EGI OpenIR
基于 CMIP5 模式中亚地区降水与气温模拟不确定性分析与订正
Alternative TitleError Analysis and Correction for Precipitation and Air Temperature over Central Asia based on CMIP5 Models
黄芳
Subtype硕士
Thesis Advisor甘淼 ; 于瑞德
2020-08-30
Degree Grantor中国科学院大学
Place of Conferral北京
Degree Discipline理学硕士
KeywordCMIP5 CRU EOF 分解 气候漂移 多元线性回归 CMIP5 CRU Empirical Orthogonal Function Climate drift Multiple linear regression
Abstract全球气候变化一直以来是人们所关注的热点,又以全球气候变暖对生态环境以及人类社会发展影响深远。全球气候模式是进行气候模拟及预估的重要工具,近二十年来已有大量学者利用全球气候模式对全球气候变化进行模拟及预估研究,并得到许多有意义的结论。而由于气候系统内部变率复杂,以及气候模式的不确定性,在利用气候模式对未来气候变化预估中仍存在许多的不确定性。 如何减少模式预估的不确定性,提高模式模拟能力成为气候变化研究中亟待研究解决的核心问题。本文主要分析全球气候模式对中亚降水及气温模拟的不确定性,同时通过气候漂移以及多元线性回归订正法尝试对其模拟误差进行订正。最终得出以下结论:CMIP5 多模式集合平均对中亚降水模拟具有较大不确定性, 对中亚年均以及各季节模拟呈现大范围正偏差。 参考时段的降水模拟误差经验正交函数(EOF)分解显示两个主要空间模态,第一模态为误差气候漂移部分,第二模态属于误差的非定常部分。 气候漂移和多元线性回归订正对降水误差有明显的改善, 同时降水年际变率 CV 值显示,夏、 秋季降水订正效果较冬、 春季更为显著, 对中亚南部订正效果好于北部。各模式对气温的模拟能力均较好,多模式集合对中亚大部分地区年均气温模拟偏高,季节误差空间分布中夏、 冬季也呈大范围正偏差。 参考时段平均气温误差时空分解显示, 误差主要模态为地形引起的气候漂移部分。订正对平均气温误差有很好的改善, 很好的去除了地形等因素引起的误差。多模式集合对中亚最高气温年均以及各季节模拟也呈大范围正偏差。 参考时段模拟误差时空分解显示第一模态既有模式集合的气候漂移部分,也有误差的非定常部分。订正后最高气温误差有明显的改善,各季节最高气温模拟与观测数据的空间相关系数均达到 0.99,尤其冬季最高气温订正效果最为显著。模式对中亚最低气温的模拟能力,较平均气温与最高气温弱, 多模式集合对中亚最低气温年均以及春、秋季模拟呈现大范围负偏差,夏、冬季模拟呈大范围正偏差。 参考时段最低气温误差时空分解显示, 第一模态主要属于模式集合的气候漂移部分。通过气候漂移和多元线性回归订正,最低气温与观测数据的空间相关系数有了明显提高。模式对中亚降水以及气温模拟误差特征分析显示模拟预估误差可能多来自模式自身的问题,如模式分辨率过低,对降水物理过程描述不足等。气候漂移订正以及多元线性回归订正,对中亚降水以及气温有较好的订正效果,但对局部区域订正并不明显,因此为更加准确地预估中亚未来气候变化需要更加细致的分析影响中亚气温、降水模拟的不确定性因素,实现更为有效的误差订正。
Other AbstractGlobal climate change has always been a hotspot of attention, and globalwarming has a profound impact on the ecological environment and the developmentof human society. The global climate models are the important tools for climatesimulation and prediction. In the past two decades, a large number of scholars haveused the global climate models to simulate and predict global climate change andhave drawn many meaningful conclusions. However, due to the complexity of theinternal variability of the climate system and the uncertainty of climate models, thereare many uncertainties if directly using the climate models to predict future climatechange. How to reduce the uncertainty and improve model simulation ability hasbecome an urgent problem in research. This paper mainly analyzes the uncertainty ofthe global climate models for the simulation of precipitation and temperature inCentral Asia. At the same time, it attempts to correct the uncertainty through climatedrift and multiple linear regression correction methods. The main conclusions are asfollows:The multi-mode ensemble of CMIP5 has greater uncertainty for Central Asianprecipitation simulation, and presents a wide range of positive deviations for theannual average of Central Asia and the seasonal simulations. The EmpiricalOrthogonal Function (EOF) analysis of the reference period error shows two mainspatial modes. The first mode is the part of climate drift. The second mode is theunsteady part of the error. The correction of climate drift and multiple linearregression has significantly reduced the precipitation error. At the same time, thecoefficient of variation (CV) value shows that the effect of summer and autumnprecipitation correction is more obvious than that of winter and spring, and the overallcorrection effect in southern Central Asia is better than North.Model has a good ability to simulate air temperature. The multi-mode ensembleset simulates a high average annual temperature in most parts of Central Asia, and thespatial distribution of seasonal errors also shows a large positive deviation in summerand winter. The main modes of errors are mainly caused by climate drift. The correction can well reduce the error of the average temperature, and it removes theerror caused by terrain and other factors.The multi-mode ensemble also shows a large range of positive deviations of themaximum temperature in Central Asia. The reference period error decompositionshows that the first mode has both the climate drift and the unsteady part of the error.After correction, the maximum temperature error has been significantly reduced. Thespatial correlation coefficient of the maximum temperature in all seasons of CentralAsia has reached 0.99, especially in winter.The ability of the model to simulate the minimum temperature is weaker than theaverage temperature and the maximum temperature. The model collection shows alarge range of negative deviations in the annual average of Central Asia's minimumtemperature and spring and autumn simulations, and a large range of positivedeviations in summer and winter simulations. The first mode of the error of theminimum temperature mainly belongs to the climate drift. Through the correction ofclimate drift and multiple linear regression, the spatial correlation coefficient of theminimum temperature has been significantly improved.The error analysis of precipitation and temperature in Central Asia by the modelshows that the estimated error may mostly come from the problems of the model itself,such as the low resolution of the model and the insufficient description of the physicalprocess of precipitation. The correction of climate drift and multiple linear regressionhas a good effect on precipitation and temperature in Central Asia, but the correctionis not obvious for some regions. Therefore, in order to more accurately predict thefuture climate change of Central Asia, need more detailed analysis. In order to find themost appropriate correction method, Influencing factors of temperature andprecipitation is required.
Subject Area自然地理学
Language中文
Document Type学位论文
Identifierhttp://ir.xjlas.org/handle/365004/15424
Collection中国科学院新疆生态与地理研究所
研究系统
Affiliation中国科学院新疆生态与地理研究所
First Author Affilication中国科学院新疆生态与地理研究所
Recommended Citation
GB/T 7714
黄芳. 基于 CMIP5 模式中亚地区降水与气温模拟不确定性分析与订正[D]. 北京. 中国科学院大学,2020.
Files in This Item:
There are no files associated with this item.
Related Services
Recommend this item
Bookmark
Usage statistics
Export to Endnote
Google Scholar
Similar articles in Google Scholar
[黄芳]'s Articles
Baidu academic
Similar articles in Baidu academic
[黄芳]'s Articles
Bing Scholar
Similar articles in Bing Scholar
[黄芳]'s Articles
Terms of Use
No data!
Social Bookmark/Share
All comments (0)
No comment.
 

Items in the repository are protected by copyright, with all rights reserved, unless otherwise indicated.