AI-Driven Flood Mapping and Precision Rice Monitoring in Morocco Using Sentinel Satellite Data

This study presents a multi-sensor approach for precision agriculture, combining Sentinel-1 synthetic aperture radar (SAR) and Sentinel-2 optical imagery to map rice fields and monitor crop phenology in the Gharb plain. Rice fields were delineated using k-means clustering on fused multi-temporal datasets, followed by morphological filtering, with the rice cluster manually selected based on NDVI time series from visually identified parcels. Although formal computation of the Kappa coefficient was not possible due to the manual selection, qualitative validation through NDVI profiles and visual inspection confirms accurate detection of rice parcels and effective exclusion of non-rice areas. Temporal dynamics captured by NDVI and SAR backscatter reflect key rice growth stages and flooding events, demonstrating the agronomic relevance of the method. Limitations from cloud cover, SAR signal complexity, and mixed pixels highlight the need for higher temporal resolution data, advanced machine learning models, and expanded field validation in future work. This study illustrates the utility of Google Earth Engine for scalable, timely, and accurate monitoring of irrigated crops, supporting sustainable water management and precision agriculture practices.