Search

• IMERG satellite and MRMS radar precipitation applied for integrated hydrologic validation. • Discrete flood characteristics assessed with respect to USGS stream gauges. • Hydrologic model impacts on flood discharge, duration, and timing are investigated. • IMERG-E discharge errors are higher than MRMS, but IMERG-E duration errors are lower. • Flood timings are better simulated at smaller basin sizes with low systematic error. This study investigates the hydrologic utility of satellite precipitation estimates from the Global Precipitation Measurement mission by comparing flood signals produced across the Continental United States by a ten-year span of in-situ, ground-based radar and satellite-based precipitation data. The flood characteristics generated with radar and satellite precipitation through a distributed hydrologic model are contrasted against reference stream gauge data as a method of integrated validation to assess and quantify error budgets between precipitation products by highlighting precipitation products' accuracy, hydrologic scaling effects, and the impact of the hydrologic model. It is found that systematic and random errors associated with flood characteristics behave similarly to trends previously seen in precipitation rate errors between precipitation products, establishing a clear link through propagation of errors into the water cycle. Additionally, behaviors associated with both water balance and routing schemes within the hydrologic model were shown to affect outputs. Errors generated by water balance tend to cause overestimation of peak discharge values, while errors associated with routing tend to cause underestimation of flood durations and push flood timings earlier than the stream gauge reference. [ABSTRACT FROM AUTHOR]

• A novel framework is developed for integrated hydrologic validation of satellites. • Assessment focuses on flood characteristics in favor of bulk metrics. • The effects of IMERG resolution are seen on flood peak discharge. • Basin size is also found to play a role in over or underestimation of peak discharge. • IMERG simulated floods events start earlier and end later than the MRMS benchmark. In recent years, a great amount of research has been done towards evaluating precipitation data generated by satellites, but less has focused on how these estimates and their uncertainties manifest further into the water cycle. In this study, ten years of satellite-based and ground-based radar data are used as forcings for a distributed hydrologic model across the Continental United States. They are compared using a methodology designed to assess the flood signals and characteristics generated by the model. By focusing on how well the model reproduces flood characteristics rather than fits traditional bulk statistics, this research provides robust insights into satellite precipitation deficiencies. It is found that satellite data has greater success at resolving lower magnitude flood events while tending to generate floods of longer durations. Additionally, flood managers should note that satellites tend to generate floods that characteristically both begin earlier and end later than the ground radar reference. Subsequent research is recommended into other satellite data products in order to better understand these discrepancies and mitigate or plan for them in the future. In recent years, satellites have been increasingly used to provide valuable insights on rainfall across the globe, especially in locations where radars are unable to be installed on the ground. By nature, however, the rainfall data provided by satellites has uncertainties. Large amounts of research has gone into the difference in accuracy between satellite radar and ground references, but less has focused on how these differences impact flood simulations, respectively. In this study it was found that the rainfall data from satellites tends to predict floods that are more severe (have higher magnitudes), last longer (have longer durations), and have different timings (start earlier and end later) than the reference. It is also found that satellites struggle with predictions involving extreme rainfall values. Further research is suggested to better understand these issues, as well as compare against other forms of satellite rainfall data. [ABSTRACT FROM AUTHOR]