A Piecewise-Integration Method For Climate Simulation And Its Application On Desertification Sensitivity Experiments

The initialization error and model error can cause large inaccuracy in climate simulations, and the inaccuracy can not be overcome. To investigate this problem, we study on the simulation of climate change caused by the external forcing changes by the regional climate model. The experiments show, the regional climate model has the ability to reappear the present climate. However, there are incredible errors in the simulation which make the results of the sensible experiments about external forcing change to be uncertain. To reduce the uncertainty, we propose a piecewise-integration method in which the observation and analysis is absorbed into the simulation by the four-dimension data assimilation method, and that make the simulation of current state of atmosphere more close to the reality, and the sensitivity experiment more believable. Then, we use the method to study the influence of desertification in North-West China. The major works are summarized as follows:(1) Based on the traditional method, we study on the influence of desertification in North-West China by the regional climate model (RegCM3), and also study on the influence of initialization error and model error, which are small in present climate simulation and become opposite large in the sensible experiments. Furthermore, the larger influence cannot be reduced by the long-term run of model in the sensitivity experiments.(2) Based on the data assimilation technique, the theory and method of the Piecewise-Integration Method (PIM) is described. The sensitivity experiments by a low-order spectral model show that, the model output may be turn to another equilibrium status and occurs "climate drift" along with accumulation of model error. We continuously replace the initial filed in the new method to reducing the accumulation. It is found that the new method is feasible, which can reduce the accumulation of model error well and prevent "climate drift" occurring.(3) Focused on the "climate drift" caused by the accumulation of model error, the comparison between Piecewise-Integration Method and traditional Continuously-Integration Method is examined again by the complex model (MM5). The Piecewise-Integration Method can reduce the accumulation of model error well, and make the simulation more close to the current state of atmosphere, which ensures the results of sensible experiments in Piecewise-Integration Method more credible than in traditional method.(4) On the basic of desertification speed of China in 1990s, enlarge the area of desert in North-West China, and study on the influences of desertification in summer by the MM5 model in the Piecewise-Integration Method. It is found that, the precipitation reduces in most of northern China, and increases in southern China after desertification expanding, which will aggravates the precipitation difference between the south and the north in China.