Erosion and flood susceptibility evaluation in a catchment of Kopet-Dagh mountains using EPM and RFM in GIS

Erosion and flood events can damage soils, water, quality, and sediment transportation, causing many cumulative hazards. In developing countries, such as Iran, the empirical models, which are low-cost procedures to mitigate environmental hazards, are necessary to plan the watersheds. Hence, the main aim of this study is to evaluate erosion and flood susceptibility using empirical models of erosion potential method (EPM) and rational flood model (RFM) to prioritize the GIS-based prone zones in a catchment of the Kopet-Dagh Mountains. The results revealed that the heavy classes of erosion and flood susceptibility include 40.4–58.2% of the total study area, dominantly in the upstream catchments. The correlation test revealed a strong, significant, and direct association (R equal to 0.705) between W and Qp at the 99% confidence level. Consequently, the results of our research indicated the prioritization of the three sub-catchments based on their slight sensitivity and susceptibility to occurrences of soil erosion and flood events through future spatial developments. Ultimately, the model validity explained the AUC (area under the curve) values averagely equal to 0.898 and 0.917 for erosion and flood susceptibility evaluations (i.e., EPM and RFM), explaining the very good performance of the models and excellent sensitivities.


Introduction
Investigating the nature of the catchments and studying the drainage basin from many aspects, such as flooding, erosion, and sedimentation, have an important role in environmental planning (Brierley and Fryirs 2006;Ahmadi et al. 2020). Erosion and flood events can trigger damage to soils, water, quality, and sediment transportation, causing many cumulative hazards such as land creep and mudflow (Ebrahimi et al. 2021). Evaluation of soil erosion and a flash flood is a planning procedure to combat the threats in each watershed (Pandey et al. 2007). Both soil erosion and floods cause billions of dollars of damage to natural resources and agriculture, particularly in developing countries, where agriculture is the foundation of their economy (Mosavi et al. 2020a).
Floods are often accompanied by severe soil erosion and high sediment concentrations, leading to higher sediment load (Coppus and Imeson 2002;Dai and Lu 2010). Several studies have modeled the susceptibility evaluation and mapping for the flood hazard (e.g., Bui et al. 2018;Zhao et al. 2018a;Janizadeh et al. 2019;Hosseini et al. 2020;Pham et al. 2020) and soil erosion (e.g., Avand et al. 2019;Gayen et al. 2019;Moradi et al. 2019;Amiri and Pourghasemi 2020;Mosavi et al. 2020b). Susceptibility models significantly contribute to accurately identifying the prone zones to implement and develop watershed programs such as flood and soil erosion control (Azareh et al. 2019). Susceptibility evaluation attempts to provide insight for prevention, planning the mitigation actions, and adaptation to extremes to conserve land and water resources in the watersheds (Mosavi et al. 2020a).
In this regard, the erosion susceptibility models are used to evaluate erosion-prone areas concerning land characteristics and the sedimentary potentials of the watersheds (Feng et al. 2010), where water conservation programs are proposed (Fox et al. 2006). Many empirical methods have been developed on the scale of the catchments to evaluate erosion 1 3 490 Page 2 of 15 the sedimentation (Noori et al. 2016), such as the universal soil loss equation (USLE) (Auerswald et al. 2014;Ebrahimi et al. 2021), revised or modified USLE (Baskan et al. 2010;Zhao et al. 2018b), water erosion prediction project (WEPP) (Singh et al. 2011;Srivastava et al. 2015), pacific southwest interagency committee (PSIAC) (Daneshvar and Bagherzadeh 2012;Bagherzadeh and Daneshvar 2013) and erosion potential method (EPM) (Bagherzadeh and Daneshvar 2011;Ahmadi et al. 2020). Among the abovementioned methods, the applicability of the EPM in analyzing erosion potential and spatial data manipulation in GIS is appropriate (Mohammadi et al. 2021). Besides, flood susceptibility models are the proper way to mitigate human and economic damages (Tingsanchali 2012), utilizing the empirical and rational procedures (Su 2017), where the increasing trend of flash floods are observed under the abrupt changes of the hydro-climatic condition (Huang et al. 2017;Ahmadi and Moradkhani 2019;Mihu-Pintilie et al. 2019). For this purpose, the rational flood models (RFM) belong to a group of hydrological models, which concludes the unit of analysis as an integrated value (Aja et al. 2020).
The aforementioned empirical procedures, which are based on natural descriptors of the catchments, have advantages for research because of validation of the model estimations and fewer input-data parameters compared with the process-based models (Schoenau et al. 2008;Lazzari et al. 2015). In developing countries like Iran, direct erosion and flood measurement and analysis in the watersheds are very time-consuming and costly (Spalevic et al. 2020;Mohammadi et al. 2021). Therefore, using empirical models is very necessary to environmental plan the watersheds. Hence, the main aim of this study is to evaluate erosion and flood susceptibility using empirical models of EPM and RFM to prioritize the GIS-based prone zones in a catchment of the Kopet-Dagh Mountains. The study area has some natural and cultural touristic attractions and nowadays has been selected as the main destination of eco-tourism and recreation activities in northeastern Iran, particularly in Mashhad city. The important characteristic of the study area is its geological setting of limestone units, carbonate rocks, steep slopes, and sensitive Karst and water dissolution (Ebrahimi et al. 2019), which are susceptible to erosion and flood events. The susceptibility evaluation of the natural hazards is a way to reduce social and economic damages (Bathrellos et al. 2012). Integrated procedures and multi-criteria methods to model the natural hazards of erosion and flood events allows the preposition and the application of mitigation measures on local and regional scales to increase future social and environmental resilience (see Rozos et al. 2013;Bathrellos et al. 2016). Hence, the present study should expose the prone zones of the study area for future spatial planning and development. We anticipate that this research's output can be considered a low-cost procedure to mitigate the environmental hazards and losses caused by flood and erosion events.

