Towards Safe and Sustainable Agricultural Reconstruction in Ukraine: Insights from Satellite-Based Crop Classification Maps

Ukraine, a significant contributor to the global agricultural market, has been severely affected by the ongoing war. Accurate information on land use changes, particularly in the agricultural sector, is crucial for assessing the extent of damage and guiding post-war reconstruction efforts. This study presents a robust methodology for crop classification and monitoring in Ukraine, utilizing a combination of Sentinel-1 and Sentinel-2 satellite data and ground truth datasets collected annually from 2016 to 2023. A Random Forest classifier implemented within the Google Earth Engine platform was employed, achieving overall accuracies exceeding 95% and F1 scores surpassing 92% for major crops. The resulting annual crop classification maps provide detailed information on the spatial distribution of various crop types and the impact of the war on Ukraine's agricultural sector. Comparison with global land cover products highlights our maps' superior consistency, detail, and reliability for comprehensive agricultural monitoring. The high correlation with official statistics underscores the potential of this methodology to inform policy decisions and support reconstruction efforts in Ukraine and other regions facing similar challenges.