全球变化知识资源中心

Model-based sensitivity experiments are a widely used method for studying climate change attributions. In traditional climate sensitivity studies, climate drift occurs because of the accumulation of model errors during long-term integrations, both in actual and hypothetical states. A more accurate piecewise-integration method is used to reduce the model errors, dividing the continuous simulation into a series of sequential short-term simulations. The model fields are updated at the end of each subinterval with analysis data for the actual state run, and with the sum of the analysis data and the difference between hypothetical and actual states for the hypothetical state run. This paper conducts sensitivity experiments with the continuous-integration method and the piecewise-integration method to evaluate the impacts of the central and eastern Pacific Ocean sea surface temperature anomalies (SSTAs) on the severe drought that occurred in southwestern China (SWC) in the winter of 2009/2010. The results show the following. (a) Model errors produced by the piecewise-integration method for the actual state simulation are significantly less than those produced by the continuous-integration method. (b) The intensity of the drought is significantly overestimated in continuous-integration experiments. However, the drought can be accurately simulated in its spatial distribution and intensity via the piecewise-integration method. Thus, climate sensitivity to changes in external forcing can be studied with greater credibility. (c) The warm central Pacific Ocean SSTAs may be the main cause of the drought over SWC. Adequate precipitation occurred over SWC when the centre of the SSTAs shifted to the east. (d) Warm SSTAs over the central Pacific Ocean influenced precipitation over SWC by weakening the water vapour transport branch from South China Sea to SWC. With a deficit of water vapour, the pronounced subsidence and warm temperatures were the main dynamic and thermodynamic factors that caused and maintained the drought.