English Extended Abstract
Introduction
Floods are among the most destructive natural hazards worldwide, causing substantial economic losses, environmental degradation, infrastructure damage, and threats to human life each year. In recent decades, the increasing frequency and intensity of extreme climatic events, combined with rapid land-use changes, urban expansion, and vegetation degradation, have significantly increased watershed vulnerability to flood occurrence. Consequently, flood susceptibility assessment and the identification of flood-prone areas have become essential components of watershed management, disaster risk reduction, and sustainable land-use planning. Flood generation is controlled by a complex interaction of geomorphological, hydrological, geological, soil, and land-cover characteristics. Factors such as slope gradient, drainage patterns, elevation, soil permeability, lithology, vegetation cover, and land use directly influence runoff production, flow concentration, and flood propagation processes. Recent advances in Geographic Information Systems (GIS) and Remote Sensing (RS) technologies have greatly enhanced the ability to integrate and analyze these spatial factors, providing effective tools for flood hazard mapping and susceptibility assessment. Among the available approaches, the Modified Flash Flood Potential Index (MFFPI) has emerged as a practical and efficient method for evaluating flash flood susceptibility. The model combines multiple physiographic and environmental variables within a GIS framework and has been successfully applied in various mountainous and semi-arid watersheds. Despite numerous flood susceptibility studies worldwide, comprehensive assessments integrating geomorphological indicators with MFFPI remain limited in many Iranian watersheds. Therefore, the present study aimed to evaluate flash flood susceptibility in the Balikhli chay Watershed, Ardabil Province, Iran, using the MFFPI model and to investigate the relationships between geomorphological indices and flood potential.
Material and Methods
The study was conducted in the Balikhli chay Watershed, one of the most important watersheds in Ardabil Province, characterized by complex topography, diverse geological formations, extensive agricultural activities, and significant hydrological variability. The MFFPI model was implemented within a GIS environment by integrating six thematic layers including slope, flow accumulation, land use/land cover (LULC), slope curvature, soil texture, and lithology. A 30-m resolution ASTER Digital Elevation Model (DEM) was used to derive slope, curvature, and flow accumulation layers. Land-use data were obtained from the GLC-FCS30D database through Google Earth Engine, while geological information was derived from 1:100,000 geological maps. Soil texture maps and Sentinel-2 satellite imagery were utilized for soil and vegetation analyses. All thematic layers were converted into raster format and reclassified into five classes according to their hydrological response and contribution to flood generation. The weighted layers were subsequently integrated using the MFFPI equation within ArcGIS. To provide a more comprehensive geomorphological assessment, seven additional geomorphometric indices were calculated, including Topographic Wetness Index (TWI), Elevation (DEM), Terrain Ruggedness Index (TRI), Length-Slope Factor (LSF), Topographic Position Index (TPI), Normalized Difference Vegetation Index (NDVI), and Aspect. These indices were extracted from DEM and Sentinel-2 data using ArcGIS 10.8. To quantify the relationship between geomorphological characteristics and flood susceptibility, 300 random sample points were extracted from the spatial intersection of the MFFPI map and the geomorphological layers. Spearman’s rank correlation coefficient was calculated in R 4.3 software. The Shapiro–Wilk test confirmed non-normal distributions for most variables; therefore, the non-parametric Spearman approach was selected. Multiple testing effects were controlled using the Benjamini–Hochberg false discovery rate correction.
Results and Discussion
The resulting MFFPI map classified the watershed into five flood susceptibility categories: very low, low, moderate, high, and very high. The high susceptibility class occupied the largest area, covering 444 km² (36.75%) of the watershed. Low and very low susceptibility classes accounted for 246 km² (20.36%) and 241 km² (19.95%), respectively, while the moderate class covered 218 km² (18.04%). The very high susceptibility class represented 59 km² (4.88%) of the watershed. Overall, approximately 503 km² (41.63%) of the watershed fell within the high and very high flood susceptibility classes, indicating considerable vulnerability to flash flood events. These areas are primarily concentrated in the middle and downstream sections of the watershed, particularly along the Balikhli chay River corridor and associated floodplains. Their high susceptibility can be attributed to low elevations, gentle slopes, high flow accumulation, elevated topographic wetness values, and extensive agricultural land use. Conversely, approximately 40.31% of the watershed exhibited low or very low susceptibility, mainly within the mountainous northern and northwestern sectors characterized by steep slopes, higher elevations, greater infiltration potential, and relatively natural vegetation cover. Geomorphological analysis revealed significant spatial variability across the watershed. Elevation ranged from 1,316 m in downstream regions to 4,788 m in mountainous headwaters, producing substantial hydraulic gradients that facilitate rapid runoff transfer. TWI values varied between 7.02 and 14.45, with the highest values concentrated in the middle portions of the watershed, indicating zones prone to water accumulation and soil saturation. LSF values reached 33.47 in steep northern slopes, highlighting areas with high sediment transport and erosion potential. Correlation analysis demonstrated that five of the seven geomorphological indices exhibited statistically significant relationships with MFFPI. TWI showed the strongest positive correlation (ρ = 0.7674), confirming the critical role of moisture accumulation and runoff concentration in flash flood generation. Elevation displayed the strongest negative correlation (ρ = −0.6204), indicating increased flood susceptibility in lower-lying areas. TRI (ρ = −0.3173) and LSF (ρ = −0.3011) also exhibited moderate negative relationships, suggesting that relatively smoother and less rugged terrains are more susceptible to flood accumulation. NDVI showed a weak but significant negative correlation (ρ = −0.1898), emphasizing the protective role of vegetation cover in reducing runoff generation. In contrast, TPI and Aspect displayed weak and statistically insignificant relationships with flood susceptibility. The correlation matrix further demonstrated the absence of multicollinearity among independent variables, as no correlation coefficient exceeded the critical threshold of 0.80. These findings highlight the dominant influence of hydrological and topographic factors, particularly TWI and elevation, in controlling flood susceptibility patterns within the watershed.
Conclusion
This study demonstrated the effectiveness of integrating the Modified Flash Flood Potential Index (MFFPI) with geomorphological analysis within a GIS framework for flash flood susceptibility assessment in the Balikhli chay Watershed. The results revealed that more than 41% of the watershed is exposed to high and very high flood susceptibility, primarily concentrated within downstream floodplains and river corridors. Among the investigated geomorphological factors, Topographic Wetness Index and elevation emerged as the most influential predictors of flood susceptibility, while vegetation cover contributed to reducing flood risk through enhanced infiltration and runoff regulation. The findings confirm that the combined effects of topography, hydrological accumulation processes, land use, soil properties, and lithological characteristics govern flood generation and spatial flood patterns in the watershed. The methodology presented in this study provides a reliable, cost-effective, and data-efficient approach for flood hazard assessment in mountainous and semi-arid watersheds. The resulting susceptibility maps can support watershed management, flood mitigation planning, land-use regulation, and the implementation of early warning systems. Future studies should incorporate climate change scenarios, hydrodynamic modeling, and machine learning algorithms to further improve flood susceptibility prediction and risk management strategies.
Keywords: Flash Flood, Geomorphological Indices, Spearman Correlation; Google Earth Engine (GEE), Balikhli chay Watershed.