Validation of the Air Force Weather Agency Ensemble Prediction Systems
Abstract
Air Force Weather Agency's (AFWA) Ensemble Prediction Systems (EPS), Global Ensemble Prediction System (GEPS), 20km Mesoscale Ensemble Prediction System (MEPS20) and 4km Mesoscale Prediction System (MEPS4), were evaluated from April to October 2013 for 10 locations around the world to determine how accurately forecast probabilities for wind and precipitation thresholds and lightning occurrence match observed frequencies using Aerodrome Routine Meteorological Reports (METARs) and Aerodrome Special Meteorological Reports (SPECIs). Reliability diagrams were created for each forecast hour detailing the Brier skill score (BSS) to depict EPS performance compared to climatology for each site and score composition through reliability, resolution and uncertainty. To illustrate how the BSS changed, the score and its decomposition were plotted for all forecast hours. This study showed that all three EPS suffered from a lightning overforecasting bias at all locations and most forecast hours. For wind speeds, it was clear that decreased model grid spacing allowed better resolution of terrain features, producing a better BSS. Likewise, precipitation was better resolved with increased horizontal resolution as explicit resolution of precipitation processes outperformed cumulus parameterization schemes.
Document Details
- Document Type
- Technical Report
- Publication Date
- Mar 27, 2014
- Accession Number
- ADA598471
Entities
People
- William B. Clements
Organizations
- Air Force Institute of Technology