[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"detail-sidebar-cat-0-en-105":3,"doc-seo-290516-105":59,"doc-detail-290516-en":130},{"code":4,"msg":5,"data":6},0,"success",[7,13,18,23,28,33,38,43,48,51,55],{"id":8,"doc_module":4,"doc_module_name":9,"category_name":10,"show_sort_weight":11,"slug":12},1,"Document","Story & Novel",90,"story-novel",{"id":14,"doc_module":4,"doc_module_name":9,"category_name":15,"show_sort_weight":16,"slug":17},2,"Literature",80,"literature",{"id":19,"doc_module":4,"doc_module_name":9,"category_name":20,"show_sort_weight":21,"slug":22},4,"Exam",70,"exam",{"id":24,"doc_module":4,"doc_module_name":9,"category_name":25,"show_sort_weight":26,"slug":27},5,"Comic",60,"comic",{"id":29,"doc_module":4,"doc_module_name":9,"category_name":30,"show_sort_weight":31,"slug":32},6,"Technology",50,"technology",{"id":34,"doc_module":4,"doc_module_name":9,"category_name":35,"show_sort_weight":36,"slug":37},7,"Healthcare",40,"healthcare",{"id":39,"doc_module":4,"doc_module_name":9,"category_name":40,"show_sort_weight":41,"slug":42},8,"Research & Report",30,"research-report",{"id":44,"doc_module":4,"doc_module_name":9,"category_name":45,"show_sort_weight":46,"slug":47},9,"Religion & Spirituality",20,"religion-spirituality",{"id":46,"doc_module":4,"doc_module_name":9,"category_name":49,"show_sort_weight":46,"slug":50},"World Cup","world-cup",{"id":52,"doc_module":4,"doc_module_name":9,"category_name":53,"show_sort_weight":52,"slug":54},10,"Lifestyle","lifestyle",{"id":56,"doc_module":4,"doc_module_name":9,"category_name":57,"show_sort_weight":24,"slug":58},19,"General","general",{"code":4,"msg":60,"data":61},"ok",{"site_id":62,"language":63,"slug":64,"title":65,"keywords":66,"description":67,"schema_data":68,"social_meta":123,"head_meta":125,"extra_data":127,"updated_unix":129},105,"en","a-metamodel-approach-integrating-machine-learning-based-weather-prediction-frameworks","A Metamodel approach integrating Machine Learning-based Weather Prediction frameworks","","This master’s thesis investigates how machine learning models can generate realistic short-term global weather forecasts that can rival numerical, physics-based approaches. While numerical methods achieve accuracy through partial differential equation approximations, they require extensive computations across many CPU nodes. The study contrasts this with GPU-friendly ML inference, which reduces computational complexity and energy use. It evaluates PanguWeather, SFNO, FourCastNet, and DLWP and proposes a combined meta-model producing hybrid predictions focused on Italian weather pattern analysis.",{"@graph":69,"@context":122},[70,84,105],{"@type":71,"itemListElement":72},"BreadcrumbList",[73,77,79,82],{"item":74,"name":75,"@type":76,"position":8},"https://docshare.wps.com","Home","ListItem",{"item":78,"name":9,"@type":76,"position":14},"https://docshare.wps.com/document/",{"item":80,"name":40,"@type":76,"position":81},"https://docshare.wps.com/document/research-report/",3,{"item":83,"name":65,"@type":76,"position":19},"https://docshare.wps.com/document/a-metamodel-approach-integrating-machine-learning-based-weather-prediction-frameworks/290516/",{"url":83,"name":65,"@type":85,"image":86,"author":91,"headline":65,"publisher":94,"fileFormat":97,"inLanguage":63,"description":67,"dateModified":98,"datePublished":99,"encodingFormat":97,"isAccessibleForFree":100,"interactionStatistic":101},"DigitalDocument",{"url":87,"@type":88,"width":89,"height":90},"https://docshare.wps.com/thumbnails/a-metamodel-approach-integrating-machine-learning-based-weather-prediction-frameworks/290516.png","ImageObject",300,407,{"name":92,"@type":93},"Noah","Person",{"url":74,"name":95,"@type":96},"DocShare","Organization","application/pdf","2026-11-04","2026-09-17",true,{"@type":102,"interactionType":103,"userInteractionCount":52},"InteractionCounter",{"@type":104},"ViewAction",{"@type":106,"mainEntity":107},"FAQPage",[108,114,118],{"name":109,"@type":110,"acceptedAnswer":111},"What problem does the thesis address in weather forecasting?","Question",{"text":112,"@type":113},"It examines whether machine learning models can produce realistic weather forecasts comparable to numerical physics-based models, focusing on short-term deterministic prediction at global scale.","Answer",{"name":115,"@type":110,"acceptedAnswer":116},"Why are machine learning methods proposed as an alternative to numerical models?",{"text":117,"@type":113},"Numerical methods depend on approximating atmospheric partial differential equations and require lengthy computation on multiple CPU nodes, while trained ML models can deliver comparable predictions in minutes with lower energy use during inference on GPUs.",{"name":119,"@type":110,"acceptedAnswer":120},"Which models are evaluated and how does the thesis improve results?",{"text":121,"@type":113},"The thesis evaluates PanguWeather, SFNO, FourCastNet, and DLWP, then proposes a meta-model that combines them to generate hybrid forecasts, leveraging each model’s strengths and emphasizing patterns in Italy.","https://schema.org",{"og:url":83,"og:type":124,"og:title":65,"og:site_name":95,"og:description":67},"article",{"robots":126,"canonical":83},"index,follow",{"doc_id":128,"site_id":62},290516,1790401592,{"code":4,"msg":5,"data":131},{"doc_id":128,"user_id":132,"nickname":92,"user_avatar":133,"doc_module":4,"category_id":39,"category_name":40,"doc_title":65,"doc_description":67,"doc_content":134,"file_id":135,"file_url":136,"file_type":137,"file_size":138,"view_count":52,"is_deleted":4,"is_public":8,"is_downloadable":8,"audit_status":8,"page_count":139,"language":140,"language_code":63,"site_id":62,"html_lang":63,"table_of_contents":141,"faqs":142,"seo_title":143,"seo_description":67,"update_tm":144,"read_time":145,"preview_page_count":4},8796095462418,"https://ap-avatar.wpscdn.com/avatar/80000253c1241d02b47?x-image-process=image/resize,m_fixed,w_180,h_180&k=1778826106357471780","A