We present a study using a class of post-hoc local explanation methods i.e., feature importance methods for "understanding" a deep learning (DL) emulator of climate. Specifically, we consider a multiple-input-single-output emulator that uses a DenseNet encoder-decoder architecture and is trained to predict interannual variations of sea surface temperature (SST) at 1, 6, and 9 month lead times using the preceding 36 months of (appropriately filtered) SST data. First, feature importance methods are employed for individual predictions to spatio-temporally identify input features that are important for model prediction at chosen geographical regions and chosen prediction lead times. In a second step, we also examine the behavior of feature importance in a generalized sense by considering an aggregation of the importance heatmaps over training samples. We find that: 1) the climate emulator's prediction at any geographical location depends dominantly on a small neighborhood around it; 2) the longer the prediction lead time, the further back the "importance" extends; and 3) to leading order, the temporal decay of "importance" is independent of geographical location. An ablation experiment is adopted to verify the findings. From the perspective of climate dynamics, these findings suggest a dominant role for local processes and a negligible role for remote teleconnections at the spatial and temporal scales we consider. From the perspective of network architecture, the spatio-temporal relations between the inputs and outputs we find suggest potential model refinements. We discuss further extensions of our methods, some of which we are considering in ongoing work.
High-Resolution Climate Projections Using Diffusion-Based Downscaling of a Lightweight Climate Emulator
Information Sciences and Technology, The Pennsylvania State University(宾夕法尼亚州立大学信息科学与技术系);Department of Applied Mathematics, University of California, Santa Cruz(加州大学圣克鲁兹分校应用数学系);Allen Institute for Artificial Intelligence (AI2)(人工智能研究所);School of Mechanical Engineering, Purdue University(普渡大学机械工程学院)
How Can Machine Learning Emulators Best Support Climate Science?
Cluster of Excellence Machine Learning University of Tübingen(图宾根大学机器学习卓越中心);Institute for Atmospheric and Climate Science ETH Zurich(大气与气候科学研究所,苏黎世联邦理工学院)
Department of Aeronautics and Astronautics, Massachusetts Institute of Technology(航空与航天系,麻省理工学院);Center for Sustainability Science and Strategy, Massachusetts Institute of Technology(可持续科学与战略中心,麻省理工学院);Department of Earth, Atmospheric, and Planetary Sciences, Massachusetts Institute of Technology(地球、大气与行星科学系,麻省理工学院);Brahmal Vasudevan Institute for Sustainable Aviation, Department of Aeronautics, Imperial College London(可持续航空研究所,帝国理工学院伦敦校区);Institute for Data, Systems, and Society, Massachusetts Institute of Technology(数据、系统与社会研究所,麻省理工学院)
MERCURY: A fast and versatile multi-resolution based global emulator of compound climate hazards
University of Oxford(牛津大学);Montreal Polytechnique(蒙特利尔理工学院);Columbia Climate School(哥伦比亚气候学院);International Institute for Applied Systems Analysis(国际应用系统分析研究所);Universität Leipzig(莱比锡大学);IRI THESys;Humboldt-Universität zu Berlin(柏林洪堡大学);Institute Pierre-Simone-Laplace(皮埃尔-西蒙·拉普拉斯研究所)
Building Interpretable Climate Emulators for Economics
Institute of Computing, Università della Svizzera italiana(瑞士大学计算研究所);Institute for Atmospheric and Climate Science, ETHZ(大气与气候科学研究所);ifo Institute, Ludwig-Maximilian University of Munich(ifo研究所,慕尼黑莱布尼茨大学);Department for Banking and Finance, University of Zürich(金融学院,苏黎世大学);Department of Economics, University of Lausanne(经济学系,洛桑大学);Grantham Research Institute, London School of Economics(格兰塔姆研究所,伦敦政治经济学院)
LUCIE-3D: A three-dimensional climate emulator for forced responses
Pennsylvania State University(宾夕法尼亚州立大学);Allen Insitute of Artificial Intelligence(人工智能 Allen 机构);Argonne National Laboratory(阿贡国家实验室);University of California, Santa Cruz(加州大学圣克鲁兹分校)
Assessing the Robustness of Climate Foundation Models under No-Analog Distribution Shifts
Centre for Artificial Intelligence(人工智能中心);Department of Computer Science(计算机科学系);University College London(伦敦大学学院);Division of Emerging Interdisciplinary Areas(新兴交叉学科领域);Hong Kong University of Science and Technology(香港科技大学);University of Oxford(牛津大学)
Samudra 2: Scaling Ocean Emulators across Resolutions
Courant Institute School of Mathematics, Computing, and Data Science, New York University(Courant学院数学、计算与数据科学系,纽约大学);Open Athena AI Foundation, Inc.(开放Athena人工智能基金会);Program in Atmospheric and Oceanic Sciences, Princeton University(大气与海洋科学项目,普林斯顿大学)
Building high accuracy emulators for scientific simulations with deep neural architecture search
University of Oxford(牛津大学);University of Rochester(罗切斯特大学);Yale University(耶鲁大学);University of Seville(塞维利亚大学);Max-Planck-Institut für Plasmaphysik(马克斯·普朗克等离子体物理研究所)
Extreme heatwave sampling and prediction with analog Markov chain and comparisons with deep learning
Laboratoire des Sciences du Climat et de l’Environnement, UMR 8212 CEA-CNRS-UVSQ, Université Paris-Saclay & IPSL(气候与环境科学实验室,UMR 8212 CEA-CNRS-UVSQ,巴黎-萨克雷大学与皮埃尔·西蒙·拉普拉斯研究所)
DiffESM: Conditional Emulation of Temperature and Precipitation in Earth System Models with 3D Diffusion Models
Utah State University(犹他州立大学);Joint Global Change Research Institute(全球变化联合研究所);Pacific Northwest National Laboratory(西北太平洋国家实验室);Western Washington University(西华盛顿大学);Indiana University(印第安纳大学)
Artificial intelligence pathways from weather to climate
University of California, Los Angeles(加州大学洛杉矶分校);Hong Kong University of Science and Technology(香港科技大学);University of Lausanne(洛桑大学);Allen Institute for AI(艾伦人工智能研究所);University of Colorado Boulder(科罗拉多大学博尔德分校);California Institute of Technology(加州理工学院);NOAA/Geophysical Fluid Dynamics Laboratory(NOAA地球物理流体动力学实验室);Google Research(谷歌研究院);New York University(纽约大学)