Applicability Assessment And Improvement Of PRECIS Model On The Loess Plateau

Under the background of global warming, accurate prediction of climate change is of great importance for regional sustainable development, which can provide information for adaptive strategies of climate change. Due to the low resolution fo global climate models(GCM), it is necessary to downscale GCM to provide high resolution projection. Regional climate model(RCM) is an important tool for GCM downscaling, but its simulation still has some deviations. As the accuracy of RCM projection directly influences the results impacts assessment, RCM should be thoroughly evaluated and corrected to provide accurate simulation and projection. Using ERA40 reanalysis data as the boundary conditions to drive PRECIS(Providing Regional Climates for Impacts Studies), this study assessed the capability of PRECIS to simulate precipitation and temperature on the Loess Plateau during 1960–2000. The performances of three correction method were compared to determine the best approach for model optimization. The precipitation and temperature change scenarios on the Loess Plateau for the period of 2070-2099 were presented based on the above evaluation and correction. The main conclusions are as follows:(1) PRECIS well simulates the overall spatial distribution and tempora change of each variable; however, diviations exist between the observation and simulation. The gradient distribution from southeast to northwest and the temporal changes on different time scales are well simulated, but those simulated are generally greater than the observed. The number of days with light or heavy rainfall is underestimated, and model performs better in simulating rainfall frequency for non-flood seasons than for flood seasons. Due to the overestimated wet days, the rainfall duration and drought duration is respectively overestimated and underestimated. PRECIS is superior for temperature simulation than precipitation since the values and change trends are similar as those observed.(2) The inter-site and inter-variable correlation are not well simulated by PRECIS. The spatial correlation of temperature is overestimated while that of precipitation underestimated. As for the inter-variable correlation, PRECIS underestimates the correlation between precipitation and termperature; however, it well reproduces the inter-variable correlation among the mean, maximum and minimum temperature.(3) The three correction method can effectively reduce the deviation between the simulated and observed climate; however, the performances of different correction method are different. The Linear Scaling method outperforms the other methos for correction of temperature, while the Quantile Mapping method is superior to correcting precipitation. Therefore, the selection of correction method should be cautious.(4) Under the medium-low emission scenario B2, the climate during 2070-2099 tends to be wetter and warmer on the Loess Plateau. Compared with the baseline period of 1960-2000, temperature will keep increasing with a lower change rates. The change rate of precipitation will be changed from-1.8 mm a-1 to 3.2 mm a-1. The gradient distribution of temperature and precipitation from southeast to northwest will be kept; however, the regional differences will be intensified.