [31], a small catchment in western Norway [32], and the Adyar Basin in Chennai, India [33]. RoG models have been evaluated against flow hydrographs for urban flood simulations, demonstrating their effectiveness in depicting overland flow and supporting urban drainage planning [34].

The principle of RoG is to calculate the water surface elevation and velocity based on the Saint-Venant equations (SVEs), also known as the shallow water equations (SWEs). These equations describe the conservation of mass and momentum in a fluid flow and are derived from the fundamental principles of fluid mechanics. The SVE is particularly relevant for modeling diffusive flood waves (DFWs), an approximation of a full hydrodynamic wave model that has been extensively applied in various field studies [35,36,37,38,39]. The simplified 2D SVE is shown below [40].

$${\displaystyle \frac{\mathsf{\partial }H}{\mathsf{\partial }t}}+{\displaystyle \frac{\mathsf{\partial }\left(hu\right)}{\mathsf{\partial }x}}+{\displaystyle \frac{\mathsf{\partial }\left(hv\right)}{\mathsf{\partial }y}}=r-i$$

(4)

$${\displaystyle \frac{\mathsf{\partial }u}{\mathsf{\partial }t}}+u{\displaystyle \frac{\mathsf{\partial }u}{\mathsf{\partial }x}}+v{\displaystyle \frac{\mathsf{\partial }u}{\mathsf{\partial }y}}=-g{\displaystyle \frac{\mathsf{\partial }H}{\mathsf{\partial }x}}+v_t\left({\displaystyle \frac{{\mathsf{\partial }}^{2}u}{{\mathsf{\partial }x}^2}}+{\displaystyle \frac{{\mathsf{\partial }}^{2}u}{{\mathsf{\partial }y}^2}}\right)-c_{f}u$$

(5)

$${\displaystyle \frac{\mathsf{\partial }v}{\mathsf{\partial }t}}+u{\displaystyle \frac{\mathsf{\partial }v}{\mathsf{\partial }x}}+v{\displaystyle \frac{\mathsf{\partial }v}{\mathsf{\partial }y}}=-g{\displaystyle \frac{\mathsf{\partial }H}{\mathsf{\partial }y}}+v_t\left({\displaystyle \frac{{\mathsf{\partial }}^{2}v}{{\mathsf{\partial }x}^2}}+{\displaystyle \frac{{\mathsf{\partial }}^{2}v}{{\mathsf{\partial }y}^2}}\right)-c_{f}v$$

(6)

$$c_f={\displaystyle \frac{n^{2}g\left|V\right|}{h^{\frac{4}{3}}}}$$

(7)

where h is the water depth, H is the water surface elevation (sum of the bed elevation and water depth), t is the time, u and v are the velocities in the x and y directions, r is the rainfall, i is the infiltration (delta between r and i as net precipitation), g is the gravitational acceleration, $v_t$ is the horizontal eddy viscosity, $c_f$ is the bottom friction coefficient, n is the Manning coefficient, and $\left|V\right|$ = resultant velocity $\left(\left|V\right|=\sqrt{u^2+v^2}\right)$. Equation (1) has the mass continuity as a source term to obtain the water depth H, while Equations (2) and (3) have the momentum conservation to calculate the velocities u and v at each grid.

The RoG model is available in many software packages, like HEC-RAS. The RoG procedure in HECRAS starts with inputting DEM data as terrain with a projection of a 37N UTM. The shape files of the Wadi Ibrahim watershed use 2D flow areas in the perimeter section. It is necessary to set up an unsteady flow data normal depth of 0.01 m, boundary conditions, and precipitation data. The soil data, LC, curve number (CN), hydrologic soil group (HSG), and Manning value of the study area are parameters that are needed to estimate the rainfall losses and are used as parameter settings for the infiltration value using the SCS method in HEC-RAS [41].

3.2.3. Flood Risk Matrix

Risk indices and risk matrices have become prevalent tools for risk assessment and prioritizing remediation alternatives. The popularity of risk matrices is owing to their ability to quickly assess risks using simple, cost-effective solutions. These matrices offer a clear and efficient way to evaluate potential risks by categorizing them based on their likelihood and impact. A typical risk matrix, as shown in

Table 4. Flood risk matrix adapted from the Federal Emergency Management Agency (FEMA) (2004).

