{"id":1625,"date":"2026-06-30T14:49:00","date_gmt":"2026-06-30T13:49:00","guid":{"rendered":"https:\/\/research.reading.ac.uk\/palaeoclimate\/?p=1625"},"modified":"2026-06-30T14:49:47","modified_gmt":"2026-06-30T13:49:47","slug":"improving-wildfire-prediction-with-machine-learning-and-firebreaks","status":"publish","type":"post","link":"https:\/\/research.reading.ac.uk\/palaeoclimate\/improving-wildfire-prediction-with-machine-learning-and-firebreaks\/","title":{"rendered":"Improving Wildfire Prediction with Machine Learning and Firebreaks"},"content":{"rendered":"

Introduction<\/strong><\/p>\n

Wildfire management involves a range of approaches, including fuel load reduction, community preparedness, and fire suppression once a fire has started. An important management tool is the use of firebreaks, which can be either permanent features created by clearing vegetation or temporary barriers deployed during active fires using water or fire-retardant chemicals. Despite their widespread use, firebreaks are rarely incorporated into models used to predict wildfire spread.<\/p>\n

Accurately simulating wildfire behaviour remains a major challenge for scientists and land managers. Although a range of predictive wildfire models exist, many do not explicitly account for the influence of firebreaks on fire propagation. To address this gap, a team of researchers, including SPECIAL Group PI Sandy Harrison<\/a> and colleagues from the Leverhulme Centre for Wildfires<\/a>, have developed a proof-of-concept machine learning framework that integrates firebreaks into wildfire spread modelling. Their results demonstrate that including firebreak information can improve the accuracy of wildfire predictions.<\/p>\n

_______________________________________________________________________________________________________________________________________________________________________________________________________________________________<\/p>\n

Predicting spatio-temporal wildfire propagation with dynamic firebreaks – Zheng et al. 2026<\/a><\/p>\n

_______________________________________________________________________________________________________________________________________________________________________________________________________________________________<\/p>\n

A Multi-Model Approach<\/strong><\/p>\n

Machine learning techniques are increasingly used to model dynamic systems, including wildfires. In their 2026 study, Zheng et al. combined two modelling approaches to improve both the accuracy and speed of wildfire simulations.<\/p>\n

The first is a Cellular Automata (CA) model, a grid-based approach in which each cell changes state according to a set of rules based on its surrounding cells and environmental conditions. The model simulates wildfire spread using information on weather conditions, topography, vegetation density, and firebreak placement. These simulations generate large amounts of wildfire propagation data that are then used to train a second model. The CA framework is based on the well-established model of Alexandridis et al. (2008), which has previously been validated for wildfire applications.<\/p>\n

The second model is a Convolutional Long Short-Term Memory (ConvLSTM) network, an advanced form of recurrent neural network (RNN). RNNs have been shown to work well for dynamics fire modelling as they capture time-sequential patterns. The model is trained on the simulations produced by the CA model to produce a data-driven model of wildfire propagation.<\/p>\n

Both modelling approaches simulate wildfire behaviour through a series of time steps, with cells transitioning between different states. These states include unburned, burning, burned, permanent firebreak, and temporary firebreak, with temporary firebreaks progressively losing effectiveness over time. The CA model also includes an additional state representing areas that cannot burn.<\/p>\n

\"\"<\/p>\n

Fig 1: Flow chart showing potential states the cellular automata model can move through. This particular path outlines the CA model transitions when a neighbouring cell is burning.<\/em><\/p>\n

Key Findings<\/strong><\/p>\n

To evaluate the framework, the researchers tested it using three historical California wildfires:<\/p>\n

    \n
  1. Chimney Fire (2016)<\/li>\n
  2. Ferguson Fire (2018)<\/li>\n
  3. Bear Fire (2020)<\/li>\n<\/ol>\n

    Multiple simulations were performed with different firebreak configurations, as well as scenarios without firebreaks, to assess both model accuracy and computational performance.<\/p>\n

    The results showed that the ConvLSTM model successfully reproduced wildfire spread patterns while accounting for the effects of firebreaks. Simulations that incorporated firebreak information produced more accurate predictions than those that did not. This improvement likely reflects the substantial influence that firebreaks can have on fire behaviour by altering factors such as rate of fire spread.<\/p>\n

    Conclusions<\/strong><\/p>\n

    This study used wildfire simulations that included wind and landscape slope to create training data, but kept wind conditions the same throughout each simulation to make the modelling process simpler. While this reduces how closely the simulations match real-world conditions, it helps test the model more clearly. The results showed that the ConvLSTM model can accurately predict how wildfires spread and how effective different firebreaks are. It also runs much faster than the traditional cellular automata (CA) model, making it more suitable for real-time wildfire management. Future work will improve the model by including changing wind conditions, more detailed terrain information, and additional factors that affect how fires burn. The presented model, and suggested improvement to more accurately mimic real world conditions, will help land managers to more accurately predict real-time fire propagation taking into consideration the management decisions that are being made.<\/p>\n

