北京大学学报自然科学版 ›› 2021, Vol. 57 ›› Issue (4): 632-644.DOI: 10.13209/j.0479-8023.2021.042

上一篇    下一篇

全球变暖背景下内蒙古地区沙尘暴频次变化的预估

杨诗妤1,2, 闻新宇1,†   

  1. 1. 北京大学物理学院大气与海洋科学系, 北京 100871 2. 全球变化与中国绿色发展协同创新中心, 北京 100875
  • 收稿日期:2020-04-30 修回日期:2020-07-03 出版日期:2021-07-20 发布日期:2021-07-20
  • 通讯作者: 闻新宇, E-mail: xwen(at)pku.edu.cn
  • 基金资助:
    国家自然科学基金(41875088, 41630527)和中央高校基本科研业务费专项资金资助

Prediction of Dust Storm Frequency Variation in Inner Mongolia Region under Global Warming Scenarios

YANG Shiyu1,2, WEN Xinyu1,†   

  1. 1. Department of Atmospheric and Oceanic Sciences, School of Physics, Peking University, Beijing 100871 2. Joint Center for Global Change Studies (JCGCS), Beijing 100875
  • Received:2020-04-30 Revised:2020-07-03 Online:2021-07-20 Published:2021-07-20
  • Contact: WEN Xinyu, E-mail: xwen(at)pku.edu.cn

摘要:

构建一个基于BP神经网络的统计模型, 利用CMIP5模式中历史情景和未来情景的预估数据, 重建1860—2100年内蒙古地区春季沙尘暴频次(分辨率达到日尺度)序列。在此基础上, 研究内蒙古地区沙尘暴未来长期变化特征。结果表明, 在未来情景RCP2.6 和RCP8.5中, 与历史时期(1860—2005年)相比, 内蒙古地区沙尘暴频次持续减少; 影响范围较大的沙尘暴事件占比也持续减少; 在增温更多的RCP8.5情景中, 沙尘暴的减少更加显著; 春季沙尘暴的季节性锁相特征(4月沙尘暴频次达到峰值)不随全球变暖而变化。

关键词: 沙尘暴, 神经网络, CMIP5, 全球变暖, 内蒙古地区

Abstract:

The authors reconstruct a new dust storm series on daily basis for Inner Mongolia Region covering the period 1860?2100, by applying a newly developed artificial neural network model onto CMIP5 results from historical and RCP runs. The authors investigate the new series and suggest that the number of China’s dust storms keeps decreasing in both RCP 2.6 and 8.5 scenarios; the proportion of dust storms with large impact area also decrease; the decreasing is more evident in the warmer RCP8.5 scenario than that in RCP2.6; the phase-lock features, i.e. maximum occurrence frequency of dust storms in April every year, remain unchanged in both global warming scenarios. 

Key words: dust storm, neural network, CMIP5, global warming, Inner Mongolia