Performance of the NCEP Global Forecast System (GFS) and ECMWF High Resolution (HRES) forecast models during NOAA's 2016 El Niño Rapid Response Campaign

​Juliana Dias

NOAA/ESRL PSL

Tuesday, Feb 21, 2017, 2:00 pm
DSRC Room 2A305

Abstract

We assess the quantitative precipitation forecasts (QPF) skill of the NCEP Global Forecast System (GFS) and ECMWF High Resolution (HRES) forecast models during NOAA’s El Niño Rapid Response (ENRR) field campaign. ENRR targeted equatorial Pacific atmospheric convective activity during January-March 2016 through enhanced observations using dropsondes from the NOAA G-IV aircraft and radiosonde observations from Kiritimati (Christmas) Island and the NOAA research ship the Ronald H. Brown. During the field campaign GFS and HRES model output were used for guidance in planning flight operations from Honolulu to equatorial convective regions 24 hours and more in advance. Our impressions during these planning sessions was that extratropical forecast skill was far superior to that in the tropics at all forecast lead times. We obtained global GFS and HRES precipitation forecasts out to 240 hours on a .25 degree grid for the entire December-March 2015/16 period, and compared these to satellite derived Global Precipitation Measurement (GPM) precipitation estimates for the same period. Forecast skill was quantified using a variety of metrics including space-time correlations for various latitude bands, mean-square errors of precipitation forecasts over the global tropics and extratropics as well as the Pacific domain, and Brier skill scores. Results reveal that in general initial conditions were reasonably well estimated in both forecast systems, as indicated by relatively good QPF scores for the 6-12 hour forecasts. In the tropics, a well-documented overestimation of light precipitation develops almost immediately, being more prevalent in the GFS than in the HRES. In general, extratropical Northern Hemisphere QPF in the GFS was superior to the HRES at all lags and for all sectors, but this was reversed in the tropics, with HRES scoring somewhat better. However, overall, the tropical QPF in both systems was not considered useful by typical metrics much beyond 6 days using typical metrics, although this is situationally dependent, as seen by the increased skill of especially the HRES system during the Madden-Julian Oscillation (MJO) event of December 2015. In both systems, tropical skill over the eastern Pacific was increased during mid- to late February 2016, when a strong “westerly duct” formed over the region, allowing the penetration of extratropical Rossby wave energy into the equatorial region​.​