			Extent of Consequences
Probability of occurrence		<0.1 m	0.1–0.5 m	0.5–1 m	1–2 m	>2 m
		Low	Minor	Major	Severe	Catastrophic
	Possible (1 in 50 years)					High Risk
	Unlikely (1 in 100 years)	Low Risk		Medium Risk
	Rare (1 in 200 years)

, categorizes risks on the y-axis using ordinal descriptors such as rare, unlikely, and possible, whereas the x-axis classifies the potential consequences with terms such as minor, major, severe, or catastrophic [42,43]. The matrix is then populated with the corresponding risk values (low, medium, and high) or color-coded representations (green, yellow, and red). In this study, flood risk is defined based on the inundation depth derived from hydrodynamic simulations using the 2D HEC-RAS RoG model. Areas are categorized into three risk levels: low risk (0–0.1 m), medium risk (0.1–2 m), and high risk (>2 m) of flooding. This classification provides a practical measure of the potential flood impact severity on urban infrastructure and population.

4. Results

The flood simulation and depth comparison across the three return periods for the different scenarios are illustrated in Figure 11, emphasizing the variations at a specific location along profile line A-A’. Figure 11A displays the model output, Figure 11B shows the conditions before flooding, Figure 11C shows after flooding, and Figure 11D provides a comparison of the inundation depth along profile line A-A’ for all the scenarios. The horizontal distances, approximately 447 m, represent positions within a flood-prone area. The flood depth increases with the return period and severity of climate change. Under the current scenario, the depths increase by 0.34 m (50–100 years) and 0.53 m (100–200 years). Under RCP 2.6, the respective increases are 0.75 m and 0.57 m. In RCP 4.5, the depth increases are 1.34 m and 0.87 m, while in RCP 8.5, they reach 1.03 m and 1.04 m, respectively. These increasing depths across the return periods indicate escalating flood severity under future climate conditions. Compared to the current scenario for the 50-year return period, RCP 2.6 results in an additional depth of 1.5 m, RCP 4.5 leads to an increase of 2.1 m, and RCP 8.5 shows the highest increase of 3.3 m. These findings underscore the significant impact of climate change on flood hazards, particularly under high-emission scenarios.

Figure 12 presents the flood risk maps for the return periods of 50, 100, and 200 years under the current climate scenario, as well as for RCP 2.6, RCP 4.5 and RCP 8.5. The risk levels are categorized into three classes based on the matrix in Table 3: low risk (<0.5 m), medium risk (0.5–2 m), and high risk (>2 m). These categories are visually represented by a color gradient from green to red. Red zones indicate high-priority areas requiring immediate flood mitigation measures, whereas yellow zones represent relatively lower-priority areas. Green zones do not pose a significant hazard. The flood risk classification provides valuable information for flood management and mitigation planning.

The analysis of the flood parameters across the different return periods and climate scenarios (current, RCP 2.6, RCP 4.5, and RCP 8.5) highlights a significant increase in flood severity under the future climate projections. The inundation volume increases notably with both the return period and the climate change intensity. Under current conditions, it ranges from 18,919 × 10³ m³ (50-years) to 24,821 × 10³ m³ (200-years). Under RCP 2.6, it increases to 28,793 × 10³ m³ (50-year) and 38,927 × 10³ m³ (200-year). In RCP 4.5, the volumes nearly double, reaching 33,407 × 10³ m³ (50-year) and 64,947 × 10³ m³ (200-year). The largest increases are observed under RCP 8.5, with the inundation volumes rising to 44,528 × 10³ m³ (50-year) and 86,061 × 10³ m³ (200-year), underscoring the escalating flood risks under more severe climate conditions.

The flood depth also exhibits a clear upward trend, with the average depth increasing from 0.2 m (200 years) under current conditions to 0.3 m under RCP 2.6, 0.6 m under RCP 4.5, and 0.8 m under RCP 8.5, indicating progressively deeper flooding in future scenarios. Similarly, the peak discharge and runoff volume show significant increases under the future climate scenarios, reflecting the intensifying severity of flood events. Compared to the current conditions, the projected changes under RCP 2.6, RCP 4.5 and RCP 8.5 indicate a substantial rise in the discharge rates, particularly for higher return periods. This suggests that extreme rainfall events lead to more intense and prolonged flooding. Details are presented in

Table 5. Summary of the inundation depth for each return period under the current climate, RCP 2.6, RCP 4.5 and RCP 8.5 scenarios.

	Current Climate			RCP 26			RCP 45			RCP 85
Parameters	Return Period (Years)			Return Period (Years)			Return Period (Years)			Return Period (Years)
	50	100	200	50	100	200	50	100	200	50	100	200
Flood vol (1000 m3)	18,919	21,845	24,821	28,793	34,091	38,927	33,407	47,572	64,947	44,528	63,583	86,061
Max depth (m)	11.8	12.5	12.7	12.9	13.3	13.5	13.2	15.7	18.4	14.9	18.3	21.5
Average depth (m)	0.17	0.19	0.22	0.25	0.30	0.34	0.29	0.42	0.57	0.39	0.56	0.76
Peak discharge (m3/s)	583	701	823	1045	1283	1536	1245	2032	3166	1844	3111	4999
Runoff vol (1000 m3)	6159	7335	8533	10,701	13,000	15,327	12,655	19,829	30,014	18,129	29,524	46,428

.

