Evaluation of MM5 High-Resolution Real-Time Forecasts over the Urban Area of Athens, Greece

Abstract

In this paper the fifth-generation Pennsylvania State University?National Center for Atmospheric Research Mesoscale Model (MM5) forecast skill over an area of complex terrain is evaluated. Namely, the model is verified over a period of 1 yr (2002) over the greater area of Athens, Greece, for its near-surface temperature and wind forecasts, at 8- and 2-km grid spacing, but also over a 15-day period for the summer thunderstorm activity forecasts. For the near-surface temperature a cold bias is evident. The model is, in general, unable to reproduce the summer heat waves observed in the area. The increase of the grid resolution, from 8 to 2 km, results in an improvement of the forecast skill. Postprocessing of the forecasts by applying a Kalman-filtering correction method was very effective for both the 8- and the 2-km forecasts. For the forecast skill of wind, the analysis showed that there is not any net increase of the errors with increasing forecast time for the 48-h forecast period, the mean absolute errors, in general, present the lowest values at noontime, and the increase in resolution, from 8 to 2 km, results in a slight decrease of these errors. The analysis of the model skill to accurately forecast summertime precipitation showed that the 2-km simulations, without activation of the convective parameterization scheme, were unable to reproduce the observed thunderstorm activity. Sensitivity tests for the same period with simulations in which the convective parameterization was not activated for both the 8- and the 2-km simulations were still inaccurate, while activation of the convective parameterization scheme at all grids (even at 2 km) considerably increased the precipitation forecast skill.