Geographical setting of the study area
The Qarasu watershed with an area of 14.2 km 2 between latitudes 36°56΄-36°59΄ N and longitudes 59°39΄-59°43΄ E is located on the northern side of Kopet-Dagh Mountains, in the Kalat region, northeastern Iran. On a regional scale, the Qarasu catchment and Kalat region, in the vicinity of the border of Turkmenistan and Iran, are highland areas with many environmental phenomena, including geo parks, reserves, eco heritages, and historical touristic sites (Ebrahimi et al. 2019). Nowadays, it has been selected as the main destination for eco-tourism and recreation activities in northeastern Iran around Mashhad city (Fig. 1). From the natural viewpoint, the Kopet-Dagh Mountains are from a basin of sedimentary succession, which was formed by the convergence of Central Iran and Eurasian plates following the Middle Triassic (Berberian and King 1981;Alavi et al. 1997), and continuous sediment depositions took place from the Jurassic and Cretaceous (Mahboubi et al. 2006). According to the digital elevation model (DEM) data, the topographical elevation varies between 2500 m a.s.l in the south and 1000 m a.s.l in the north with > 75% of the area > 1400 m (Fig. 2). Based on the long-term climatic data (1970-2020) acquired from IRIMO (2021) and re-analyzed based on topographic gradients, the study area has a semiarid climate with a mean annual temperature of 13 °C and annual precipitation of 400 mm. In this regard, an adjusted rescaled-plot of temperature and precipitation of the study area is shown in Fig. 3 based on the monthly scale (average from 1970 to 2020). On this basis, the rainiest and driest periods are observed in March and July, with the mean precipitation (temperature) of > 100 mm (< 10 °C) and ~ 0 mm (~ 25 °C), respectively. Meanwhile, in the coldest month of January (averagely < 0 °C), approximately 50 mm of the total precipitation could fall in as snow instead of rainfall. The catchment was divided into 14 homogeneous terrain units of sub-catchments based on the visual interpretation of satellite images and DEM data (Fig. 4). The surface area in the sub-catchments varies between 0.4 km 2 (no. 11) and 1.8 km 2 (no. 6).

