人工智能在气候研究和服务中的机遇与挑战  

Opportunities and challenges of artificial intelligence in climate research and services

作　　者：陈德亮 谭显春[3,4,5,6] 彭喆 闫洪硕[3,4,5] 程永龙 CHEN Deliang;TAN Xian-Chun;PENG Zhe;YAN Hong-Shuo;CHENG Yong-Long(Department of Earth System Science,Tsinghua University,Beijing 100084,China;Department of Earth Sciences,University of Gothenburg,40530,Sweden;Institutes of Science and Development,Chinese Academy of Sciences,Beijing 100190,China;School of Public Policy and Management,University of Chinese Academy of Sciences,Beijing 100049,China;Center for Carbon Neutrality Strategy,Institutes of Science and Development,Chinese Academy of Sciences,Beijing 100190,China;Institute of Intelligent Social Governance,Chongqing University,Chongqing 400044,China)

机构地区：[1]清华大学地球系统科学系,北京100084 [2]瑞典哥德堡大学地球科学系,瑞典哥德堡40530 [3]中国科学院科技战略咨询研究院,北京100190 [4]中国科学院大学公共政策与管理学院,北京100049 [5]中国科学院科技战略咨询研究院碳中和战略研究中心,北京100190 [6]重庆大学智能社会治理研究院,重庆400044

出　　处：《气候变化研究进展》2024年第6期669-681,共13页Climate Change Research

基　　金：国家自然科学基金专项项目“面向碳中和的国家气候治理体系变革与创新研究”(72140007)。

摘　　要：人类活动引起的气候变化是自然生态和人类社会面临的严峻挑战。作为应对气候变化的重要工具,气候模式为理解过去气候系统演变机理、气候服务及预估未来全球变化情景提供重要支撑。然而,气候模式在过程表达、次网格参数化及空间分辨率等方面的问题严重制约了气候模式的模拟与预测能力。随着经济社会的发展,人工智能技术得以广泛应用,其具有的整合多源数据、识别潜在信息、学习已有“经验”等优势,有望为气候研究和应用提供新的助力。在此背景下,文中首先对气候模式的发展进行简要回顾,接着在此基础上结合人工智能技术特征,阐述人工智能在气候研究和服务中的应用并分析现阶段面临的挑战。结果表明,已有观测数据与模式输出结果能够为人工智能应用提供必要数据基础,人工智能则据此对气候模式进行优化和集成,进而提升了气候模拟与预测结果的准确性。展望未来,应当聚焦人工智能与气候模式的耦合问题并拓展人工智能在应对气候变化各领域的应用,从而更有效地应对全球气候变化带来的挑战。Climate change is a serious challenge to natural ecology and human society.As an important tool for addressing climate change,climate models provide important support for understanding the mechanism of the past climate system,climate services and projections of future global change scenarios.However,problems in process representation,subgrid parameterization and spatial resolution of climate models have severely limited their simulation and prediction capabilities.With the development of economy and society,artificial intelligence(AI)technology has been widely used,and its advantages of integrating multi-source data,identifying potential information,and learning from existing“experience”are expected to provide new assistance for climate research and application.Against this background,this paper firstly provides a concise review of the development of climate models,then combines the characteristics of AI on this basis to examine the application of AI in climate research and services,and analyzes the challenges faced at this stage.The results show that existing observational data and model outputs can provide a sufficient data base for AI applications,and AI can optimize and integrate climate models accordingly,thus improving the accuracy of climate simulation and prediction results.In the future,it is important to focus on the coupling of AI and climate models and to expand the application of AI in various areas of climate change,so as to more effectively address the challenges posed by global climate change.

关 键 词：气候模式 人工智能 气候变化 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程] P467[自动化与计算机技术—控制科学与工程]