Performance evaluation of cumulus parameterization schemes from Weather Research and Forecasting model (WRF) for upper air forecasting

Abstract

In order to forecast upper air (air temperature, relative humidity, wind speed, and wind direction) using a short-term weather forecasting model (Weather Research and Forecasting, WRF), the cumulus parameterization must be specified. This paper evaluates the performance of three cumulus parameterization schemes which are Betts-Miller-Janjic Scheme (BMJ), Grell 3D Ensemble Scheme (G3), and Non-Cumulus Scheme when using in the WRF model at resolution 3 × 3 kilometers to forecast upper air over Thailand during the Tropical Storm SONCA on 26-28 July 2017. The upper air forecasts are compared to radiosonde observations of upper air conditions collected from 4 stations by the Department of Royal Rainmaking and Agricultural Aviation (DRRAA) at air pressure of 1,000 – 100 Hectopascal (hPa), 00 UTC. The accuracy of the forecast is measured by the Pearson correlation coefficient and the root mean square error (RMSE). The results from the statistical analysis of the upper air forecasts from WRF model using different cumulus parameterization schemes are indicated that based on Pearson correlation coefficient and RMSE, G3 Scheme was the best for air temperature, and relative humidity forecasts, whereas the Non-Cumulus Scheme was the most suitable for the wind speed, and wind direction variable giving forecasts closest to the observation data.