A Multi-Platform Dataset Assessment of Observational Uncertainty in Extreme Precipitation Events over the Continental United States

Extreme precipitation events are associated with numerous societal and environmental impacts. Furthermore, anthropogenic climate change is projected to alter trends in precipitation intensity across portions of the Continental United States (CONUS). Therefore, a spatial understanding and intuitive means of monitoring extreme precipitation over time is critical. Towards this end, we apply an event-based indicator, developed in support of the ongoing efforts of the US National Climate Assessment, which assigns categories to extreme precipitation events based on 3-day storm totals as a basis for dataset intercomparison. To assess observational uncertainty across a wide range of historical precipitation measurement approaches, we intercompare in situ station data, satellite derived precipitation data from NASA’s Tropical Rainfall Measuring Mission (TRMM) product, gridded in situ station data from the Parameter-elevation Relationships on Independent Slopes Model (PRISM), global reanalysis from NASA’s Modern Era Retrospective-Analysis version 2 (MERRA 2), and regional reanalysis with gauge data assimilation from NCEPs North American Regional Reanalysis (NARR). Results suggest considerable variability across the five platform suite in the frequency, spatial extent, and magnitude of extreme precipitation events. Higher resolution datasets were found to capture a greater frequency of high-end extreme events compared to lower spatial resolution datasets. Differences persist after datasets are spatially aggregated to a standard grid. PRISM maintains a high relative occurrence of extreme events even after spatial interpolation, while TRMM tends to underestimate extreme events over complex topography. MERRA 2 reanalysis tends to systematically underestimate event frequency and magnitude relative to the other datasets. The degree of dataset agreement varies regionally, however all datasets successfully capture the seasonal cycle of precipitation extremes across the CONUS. These intercomparison results provide additional insight about observational uncertainty and the ability of multi-platform precipitation estimates to capture extreme events.