No abstract

Abstract. Deriving flood hazard maps for ungauged basins typically requires simulating a long record of annual maximum discharges. To improve this approach, precipitation from global reanalysis systems must be downscaled to a spatial and temporal resolution applicable for flood modeling. This study evaluates such downscaling and error correction approaches for improving hydrologic applications using a combination of NASA's Global Land Data Assimilation System (GLDAS) precipitation data set and a higher resolution multi-satellite precipitation product (TRMM). The study focuses on 437 flood-inducing storm events that occurred over a period of ten years (2002)(2003)(2004)(2005)(2006)(2007)(2008)(2009)(2010)(2011) in the Susquehanna River basin located in the northeastern United States. A validation strategy was devised for assessing error metrics in rainfall and simulated runoff as function of basin area, storm severity, and season. The WSR-88D gauge-adjusted radar-rainfall (stage IV) product was used as the reference rainfall data set, while runoff simulations forced with the stage IV precipitation data set were considered as the runoff reference. Results show that the generated rainfall ensembles from the downscaled reanalysis product encapsulate the reference rainfall. The statistical analysis consists of frequency and quantile plots plus mean relative error and root-mean-square error statistics. The results demonstrated improvements in the precipitation and runoff simulation error statistics of the satellite-driven downscaled reanalysis data set compared to the original reanalysis precipitation product. Results vary by season and less by basin scale. In the fall season specifically, the downscaled product has 3 times lower mean relative error than the original product; this ratio increases to 4 times for the simulated runoff values. The proposed downscaling scheme is modular in design and can be applied on any gridded satellite and reanalysis data set.

No abstract

Abstract. Deriving flood hazard maps for ungauged basins typically requires simulating a long record of annual maximum discharges. To improve this approach, precipitation from global reanalysis systems must be downscaled to a spatial and temporal resolution applicable for flood modeling. This study evaluates such downscaling and error correction approaches for improving hydrologic applications using a combination of NASA's Global Land Data Assimilation System (GLDAS) precipitation dataset and a higher resolution multi-satellite precipitation product (TRMM). The study focuses on 437 flood-inducing storm events that occurred over a period of ten years (2002–2011) in the Susquehanna River basin located in the northeast US. A validation strategy was devised for assessing error metrics in rainfall and simulated runoff as function of basin area, storm severity and season. The WSR-88D gauge-adjusted radar-rainfall (stage IV) product was used as the reference rainfall dataset, while runoff simulations forced with the stage IV precipitation dataset were considered as the runoff reference. Results show that the generated rainfall ensembles from the downscaled reanalysis products encapsulate the reference rainfall. The statistical analysis, including frequency and quantile plots plus mean relative error and root mean square error statistics, demonstrated improvements in the precipitation and runoff simulation error statistics of the satellite-driven downscaled reanalysis dataset compared to the original reanalysis precipitation product. Results vary by season and less by basin scale. In the fall season specifically, the downscaled product has three times lower mean relative error than the original product; this ratio increases to four times for the simulated runoff values. The proposed downscaling scheme is modular in design and can be applied on gridded satellite and reanalysis dataset.

Each year throughout the contiguous United States (CONUS), flood hazards cause damage amounting to billions of dollars in homeowner insurance claims. As climate change threatens to raise the frequency and severity of flooding in vulnerable areas, the ability to predict the number of property insurance claims resulting from flood events becomes increasingly important to flood resilience. Based on random forest, we develop a flood property Insurance Claims model (iClaim) by fusing records from the National Flood Insurance Program (NFIP), including building locations, topography, basin morphometry, and land cover, with data from multiple sources of hydrometeorological variables, including flood extent, precipitation, and operational river-stage and oceanic water-level measurements. The model utilizes two steps—damage level classification and claim number regression—and subsampling strategies designed accordingly to reduce overfitting and underfitting caused by the flood claim samples, which are unevenly distributed and widely ranged. We evaluate the model using 446,446 grid samples identified from 589 flood events occurring from 2016 to 2019 over CONUS, overlapping 258,159 claims out of a total of 287,439 NFIP records of the same period. Our rigorous validation yields acceptable performance at the grid/event, county/event, and event accumulative level, with R2 over 0.5, 0.9, and 0.95, respectively. We conclude that the iClaim model can be used in many application scenarios, including assessing flood impact and improving flood resilience.