Data preparation
The required data for soil erosion and flood susceptibility evaluation were remotely obtained from some global datasets with spatial grid pixels and time series, focusing on the geographical coordination of the study area equal to 36-37° N and 59-60° E. The geological data were extracted from the drawing sheets at the 1:100,000 scale via the Geological Survey of Iran (GSI 2015). Therefore, topographical layouts were derived from a global digital elevation model (DEM) via the National Aeronautics and Space Administration (NASA 2011). The soil units of the study area were extracted from the global soil-grid dataset via https:// soilg rids. org, and land-use types and covers were considered from the global land-use/ land-cover (LULC) database via https:// lpdaac. usgs. gov, retrieved from satellite products in 2010. The time series of daily-rainfall data was collected from the geospatial interactive online visualization and analysis infrastructure (GIOVANNI) program for 2016-2020 via https:// giova nni. gsfc. nasa. gov. The aforementioned data were processed in GIS ver. 10.4 and SPSS ver. 16.1 to produce the spatial layers and statistical attributes of the effective parameters of the EPM and RFM equations through the soil erosion and flood susceptibility evaluation. Meanwhile, several fieldwork operations were carried out in the summer months of 2020 along the watershed to observe and record the geological and topographical features and surface covers to associate with the remotely sensed maps of geology and land cover. In this regard, the contact layers between the location of main types of land covers, geological formations, and sub-catchment boundaries were routed along the Qarasu stream and some negligible GPS-based corrections were inferred in the GIS data. Furthermore, several photographs were taken during the fieldwork to represent actual forms and functions (e.g., Figs. 6 and 7 in the next parts).

Erosion model
The erosion potential method (EPM), which has been developed by Gavrilović (1988), can qualify the erosion where Y is the soil erodibility coefficient based on the soil data, X is the soil protection coefficient based on the landuse types, φ is the coefficient of the type of erosion processes based on the remotely sensed observation and surface geology, and I is the average slope gradient of the catchment (%). According to the method proposed by Gavrilović (1988), the coefficients of soil erodibility (Y) and soil protection (X), and the type of erosion process (φ) could be considered using descriptive and numerical evaluations represented in Table 1, which have been illustrated by de Vente and Poesen (2005), Haghizadeh et al. (2009), andDragičević et al. (2018).
After that, the model can estimate the watershed sediment production (W) in cubic meters per year (m 3 /yr) using the below Equation (Dragičević et al. 2018;Berteni and Grossi 2020): where T is the temperature coefficient (equal to 1.18) that is calculated based on the mean annual temperature in the study area (13 °C) using follows (Berteni and Grossi 2020): where H is the mean annual precipitation (400 mm), π is equal to 3.14, Z is the erosion coefficient calculated in Eq.
(1), and A is the study area (km 2 ). Ultimately, the value of W can estimate the total production of soil erosion and

Flood model
The rational flood model (RFM) is the empirical equation, which has been interpreted in several studies, e.g., Thompson (2006), Devi et al. (2019), and Cheah et al. (2019), to estimate the runoff coefficient and peak flood discharge using the land use and land terrain characteristics (Aja et al. 2020). The RFM is expressed by the below Equation (Parak and Pegram 2006): where Qp is the flood peak discharge in cubic meters per second (m3/s), RC is the runoff coefficient (unitless), PI is the precipitation intensity in millimeters per hour (mm/h) that is determined based on the time series linear trend (Shanableh et al. 2018), and A is the catchment surface area in squared kilometers (km 2 ). According to the precipitation data (2016-2020) and its coefficient of determination (R2 > 0.69), the mean annual precipitation of the region was estimated as 400 mm, and the maximum anomaly of hourly precipitation was obtained as 37.8 mm, which can be considered as precipitation intensity (PI) in the study area (Table 2). Furthermore, the runoff coefficient (RC) can be determined in each catchment by overlapping the aforementioned intensity classes of soil units, land-use/land-cover types, rainfall rates, and longitudinal slope ranges, which are shown in Table 3 (e.g., Mousavi et al. 2019;Aja et al. 2020). Using remotely sensed data and the Zonal Statistics extension in GIS, the land cover, longitudinal slope range, and soil unit layers are analyzed for each catchment to estimate the RC value. RC value is an important parameter for flood control projects (Zeinali et al. 2019).

