This study proposes a flood susceptibility assessment method in urban areas using a Deep Neural Network (DNN) optimized with genetic algorithms (GAs) and remote sensing data from Ibadan, Nigeria, and Metro Manila, Philippines. The DNN model significantly improves flood risk assessment accuracy, achieving 0.98 for Ibadan compared to 0.38 for Manila when using only location and precipitation data. Incorporating soil data and reducing the number of classes enhances flood risk prediction accuracy, providing insights for proactive flood mitigation and urban planning. The study highlights the importance of using diverse features, such as topographic and hydrological parameters, to improve model performance. The research involved developing predictive models for flood susceptibility using datasets from Ibadan and Manila. The Ibadan dataset included features like X, Y coordinates, slope, aspect, Topographic Wetness Index (TWI), Flow Accumulation (FA), drainage, and rainfall, while the Manila dataset included latitude, longitude, elevation, precipitation, and flood height. The DNN model was optimized using GAs to find the best hyperparameters, leading to improved accuracy. The results showed that the Ibadan dataset achieved an accuracy of 0.98, while the Manila dataset had lower accuracy due to insufficient features. Reducing the number of classes in the Manila dataset to five improved accuracy to nearly 0.99. The study discusses the effectiveness of GA-optimized DNN techniques for urban flood assessment, emphasizing the need for comprehensive datasets and diverse features. It also compares DNN-based flood assessment with traditional hydraulic modeling, highlighting the advantages of DNNs in speed, efficiency, scalability, and ability to leverage large datasets. DNNs are more efficient and scalable than traditional models, which are computationally intensive and require extensive data inputs. The study also addresses challenges in interpreting DNN models, emphasizing the need for explainable AI (XAI) techniques to enhance transparency and trust in model outputs. The limitations of the model include geographic specificity and data quality, which can affect performance in different regions. Future research directions include improving spatial resolution data, integrating climatic and social data, and developing automated monitoring systems for real-time flood warnings. The study underscores the importance of using diverse and comprehensive data to enhance the accuracy and applicability of flood risk assessment models in urban areas.This study proposes a flood susceptibility assessment method in urban areas using a Deep Neural Network (DNN) optimized with genetic algorithms (GAs) and remote sensing data from Ibadan, Nigeria, and Metro Manila, Philippines. The DNN model significantly improves flood risk assessment accuracy, achieving 0.98 for Ibadan compared to 0.38 for Manila when using only location and precipitation data. Incorporating soil data and reducing the number of classes enhances flood risk prediction accuracy, providing insights for proactive flood mitigation and urban planning. The study highlights the importance of using diverse features, such as topographic and hydrological parameters, to improve model performance. The research involved developing predictive models for flood susceptibility using datasets from Ibadan and Manila. The Ibadan dataset included features like X, Y coordinates, slope, aspect, Topographic Wetness Index (TWI), Flow Accumulation (FA), drainage, and rainfall, while the Manila dataset included latitude, longitude, elevation, precipitation, and flood height. The DNN model was optimized using GAs to find the best hyperparameters, leading to improved accuracy. The results showed that the Ibadan dataset achieved an accuracy of 0.98, while the Manila dataset had lower accuracy due to insufficient features. Reducing the number of classes in the Manila dataset to five improved accuracy to nearly 0.99. The study discusses the effectiveness of GA-optimized DNN techniques for urban flood assessment, emphasizing the need for comprehensive datasets and diverse features. It also compares DNN-based flood assessment with traditional hydraulic modeling, highlighting the advantages of DNNs in speed, efficiency, scalability, and ability to leverage large datasets. DNNs are more efficient and scalable than traditional models, which are computationally intensive and require extensive data inputs. The study also addresses challenges in interpreting DNN models, emphasizing the need for explainable AI (XAI) techniques to enhance transparency and trust in model outputs. The limitations of the model include geographic specificity and data quality, which can affect performance in different regions. Future research directions include improving spatial resolution data, integrating climatic and social data, and developing automated monitoring systems for real-time flood warnings. The study underscores the importance of using diverse and comprehensive data to enhance the accuracy and applicability of flood risk assessment models in urban areas.