This study presents two machine learning models, namely, the light gradient boosting machine (LightGBM) and categorical boosting (CatBoost), for the first time for predicting flash flood susceptibility (FFS) in the Wadi System (Hurghada, Egypt). A flood inventory map with 445 flash flood sites was produced and randomly divided into two groups for training (70%) and testing (30%). Fourteen flood controlling factors were selected and evaluated for their relative importance in flood occurrence prediction. The performance of the two models was assessed using various indexes in comparison to the common random forest (RF) method. The results show areas under the receiver operating characteristic curves (AUROC) of above 97% for all models and that LightGBM outperforms other models in terms of classification metrics and processing time. The developed FFS maps demonstrate that highly populated areas are the most susceptible to flash floods. The present study proves that the employed algorithms (LightGBM and CatBoost) can be efficiently used for FFS mapping.

This study aims to examine three machine learning (ML) techniques, namely random forest (RF), LightGBM, and CatBoost for flooding susceptibility maps (FSMs) in the Vietnamese Vu Gia-Thu Bon (VGTB). The results of ML are compared with those of the rainfall-runoff model, and different training dataset sizes are utilized in the performance assessment. Ten independent factors are assessed. An inventory map with approximately 850 flooding sites is based on several post-flood surveys. The inventory dataset is randomly split between training (70%) and testing (30%). The AUC-ROC results are 97.9%, 99.5%, and 99.5% for CatBoost, LightGBM, and RF, respectively. The FSMs developed by the ML methods show good agreement in terms of an extension with flood inundation maps developed using the rainfall-runoff model. The models’ FSMs showed 10–13% of the total area to be highly susceptible to flooding, consistent with RRI's flood map. The FSMs show that downstream areas (both urbanized and agricultural) are under high and very high levels of susceptibility. Additionally, different sizes of the input datasets are tested to determine the least number of data points having acceptable reliability. The results demonstrate that the ML methods can realistically predict FSMs, regardless of the number of training samples.

The generation and processes of wadi flash floods are very complex and are not well understood. In this paper, we investigate the relationship between variations in geomorphometric and rainfall characteristics and the responses of wadi flash floods. An integrated approach was developed based on geomorphometric analysis and hydrological modeling. The Wadi Qena, which is located in the Eastern Desert of Egypt, was selected to validate the developed approach and was divided into 14 sub-basins with areas ranging from 315 to 1488 km2. The distributed Hydrological River Basin Environment Assessment Model (Hydro-BEAM) was used to obtain a good representation of the spatial variability of the rainfall and geomorphology in the basin. Thirty-eight geomorphometric parameters representing the topographic, scale, shape and drainage characteristics of the basins were considered and extracted using geographic information system (GIS) techniques. A series of flash flood events from 1994, 2010, 2013, and 2014, in addition to synthetic virtual storms with different durations and intensities, were selected for the application of this study. The results exhibit strong correlations between scale and topographic parameters and the hydrological indices of the wadi flash floods, while the shape and drainage network metrics have smaller impacts. The total rainfall amount and duration significantly impact the relationship between the hydrologic response of the wadi and its geomorphometry. For most of the parameters, we found that the impact of the wadi geomorphometry on the hydrologic response increases with increasing rainfall intensity.

Urban growth, extreme climate, and mismanagement are crucial controlling factors that affect flood vulnerability at wadi catchments. Therefore, this study attempts to understand the impacts of these three factors on the flash flood vulnerability in different climatic regions in Egypt. An integrated approach is presented to evaluate the urban growth from 1984 to 2019 by using Google Images and SENTINEL-2 data, and to develop hazard maps by using a rainfall-runoff-inundation model (RRI). Annual rainfall trend analysis was performed to evaluate the temporal variability trend. The hazard maps that were created were classified into three categories (low, medium, and high) and integrated with the urban growth maps to evaluate the impacts on the flood-vulnerable areas. The results show a significant increase in urban growth resulting in an increase of prone areas for flood hazards over time. However, the degree of this hazard is mainly related to growth directions. Mismanagement affects urban growth directions in both planned and unplanned growth, whether by loss of control over unplanned growth or by deficiencies in approved plans. The rainfall analysis showed that there is no explicit relationship to increases or decreases in the flood vulnerable areas. An urban planning approach is recommended for risk reduction management based on a comprehensive study considering such factors.

Flood risk mapping forms the basis for disaster risk management and the associated decision-making systems. The effectiveness of this process is highly dependent on the quality of the input data of both hazard and vulnerability maps and the method utilized. On the one hand, for higher-quality hazard maps, the use of 2D models is generally suggested. However, in ungauged regions, such usage becomes a difficult task, especially at the microscale. On the other hand, vulnerability mapping at the microscale suffers limitations as a result of the failure to consider vulnerability components, the low spatial resolution of the input data, and the omission of urban planning aspects that have crucial impacts on the resulting quality. This paper aims to enhance the quality of both hazard and vulnerability maps at the urban microscale in ungauged regions. The proposed methodology integrates remote sensing data and high-quality city strategic plans (CSPs) using geographic information systems (GISs), a 2D rainfall-runoff-inundation (RRI) simulation model, and multicriteria decision-making analysis (MCDA, i.e., the analytic hierarchy process (AHP)). This method was implemented in Hurghada, Egypt, which from 1996 to 2019 was prone to several urban flood events. Current and future physical, social, and economic vulnerability maps were produced based on seven indicators (land use, building height, building conditions, building materials, total population, population density, and land value). The total vulnerability maps were combined with the hazard maps based on the Kron equation for three different return periods (REPs) 50, 10, and 5 years to create the corresponding flood risk maps. In general, this integrated methodology proved to be an economical tool to overcome the scarcity of data, to fill the gap between urban planning and flood risk management (FRM), and to produce comprehensive and high-quality flood risk maps that aid decision-making systems.