River Flood Modeling and Remote Sensing Across Scales: Lessons from Brazil

Abstract In Brazil, a substantial understanding of flooding regimes in large natural wetlands, as in the Amazon and Pantanal regions, has been promoted through remote sensing (RS) and river flood modeling. However, less research attention has been given to the floods with socioeconomic impacts. In the last decades, RS has provided new opportunities for improving flood models from local to global scales, especially in regions with large and sparsely gauged river systems. Here we present some recent lessons from Brazil regarding the use of RS in improving flood models across scales. A systematic literature review of current flood model applications in the country using RS showed that flood extent and satellite altimetry data have been underused, in particular at local scales. Models have been validated with remotely sensed water levels and flood extent mainly for large natural wetlands in the Amazon. Then, some examples of recent advances on the use of RS data for improving models are presented. Innovative methods include estimation of river cross-section parameters with data assimilation and genetic calibration algorithms, and floodplain topography estimation based on detailed in situ survey as well as on a combination of water mask and water level time series. Cross-scale comparisons between global, regional, and local flood models in Brazilian rivers also provide valuable insights on the capabilities of current models, showing, for example, that more distributed information of cross-sections are needed to achieve better predictions. We finish by summarizing some current efforts by national and international organizations to estimate flood hazard as well as to monitor and forecast floods in real-time, and discussing perspectives on how current and future satellite missions, in combination with models, could help to mitigate flood related disasters in Brazil.