Flood detection using Gravity Recovery and Climate Experiment (GRACE) terrestrial water storage and extreme precipitation data

Abstract

A complete global flood event record would aid researchers to analyze the distribution of global floods and, thus, better formulate and manage disaster prevention and reduction policies. This study used Gravity Recovery and Climate Experiment (GRACE) terrestrial water storage and precipitation data combined with high-frequency filtering, anomaly detection and flood potential index methods to successfully extract historical flood days globally between 1 April 2002 and 31 August 2016; these results were then further compared and validated with Dartmouth Flood Observatory (DFO) data, Global Runoff Data Centre (GRDC) discharge data, news reports and social media data. The results showed that GRACE-based flood days could cover 81% of the flood events in the DFO database, 87% of flood events extracted by MODIS and supplement many additional flood events not recorded by the DFO. Moreover, the probability of detection greater than or equal to 0.5 reached 62% among 261 river basins compared to flood events derived from the GRDC discharge data. These detection capabilities and detection results are both good. Finally, we provided flood day products with a 1 spatial resolution covering the range between 60S and 60N from 1 April 2002 to 31 August 2016; these products can be obtained from 10.5281/zenodo.6831384 (Zhang et al., 2022b). Thus, this research contributes a data foundation for the mechanistic analysis and attribution of global flood events.