Figure 13 shows a comparison of the flood risk area coverage across the three scenarios. The analysis of the risk level coverage across the different return periods and climate scenarios (current climate, RCP 2.6, RCP 4.5, and RCP 8.5) shows significant changes in the flood risks, with an increasing trend at more severe flood depths. Under the current climate, the flood risk distribution remains relatively stable across the return periods. Low-risk areas (0–0.5 m depth) cover 13.9 km² for the 50-year return period and slightly decrease to 13.8 km² for the 200-year period. Medium-risk areas (0.5–2 m) expand modestly from 6.5 km² to 7.0 km², while high-risk areas (>2 m) increase from 7.2 km² to 9.8 km². Consequently, the total flood-prone area rises from 27.6 km² to 30.6 km².

Under RCP 2.6, the low-risk areas decline from 13.6 km² (50-year) to 13.0 km² (200-year), while the medium-risk zones expand from 14.5 km² to 16.2 km². The high-risk areas increased from 4.3 km² to 6.5 km², raising the total affected area from 32.4 km² to 35.7 km², indicating a moderate increase in flood severity. Under RCP 4.5, the flood risk shifts more substantially. The low-risk areas decrease from 13.3 km² to 11.5 km², the medium-risk areas expand from 15.3 km² to 19.5 km², and the high-risk areas nearly double from 5.3 km² to 12.0 km². The total flood-prone area increases from 34.0 km² to 43.0 km².

RCP 8.5 shows the most pronounced escalation in the flood risk. The low-risk areas shrink from 12.0 km² to 10.2 km², while the medium-risk areas remain relatively stable at 20.3 km² and 20.8 km², respectively. The high-risk zones experience the largest expansion, growing from 9.5 km² to 16.6 km². Consequently, the total flood-prone area increases from 41.8 km² to 47.7 km². These findings highlight a progressive increase in the flood risk as the climate scenarios shift from the current condition to RCP 2.6, RCP 4.5 and RCP 8.5. The expansion of the medium- and high-risk areas underscores the escalating flood hazards, emphasizing the urgency of adaptive flood management strategies.

5. Discussion

5.1. Envelope Curves for $Q_p$, A, and V

The verification of the flood parameter calculations, including $Q_p$ and V, can be analyzed using Figure 14. This figure presents the envelope curves for the Wadi Ibrahim watershed under different climate scenarios and return periods: (a) the relationship between $Q_p$ and A compared with established global and KSA maximum envelope curves; (b) the relationship between V and A; and (c) the relationship between $Q_p$ and V. The results in Figure 14A indicate that under current climate conditions, the $Q_p$ values remain below the KSA maximum envelope curve, suggesting that the existing flood conditions in Wadi Ibrahim watershed fall within the expected historical limits for the region. However, under future climate change scenarios (RCP 2.6, RCP 4.5 and RCP 8.5), the $Q_p$ values exceed the KSA envelope curve, particularly for the RCP 4.5 200-year, the RCP 8.5 100-year and 200-year return periods. This finding suggests that the projected climate changes may lead to extreme flood events surpassing historical events.

Figure 14B illustrates the relationship between V and A, where the red and green curves represent the maximum and minimum flood volumes, respectively. The results show that all the calculated V values exceed the established envelope, indicating discrepancies between the model predictions and the historical flood records for all the scenarios. It is important to note that previous envelope curves were developed using past measurements based on the Francou–Rodier empirical approach in 1967 [44], whereas the present study incorporates future projections under two climate scenarios. Figure 14C demonstrates that the calculated V values align well within the established envelope, confirming the consistency of the modeled flood estimates with the empirical flood event boundaries. This approach enables a comprehensive assessment of the flood volume variations across different catchment sizes and climate scenarios, thereby reinforcing the robustness of the flood risk analysis.

5.2. Practical Implications of the Flood Risk in an Arid Environment

Although Wadi Ibrahim is located in an arid catchment with low average annual rainfall, it has experienced multiple significant flash flood events, highlighting the paradox of severe flood hazards in dry regions. Historical records show major floods in 1969, 2005, 2006, 2010, 2018, and 2021, with the 1969 event inundating the Holy Mosque with water reaching 2.5 m above the floor following rainfall exceeding 269 mm [6]. During the flash flood of December 2010, substantial water flows were observed moving along municipal streets [18], revealing potential deficiencies in the existing drainage infrastructure. These floods have caused widespread damage to roads, buildings, and urban infrastructure. Such extreme hydrological events are exacerbated by the region’s steep terrain, low infiltration capacity, and rapid urban expansion into natural drainage pathways. In recent decades, anthropogenic activities have significantly altered the hydrological behavior of the watershed, reducing infiltration, increasing surface runoff, and disrupting natural stream networks [45].

