Verification of the applicability of PRECIS-simulated temperature on the Loess Plateau of China

Temperature is a critical variable in plant growth and water cycle. Under the background of global warming, projection of potential changes in temperature can provide important information for related issues. Although Regional Climate Models (RCMs) are popular for climate projection due to higher resolution compared with General Circulation Models (GCMs), their simulation accuracy should still be assessed in details. By using ERA40 reanalysis data as the boundary conditions to run PRECIS, this study assessed the ability of PRECIS to simulate temperature on the Loess Plateau of China. The spatial distribution and temporal changes of mean, maximum and minimum temperature as well as extreme temperature events from PRECIS were compared with those observed. Results show that the spatial distributions of the observed mean and minimum temperature are simulated with an absolute error of < 2 °C in most regions; however, the spatial patterns of maximum temperature are not reproduced well since the absolute error is > 2 °C. The temporal trends of three temperature variables are presented similarly as the observed, among which the mean temperature is best simulated; however, the performance differs at different time scales. Although the deviations of monthly temperature are different for each season, the change rates of annual mean temperature are better simulated with deviation of 0.011 °C a− 1. For extreme temperature events, most of the simulated indices have similar spatial distribution and temporal trend as the observed, but the deviations of change rates are overvalued . Therefore, the PRECIS-simulated temperature cannot be directly used and further correction should be carried out on the Loess Plateau.