898 NWP Forecast Performance for Significant Precipitation Events over Southwest British Columbia

Wednesday, 10 January 2018
Exhibit Hall 3 (ACC) (Austin, Texas)
Anthony Di Stefano, Univ. of British Columbia, Vancouver, BC, Canada; and D. M. Siuta, G. L. West, and R. B. Stull

Accurate forecasting of precipitation intensity and timing are critical for many users, including hydroelectric generation planners. Reservoirs in the complex terrain of southwestern British Columbia (BC) are affected by significant precipitation events produced by cold-season low-pressure systems and atmospheric river events originating over the North Pacific Ocean. The goal of this research is to determine the ability of various models to predict moderate, heavy, and extreme precipitation events.

Sub-daily and daily forecasts from two ensemble prediction systems are evaluated for two past cool seasons over southwest BC. The ensembles are: 1) the short-range ensemble forecast (SREF) run by the University of British Columbia (UBC), and 2) the North American Ensemble Forecast System (NAEFS) run jointly by the U.S. National Centers for Environmental Prediction and the Canadian Meteorological Centre. The UBC-SREF consists of mesoscale-model forecasts produced using the Fifth Generation PSU/NCAR Mesoscale Model (MM5) and the Weather Research and Forecasting (WRF) model, with both the Advanced Research WRF and the Nonhydrostatic Mesoscale Model dynamical cores. UBC-SREF ensemble members are initialized by the Global Forecast System (GFS), North American Mesoscale Model (NAM), Canadian Global Deterministic Prediction System (GDPS), and the Fleet Numerical Meteorology and Oceanography Center Navy Global Environmental Model (NAVGEM).

Mean absolute error (MAE) and correlation statistics are computed across 18 reservoir locations in complex terrain to evaluate deterministic ensemble-member and ensemble-mean forecast performance for moderate, heavy, and extreme precipitation events at sub-daily and daily accumulation intervals. For hourly precipitation forecasts, results for all event categories showed that UBC-SREF ensemble members initialized off the GFS, regardless of the dynamical core, were best for our study period. Coarser ensemble members from the UBC-SREF often produced better forecasts than their fine-resolution nests. The NAEFS ensemble-mean forecast was generally more skilled than any of the mesoscale-model members of the UBC-SREF, except when verifying over heavy precipitation events. Performance of 3-, 6-, 12-, and 24-hourly precipitation accumulations are also presented.

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner