[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-117343-en":3,"doc-seo-117343-105":30,"detail-sidebar-cat-0-en-105":91},{"code":4,"msg":5,"data":6},0,"success",{"doc_id":7,"user_id":8,"nickname":9,"user_avatar":10,"doc_module":4,"category_id":11,"category_name":12,"doc_title":13,"doc_description":14,"doc_content":15,"file_id":16,"file_url":17,"file_type":18,"file_size":19,"view_count":4,"is_deleted":4,"is_public":20,"is_downloadable":20,"audit_status":20,"page_count":21,"language":22,"language_code":23,"site_id":24,"html_lang":23,"table_of_contents":25,"faqs":26,"seo_title":27,"seo_description":14,"update_tm":28,"read_time":29},117343,1099514068035,"Ezra","https://ap-avatar.wpscdn.com/davatar_276721f389ce27ea32af1340a28f341c",8,"Research & Report","Machine learning for stochastic parametrization - Position paper","Atmospheric models for weather and climate prediction are commonly built as deterministic systems: a most-likely subgrid forcing is inferred from resolved variables to advance the large-scale flow. Limited scale separation makes this assumption a major forecast error source. Stochastic techniques have therefore become widely used to represent uncertainty in small-scale processes. At the same time, machine learning is increasingly used to replace parametrization schemes, though mostly in deterministic form. This position paper connects both directions by discussing data-driven probabilistic approaches, early studies, and remaining challenges.","Environmental Data Science (2024), 3: e38, 1–12  \ndoi:10.1017/eds.2024.45  \nPOSITION PAPER  \nMachine learning for stochastic parametrization  \nHannah M. Christensen 1 , Salah Kouhen 1 , Greta Miller1  and Raghul Parthipan2,3   \n1Department of Physics, University of Oxford, Oxford, UK  \n2Department of Computer Science, University of Cambridge, Cambridge, UK 3British Antarctic Survey, Cambridge, UK  \nCorresponding author: Hannah M. Christensen; [Email: hannah.christensen@physics.ox.ac.uk](Email: hannah.christensen@physics.ox.ac.uk)  \nReceived: 24 June 2024; Accepted: 28 September 2024  \nKeywords: machine learning; stochastic parametrization; uncertainty quantification; model error  \nAbstract  \nAtmospheric models used for weather and climate prediction are traditionally formulated ina deterministic manner. In other words, given a particular state of the resolved scale variables, the most likely forcing from the subgrid scale processes is estimated and used to predict the evolution of the large-scale flow. However, the lack of scale separation in the atmosphere means that this approach is a large source of error in forecasts. Over recent years, an alternative paradigm has developed: the use of stochastic techniques to characterize uncertainty in small-scale processes. These techniques are now widely used across weather, subseasonal, seasonal, and climate timescales. In parallel, recent years have also seen significant progress in replacing parametrization schemes using machine learning(ML) . This has the potential to both speed up and improve our numerical models. However, the focus to date has largely been on deterministic approaches. In this position paper, we bring together these two key developments and discuss the potential for data-driven approaches for stochastic parametrization. We highlight early studies in this area and draw attention to the novel challenges that remain.  \nImpact Statement  \nWeather and climate predictions are relied on by users from industry, charities, governments, and the general public. The largest source of uncertainty in these predictions arises from approximations made when building the computer model used to make them. In particular, the representation of small-scale processes such as clouds and thunderstorms is a large source of uncertainty because of their complexity and their unpredictability. Machine learning (ML) approaches, trained to mimic high-quality datasets, present an unparalleled opportunity to improve the representation of these small-scale processes in models. However, it is important to account for the unpredictability of these processes while doing so. In this article, we demonstrate the untapped potential of such probabilistic ML approaches for improving weather and climate prediction.  \n1. Introduction  \nWeather and climate models exhibit long-standing biases in mean state, modes of variability, and the representation of extremes. These biases hinder progress across the World Climate Research Programme grand challenges. Understanding and reducing these biases is a key focus for the research community.  \nAt the heart of weather and climate models are the physical equations of motion, which describe the atmosphere and ocean systems. To predict the evolution of the climate system, these equations are discretized in space and time. The resolution ranges from order 100 km and 30–60 minutes in a typical  \n©The Author(s), 2024. Published by Cambridge University Press. This is an Open Access article, distributed under the terms ofthe Creative Commons Attribution licence ([http://creativecommons.org/licenses/by/4.0](http://creativecommons.org/licenses/by/4.0)), which permits unrestricted re-use, distribution and reproduction, provided the original article is properly cited.  \ne38-2 Hannah M. Christensen et al.  \nclimate model, through 10 km and 5–10 minutes in global numerical weather prediction (NWP) models, to one km and a few tens of seconds for state-of-the-art convection-permitting","cbCaii2XLgt6rBpi","https://ap.wps.com/l/cbCaii2XLgt6rBpi","pdf",2133407,1,12,"English","en",105,"# Introduction\n## Deterministic biases in weather and climate models\n## Parametrization schemes and unresolved scales\n## Machine learning emulators and opportunities\n## Key challenges for stochastic ML parametrization","[{\"question\":\"Why are deterministic parametrization approaches a significant error source?\",\"answer\":\"Atmospheric models lack strong scale separation, so the inferred subgrid forcing is sensitive to unresolved processes. This leads to substantial biases and errors in forecasts.\"},{\"question\":\"What is the goal of stochastic parametrization in these models?\",\"answer\":\"Stochastic techniques aim to characterize uncertainty in small-scale processes rather than selecting a single most-likely forcing, improving representation across multiple time scales.\"},{\"question\":\"How can machine learning improve weather and climate models?\",\"answer\":\"Machine learning can emulate or replace existing parametrization schemes, potentially speeding up computation and improving accuracy. The paper emphasizes that probabilistic ML must account for the inherent unpredictability of these processes.\"}]","Machine learning for stochastic parametrization - Position paper | PDF",1785675285,30,{"code":4,"msg":31,"data":32},"ok",{"site_id":24,"language":23,"slug":33,"title":13,"keywords":34,"description":14,"schema_data":35,"social_meta":86,"head_meta":88,"extra_data":90,"updated_unix":28},"machine-learning-for-stochastic-parametrization-position-paper","",{"@graph":36,"@context":85},[37,54,68],{"@type":38,"itemListElement":39},"BreadcrumbList",[40,44,48,51],{"item":41,"name":42,"@type":43,"position":20},"https://docshare.wps.com","Home","ListItem",{"item":45,"name":46,"@type":43,"position":47},"https://docshare.wps.com/document/","Document",2,{"item":49,"name":12,"@type":43,"position":50},"https://docshare.wps.com/document/research-report/",3,{"item":52,"name":13,"@type":43,"position":53},"https://docshare.wps.com/document/machine-learning-for-stochastic-parametrization-position-paper/117343/",4,{"url":52,"name":13,"@type":55,"author":56,"headline":13,"publisher":58,"fileFormat":61,"inLanguage":23,"description":14,"dateModified":62,"datePublished":62,"encodingFormat":61,"isAccessibleForFree":63,"interactionStatistic":64},"DigitalDocument",{"name":9,"@type":57},"Person",{"url":41,"name":59,"@type":60},"DocShare","Organization","application/pdf","2026-08-02",true,{"@type":65,"interactionType":66,"userInteractionCount":4},"InteractionCounter",{"@type":67},"ViewAction",{"@type":69,"mainEntity":70},"FAQPage",[71,77,81],{"name":72,"@type":73,"acceptedAnswer":74},"Why are deterministic parametrization approaches a significant error source?","Question",{"text":75,"@type":76},"Atmospheric models lack strong scale separation, so the inferred subgrid forcing is sensitive to unresolved processes. This leads to substantial biases and errors in forecasts.","Answer",{"name":78,"@type":73,"acceptedAnswer":79},"What is the goal of stochastic parametrization in these models?",{"text":80,"@type":76},"Stochastic techniques aim to characterize uncertainty in small-scale processes rather than selecting a single most-likely forcing, improving representation across multiple time scales.",{"name":82,"@type":73,"acceptedAnswer":83},"How can machine learning improve weather and climate models?",{"text":84,"@type":76},"Machine learning can emulate or replace existing parametrization schemes, potentially speeding up computation and improving accuracy. The paper emphasizes that probabilistic ML must account for the inherent unpredictability of these processes.","https://schema.org",{"og:url":52,"og:type":87,"og:title":13,"og:site_name":59,"og:description":14},"article",{"robots":89,"canonical":52},"index,follow",{"doc_id":7,"site_id":24},{"code":4,"msg":5,"data":92},[93,97,101,105,110,115,120,122,127,130,134],{"id":20,"doc_module":4,"doc_module_name":46,"category_name":94,"show_sort_weight":95,"slug":96},"Story & Novel",90,"story-novel",{"id":47,"doc_module":4,"doc_module_name":46,"category_name":98,"show_sort_weight":99,"slug":100},"Literature",80,"literature",{"id":53,"doc_module":4,"doc_module_name":46,"category_name":102,"show_sort_weight":103,"slug":104},"Exam",70,"exam",{"id":106,"doc_module":4,"doc_module_name":46,"category_name":107,"show_sort_weight":108,"slug":109},5,"Comic",60,"comic",{"id":111,"doc_module":4,"doc_module_name":46,"category_name":112,"show_sort_weight":113,"slug":114},6,"Technology",50,"technology",{"id":116,"doc_module":4,"doc_module_name":46,"category_name":117,"show_sort_weight":118,"slug":119},7,"Healthcare",40,"healthcare",{"id":11,"doc_module":4,"doc_module_name":46,"category_name":12,"show_sort_weight":29,"slug":121},"research-report",{"id":123,"doc_module":4,"doc_module_name":46,"category_name":124,"show_sort_weight":125,"slug":126},9,"Religion & Spirituality",20,"religion-spirituality",{"id":125,"doc_module":4,"doc_module_name":46,"category_name":128,"show_sort_weight":125,"slug":129},"World Cup","world-cup",{"id":131,"doc_module":4,"doc_module_name":46,"category_name":132,"show_sort_weight":131,"slug":133},10,"Lifestyle","lifestyle",{"id":135,"doc_module":4,"doc_module_name":46,"category_name":136,"show_sort_weight":106,"slug":137},19,"General","general"]