5个CMIP5模式对低纬高原气温的模拟和预估

Simulations and estimates of low latitude highland temperature by 5 CMIP5 models

利用CMIP5中5个气候模式1961—2005年的月平均气象要素资料、低纬高原区高分辨的月平均温度格点资料,采用BP神经网络、EOF分析以及多元回归方法对低纬高原地区2006年1月至2099年12月期间的温度进行预估研究.结果表明,各温度模型集合模拟的低纬高原地区未来94年温度在RCP2.6、RCP4.5、RCP8.5三种排放情景下均呈明显的上升趋势.至2099年,与2005年平均温度相比,RCP2.6排放情景下,温度升高0.9℃,RCP4.5排放情景下,温度升高1.8℃,RCP8.5排放情景下,温度升高3.3℃.

The research applies monthly average data of 5 models on CMIP5 from 1961 to 2005 and temperature of the low latitude highland in high- resolution grid,followed by BP neural network,EOF analysis and Multiple Linear Regression to the area of low latitude highland between January 2006 and December 2099. The result shows that the temperature estimate of the future 94 years from the ensemble model simulation presents an apparent rising trend,in all emission scenario of RPC2. 6,RCP4. 5 and RCP8. 5. In 2099,compared with the average temperature in 2005,temperature is likely to increase by 0. 9℃ on RCP2. 6 emissions scenario,to increase by 1.8 ℃ on RCP4. 5,and to increase by 3. 3 ℃ on RCP8. 5.