Performance of the Regional Climate Model version 4 (RegCM4) with different physical parameterizations over Iran: A case study in 2010

Providing reliable monthly to seasonal forecasts of the climate system using regional climate models with a relatively high spatial resolution is essential to reduce the socio-economic impacts of extreme climate events. In this study, performance of the Regional Climate Model version 4 (RegCM4) with four different sub-grid scale parameterization schemes in simulating 2-m temperature and precipitation over Iran against the observational Climate Research Unit (CRU) dataset in 2010 is evaluated. The climate forecast system version 2 (CFSv2) data are used as initial and boundary conditions for RegCM4. Analysis of 2-m temperature biases of these four simulations by RegCM4 indicated that the largest negative bias is located in the southern coastal plains of the Caspian Sea and over the Alborz Mountain, while the largest positive bias is located over Dast-e Lut and southern Iran. It is found that the largest temperature biases over Iran mostly occur in late spring, during summer and early autumn. Analysis of absolute monthly mean 2-m temperature biases indicated that in two of the conducted simulations the bias is lower, in both of which the Holstlag boundary-layer scheme is used. The root mean square errors (RMSE) of 2-m temperature for these four simulations are also examined and it is found that the Holstlag boundary-layer scheme and UW PBL perform better in cold and warm months, respectively. Using a combination of the Tiedtke convection scheme and the Holstlag boundary-layer scheme significantly reduces the RMSE of 2-m temperature in warm months of the year and leads to the least bias during the whole year over Iran. Thus, to conduct simulations with RegCM4 over Iran, the Holstlag boundary-layer scheme in cold months of the year and the same boundary-layer scheme along with the Tiedtke convection scheme during the whole year are recommended in order to have the least 2-m temperature biases. In the conducted four simulations, positive precipitation biases are observed over Iran in most months of the year, suggesting that RegCM4 generally overestimates precipitation over Iran. Results also indicated that the largest correlation between the observed and simulated precipitation is seen over southeastern, eastern, central, western and northwestern Iran, while the least correlation is seen over the Alborz Mountain, western foothills of the Zagros Mountains, western parts of southern Iran and some parts of Dasht-e Kavir. It is also found that in the RegCM4 simulation with the Tiedtke convection scheme, there are larger spatial and temporal correlations between the observed and simulated precipitation.