HAPpy Hour Seminar - Addressing Satellite Precipitation Uncertainty Across Modeling Applications

This talk will review research work that incorporates satellite precipitation uncertainty into three distinct modeling applications: landslide hazard assessment, hydrologic modeling for flood forecasts, and land surface modeling predictions. The unique challenges to characterizing and incorporating precipitation uncertainty into these models, as well as the potential benefits of doing so, will be discussed. One such challenge, that of representing precipitation error across arbitrary space-time scales, is addressed by developing the Space-Time Rainfall Error and Autocorrelation Model (STREAM), which simulates the nonstationary, anisotropic autocorrelation structure of precipitation error independently of ground reference data. Results demonstrate the integral role that error autocorrelation plays in estimating satellite precipitation uncertainty across scales. The STREAM approach presented in this work is promising for future applications because it can ‘stitch together’ precipitation uncertainty estimates without reliance on ground reference data and is adaptable to other precipitation products, such as error-prone numerical weather model fields.