中文摘要：

文中利用区域气候模式RegCM3，以1998年夏季中国异常气候事件为例，对季节尺度区域气候模拟适应调整时间的选取问题进行了数值研究。共做了11个试验，每个试验的适应调整时间最短为10d，最长为6个月，以检验适应调整时间长短对夏季中国异常气候事件模拟结果的影响。结果表明：对于大气变量而言，模式通常在经过4—8d的适应调整时间后，就进入“气候模态”运行，此后模拟误差主要是由于模式对区域内大气过程描述能力不足造成的，对适应调整时间选取不再敏感，这进一步证实了区域气候模拟是一个边值问题的观点。各气候区平均降水量模拟结果受适应调整时间影响也不大，但不同的适应调整时间对降水分布格局模拟将产生一定影响，降水分布模拟结果随适应调整时间的不同存在一定的不确定性，这种不确定性通常出现在强降水发生区域。总之，对于季节尺度降水模拟，适应调整时间大于2个月效果更好。对降水分布格局模拟误差和东亚夏季风系统进退过程之间关系的进一步分析发现，模式对受夏季风系统影响比较大的区域模拟的降水相关系数变化性也比较大，因此，发展合适的积云对流参数化方案以提高受夏季风系统直接影响区域强降水过程的描述能力是改进区域气候模式对中国区域夏季气候模拟效果的有效途径。

英文摘要：

The numerical experiments on the issue of spin-up time for seasonal-scale regional climate modeling were conducted with the newly Regional Climate Model （RegCM3）, in the case of the abnormal climate event in the summer of 1998 in China. To test the effect of spin-up time on the regional climate simulation results for such an abnormal climate event in summer, a total of 11 experiments were performed with different spin-up time from 10 days to 6 months, respectively. The simulation results show that, for meteorological variables in the atmosphere, the model would be running in ＂climate mode＂ after 4-8 day spin-up time, and then, independent to the spin-up time basically, and the simulation errors would be mainly caused by the model＇s failure in describing the atmospheric processes over the model domain. This again verifies the point of view that the regional climate modeling is indeed a lateral boundary condition problem as demonstrated by earlier research works. The simulated mean precipitation rate over each subregion is not sensitive to the spin-up time, but the precipitation pattern is somewhat different for experiments with different spin-up time, which shows that there exists uncertainty in the simulation of precipitation pattern, and such uncertainty exhibits more over the area where heavy rainfall happened. Generally, for monthly-scale precipitation simulation, one month spin-up time is enough, whereas 2 months spin-up time is better for seasonal-scale one. Furthermore, the relationship between the error of precipitation pattern and the advance/withdraw of East Asian summer monsoon were analyzed. It is found that the variability of correlation coefficient for precipitation is more significant over the area where the summer monsoon is predominant. Therefore, the model＇s capability in reproducing precipitation features is related to the heavy rainfall processes associated with the advance/withdraw of East Asian summer monsoon, which suggests that it is necessary to develop a more reliable para