土壤湿度年际变化对中国区域极端气候事件模拟的影响研究 I. 基于CAM3.1的模式评估

利用NCAR大气模式CAM3.1对中国区域近40年的极端气候事件进行了模拟试验;在此基础上, 利用1961～2000年中国区域452站的逐日最高、最低气温和降水资料, 从气候平均、年际变化和长期变化趋势等方面全面评估了该模式对中国极端气候事件的模拟能力。结果表明:(1)模式对中国区域极端气候指数气候平均态的大尺度空间分布特征具有一定的模拟能力;模式对极端降水指标空间分布的模拟能力较好, 而对极端气温指标的模拟较差;模式对极端气候指标的模拟存在系统性的偏差, 模拟的极端降水的系统性偏差要远大于对极端温度的模拟。(2)模式对极端气温指数的年际变化特征具有较强的模拟能力, 而对极端降水指数的年际变化基本没有模拟能力;模式模拟的各极端降水指标的年际变幅与观测存在较大的偏差。(3)模式较好地模拟出了暖夜和暖昼指数在中国大部分区域的增加趋势, 但变幅较实测偏小;模式对热浪持续指数长期趋势的模拟则相对略差。模式对极端气温指标长期趋势的模拟能力总体优于对极端降水指标的模拟。模式对极端降水频次和中雨日数长期趋势的模拟尚可, 但对持续湿期长期趋势的空间分布模拟较差。研究结果可为该模式用于极端气候的模拟研究提供一定参考。

Impact of Interannual Soil Moisture Anomaly on Simulation of Extreme Climate Events in China. Part I: Model Evaluation of CAM3.1

Extreme climate events over China in recent 40 years are simulated by using NCAR Community Atmosphere Model (CAM3.1). Based on the observed daily data of maximum/minimum temperature and precipitation at 452 stations from 1961 to 2000 in China, the performance of CAM3.1 is evaluated from three aspects, i.e., climatology, interannual variations, and long-term trends. Results show that: 1)CAM3.1 can generally reproduce the basic features of the large-scale spatial patterns of the annual-mean extreme climate indices. The model performs better in simulating the spatial patterns of extreme precipitation than extreme temperature simulation. Systematic bias is found in the simulation of extreme climate events, and the bias in extreme precipitation simulation is evidently larger than that in the extreme temperature simulation on the whole. 2) CAM3.1 can ideally reproduce the interannual variations of the temperature extreme indices, but has poor performance in simulating the interannual variations of the precipitation extreme indices. Large bias is found in the amplitude of the interannual variations between the simulated and observed extreme precipitation events. 3) Increasing trends of both Tn95p (warm nights) and Tx95p (warm days) over most areas of China are well reproduced by the model, but the observed trends are underestimated to some extent. In contrast, the capability of the model in simulating the long-term trends of HWDI (heat wave duration) is poor. Overall, the capability of the model in simulating the long-term trends of precipitation extremes is poorer than those of temperature extremes. CAM3.1 can also capture the long-term trends of extreme precipitation events such as P95p (frequency of extreme heavy precipitation) and R10 (number of days with precipitation greater than 10 mm) in some regions of China, but could not reproduce the long-term trends of CWD (consecutive wet days) very well. Results can provide some references for using CAM3.1 in extreme climate simulation.