Metamodel approach integrating Machine Learning-based Weather Prediction frameworks  \nTesi di Laurea Magistrale in  \nHigh Performance Computing EngineeringIngegneria del Calcolo ad Alte Prestazioni  \nAuthor: Filippo Desantis  \nStudent ID: 222261  \nAdvisor: Prof. Edie Miglio, Prof. Luca Formaggia, Dr. Tommaso Benacchio  \nCompany referent: Dr. Giovanni Viciconte  \nAcademic Year: 2024-25  \ni  \nAbstract  \nHumans have endeavored to forecast the weather since the dawn of their existence. While for decades computers have primarily been employed to run numerical weather prediction models, there is now a growing tendency to leverage artificial intelligence techniques for the same purpose. This thesis aims to investigate the capabilities of Machine Learning models, developed by leading global companies and research institutions, in generating realistic weather forecasts that rival those produced by the existing numerical, physicsbased models.  \nNumerical methods, which rely on approximations of the Partial Differential Equations governing atmospheric physics, are known for their accuracy and reliability. Nonetheless, these methods necessitate lengthy computations across multiple CPU nodes. In contrast, Machine Learning models, once trained, can produce comparable predictions within minutes and with significantly lower energy consumption. This efficiency is attributable to their lower computational complexity during the inference phase, which allows for straightforward management and parallelization on GPU architectures. The strategies for weather prediction can either be deterministic or stochastic, focusing on short-term or long-term forecasts, extreme weather events, and enhancing spatial resolution.  \nIn this thesis, the focus is on short-term deterministic predictions on a global scale encompassing the entire planet. To address this research objective, the study evaluates the performance of four prominent machine learning weather prediction models: PanguWeather, SFNO, FourCastNet, and DLWP. Recognizing the strengths and weaknesses of each model, the thesis proposes the development of a meta-model that combines them to achieve optimal results. This meta-model generates hybrid forecasts that leverage the strengths of the four individual models, with a particular focus on the analysis of weather patterns in Italy. By doing so, the dataset size is reduced, enabling computations to be more manageable within computing environments similar to those used for local weather predictions.  \nKeywords: Weather Forecasting, Machine Learning, Numerical Models, Deterministic Prediction, Meta-Model, GPU Computing.  \nAbstract in lingua italiana  \nSin dagli albori della loro esistenza, gli esseri umani hanno cercato di prevedere il tempo atmosferico. Se per decenni i computer sono stati impiegati principalmente per l’esecuzionedi modelli numerici di previsione meteorologica, oggi si registra una crescente tendenza asfruttare tecniche di intelligenza artificiale per lo stesso scopo. Questa tesi si propone di indagare le potenzialità dei modelli di Machine Learning, sviluppati da aziende e istituzionidi ricerca di livello internazionale, nel generare previsioni meteorologiche realistiche, ingrado di competere con quelle prodotte dagli attuali modelli numerici basati sulla fisica.  \nI metodi numerici, che si fondano su approssimazioni delle Equazioni Differenziali alle Derivate Parziali che governano la fisica dell’atmosfera, sono noti per precisione e affidabilità . Tuttavia, tali metodi richiedono lunghe elaborazioni su più nodi CPU. Al contrario, i modelli di Machine Learning, una volta addestrati, sono in grado di produrreprevisioni comparabili in pochi minuti e con un consumo energetico significativamente inferiore. Questa efficienza è attribuibile alla minore complessità computazionale durante la fase di inferenza, che consente una gestione agevole e una parallelizzazione ottimalesulle architetture GPU. Le strategie di previsione meteorologica pos","cbCailpKGPCtSihW","https://ap.wps.com/l/cbCailpKGPCtSihW","pdf",9907382,116,"English","# Introduction\n## Machine Learning-based weather prediction models\n## Inferences and comparisons among different models","[{\"question\":\"What problem does the thesis address in weather forecasting?\",\"answer\":\"It examines whether machine learning models can produce realistic weather forecasts comparable to numerical physics-based models, focusing on short-term deterministic prediction at global scale.\"},{\"question\":\"Why are machine learning methods proposed as an alternative to numerical models?\",\"answer\":\"Numerical methods depend on approximating atmospheric partial differential equations and require lengthy computation on multiple CPU nodes, while trained ML models can deliver comparable predictions in minutes with lower energy use during inference on GPUs.\"},{\"question\":\"Which models are evaluated and how does the thesis improve results?\",\"answer\":\"The thesis evaluates PanguWeather, SFNO, FourCastNet, and DLWP, then proposes a meta-model that combines them to generate hybrid forecasts, leveraging each model’s strengths and emphasizing patterns in Italy.\"}]","A Metamodel approach integrating Machine Learning-based Weather Prediction frameworks | PDF",1789641916,292]