Wednesday, 15 July 2020: 2:40 PM
Virtual Meeting Room
Atmospheric River (AR) events can cause flooding and deaths and are responsible for nearly half of California's annual precipitation. We evaluate forecasts of numerous AR events that have occurred over the past few years by running retrospective simulations and downloading existing output of the RAP/HRRR mesoscale weather model, and compare quantitative precipitation forecast (QPF) and other model parameters to available observations. We evaluate multiple versions of the RAP/HRRR model (HRRR versions 2, 3, and 4) as well as simulations of HRRRv4 with and without X-band radar data included in the data simulation. Model forecasts are compared to widely-used quantitative precipitation estimation (QPE) products such as the Stage IV analysis, rain gauges, and radars, via several statistical techniques including closest-grid-point and neighborhood maximum. While the HRRR model versions have some differences in spatial/temporal QPF accuracy, all model versions predict QPF reasonably well, but have some consistent spatial biases, such as overprediction of rainfall in the Sierra Nevadas, and underprediction of rainfall in the Bay Area, based on the few cases studied. We explore possible causes of these biases.
- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner