A Cloud-Based Geo-AI Framework for Automated High-Resolution Flood Mapping with Explainable Machine Learning

This study presents a hybrid cloud-local Geo-AI framework for high-resolution flood mapping. It combines multi-source data processing in Google Earth Engine with local machine-learning benchmarking and interpretation. Applied to the 2022 Pakistan floods, the workflow derives training samples from Sentinel-1 time-series statistics, compares 17 algorithms, and uses SHAP to interpret influential predictors. Spatial comparison with an external flood product identifies areas of agreement and discrepancy. Further evaluation across regions and of end-to-end operational latency remains necessary.