12:00 PM - 12:15 PM
[AAS04-12] Improving Ensemble Forecasts by Applying Blending Short-term Forecasts
Keywords:Ensemble forecasts, Nowcasting, short-term forecasts
The time-lagged ensemble forecasts combine previous model runs, can reflect the latest observational data by assimilating, and are suitable for short-term prediction. Successful digestion of observational data represented improved skill scores in high-resolution-short-term forecasts. To represent probability of severe weather in CONUS, time-lagged ensemble forecasts are suggested by combining the maps of the mosaics of column maximum reflectivity (CMAX) from WSR-88D, EXT, HRRR, and blending of HRRR and EXT (i.e., SAL). Three ensemble forecasts of 23 and 40 dBZ (ensEXT, ensHRRR, and ensSAL) are predicted using different numbers of previous forecasts (i.e., 1 to 7) and initializations (00, 06, 12, and 18UTC) from the data of mid-May to mid-June 2015. The skill scores of the ensemble forecasts were evaluated as Brier Skill Scores (BSSs), reliabilities, and resolutions (indicated as areas under Relative Operating Characteristics curves). The ensSAL of 40 dBZ showed the highest BSSs in 1 to 2 h forecasts. Reliabilities and resolutions of ensSAL of 40 dBZ indicated the best results except 18Z (ensHRRR showed the best scores). The ensSAL would be the best option for the short term forecasts (1 - 2 h) by obtaining useful information about high-impact weather (over 40 dBZ).
This work was funded by the Korea Meteorological Industry Promotion Agency under Grant KMIPA 2015-1050. This work was financially supported by the BK21 plus Project of the Graduate School of Earth Environmental Hazard System.