中文摘要：

介绍了利用全球海气耦合气候模式ECHO—G进行的小冰期以来的长时间积分气候模拟试验，并与中国区域温度重建资料作对比。共做了2个模拟试验：控制试验和真实强迫试验。首先将模拟结果与器测资料作对比，验证该模式模拟中国气候的能力；然后将模拟结果与中国10个区域重建的500年温度序列进行比较分析。均值、方差及EOF分析表明，对于1550年至今的时段，除了东北、新疆、西南地区外，其它地区模拟与重建序列的相关性尚好，相关的置信度超过90％；而对于1760年以来的时期，除了华南、西北、西南地区外，其它地区二者较为一致，相关的置信度均超过90％，表明气候模式ECHO-G能捕捉到中国大部分地区温度的趋势及低频变化特征，也说明上述强迫条件是近500年来气候变化的主要控制因子。然而模拟的温度距平的空间差异性比重建资料的小，对于年代际、年际等短时间尺度的温度变化模拟与重建结果的吻合度较差。误差来自于重建和模拟两个方面。在重建资料方面，需要提高代用资料的代表性、精确性和可靠性；在模拟方面，应提高各强迫条件序列的精确性，同时应引入更多的强迫因子，如下垫面植被及工业气溶胶等。这样从两方面努力，才能更深入地刻划和认识中国历史气候演变及其成因机制问题。

英文摘要：

Climatic simulation experiments of long-term integration since the Little Ice Age by the use of the global atmosphere-ocean coupled climate model ECHO-G are introduced in this paper. The simulated temperature series were compared with the reconstructed temperature of China. There were two modeling experiments： control run and forced run. First, modeling result was compared with the observed data for examining the model ability of simulating the climate of China. Then the simulated temperature series were compared with the reconstructed 500- year temperature series in 10 regions of China. The analysis of mean value, variance and EOF showed that for the period from 1550 to the present, there are good correlations between the simulated and reconstructed series except for the regions of the Xinjiang, the Northeastern and the Southwestern China. The correlation confidences are greater than 90% for this period. For the period since 1760, most regions have good correlations with more than 90% confidence levels between simulations and reconstructions except for south, northwest, and southwest of China. This indicates that the climate model ECHO-G can simulate the temperature characteristics of the trends and low frequency changes in most regions of China, and the forcing factors used in the simulations are the main controlling factors for climaie change during the last 500 years. However, the spatial differences of simulated temperature anomalies are less than that of reconstructions, and for shorter term variations, such as decadal ＇and interannual changes , the model results are not consistent with the reconstructions very well. The errors may come from both the simulation and the reconstruction. For the reconstruction, the representation, accuracy, and reliability of proxies need to be improved. And for the simulation, more accurate forcing series and more forcing factors, such as land surface vegetation and industrial aerosols need to be drawn into the modeling. If much more work both in simulation and reconstruc