Identifying Paddy Rice Fields in the U.S. from the Operational VIIRS Flood Products

Operational satellite-based flood products are generated by comparing water classification maps from satellite imagery with permanent or normal water masks. This approach may misclassify some water bodies—such as irrigated paddy rice fields—as floodwaters because they are not masked as permanent or normal water sources. Due to the importance of paddy fields for food security, in this study, methodologies based on the long-time duration of water presence combined with paddy rice phenological algorithms and change detection analysis are developed to extract paddy rice fields from the operational VIIRS (Visible Infrared Imaging Radiometer Suite) flood products. This method is also compared with the regression analysis and the Mann–Kendall analysis. Evaluations are performed through confusion matrix analysis by comparing with the USDA rice data. The three paddy rice extraction algorithms show good agreement and can achieve an accuracy of 93% with an F1-score exceeding 80%.