Hard-Constrained Deep Learning for Climate Downscaling
Fraunhofer ITWM（弗劳恩霍夫信息技术与数学研究所）；Mila Quebec AI Institute（米拉魁北克人工智能研究所）；University of Montreal（蒙特利尔大学）；New York University（纽约大学）；IBM Research（IBM研究院）；McGill University（麦吉尔大学）
Paula Harder, Alex Hernandez-Garcia, Venkatesh Ramesh, Qidong Yang, Prasanna Sattigeri, Daniela Szwarcman, Campbell Watson, David Rolnick
The availability of reliable, high-resolution climate and weather data is important to inform long-term decisions on climate adaptation and mitigation and to guide rapid responses to extreme events. Forecasting models are limited by computational costs and, therefore, often generate coarse-resolution predictions. Statistical downscaling, including super-resolution methods from deep learning, can provide an efficient method of upsampling low-resolution data. However, despite achieving visually compelling results in some cases, such models frequently violate conservation laws when predicting physical variables. In order to conserve physical quantities, here we introduce methods that guarantee statistical constraints are satisfied by a deep learning downscaling model, while also improving their performance according to traditional metrics. We compare different constraining approaches and demonstrate their applicability across different neural architectures as well as a variety of climate and weather data sets. Besides enabling faster and more accurate climate predictions through downscaling, we also show that our novel methodologies can improve super-resolution for satellite data and natural images data sets.
Projection of changes in extreme indices of climate variables such as temperature and precipitation are critical to assess the potential impacts of climate change on human-made and natural systems, including critical infrastructures and ecosystems. While impact assessment and adaptation planning rely on high-resolution projections (typically in the order of a few kilometers), state-of-the-art Earth System Models (ESMs) are available at spatial resolutions of few hundreds of kilometers. Current solutions to obtain high-resolution projections of ESMs include downscaling approaches that consider …
University of Cambridge（剑桥大学）；British Antarctic Survey（英国南极调查局）；The Alan Turing Institute（艾伦·图灵研究所）
A new model is presented for multisite statistical downscaling of temperature and precipitation using convolutional conditional neural processes (convCNPs). ConvCNPs are a recently developed class of models that allow deep learning techniques to be applied to off-the-grid spatio-temporal data. This model has a substantial advantage over existing downscaling methods in that the trained model can be used to generate multisite predictions at an arbitrary set of locations, regardless of the availability of training data. The convCNP model is shown to outperform an ensemble of existing downscaling …
Multi-variable Hard Physical Constraints for Climate Model Downscaling
Instituto de Física de Cantabria (IFCA), CSIC-UC（坎塔布里亚物理研究所（IFCA），西班牙国家研究委员会-坎塔布里亚大学）；Mila Quebec AI Institute（米拉魁北克人工智能研究所）；University of Montreal（蒙特利尔大学）；Fraunhofer ITWM（弗劳恩霍夫工业数学研究所）；University of Kaiserslautern（凯泽斯劳滕大学）；McGill University（麦吉尔大学）
Fourier Neural Operators for Arbitrary Resolution Climate Data Downscaling
Mila Quebec AI Institute（米拉魁北克人工智能研究所）；New York University（纽约大学）；University of Montreal（蒙特利尔大学）；Fraunhofer ITWM（弗劳恩霍夫工业数学研究所）；IBM Research（IBM研究院）；McGill University（麦吉尔大学）
Diffusion Model-based Probabilistic Downscaling for 180-year East Asian Climate Reconstruction
Nanjing University of Information Science and Technology（南京信息工程大学）；Japan Agency for Marine-Earth Science and Technology（日本海洋研究开发机构）；Shanghai AI Laboratory（上海人工智能实验室）；Shanghai Jiao Tong University（上海交通大学）；Institute of Atmospheric Physics, Chinese Academy of Sciences（中国科学院大气物理研究所）；Tokyo Institute of Technology（东京工业大学）；RIKEN Center for Advanced Intelligence Project（理化学研究所革新智能综合研究中心）
