Assessing and attributing flood potential in Brazil using GPS 3D deformation

Global Positioning System (GPS) instruments capture the daily crustal 3D deformation responding elastically to terrestrial water storage (TWS) variations, providing a powerful tool for hydrological studies. Here, we further expand the application of GPS in flood potential assessment. GPS vertical and horizontal crustal deformation are inverted into TWS variations using a 3D-Inversion model, and then a novel GPS-based modified flood potential index (GMFPI) is developed to assess and attribute the spatiotemporal patterns of flood potential in Brazil. The 3D-Inversion-derived TWS estimates show more spatial details compared to those derived from vertical deformation (1D-Inversion), with annual water thickness amplitudes of about 900 mm in the middle Amazon River, which is consistent with the Gravity Recovery and Climate Experiment Mascon solutions but is greater than the 1D-Inversion estimates. The comparison between in-situ discharge data and GMFPI indicates that GMFPI performs well in monitoring flood potential, showing Accuracy (ACC) values greater than 0.8 at basin scales. In four reported flood events, the spatial patterns of GMFPI and discharge show that the locations of those floods are accurately characterized by GMFPI. The attribution analysis of flood dynamics shows that precipitation in coastal regions can rapidly increase flood potential, while a large amount of precipitation in inland regions first replenishes unsaturated soil water and groundwater. Additionally, the daily GMFPI exhibits good consistency with daily discharge, demonstrating a capacity for monitoring floods at a sub-monthly scale. Our study highlights the improvement of 3D-Inversion to TWS estimates and the novel application of GPS in flood potential assessing with high spatiotemporal resolution, providing valuable insights for flood early warning and prevention.