Description of the environmental parameters
The geological surface of the study region is covered mainly by limestone formation and lime members (> 75%), such as dolomite and lime-shale, and lime-sandstone (Fig. 5). The oldest rock units observed in the study area are Jurassic dolomites of the Mozduran formation (Jmz), dominantly in the sub-catchments 2, 3, 4, 5, 6, and 8, and the youngest are the Cretaceous-aged formation of Atamir (Kat) and Abderaz   GSI 2015). In the study area, sub-catchment 9 totally is covered by coarse and thick-bedded limestone of Tirgan (Kt) formation, while the sub-catchments 10, 11, and 12 dominantly are located above feeble-cement formations of shale-marl (Sarcheshmeh: Ksr) and shale-siltstone (Sanganeh: Ksn). Besides, sub-catchments 1 and 7 are located in the multiple contact layers of dolomite (Mozduran: Jmz), thick limestone (Tirgan: Kt), and lime-sandstone (Shurijeh: Ksh), which involves the main fault line of the study area based on the sectional inclining and fracturing of the aforementioned formations (Fig. 6). The anticline and fault lines, formed mainly by alternations between the lime-sandstone of Shurijeh and the dolomite rock of Mozduran (such as the Istisu fault transferred from the middle part of the catchment) are observed crossing the Qarasu watershed. The network of junctions on carbonate rocks with high porosity, such as Karst shafts and waterfalls over Tirgan formation, are surveyed in subcatchment 9 (Fig. 7). Karst valleys in the eastern part of Kopet-Dagh, such as the Kalat region, are combined with layers of lime members (Daneshvar et al. 2014).
The general physiographic trend of the catchment extends in the southwestern-northeastern direction with an average of 8.5 km length from the upstream to downstream, and the slope classes over 15% are covered > 75% of the total area (Fig. 8). The main soil units (based on global soil grids) are classified into Mollisols and Inceptisols (in the elevations with permeable units of sandy loams) and Entisols (in the downstream with the impervious texture of clay loams) (Fig. 9). Land covers are categorized as the forest of Junipers (~ 38% of the catchment upstream), bare land of rock faces (~ 31% of the catchment upstream), pasture land of Artemisia and Agropyron (~ 23% of the catchment in the middle parts), and farmland of dry farming (~ 8% of the catchment in the downstream) (Fig. 10). Based on the fieldwork operations, the low-density forests together with bare land of rock faces dominantly are observed in the geological unit of Mozduran formation (i.e., hard limestone and dolomites). Data-layer values for different classes of the environmental parameters in the study area are shown in Table 4.