    Read more about this study in the full paper to find out about model configurations and specific results from the simulations!<\/p>\n

    Zheng, J., Xu, Z., Arcucci, R., Harrison, S.P.<\/strong>, Xu, L.L. & Cheng, S. 2026. Predicting spatio-temporal wildfire propagation with dynamic firebreaks. Natural Hazards Earth System Science<\/em>, 26, 2871\u20132895, https:\/\/doi.org\/10.5194\/nhess-26-2871-2026.<\/p>\n","protected":false},"excerpt":{"rendered":"

    Introduction Wildfire management involves a range of approaches, including fuel load reduction, community preparedness, and fire suppression once a fire has started. An important management tool is the use of…Read More ><\/a><\/p>\n","protected":false},"author":959,"featured_media":1626,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"_acf_changed":false,"_monsterinsights_skip_tracking":false,"_monsterinsights_sitenote_active":false,"_monsterinsights_sitenote_note":"","_monsterinsights_sitenote_category":0,"__cvm_playback_settings":[],"__cvm_video_id":"","footnotes":""},"categories":[22],"tags":[],"class_list":["post-1625","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-blog"],"acf":[],"yoast_head":"\nImproving Wildfire Prediction with Machine Learning and Firebreaks - SPECIAL Palaeoclimate<\/title>\n<meta name=\"robots\" content=\"index, follow, max-snippet:-1, max-image-preview:large, max-video-preview:-1\" \/>\n<link rel=\"canonical\" href=\"https:\/\/research.reading.ac.uk\/palaeoclimate\/improving-wildfire-prediction-with-machine-learning-and-firebreaks\/\" \/>\n<meta property=\"og:locale\" content=\"en_GB\" \/>\n<meta property=\"og:type\" content=\"article\" \/>\n<meta property=\"og:title\" content=\"Improving Wildfire Prediction with Machine Learning and Firebreaks - SPECIAL Palaeoclimate\" \/>\n<meta property=\"og:description\" content=\"Introduction Wildfire management involves a range of approaches, including fuel load reduction, community preparedness, and fire suppression once a fire has started. An important management tool is the use of...Read More >\" \/>\n<meta property=\"og:url\" content=\"https:\/\/research.reading.ac.uk\/palaeoclimate\/improving-wildfire-prediction-with-machine-learning-and-firebreaks\/\" \/>\n<meta property=\"og:site_name\" content=\"SPECIAL Palaeoclimate\" \/>\n<meta property=\"article:published_time\" content=\"2026-06-30T13:49:00+00:00\" \/>\n<meta property=\"article:modified_time\" content=\"2026-06-30T13:49:47+00:00\" \/>\n<meta property=\"og:image\" content=\"https:\/\/research.reading.ac.uk\/palaeoclimate\/wp-content\/uploads\/sites\/78\/2026\/06\/zheng_Et_al_1.png\" \/>\n\t<meta property=\"og:image:width\" content=\"732\" \/>\n\t<meta property=\"og:image:height\" content=\"372\" \/>\n\t<meta property=\"og:image:type\" content=\"image\/png\" \/>\n<meta name=\"author\" content=\"Sophia Cain\" \/>\n<meta name=\"twitter:card\" content=\"summary_large_image\" \/>\n<meta name=\"twitter:label1\" content=\"Written by\" \/>\n\t<meta name=\"twitter:data1\" content=\"Sophia Cain\" \/>\n\t<meta name=\"twitter:label2\" content=\"Estimated reading time\" \/>\n\t<meta name=\"twitter:data2\" content=\"4 minutes\" \/>\n<script type=\"application\/ld+json\" class=\"yoast-schema-graph\">{\"@context\":\"https:\/\/schema.org\",\"@graph\":[{\"@type\":\"WebPage\",\"@id\":\"https:\/\/research.reading.ac.uk\/palaeoclimate\/improving-wildfire-prediction-with-machine-learning-and-firebreaks\/\",\"url\":\"https:\/\/research.reading.ac.uk\/palaeoclimate\/improving-wildfire-prediction-with-machine-learning-and-firebreaks\/\",\"name\":\"Improving Wildfire Prediction with Machine Learning and Firebreaks - SPECIAL