Assessing Bias Correction Methods in Support of Operational Weather Forecast in Arid Environment

In this study, the Weather Research and Forecasting (WRF) model is employed for operational forecasting over the United Arab Emirates (UAE). The goal of this study is to assess two bias correction methods, namely the multiplicative Ratio Correction (RC) and Kalman Filter (KF), in support of operational mesoscale forecasts in the UAE. These techniques are applied to the 2-m temperature with the corrected temperature subsequently used to update the Relative Humidity (RH) predictions. The simulation covers the 2-year period 1st January 2017 to 31st December 2018. To evaluate the WRF performance, Meteorological Aerodrome Reports (METARs) observations at five airport stations are used. It is concluded that when any of the bias correction techniques are applied, there is a significant reduction of the bias and Root-Mean-Square-Error (RMSE). This is particularly true in the summer season and during nighttime and early morning hours, when WRF has a systematic cold bias of up to 2 °C. In addition, the bias distribution is more symmetric with a reduced spread, skewness and kurtosis values. The RC technique is found to give the best scores, with the observed and modelled temperatures generally within 0.25 °C for the first two forecast days. In addition, it successfully removes the model tendency of underperforming in the warm season. A similar improvement in the skill scores is seen in the RH forecasts albeit with smaller magnitudes. The KF and RC techniques used here have been employed successfully in operational forecasts with the potential to expand them to other model variables.