The conversion of alluvial plains into urban areas introduces complex hydrological challenges. Therefore, supplementary flood mitigation measures are recommended, such as constructing a series of small dams at the base of tributary channels to retain runoff and sediment. These structures could reduce the peak discharge and runoff velocity while enhancing the groundwater recharge through infiltration into the alluvial deposits. Ideally, excess runoff should be directed to alluvial zones underlain by deep, uncontaminated aquifers that are hydrologically isolated from waste disposal and sewage discharge sites, thereby maximizing the groundwater recharge potential while safeguarding water quality.

The findings of this study directly support flood risk management and mitigation strategies in this arid environment by providing quantitative insights into how flood hazards may evolve under current and future climate scenarios. By mapping the projected increases in the flood volume, depth, and inundation extent, this study identifies priority areas where flood defenses, drainage improvements, and land use controls are most urgently needed. The expansion of medium- and high-risk zones under climate change scenarios underscores the importance of updating urban planning policies to restrict development in vulnerable flood-prone areas and to protect natural floodplains that serve as critical buffers.

Furthermore, by analyzing multiple climate scenarios, this study provides a strong scientific basis for integrating climate projections into flood defense design, supporting adaptive infrastructure resilient against increasing flood magnitude and frequency. The frequency and intensity of extreme rainfall events in arid regions such as Wadi Ibrahim are projected to rise under future climate scenarios [46], amplifying the flood risks despite the region’s low average annual rainfall. This study’s findings can inform emergency preparedness by delineating areas at heightened risk of deep or rapid inundation, enabling the planning of targeted evacuation routes and disaster response measures. It emphasizes that even in water-scarce areas, proactive planning is essential to safeguard lives, property, and infrastructure. By translating projections into practical guidance, it offers actionable insights for developing integrated, climate-resilient flood mitigation strategies in arid urban catchments like Wadi Ibrahim.

6. Conclusions

This study assesses the flood risk in the Wadi Ibrahim watershed under current and future climate scenarios (RCP 2.6, RCP 4.5 and RCP 8.5) for return periods of 50, 100, and 200 years. The results indicate increasing flood depths, discharges, and inundation volumes, with the most severe impacts projected under extreme climatic conditions. The key findings are summarized as follows:

• The rain-on-grid (RoG) model demonstrated reliable performance and logical consistency in simulating the flood dynamics across scenarios.
• Under current climate conditions, the flood volume increased significantly from 18,919 × 10³ m³ (50-year return period) to 24,821 × 10³ m³ (200-year return period), with an average flood depth of 0.2 m.
• Under RCP 2.6, the flood volumes increased to 28,793 × 10³ m³ (50 years) and 38,927 × 10³ m³ (200 years), with the average depths rising to 0.3 m. Under RCP 4.5, the flood volumes nearly doubled, reaching 33,407 × 10³ m³ (50 years) and 64,947 × 10³ m³ (200 years), while the average depths increased from 0.3 m to 0.6 m. The most extreme increases were projected under RCP 8.5, with the flood volumes peaking at 86,061 × 10³ m³ (200 years) and the depths reaching 0.8 m.
• Flood risk mapping indicated a significant expansion of the medium- and high-risk zones. Under current climate conditions, the low-risk areas (0–0.5 m) decreased slightly from 13.9 km² (50-year) to 13.8 km² (200-year); the medium-risk areas (0.5–2 m) expanded from 6.5 km² to 7.0 km²; and the high-risk areas (>2 m) increased from 7.2 km² to 9.8 km².
• Under RCP 2.6, the high-risk areas increased from 4.3 km² (50 years) to 6.5 km² (200 years); under RCP 4.5, from 5.3 km² to 12.0 km²; and under RCP 8.5, from 9.5 km² to 16.6 km².
• The projected increases in the peak discharge and runoff volume under future climate scenarios underscore the escalating flood risks, especially for higher return periods.

These findings highlight the escalating flood risks driven by climate change and emphasize the need for adaptive flood management strategies in the Wadi Ibrahim watershed. In addition, based on the flood risk maps, the results offer valuable insights for prioritizing flood defenses, enhancing early flood warning systems, and ensuring future flood preparedness by identifying vulnerable areas under various climate scenarios. Ultimately, this study emphasizes that even in water-scarce environments, proactive and integrated flood management is vital to protect lives, property, and infrastructure. By translating climate projections into actionable insights, this research supports urban planners, engineers, and policymakers in developing sustainable flood mitigation strategies tailored to the unique challenges of arid urban catchments like Wadi Ibrahim.