Xiaoduo Pan , Deliang Chen , Baoxiang Pan , Xiaozhong Huang , Kun Yang , Shilong Piao , Tianjun Zhou , Yongjiu Dai , Fahu Chen , Xin Li
{"title":"Evolution and prospects of Earth system models: Challenges and opportunities","authors":"Xiaoduo Pan , Deliang Chen , Baoxiang Pan , Xiaozhong Huang , Kun Yang , Shilong Piao , Tianjun Zhou , Yongjiu Dai , Fahu Chen , Xin Li","doi":"10.1016/j.earscirev.2024.104986","DOIUrl":null,"url":null,"abstract":"<div><div>Earth system models (ESMs) serve as vital tools for comprehensively simulating the intricate interplay of physical, chemical, and biological processes across the Earth system's diverse components. Here, we provide a brief overview of the historical development of ESMs and highlight key challenges posed by the intricate feedback mechanisms in the cryosphere, the nonlinear and long-term effects of the lithosphere, and the growing impacts of human activities for modeling Earth system. We then focus on the current opportunities in Earth system modeling, driven by the growing capacity for data-driven approaches such as machine learning (ML) and Artificial Intelligence (AI).</div><div>The next generation of ESMs should embrace dynamic frameworks that enable more precise representations of physical processes across a range of spatiotemporal scales. Multi-resolution models are pivotal in bridging the gap between global and regional scales, fostering a deeper understanding of local and remote influences. Data-driven methodologies including ML/AI offer promising avenues for advancing ESMs by harnessing a wide array of data sources and surmounting limitations inherent in traditional parameterization techniques. However, the integration of ML/AI into ESMs presents its own set of challenges, including the identification of suitable data sources, the seamless incorporation of ML/AI algorithms into existing modeling infrastructures, and the resolution of issues related to model interpretability and robustness. A harmonious amalgamation of physics-based and data-driven methodologies have the potential to produce ESMs that achieve greater precision and computational efficiency, better capturing the intricate dynamics of Earth system processes.</div><div>Although ESMs have made substantial progress in simulating the complex dynamics of Earth system's subsystems, there is still considerable work to be done. Prospects in the development of ESMs entail a deepened comprehension of pivotal subsystems, including the anthroposphere, lithosphere, and cryosphere. Adopting innovative technologies and methodologies, such as ML/AI and multi-resolution modeling, holds immense potential to substantially enhance our capability to anticipate and mitigate the consequences of human activities on the Earth system.</div></div>","PeriodicalId":11483,"journal":{"name":"Earth-Science Reviews","volume":"260 ","pages":"Article 104986"},"PeriodicalIF":10.8000,"publicationDate":"2024-11-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Earth-Science Reviews","FirstCategoryId":"89","ListUrlMain":"https://www.sciencedirect.com/science/article/pii/S0012825224003143","RegionNum":1,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q1","JCRName":"GEOSCIENCES, MULTIDISCIPLINARY","Score":null,"Total":0}
引用次数: 0
Abstract
Earth system models (ESMs) serve as vital tools for comprehensively simulating the intricate interplay of physical, chemical, and biological processes across the Earth system's diverse components. Here, we provide a brief overview of the historical development of ESMs and highlight key challenges posed by the intricate feedback mechanisms in the cryosphere, the nonlinear and long-term effects of the lithosphere, and the growing impacts of human activities for modeling Earth system. We then focus on the current opportunities in Earth system modeling, driven by the growing capacity for data-driven approaches such as machine learning (ML) and Artificial Intelligence (AI).
The next generation of ESMs should embrace dynamic frameworks that enable more precise representations of physical processes across a range of spatiotemporal scales. Multi-resolution models are pivotal in bridging the gap between global and regional scales, fostering a deeper understanding of local and remote influences. Data-driven methodologies including ML/AI offer promising avenues for advancing ESMs by harnessing a wide array of data sources and surmounting limitations inherent in traditional parameterization techniques. However, the integration of ML/AI into ESMs presents its own set of challenges, including the identification of suitable data sources, the seamless incorporation of ML/AI algorithms into existing modeling infrastructures, and the resolution of issues related to model interpretability and robustness. A harmonious amalgamation of physics-based and data-driven methodologies have the potential to produce ESMs that achieve greater precision and computational efficiency, better capturing the intricate dynamics of Earth system processes.
Although ESMs have made substantial progress in simulating the complex dynamics of Earth system's subsystems, there is still considerable work to be done. Prospects in the development of ESMs entail a deepened comprehension of pivotal subsystems, including the anthroposphere, lithosphere, and cryosphere. Adopting innovative technologies and methodologies, such as ML/AI and multi-resolution modeling, holds immense potential to substantially enhance our capability to anticipate and mitigate the consequences of human activities on the Earth system.
期刊介绍:
Covering a much wider field than the usual specialist journals, Earth Science Reviews publishes review articles dealing with all aspects of Earth Sciences, and is an important vehicle for allowing readers to see their particular interest related to the Earth Sciences as a whole.