Deep Neural Networks Hydrologic and Hydraulic Modeling in Flood Hazard Analysis

Natural disasters can have devastating effects on the environment and natural resources, making flood inundation mapping and hydraulic modeling essential for forecasting critical characteristics like flood depth and water surface elevation. This study investigated key flood-influencing factors including rainstorm intensity, precipitation depth, soil type, geologic setting, and topographic features, while conducting hydraulic modeling of storm flows for 50- and 100-Year return periods, which estimated water depths in Wadi Al Wala reaching 15 m during 50-Year storms and 25 m during 100-Year storms. To enhance prediction capabilities, a Deep Neural Network (DNN) model was developed using 1980–2018 meteorological data, featuring three hidden layers (10, 20, and 30 neurons) with ReLU activation, trained on 80% of data and validated on 20% using Python’s Keras library. Using relative humidity, rainfall, wind speed, and wet bulb temperature as inputs, the model achieved 92% prediction accuracy based on MSE metrics. The integration of HEC-HMS hydrological modeling, HEC-RAS hydraulic modeling, and DNN analysis produced effective flood hazard predictions that support strategic infrastructure planning and improved disaster management in vulnerable regions.