Performance of national weather service forecasts compared to operational, consensus, and weighted model output statistics

Model output statistics (MOS) guidance has been the central model postprocessing approach used by the National Weather Service since the 1970s. A recent advancement in the use of MOS is the application of consensus MOS (CMOS), an average of MOS from two or more models. CMOS has shown additional skill over individual MOS forecasts and has performed well compared to humans in forecasting contests. This study compares MOS, CMOS, and WMOS (weighting component MOS predictions by their past performance) forecasts of temperature and precipitation to those of the National Weather Service (NWS) subjective forecasts. Data from 29 locations throughout the United States from 1 August 2003 through 1 August 2004 are used. MOS forecasts from the Global Forecast System (GMOS), Eta (EMOS), and Nested Grid Model (NMOS) models are included, with CMOS being a simple average of these three forecasts. WMOS is calculated using weights determined from a minimum variance method, with varying training periods for each station and variable. Performance is analyzed at various forecast periods, by region of the United States, and by time/season, as well as for periods of large daily temperature changes or large departures from climatology. The results show that CMOS is competitive or superior to human forecasts at nearly all locations and that WMOS is superior to CMOS. Human forecasts are most skillful compared to MOS during the first forecast day and for periods when temperatures differ greatly from climatology. The implications of these results regarding the future role of human forecasters are examined in the conclusions.