Estimation of the EPM
The mean values of the erosion parameters and coefficients of the EPM model were estimated for each sub-catchment in Table 5, including surface area (A), slope gradient (I), soil protection coefficient (X), soil erodibility coefficient (Y), erosion coefficient (Z), and type of erosion processes (φ). The mean value of the slope range in the study area was calculated at 14%, where the highest and the lowest values, with 24 and 4%, belong to sub-catchments 5 and 14, respectively. The most values of soil protection coefficient (X = 0.8: bare land covers) were determined for sub-catchments in the upstream (over 1400 m a.s.l), while the most values of soil erodibility coefficient (Y = 1.3: shale and marl landforms) and type of erosion processes (φ = 0.8: gully and surface erosion) were verified for sub-catchments in the downstream (below 1400 m a.s.l). Besides, the values of erosion coefficient and watershed sediment production (W)  were estimated for the study sub-catchments, revealing the very highest erosion coefficient and sedimentation production (Z >1.5 and W >3000 m 3 /year) for sub-catchments 4, 5, and 7. Theoretically, the mentioned erosion coefficient and sedimentation production are categorized as severe and excessive erosion potential (Dragičević et al. 2018). The erosion susceptibility model was categorized in Fig. 11 through heavy, moderate, and slight classes. The heavy class of erosion susceptibility (with sediment production of 2127-4529 m 3 per year) belongs to seven catchments with 58.2% of the total study area. The moderate and slight classes of erosion susceptibility (with sediment production of 1103-2048 and 309-882 m 3 per year) belong to seven catchments with 19.0 and 22.8% of the total study area, respectively. The mean value of sediment production in all sub-catchments is estimated to equal 2120 m 3 per year ( Table 6).

Estimation of the RFM
Same as the erosion model, the mean values of the flood parameters and coefficients of the RFM model were estimated for each sub-catchment in Table 7, including surface area (A), runoff coefficient (RC), and precipitation intensity (PI). The mean value of the runoff coefficient was calculated at 0.35, and the highest and the lowest values, with 0.65 and 0.15, belong to sub-catchments 5 and 12, respectively. The most values of the runoff coefficient (RC >0.5) were determined for sub-catchments in the upstream (over 1400 m a.s.l), which have the highest slope ranges and rock outcrops over the hard dolomite landforms.
In this regard, the values of flood peak discharge (Qp) were estimated for the study sub-catchments, revealing the very highest flood discharge (> 75 m 3 /s) for sub-catchments 4, 5, 6, and 7. On this basis, the flood susceptibility model was categorized in Fig. 12 into three classes heavy, moderate, and slight. The heavy class of flood susceptibility (with a discharge of 84.37-141.52 m 3 per second) belongs to four catchments with 40.4% of the total study area. The moderate and slight classes of flood susceptibility (with a peak discharge of 25.86-65.99 and 8.57-23.81 and m 3 per year) belong to seven catchments with 44.3 and 15.3% of the total study area, respectively. The mean value of flood peak discharge in all sub-catchments is estimated to equal 53.92 m 3 per second (Table 8).

Interpretations of the susceptibility evaluation models
The statistical results revealed that the heavy classes of erosion and flood susceptibility (with sediment production of 2127-4529 m 3 per year and flood discharge of 84.37-141.52 m 3 per second) include 40.4-58.2% of the total study area dominantly in the upstream sub-catchments of the Qarasu catchment. It is worthy to be stated that in the coldest month of the winter (January), the snowfall is observed at about 50 mm particularly in the upstream sub-catchments. The melting snow could trigger the flooding events by weathering and erosion of the soil units in the upstream prone zones. Based on the spatial survey, the heaviest flood discharge and erosion potential prone zones are observed corresponding to the Mozduran and the Shurijeh formations composed mainly of dolomite, limestone, and shale on the bare landforms and steep slopes. This result is accordant with the previous works investigating the erosion susceptibility in the other catchments of the Kopet-Dagh, such as Bagherzadeh and Daneshvar (2011). Vice versa, areas with slight flood discharge and erosion potential classes are observed in the sub-catchments, which have been covered fully by forest or dense pastureland without relevant relation with rock and soil units.
The similar distribution of heavy and slight susceptibilities of flood and erosion models in the study area are examined by a correlation test to reveal possible relationships. In this regard, a correlation test using the Pearson test (p < 0.05) was assumed to analyze the relationships between EPM and RFM outputs (watershed sediment production: W and flood peak discharge: Qp). The correlation analysis was carried out based on the models' output  Table 5 Estimation of the parameters of the EPM in the sub-catchments, including surface area (A), slope gradient (I), soil protection coefficient (X), soil erodibility coefficient (Y), erosion coefficient (Z), type of erosion processes (φ), and watershed sediment production (W)    in 14 sub-catchments. Table 9 revealed a strong, significant, and direct association (R equal to 0.705) between W and Qp at the 99% confidence level. Hence, the flood and erosion susceptibility evaluations demonstrated the study area's co-related spatial and statistical outcomes.

