A Comparison of Advect Cloud Model and Fifth-Generation Mesoscale Model Total Fractional Cloud Forecasts

Abstract : Total fractional cloudiness (FC) forecasts from the advect cloud (ADVCLD) model and total FC forecasts diagnosed from the relative humidity and total cloud condensate (snow, ice, rain water, and cloud water) fields from the fifth generation mesoscale model (MM5) were statistically and subjectively compared to determine which model produced the better total FC forecasts from August through November 1997 for a forecast window centered on Bosnia. The real time nephanalysis (RTNFPH) model was used to represent the true state of the atmosphere. ADVCLD and MM5 forecasts were also compared against a persistence forecast to provide a minimal skill baseline. The Air Force Weather Agency (AFWA) provided all model data. The Kvamsttheta and Sundqvist schemes were used to transform the relative humidity forecasted at individual grid points by MM5 into a total FC forecast. Two other methods, the vertical column and layered methods, based on threshold techniques and devised by the author were used to transform the total cloud condensate forecasted by MM5 into a total FC forecast. The results indicate that ADVCLD produced the better total FC forecasts for the first 33 hours of the 36 hr forecast period despite having a tendency to produce too much cloudiness. Overall, the MM5 displayed a significant negative bias in both the relative humidity and total cloud condensate fields. The MM5 was found to significantly underforecast cloud cover over the Adriatic Sea and slightly overforecast cloudiness over mountainous regions. ADVCLD demonstrated skill against persistence throughout the forecast cycle, however, MM5 demonstrated skill only after the 12 hr forecast point.