Using multi-model ensembles to improve the simulated effects of land use/cover change on temperature: a case study over northeast China

Abstract

Rather than simulating the effects of land use and land cover change (LUCC) on the climate using one climate model, as in many previous studies, three regional climate models (Regional Climate Model, version 3; the Weather Research and Forecasting model; and the Regional Integrated Environmental Model System) were used in the present study to simulate changes in temperature due to LUCC. Two experiments (CTL and NE) were designed and run using the three regional climate models. The CTL experiment was used to compare the simulations of the different models and served to illustrate the improvement that could be achieved as a result of employing a multi-model ensemble. The NE experiment was used to evaluate the changes in temperature caused by LUCC in northeast China between 1981 and 2000. The results of the CTL simulations showed that changes in temperature were simulated well by the three regional climate models; however, the simulated temperatures were different, dependent on the model used. The multi-model ensembles [the arithmetic ensemble mean (AEM) and Bayesian model averaging (BMA)] attained better results than any individual model. Of the two ensemble methods, BMA performed better than the AEM. The effects of LUCC on the climate in northeast China were assessed by the differences between the CTL and NE simulations for every RCM and the ensemble simulations. The BMA simulations produced more reasonable results than the other simulations. Based on the results, we can state with some confidence that LUCC in northeast China over the 20-year period studied caused a decrease in temperature, because of an expansion of arable land.