Real-time Verification of Quantitative Precipitation Forecasts Produced During the 2002 International H2O Project (IHOP)

Andrew F. Loughe¹, Linda Wharton², Jennifer Luppens Mahoney², and Edward I. Tollerud²

¹Cooperative Institute for Research in the Environmental Sciences (CIRES)

²NOAA/Forecast Systems Laboratory, Boulder, CO

Contact: Andrew.Loughe@noaa.gov

During May and June 2002, the International H20 Project (IHOP) was held over the south-central plains of the United States. Numerous operational and experimental Numerical Weather Predition models were utilized as guidance to forecasters focused on predicting the onset of convection. These forecasters provided guidance to personnel manning flights to obtain high resolution profiles of the atmospheric moisture field. Precipitation forecasts from these models, which include ETA-12, RUC-20, RUC-10, WRF-22, WRF-10, LAPS/MM5-12, and LAPS/MM5-4, were evaluated by the Real-Time Verification System (RTVS) using a standard grid-to-station approach, as well as a grid-to-grid approach following the methods of Ebert and McBride, 2000.

For the first approach, model forecast values were interpolated to hourly precipitation gauge stations available from the Hydrometeorological Automated Data System (HADS), and to stations available from the Oklahoma Mesonet. Model precipitation forecasts were reduced to dichotomous forecasts at thresholds ranging from .01 to 2.0 inches, and for accumulation periods of 3, 6, and 12 hours. A relational database application was utilized to store raw forecast/observation values at each station, and this information was then used to compute a variety of skill scores for evaluation purposes. Results are displayed on the web in both graphical and tabular form, and include analysis of the Equitable Skill Statistic (or Equitable Threat Score) and the frequency bias. These results are available at http://www-ad.fsl.noaa.gov/fvb/rtvs/ihop/station.

In addition to the standard verification statistics provided through the above listed RTVS web page, an error partitioning approach, applied most notably in Australia by Ebert and McBride, was used in near real time to provide feedback directly to IHOP forecasters during the field operation (http://www-ad.fsl.noaa.gov/fvb/rtvs/ihop/grid). This diagnostic approach determines the displacement, volumetric, and pattern error of individual forecasts as compared to the newly enhanced 4km Stage IV precipitation analysis provided by the National Center for Environmental Prediction (NCEP). The initial statistical results and verification techniques generated by RTVS for the IHOP field project will be presented in this paper.