Sub-catchment
In the mountainous catchments, higher flood peaks and runoff flow can indicate high erosion and sediment yield transportation (Dragičević et al. 2019). Consequently, the results of our research can indicate the prioritization of the sub-catchments based on their sensitivity and susceptibility to occurrences of soil erosion and flood events. Hence, the study area's sub-catchments of 8, 12, and 14 are considered low susceptible zones against flood and erosion hazards. To explain this result, we should say that our model outputs depend on the susceptibility potential of each sub-catchment in the occurrences of erosion and flood hazards instead of the risk assessment of the cumulative values of flood discharge or sedimentation volume. The differences between sub-catchments relate to their physical characteristics of topography, land cover, and geological unit. The cumulative flood or sedimentation discharges of each watershed are defined after the concentration times considering the total watershed from the upstream to the outlet cross-section (Heidari et al. 2021).
In our study, we assumed the flood and erosion discharges from the upstream of each sub-catchment to its outlet. Some photos of the natural surface cover of these susceptible prone zones in the high latitude areas are shown in Fig. 13, exposing the eroded rocks resulting in alluvial fans with a high potential of soil loss and inundated hills arising from the recent flooding event, particularly in subcatchments of 4, 5, 6, and 7. Concerning the Mozduran formation with Dolomite and hard limestone rocks in the aforementioned sub-catchments, we could note that the main reason for susceptibility of the rocks depends on the chemical weathering with water dissolution effects. The existence of aragonite grains in carbonate rocks of the Mozduran formations is one of the important reasons for the great effect of dissolution and dolomitization (e.g., Adabi 2009;Jamalian and Adabi 2015). Besides, physical and chemical processes of alteration, dissolution, and fracturing affect the contact layer of the Shurijeh formation with the Mozduran formation (e.g., Moussavi-Harami et al. 2009) resulting in the susceptible zones for fault system, landslide, erosion, soil loss, and flash flood potential. In regard to the low susceptible subcatchment of 8 (purely from its upstream to its outlet) we could observe that it has a low surface area in the watershed, influencing the estimation of its slight values for W and Qp (see Eqs. 2 and 4). Besides, the mentioned area has the lowest soil protection coefficient (X) influencing the slight value of watershed sediment production (W) (see estimations in Table 5), and the lowest runoff coefficient (RC) influencing the slight value of flood peak discharge (Qp) (see estimations in Table 7). Overall, in addition to the low surface area with a low length of sub-catchment, the land-cover status of the Juniper forest seems to be the main reason for the lowest values of X and RC coefficients, resulting in the slight of W and Qp values. Based on the previous works, the runoff and erosion rates are lowest in the juniper forests due to the protection of the canopy and high levels of ground cover (Wilcox et al. 1996;Pierson et al. 2007).

