Flood Susceptibility Mapping Using Publicly Available Big Data with Google Earth Engine and Deep Learning Algorithms

Flood susceptibility mapping is essential for disaster risk management in flood-prone regions. This study utilizes point-based data from publicly available open-source earth system datasets, historical flood observation datasets, Google's cloud computing platform, and deep learning models to create flood susceptibility maps for the Nam Ngum River Basin in Lao PDR. The assessment integrates the digital elevation model (DEM), satellite-observed rainfall data, land use/land cover, and remote sensing image-derived indices such as NDVI to capture detailed flood causation factors. Deep learning techniques, including Artificial neural networks (ANN), Long short-term memory (LSTM), and Deep Neural Networks (DNN), are employed to analyze various hydro-meteorological and geomorphological parameters. The models were trained and tested using eleven flood conditioning variables and 390 locations to ensure predictive accuracy and reliability. These datasets were randomly divided into training and testing datasets in a 70:30 ratio. The developed models are evaluated based on performance metrics like accuracy, precision, and the area under the curve of Receiver Operating Characteristics (AUROC). This map identifies critical zones within the Nam Ngum River Basin at high risk of flooding, offering valuable insights for local authorities and stakeholders. This information is crucial for enhancing flood risk management, emergency planning, and mitigation strategies.