Palaeoclimate\",\"isPartOf\":{\"@id\":\"https:\/\/research.reading.ac.uk\/palaeoclimate\/#website\"},\"datePublished\":\"2026-06-30T13:49:00+00:00\",\"dateModified\":\"2026-06-30T13:49:47+00:00\",\"author\":{\"@id\":\"https:\/\/research.reading.ac.uk\/palaeoclimate\/#\/schema\/person\/6ae1715d9ded9f0caa51878ed800cdc9\"},\"breadcrumb\":{\"@id\":\"https:\/\/research.reading.ac.uk\/palaeoclimate\/improving-wildfire-prediction-with-machine-learning-and-firebreaks\/#breadcrumb\"},\"inLanguage\":\"en-GB\",\"potentialAction\":[{\"@type\":\"ReadAction\",\"target\":[\"https:\/\/research.reading.ac.uk\/palaeoclimate\/improving-wildfire-prediction-with-machine-learning-and-firebreaks\/\"]}]},{\"@type\":\"BreadcrumbList\",\"@id\":\"https:\/\/research.reading.ac.uk\/palaeoclimate\/improving-wildfire-prediction-with-machine-learning-and-firebreaks\/#breadcrumb\",\"itemListElement\":[{\"@type\":\"ListItem\",\"position\":1,\"name\":\"Home\",\"item\":\"https:\/\/research.reading.ac.uk\/palaeoclimate\/\"},{\"@type\":\"ListItem\",\"position\":2,\"name\":\"Improving Wildfire Prediction with Machine Learning and Firebreaks\"}]},{\"@type\":\"WebSite\",\"@id\":\"https:\/\/research.reading.ac.uk\/palaeoclimate\/#website\",\"url\":\"https:\/\/research.reading.ac.uk\/palaeoclimate\/\",\"name\":\"SPECIAL Palaeoclimate\",\"description\":\"Webpage of Sandy's PalaeoEnvironments and Climate Analysis research group at the University of Reading (UK)\",\"potentialAction\":[{\"@type\":\"SearchAction\",\"target\":{\"@type\":\"EntryPoint\",\"urlTemplate\":\"https:\/\/research.reading.ac.uk\/palaeoclimate\/?s={search_term_string}\"},\"query-input\":\"required name=search_term_string\"}],\"inLanguage\":\"en-GB\"},{\"@type\":\"Person\",\"@id\":\"https:\/\/research.reading.ac.uk\/palaeoclimate\/#\/schema\/person\/6ae1715d9ded9f0caa51878ed800cdc9\",\"name\":\"Sophia Cain\",\"image\":{\"@type\":\"ImageObject\",\"inLanguage\":\"en-GB\",\"@id\":\"https:\/\/research.reading.ac.uk\/palaeoclimate\/#\/schema\/person\/image\/\",\"url\":\"https:\/\/secure.gravatar.com\/avatar\/a4b6b71e445f4365e2434d381d6a4e6733a438e4ce1c59fbbd12a0e11cf74672?s=96&d=mm&r=g\",\"contentUrl\":\"https:\/\/secure.gravatar.com\/avatar\/a4b6b71e445f4365e2434d381d6a4e6733a438e4ce1c59fbbd12a0e11cf74672?s=96&d=mm&r=g\",\"caption\":\"Sophia Cain\"},\"url\":\"https:\/\/research.reading.ac.uk\/palaeoclimate\/author\/kw931720\/\"}]}<\/script>\n<!-- \/ Yoast SEO plugin. -->","yoast_head_json":{"title":"Improving Wildfire Prediction with Machine Learning and Firebreaks - SPECIAL Palaeoclimate","robots":{"index":"index","follow":"follow","max-snippet":"max-snippet:-1","max-image-preview":"max-image-preview:large","max-video-preview":"max-video-preview:-1"},"canonical":"https:\/\/research.reading.ac.uk\/palaeoclimate\/improving-wildfire-prediction-with-machine-learning-and-firebreaks\/","og_locale":"en_GB","og_type":"article","og_title":"Improving Wildfire Prediction with Machine Learning and Firebreaks - SPECIAL Palaeoclimate","og_description":"Introduction Wildfire management involves a range of approaches, including fuel load reduction, community preparedness, and fire suppression once a fire has started. An important management tool is the use of...Read More >","og_url":"https:\/\/research.reading.ac.uk\/palaeoclimate\/improving-wildfire-prediction-with-machine-learning-and-firebreaks\/","og_site_name":"SPECIAL Palaeoclimate","article_published_time":"2026-06-30T13:49:00+00:00","article_modified_time":"2026-06-30T13:49:47+00:00","og_image":[{"width":732,"height":372,"url":"https:\/\/research.reading.ac.uk\/palaeoclimate\/wp-content\/uploads\/sites\/78\/2026\/06\/zheng_Et_al_1.png","type":"image\/png"}],"author":"Sophia Cain","twitter_card":"summary_large_image","twitter_misc":{"Written by":"Sophia Cain","Estimated reading time":"4 