Validation of the models
In recent research, the cumulative amounts of erosion and flood susceptibility have exposed the susceptible zones in the downstream part of the watersheds (e.g., Dodangeh et al. 2020;Hosseini et al. 2020;Mosavi et al. 2020a). However, our study assumes the susceptibility of each sub-catchment independently to reveal the important role of runoff coefficient (RC) and erosion coefficient (Z) in the flood and erosion susceptibility evaluation. The RC value is an important parameter in flood discharge estimation, and high RC is associated with high flood discharge and susceptibility (Zeinali et al. 2019). Meanwhile, the Z value is the main model equation that gives numerical and descriptive information about the susceptibility of a given area to erosion processes (Dragičević et al. 2018). These parameters can essentially represent the actual state of the study region toward the model variable test, which can be used in the model validation procedure, namely the receiver operating Fig. 13 Natural surface cover of the susceptible prone zones in the high latitude of the study area, a eroded rocks resulted in alluvial fans with potential of soil loss and b inundated hills arising from recent flooding event characteristic curve (ROC-curve). In this section, the ROC curves for both susceptibility models were produced in Fig. 14. The model validity explained the AUC (area under the curve) values averagely equal to 0.898 and 0.917 for erosion and flood susceptibility evaluation (i.e., EPM and RFM), explaining the models' very good performance and excellent performance sensitivities.
Besides, for investigating the model outputs in the scale of each sub-catchment, adjusted rescaled-plots were produced to indicate the variation of sub-catchment susceptibility to occurrences of soil erosion and flood events, regarding the sorting of φ values (based on the geological units) and the sorting of X values (based on the land covers) (Fig. 15). On this basis, the high natural hazards susceptibilities are observed in the sub-catchments (e.g., no. 4, 5, and 6), located on the Mozduran dolomites (Jmz) with low-density flora, and bare land outcrops. Contrarily, the low and moderate natural hazards susceptibilities are observed in the sub-catchments (e.g., no. 13, 14, 2, and 8), located on the lime-shale (Kad) and lime-conglomerate (Kat) with farmland or pastureland covers.

Limitations
The limitations of the research depend on the actual accessibility and reliability of the fieldwork data within a long-term temporal window in addition to possible downscaling errors from regional to local scales, which could influence the rational results of the susceptibility evaluation models. To overcome these problems, further research needs to assume reliable and accessible global datasets for analyzing the environmental characteristics instead of the local limited data.

Conclusions
The main aim of this study was to evaluate erosion and flood susceptibility using empirical models of EPM and RFM to prioritize the GIS-based prone zones in a catchment of the Kopet-Dagh Mountains. The required soil erosion and flood susceptibility evaluation data were obtained from some global datasets with spatial grid pixels and time series. The important characteristic of the study area was its geological setting of limestone units and  carbonate rocks, particularly physical and chemical processes of alteration, dissolution, and fracturing through the Shurijeh and the Mozduran formations, which were susceptible to erosion and flood events. The results revealed that the heavy classes of erosion and flood susceptibility (with sediment production of 2127-4529 m 3 per year and flood discharge of 84.37-141.52 m 3 per second) include 40.4-58.2% of the total study area dominantly in the upstream catchments.
The similar distribution of heavy and slight susceptibilities of flood and erosion models in the study area was examined by correlation test, exposing a strong, significant, and direct association (R equal to 0.705) between W and Qp at the 99% confidence level. Therefore, the present study assumed the susceptibility of each sub-catchment independently to reveal the important role of runoff coefficient (RC) and erosion coefficient (Z) in the flood and erosion susceptibility evaluation. These parameters were represented as the actual state of the study region toward the model variable test to validate the models using the ROC-curve. The model validity explained the AUC (area under the curve) values averagely equal to 0.898 and 0.917 for erosion and flood susceptibility evaluations (i.e., EPM and RFM), explaining the models' very good performance and excellent performance sensitivities.
The practical implication of this research depends on prioritizing susceptible zones for spatial development and tourism plans of the study area by providing proper environmental insight for their managers, planners, investors, and stockholders. According to Bathrellos et al. (2017), preparing the natural hazard potential and predictive maps is very important to take into account during land-use planning. Particularly, such an approach is used to identify the prone zones to natural hazards during the early stages of the local spatial development (Skilodimou et al. 2019). Meanwhile, the theoretical implication of the research is to provide more interpretations for EPM and RFM models for localizing and generalizing the equations and coefficients in future studies. Further research can be carried out based on the combination of other environmental hazards, such as earthquakes and landslides and flood and soil erosion, in the susceptibility evaluation models of the watersheds. Some studies have shown a significant relationship between land-use change, flash floods, and soil erosion (Ferreira et al. 2015;Mohammadi et al. 2021). Hence, future research can consider the role of land-use changes in the erosion and flood susceptibility evaluation within the time series.