[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-seo-328180-105":3,"detail-sidebar-cat-0-en-105":81,"doc-detail-328180-en":131},{"code":4,"msg":5,"data":6},0,"ok",{"site_id":7,"language":8,"slug":9,"title":10,"keywords":11,"description":12,"schema_data":13,"social_meta":74,"head_meta":76,"extra_data":78,"updated_unix":80},105,"en","physics-guided-spatiotemporal-neural-models-for-fuel-density-prediction","PHYSICS-GUIDED SPATIOTEMPORAL NEURAL MODELS FOR FUEL DENSITY PREDICTION","","Physics-guided machine learning framework integrates physics constraints and domain knowledge to improve fuel density prediction accuracy and stability. Three architectures—ConvLSTM, Adaptive Fourier Neural Operator (AFNONet), and Video Vision Transformer (ViViT)—learn spatiotemporal fuel density evolution. The method adds differentiable physics-informed loss terms, including mass-conserving fuel transport and rate-of-spread estimation. Experiments averaged over independent trials show the framework surpasses data-only baselines in both accuracy and stability, enabling efficient, physically plausible fire forecasting for adaptive prescribed burn management.",{"@graph":14,"@context":73},[15,34,56],{"@type":16,"itemListElement":17},"BreadcrumbList",[18,23,27,31],{"item":19,"name":20,"@type":21,"position":22},"https://docshare.wps.com","Home","ListItem",1,{"item":24,"name":25,"@type":21,"position":26},"https://docshare.wps.com/document/","Document",2,{"item":28,"name":29,"@type":21,"position":30},"https://docshare.wps.com/document/research-report/","Research & Report",3,{"item":32,"name":10,"@type":21,"position":33},"https://docshare.wps.com/document/physics-guided-spatiotemporal-neural-models-for-fuel-density-prediction/328180/",4,{"url":32,"name":10,"@type":35,"image":36,"author":41,"headline":10,"publisher":44,"fileFormat":47,"inLanguage":8,"description":12,"dateModified":48,"datePublished":49,"encodingFormat":47,"isAccessibleForFree":50,"interactionStatistic":51},"DigitalDocument",{"url":37,"@type":38,"width":39,"height":40},"https://docshare.wps.com/thumbnails/physics-guided-spatiotemporal-neural-models-for-fuel-density-prediction/328180.png","ImageObject",300,407,{"name":42,"@type":43},"Aurelia","Person",{"url":19,"name":45,"@type":46},"DocShare","Organization","application/pdf","2026-10-31","2026-09-21",true,{"@type":52,"interactionType":53,"userInteractionCount":55},"InteractionCounter",{"@type":54},"ViewAction",12,{"@type":57,"mainEntity":58},"FAQPage",[59,65,69],{"name":60,"@type":61,"acceptedAnswer":62},"What is the goal of the physics-guided framework in this paper?","Question",{"text":63,"@type":64},"The framework aims to predict fuel density over space and time by integrating physics constraints and domain knowledge into deep learning models to improve accuracy and stability.","Answer",{"name":66,"@type":61,"acceptedAnswer":67},"Which neural network architectures are compared for fuel density modeling?",{"text":68,"@type":64},"The paper compares ConvLSTM, Adaptive Fourier Neural Operator (AFNONet), and Video Vision Transformer (ViViT) for spatiotemporal fuel density change.",{"name":70,"@type":61,"acceptedAnswer":71},"What physics-informed components are included in the training loss?",{"text":72,"@type":64},"The loss incorporates differentiable physics-informed terms, including a mass-conserving fuel transport term and a rate-of-spread estimation term.","https://schema.org",{"og:url":32,"og:type":75,"og:title":10,"og:site_name":45,"og:description":12},"article",{"robots":77,"canonical":32},"index,follow",{"doc_id":79,"site_id":7},328180,1790610306,{"code":4,"msg":82,"data":83},"success",[84,88,92,96,101,106,111,115,120,123,127],{"id":22,"doc_module":4,"doc_module_name":25,"category_name":85,"show_sort_weight":86,"slug":87},"Story & Novel",90,"story-novel",{"id":26,"doc_module":4,"doc_module_name":25,"category_name":89,"show_sort_weight":90,"slug":91},"Literature",80,"literature",{"id":33,"doc_module":4,"doc_module_name":25,"category_name":93,"show_sort_weight":94,"slug":95},"Exam",70,"exam",{"id":97,"doc_module":4,"doc_module_name":25,"category_name":98,"show_sort_weight":99,"slug":100},5,"Comic",60,"comic",{"id":102,"doc_module":4,"doc_module_name":25,"category_name":103,"show_sort_weight":104,"slug":105},6,"Technology",50,"technology",{"id":107,"doc_module":4,"doc_module_name":25,"category_name":108,"show_sort_weight":109,"slug":110},7,"Healthcare",40,"healthcare",{"id":112,"doc_module":4,"doc_module_name":25,"category_name":29,"show_sort_weight":113,"slug":114},8,30,"research-report",{"id":116,"doc_module":4,"doc_module_name":25,"category_name":117,"show_sort_weight":118,"slug":119},9,"Religion & Spirituality",20,"religion-spirituality",{"id":118,"doc_module":4,"doc_module_name":25,"category_name":121,"show_sort_weight":118,"slug":122},"World Cup","world-cup",{"id":124,"doc_module":4,"doc_module_name":25,"category_name":125,"show_sort_weight":124,"slug":126},10,"Lifestyle","lifestyle",{"id":128,"doc_module":4,"doc_module_name":25,"category_name":129,"show_sort_weight":97,"slug":130},19,"General","general",{"code":4,"msg":82,"data":132},{"doc_id":79,"user_id":133,"nickname":42,"user_avatar":134,"doc_module":4,"category_id":112,"category_name":29,"doc_title":10,"doc_description":12,"doc_content":135,"file_id":136,"file_url":137,"file_type":138,"file_size":139,"view_count":55,"is_deleted":4,"is_public":22,"is_downloadable":22,"audit_status":22,"page_count":97,"language":140,"language_code":8,"site_id":7,"html_lang":8,"table_of_contents":141,"faqs":142,"seo_title":143,"seo_description":12,"update_tm":144,"read_time":145,"preview_page_count":22,"preview_urls":146},1099514068365,"https://ap-avatar.wpscdn.com/avatar/10000253d8d9f28188e?x-image-process=image/resize,m_fixed,w_180,h_180&k=1794309523707217728","PHYSICS-GUIDED SPATIOTEMPORAL NEURAL MODELS FOR FUEL DENSITY PREDICTION  \nTolga Caglar, Jaynil Jaiswal, Saqib Azim, Yudhir Gala, Mai H. Nguyen, Ilkay Altintas  \nSan Diego Supercomputer Center  \nUniversity of California, San Diego  \n{tcaglar, [mhnguyen](mhnguyen}@ucsd.edu)[}](mhnguyen}@ucsd.edu)[@ucsd.edu](mhnguyen}@ucsd.edu)  \n8 Jul 2026  \nAbstract—This paper presents a physics-guided machine learning (PGML) framework for fuel density prediction, integrating physics constraints and domain knowledge into deep learning models to enhance model accuracy and stability. We explore three deep learning architectures — ConvLSTM, Adaptive Fourier Neural Operator (AFNONet), and Video Vision Transformer (ViViT)—to model the spatiotemporal evolution of fuel density. Our approach incorporates differentiable physics-informed terms in the loss function, including a mass-conserving fuel transport term and a rate-of-spread estimation. Experimental results, averaged across multiple independent trials, demonstrate that the proposed PGML framework outperforms purely data-driven baselines without physics constraints in both accuracy and stability. This framework enables computationally efficient, physically plausible fire forecasting to support adaptive prescribed burn management.  \nI. INTRODUCTION  \narXiv :2607 .06999v1  \ndangerous wildfires, enhance forest health, and help manage pests[2] . However, these burns are only initiated under specific conditions to ensure safety and prevent unintended consequences [3] . Fire managers rely on process-based models such as FARSITE [4] and QUIC-Fire [5] to predict the behavior and trajectory of prescribed fires under varying weather conditions (primarily wind speed and wind direction), fuel density in the area, and pre-defined ignition conditions. These processbased models simulate physical fire dynamics to assess burn risks, but their computational demands often prevent real-time decision-making [6, 7] .  \nWhile purely data-driven deep learning architectures, such as CNNs [8, 9] and U-Nets [10], offer computational speedup, they are ill-suited for complex multi-point ignitions and frequently predict non-physical behaviors like spontaneous fuel regeneration. To counteract this, existing physics-guided ML approaches attempt to constrain outputs using partial differential equations (PDEs) [11, 6] . However, these rigid mathematical formulations are often highly unstable, fail to converge, or suffer from well-documented gradient flow patholo-  \ngies [12, 13] . Our proposed surrogate framework addresses these gaps by embedding domain-specific physical constraints as ”soft” penalties, providing the guidance of physics without the numerical fragility of strict equation solving.  \nIn this paper, we present a framework to incorporate physics components into deep learning models that can work across different architectures. Specifically, we present a comparative analysis with and without physics-guided loss for three deep learning models, namely Convolutional LSTM (ConvLSTM) [14], Adaptive Fourier Neural Operator (AFNONet) [15], and Video Vision Transformer (ViViT) [16], for the task of spatiotemporal fuel density change.  \nII. DATASET AND PROBLEM FORMULATION  \nWe utilize an ensemble of QUIC-Fire simulations of fire spread, measuring fuel density (mass of fire fuel per unit volume) over time in an area. Each simulation runs for 50 seconds on a 300 × 300 spatial grid with flat grassland. The dataset comprises all possible combinations of 7 wind speed values (ranging from 1 m/s to 15 m/s), 11 wind direction values (from 230 to 330 degrees), 4 distinct ignition pattern types (aerial, inward, outward, strip north) . The wind speed and wind direction remain constant across the entire spatial grid and all timesteps. The data contains fuel densities in the range [0, 0.7kg/m2 ] . To ensure effective learning, preprocessing steps include normalization of wind speed and wind direction using min-max scaling, and incorporation ","cbCaiqGEQeQUvU4U","https://ap.wps.com/l/cbCaiqGEQeQUvU4U","pdf",7131408,"English","# Abstract\n# I. Introduction\n# II. Dataset and Problem Formulation\n# III. Methods\n## A. ConvLSTM","[{\"question\":\"What is the goal of the physics-guided framework in this paper?\",\"answer\":\"The framework aims to predict fuel density over space and time by integrating physics constraints and domain knowledge into deep learning models to improve accuracy and stability.\"},{\"question\":\"Which neural network architectures are compared for fuel density modeling?\",\"answer\":\"The paper compares ConvLSTM, Adaptive Fourier Neural Operator (AFNONet), and Video Vision Transformer (ViViT) for spatiotemporal fuel density change.\"},{\"question\":\"What physics-informed components are included in the training loss?\",\"answer\":\"The loss incorporates differentiable physics-informed terms, including a mass-conserving fuel transport term and a rate-of-spread estimation term.\"}]","PHYSICS-GUIDED SPATIOTEMPORAL NEURAL MODELS FOR FUEL DENSITY PREDICTION | PDF",1789978796,13,[147],{"page_no":22,"url":148,"width":149,"height":150},"https://res-academy.cache.wpscdn.com/upload/kdocshare/2026-11-08/e6e3e54aaa6dedcc24592fff6c3366e2.webp",983,1390]