minutes"},"schema":{"@context":"https:\/\/schema.org","@graph":[{"@type":"WebPage","@id":"https:\/\/research.reading.ac.uk\/palaeoclimate\/improving-wildfire-prediction-with-machine-learning-and-firebreaks\/","url":"https:\/\/research.reading.ac.uk\/palaeoclimate\/improving-wildfire-prediction-with-machine-learning-and-firebreaks\/","name":"Improving Wildfire Prediction with Machine Learning and Firebreaks - SPECIAL Palaeoclimate","isPartOf":{"@id":"https:\/\/research.reading.ac.uk\/palaeoclimate\/#website"},"datePublished":"2026-06-30T13:49:00+00:00","dateModified":"2026-06-30T13:49:47+00:00","author":{"@id":"https:\/\/research.reading.ac.uk\/palaeoclimate\/#\/schema\/person\/6ae1715d9ded9f0caa51878ed800cdc9"},"breadcrumb":{"@id":"https:\/\/research.reading.ac.uk\/palaeoclimate\/improving-wildfire-prediction-with-machine-learning-and-firebreaks\/#breadcrumb"},"inLanguage":"en-GB","potentialAction":[{"@type":"ReadAction","target":["https:\/\/research.reading.ac.uk\/palaeoclimate\/improving-wildfire-prediction-with-machine-learning-and-firebreaks\/"]}]},{"@type":"BreadcrumbList","@id":"https:\/\/research.reading.ac.uk\/palaeoclimate\/improving-wildfire-prediction-with-machine-learning-and-firebreaks\/#breadcrumb","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https:\/\/research.reading.ac.uk\/palaeoclimate\/"},{"@type":"ListItem","position":2,"name":"Improving Wildfire Prediction with Machine Learning and Firebreaks"}]},{"@type":"WebSite","@id":"https:\/\/research.reading.ac.uk\/palaeoclimate\/#website","url":"https:\/\/research.reading.ac.uk\/palaeoclimate\/","name":"SPECIAL Palaeoclimate","description":"Webpage of Sandy's PalaeoEnvironments and Climate Analysis research group at the University of Reading (UK)","potentialAction":[{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https:\/\/research.reading.ac.uk\/palaeoclimate\/?s={search_term_string}"},"query-input":"required name=search_term_string"}],"inLanguage":"en-GB"},{"@type":"Person","@id":"https:\/\/research.reading.ac.uk\/palaeoclimate\/#\/schema\/person\/6ae1715d9ded9f0caa51878ed800cdc9","name":"Sophia Cain","image":{"@type":"ImageObject","inLanguage":"en-GB","@id":"https:\/\/research.reading.ac.uk\/palaeoclimate\/#\/schema\/person\/image\/","url":"https:\/\/secure.gravatar.com\/avatar\/a4b6b71e445f4365e2434d381d6a4e6733a438e4ce1c59fbbd12a0e11cf74672?s=96&d=mm&r=g","contentUrl":"https:\/\/secure.gravatar.com\/avatar\/a4b6b71e445f4365e2434d381d6a4e6733a438e4ce1c59fbbd12a0e11cf74672?s=96&d=mm&r=g","caption":"Sophia Cain"},"url":"https:\/\/research.reading.ac.uk\/palaeoclimate\/author\/kw931720\/"}]}},"_links":{"self":[{"href":"https:\/\/research.reading.ac.uk\/palaeoclimate\/wp-json\/wp\/v2\/posts\/1625","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/research.reading.ac.uk\/palaeoclimate\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/research.reading.ac.uk\/palaeoclimate\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/research.reading.ac.uk\/palaeoclimate\/wp-json\/wp\/v2\/users\/959"}],"replies":[{"embeddable":true,"href":"https:\/\/research.reading.ac.uk\/palaeoclimate\/wp-json\/wp\/v2\/comments?post=1625"}],"version-history":[{"count":1,"href":"https:\/\/research.reading.ac.uk\/palaeoclimate\/wp-json\/wp\/v2\/posts\/1625\/revisions"}],"predecessor-version":[{"id":1627,"href":"https:\/\/research.reading.ac.uk\/palaeoclimate\/wp-json\/wp\/v2\/posts\/1625\/revisions\/1627"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/research.reading.ac.uk\/palaeoclimate\/wp-json\/wp\/v2\/media\/1626"}],"wp:attachment":[{"href":"https:\/\/research.reading.ac.uk\/palaeoclimate\/wp-json\/wp\/v2\/media?parent=1625"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/research.reading.ac.uk\/palaeoclimate\/wp-json\/wp\/v2\/categories?post=1625"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/research.reading.ac.uk\/palaeoclimate\/wp-json\/wp\/v2\/tags?post=1625"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}