Modeling the risk of extreme weather events in a changing climate is essential for developing effective adaptation and mitigation strategies. Although the available low-resolution climate models capture different scenarios, accurate risk assessment for mitigation and adaption often demands detail that they typically cannot resolve. Here, we develop a dynamic data-driven downscaling (super-resolution) method that incorporates physics and statistics in a generative framework to learn the fine-scale spatial details of rainfall. Our method transforms coarse-resolution ($0.25^{\circ} \times 0.25^{\…
Generating High-Resolution Regional Precipitation Using Conditional Diffusion Model
School of Computing, KAIST（韩国科学技术院计算机学院）；Data Science Group, IBS（基础科学研究院数据科学组）；Department of Climate System, PNU（釜山大学气候系统系）；Center for Climate Physics, IBS（基础科学研究院气候物理中心）
A common task in Earth Sciences is to infer climate information at local and regional scales from global climate models. Dynamical downscaling requires running expensive numerical models at high resolution which can be prohibitive due to long model runtimes. On the other hand, statistical downscaling techniques present an alternative approach for learning links between the large- and local-scale climate in a more efficient way. A large number of deep neural network-based approaches for statistical downscaling have been proposed in recent years, mostly based on convolutional architectures devel…
ORBIT-2: Scaling Exascale Vision Foundation Models for Weather and Climate Downscaling
Oak Ridge National Laboratory（奥克岭国家实验室）；AMD Research and Advanced Development（AMD研究与高级开发）；RIKEN Center for Computational Science（理化学研究所计算科学中心）；University of Alabama at Birmingham（阿拉巴马大学伯明翰分校）
Advances in climate science have rendered obsolete gridded observation data sets commonly used in macroecological analyses. Novel climate reanalysis products outperform legacy data products in accuracy, temporal resolution, and provision of uncertainty metrics. Consequently, there is an urgent need to develop a workflow through which to integrate these improved data into analyses. The ERA5 product family are the latest and most advanced global reanalysis products created by the ECMWF. These data products offer up to 83 essential climate variables at hourly intervals for the time-period of 1981…
Fast, Scale-Adaptive, and Uncertainty-Aware Downscaling of Earth System Model Fields with Generative Machine Learning
Technical University Munich（慕尼黑工业大学）；Potsdam Institute for Climate Impact Research（波茨坦气候影响研究所）；Institute of Atmospheric Physics, Chinese Academy of Sciences（中国科学院大气物理研究所）；University of Exeter（埃克塞特大学）
Downscaling with AI reveals the large role of internal variability in fine-scale projections of climate extremes
National Institute of Water and Atmospheric Research (NIWA)（国家水气和大气研究所）；ARC Centre of Excellence for Weather of the 21st Century（21世纪天气卓越研究中心）；University of New South Wales（新南威尔士大学）；Climate Change Research Centre, University of New South Wales（新南威尔士大学气候变化研究中心）；Te Aka Mātuatua School of Science, University of Waikato（怀卡托大学理学院）
CORDEX-ML-Bench: A Benchmark for Data-Driven Regional Climate Downscaling -Experiment Design and Overview
Earth Sciences New Zealand（新西兰地球科学）；Instituto de Física de Cantabria (IFCA), CSIC-Universidad de Cantabria（坎塔布里亚物理研究所（IFCA），西班牙高等科学研究委员会-坎塔布里亚大学）；School of Geographical Sciences, University of Bristol (UoB)（布里斯托尔大学地理科学学院）；Departamento de Ciencias de la Atmósfera y los Océanos, Universidad de Buenos Aires, CONICET, IFAECI/CNRS-IRD-UBA（布宜诺斯艾利斯大学大气与海洋科学系，阿根廷国家科学技术研究委员会，IFAECI/CNRS-IRD-UBA）；Abdus Salam International Centre for Theoretical Physics (ICTP)（阿卜杜斯·萨拉姆理论物理国际中心（ICTP））；Met Office（英国气象局）；Geophysical Institute, University of Bergen and Bjerknes Centre for Climate Research (UiB)（卑尔根大学地球物理研究所和比约恩斯气候研究中心（UiB））；Centre National de Recherches Météorologiques (CNRM), Université de Toulouse, Météo-France, CNRS（国家气象研究中心（CNRM），图卢兹大学，法国气象局，法国国家科学研究中心）；Global Change Institute, University of the Witwatersrand（威特沃特斯兰德大学全球变化研究所）；Swedish Meteorological and Hydrological Institute, Rossby Centre (SMHI)（瑞典气象水文研究所，罗斯贝中心（SMHI））
An intercomparison of generative machine learning methods for downscaling precipitation at fine spatial scales
Earth Sciences New Zealand（新西兰地球科学）；School of Computer Science, University of Auckland（奥克兰大学计算机科学学院）；NVIDIA AI Technology Centre, NVIDIA Corporation（英伟达人工智能技术中心，英伟达公司）；ARC Centre of Excellence for Weather of the 21st Century & Climate Change Research Centre, University of New South Wales（新南威尔士大学21世纪天气卓越研究中心及气候变化研究中心）；Commonwealth Scientific and Industrial Research Organisation (CSIRO)（联邦科学与工业研究组织（CSIRO））；Laboratory for Turbulence Research in Aerospace and Combustion, Department of Mechanical and Aerospace Engineering, Monash University（莫纳什大学机械与航空航天工程系航空与燃烧湍流研究实验室）
Kumamoto University（熊本大学）；University of California, Davis（加州大学戴维斯分校）；Sun Yat-Sen University（中山大学）
Deep learning has been utilized for the statistical downscaling of climate data. Specifically, a two-dimensional (2D) convolutional neural network (CNN) has been successfully applied to precipitation estimation. This study implements a three-dimensional (3D) CNN to estimate watershed-scale daily precipitation from 3D atmospheric data and compares the results with those for a 2D CNN. The 2D CNN is extended along the time direction (3D-CNN-Time) and the vertical direction (3D-CNN-Vert). The precipitation estimates of these extended CNNs are compared with those of the 2D CNN in terms of the root-…
Investigating the Robustness of Extreme Precipitation Super-Resolution Across Climates
Expertise Center for Climate Extremes, University of Lausanne（极端气候专家中心，洛桑大学）；Faculty of Geosciences and Environment, University of Lausanne（地球科学与环境学院，洛桑大学）；Faculty of Business and Economics (HEC), University of Lausanne（商业经济学院（HEC），洛桑大学）
There is an increasing need for high spatial and temporal resolution climate data for the wide community of researchers interested in climate change and its consequences. Currently, there is a large mismatch between the spatial resolutions of global climate model and reanalysis datasets (at best around 0.25o and 0.1o respectively) and the resolutions needed by many end-users of these datasets, which are typically on the scale of 30 arcseconds (~900m). This need for improved spatial resolution in climate datasets has motivated several groups to statistically downscale various combinations of ob…
Argonne National Laboratory（阿贡国家实验室）；Virginia Tech（弗吉尼亚理工学院）
Machine learning models have been employed to perform either physics-free data-driven or hybrid dynamical downscaling of climate data. Most of these implementations operate over relatively small downscaling factors because of the challenge of recovering fine-scale information from coarse data. This limits their compatibility with many global climate model outputs, often available between $\sim$50--100 km resolution, to scales of interest such as cloud resolving or urban scales. This study systematically examines the capability of convolutional neural networks (CNNs) to downscale surface wind s…
Generative artificial intelligence improves projections of climate extremes
Artificial Intelligence Innovation and Incubation Institute（人工智能创新与孵化院）；Fudan University（复旦大学）；Shanghai Innovation Institute（上海创新研究院）；Institute for Big Data（大数据研究院）；MOE Laboratory for National Development and Intelligent Governance（教育部国家发展与智能治理实验室）；Shanghai Academy of AI for Science（上海人工智能科学研究院）；FuXi Intelligent Computing Technology Co., Ltd.（复旦智能计算技术有限公司）
Less is more: error-distance scaling relation for data-efficient kilometer-scale downscaling of extreme heat
Concordia University（康考迪亚大学）；National Research Council Canada（加拿大国家研究委员会）；The American University in Cairo（开罗美国大学）；Mila – Quebec Artificial Intelligence Institute（Mila——魁北克人工智能研究所）；McGill University（麦吉尔大学）；Université de Montréal（蒙特利尔大学）
In an effort to provide optimal inputs to downstream modeling systems (e.g., a hydrodynamics model that simulates the water circulation of a lake), we hereby strive to enhance resolution of precipitation fields from a weather model by up to 9x. We test two super-resolution models: the enhanced super-resolution generative adversarial networks (ESRGAN) proposed in 2017, and the content adaptive resampler (CAR) proposed in 2020. Both models outperform simple bicubic interpolation, with the ESRGAN exceeding expectations for accuracy. We make several proposals for extending the work to ensure it ca…
