Project Apaw: Spatiotemporal Forecasting of River Flood Using Deep Learning

Flooding presents a growing global threat, particularly in vulnerable areas like the municipality of Nabua, Philippines, a low-lying municipality that serves as a catch basin for floodwaters from neighboring regions. This study, Project Apaw: Spatiotemporal Forecasting of River Flood using Deep Learning, aims to explore the appropriateness of Spatiotemporal Forecasting of River Flood using Deep Learning in the Municipality of Nabua. Hence, a localized weather monitoring system was implemented, consisting of an Indoor Control Unit (ICU) and an Outdoor Control Unit (OCU), which collect data on temperature, humidity, rainfall, water level, and wind speed. These units communicate via RS485 modbus, exchanging and uploading data to a webserver. Long-Short Term Memory model was used for flood prediction based on the collected data. The trained model demonstrated its ability to detect patterns and relationships in the data. Moreover, the findings demonstrate the potential of deep learning-based flood prediction models to accurately forecast river flood risks, with the constructed Spatiotemporal model achieving proficiency in capturing complex spatial and temporal correlations, as indicated by an RMSE value of 0.0101 and MAE of 0.0097. However, challenges related to data availability, quality, interpretability, and computational resources need to be addressed for widespread implementation. By incorporating social and economic data, considering spatial variability and climate dynamics, and collecting more data over an extended period, deep learning-based flood prediction models can become valuable tools for proactive flood mitigation and preparedness in vulnerable regions like Nabua, thus contributing to improved flood risk management practices.