The Development of NAEFS Statistical Post-Processing System

The North American Ensemble Forecast System (NAEFS) is a joint global ensemble forecast system (GEFS) that consists of the NCEP ensemble forecasts and the Canadian Meteorological Centre's ensemble forecasts. The objective of NAEFS is to improve the probabilistic forecast skill. Since its first implementation in 2006 at NCEP, NAEFS has been maintained and continuously enhanced through its Statistical Post-Processing (SPP) system that includes 1) bias correction process and 2) statistical downscaling processing. Currently, NAEFS provides global ensemble probability products for 1*1 degree, 5km CONUS and 6km Alaska probabilistic products 4 times per day, every 6-hour out to 16 days.

This study will focus on the future NAEFS SPP development and implementation: 1). Generate high resolution (NDGD grid) downscaled probabilistic forecast for CONUS (2.5km) and Alaska (3km) with additional variables (wind gust, cloud cover and etc.); 2). Use reforecast data to improve calibrated ensemble forecasts; 3). Introduce varied decaying coefficients (varied by lead time, domains and seasons) to optimum the 1st moment bias correction technique; 4) Apply Recursive Bayesian Model Process (RBMP) method to improve blending process (different weights) and 2nd moment; and 5). Introduce smart initialization system (developed by NCEP/EMC and ESRL/GSD) to adjust NDFD high resolution grid data before applying NAEFS statistical downscaling process. Preliminary results demonstrate that additional values are added from each stated process.