Modified flood potential index (MFPI) for flood monitoring in terrestrial water storage depletion basin using GRACE estimates

The Gravity Recovery and Climate Experiment (GRACE) and Follow-On (GRACE-FO) missions offer great potential for large-scale flood event monitoring. However, the effectiveness of current GRACE-based indices may be limited in areas with significant Terrestrial Water Storage (TWS) decline. In this study, a Modified Flood Potential Index (MFPI) was proposed to improve the monitoring capability for flood events in the Yarlung Tsangpo-Brahmaputra River Basin (YBRB). First, the GRACE/FO Terrestrial Water Storage Anomaly (TWSA) data gap was filled with a Multilayer Perceptron Artificial Neural Network Model (MLP-ANN). Based on the gap-filled data, the performance of the MFPI was compared with five selected indices: Flood Potential Index (FPI), Total Storage Deficit Index (TSDI), Modified Total Storage Deficit Index (MTSDI), Water Storage Deficit Index (WSDI), and Combined Climatologic Deviation Index (CCDI). Finally, the spatiotemporal monitoring ability of the MFPI was evaluated for typical and atypical flood events. We found that (1) MLP-ANN was able to predict the GRACE/FO data gap with a performance of 'Very Good'; (2) the MFPI outperformed the FPI, TSDI, MTSDI, WSDI, and CCDI in capturing flood events, and could also reflect their spatiotemporal processes; and (3) the MFPI avoids long-term TWSA trend effects on flood event monitoring and is available for areas with significant TWSA trends. This study can provide a reference for using GRACE data to monitor large-scale flood events.