资源科学 ›› 2019, Vol. 41 ›› Issue (6): 1141-1153.doi: 10.18402/resci.2019.06.13

• 气候资源 • 上一篇    下一篇

CMIP5模式对西北干旱区典型流域气温模拟能力评估——以开都-孔雀河为例

李晓菲1,2(), 徐长春1,2(), 李路1,2, 罗映雪1,2, 杨秋萍1,2, 杨媛媛1,2   

  1. 1. 新疆大学资源与环境科学学院,乌鲁木齐 830046
    2. 新疆大学绿洲生态教育部重点实验室,乌鲁木齐 830046
  • 收稿日期:2018-11-16 修回日期:2018-12-10 出版日期:2019-06-25 发布日期:2019-06-25
  • 作者简介:

    作者简介:李晓菲,女,新疆博乐人,硕士生,主要从事干旱区气候变化研究。E-mail: lxf_0306@163.com

  • 基金资助:
    国家自然科学基金项目(41561023);2017新疆研究生科研创新项目(XJGRI2017009)

Evaluation of air temperature of the typical river basin in desert area of Northwest China by the CMIP5 models:A case of the Kaidu-Kongqi River Basin

Xiaofei LI1,2(), Changchun XU1,2(), Lu LI1,2, Yingxue LUO1,2, Qiuping YANG1,2, Yuanyuan YANG1,2   

  1. 1. College of Resource and Environmental Science, Xinjiang University, Urumqi 830046, China
    2. Key Laboratory of Oasis Ecology of Ministry of Education, Xinjiang University, Urumqi 830046, China
  • Received:2018-11-16 Revised:2018-12-10 Online:2019-06-25 Published:2019-06-25

摘要:

西北干旱区水资源问题突出,全球变暖将进一步加剧其水资源短缺,研究未来气候变化对流域水资源合理分配和使用具有重要意义。本文利用CRU(Climate Research Unit)数据和DCHP(Downscaled CMIP3 and CMIP5 Climate and Hydrology Projections)提供的32个经BCSD降尺度的CMIP5(全球耦合模式比较计划第五阶段)模式气温数据,采用线性倾向估计、滑动平均、M-K(Mann-Kendall)检验及滑动T(MMT)等检验法,以西北干旱区典型流域开都-孔雀河流域为例,通过对1950—2005年的年平均气温、年平均最高气温与年平均最低气温3个指标的变化趋势及突变年份进行检测,评估各模式及模式集合平均对气温变化的模拟能力。研究结果表明:①12个模式能够准确模拟出1950—2005年流域内各气温指标的显著增加趋势,8个模式能够模拟出部分气温指标的增温趋势,但均低估了增温速率,集合平均也存在同样问题;②除FIO-ESM与MPI-ESM-MR能够准确模拟出气温突变时间外,绝大多数模式不能够准确模拟出。基于优选模式的集合平均PM-PLS和PM-EE对突变的模拟能力总体上优于单个模式,其中PM-PLS模拟能力更优;③对PM-PLS模式集合平均进一步评价,发现其能较好地再现流域气温线性趋势的时空变化总体特征,但仍存在增温速率低估的问题。采用气候模式进行未来气候预估仍需加强模式优选及多模式集合平均方法的深入研究。

关键词: 气温, 泰勒图, 多模式集合平均, Mann-Kendall, 突变检验, 开都-孔雀河流域

Abstract:

Global warming will result in serve water shortage, aggravating the existing outstanding water problem in the Kaidu-Kongqi River Basin. Studies on the future climate change will contribute to the rational distribution and utilization of water in the basin. Based on the CRU(Climate Research Unit) dataset and 32 BCSD-downscaled CMIP5 model air temperature dataset from DCHP (Downscaled CMIP3 and CMIP5 Climate and Hydrology Projections), the paper assessed the simulation ability of both 32 models and multi-model ensemble mean through the test of long-term trend and abrupt change of annual average, maximum and minimum air temperature in the Kaidu-Kongqi River Basin over the period of 1950-2005 by using the methods of linear trend calculation, moving average, Mann-Kendall (M-K) test and moving T-test (MMT). Results show that (1) 12 of 32 models are capable of reproducing the significant warming trend of three temperature indicators during 1950-2005, 8 of 32 models can only simulate that of some temperature indicators, but all of them underestimate the warming rate, so does the multi-model ensemble mean. (2) Most models failed to simulate the time of abrupt change accurately except two, FIO-ESM and MPI-ESM-MR. The ensemble mean of preferred models, PM-PLS and PM-EE, are superior to the individual model in simulating abrupt change. Between them, PM-PLS is better. (3) The further evaluation indicates that the multi-model ensemble PM-PLS can better capture the linear trend of spatio-temporal characteristics, but the problem of underestimating the warming rate still exists. It appeals to strengthen the study of model optimum selection and multiple models assemble in the future climate prediction using climate models.

Key words: air temperature, Taylor diagram, multi-model ensemble mean, Mann-Kendall, abrupt change, Kaidu-Kongqi River Basin