Forests　are　one　of　the　most　valuable　and　necessary　resources　to　protect　the　ecological　balance　of　the　planet,　a　natural　habitat　for　animals,　and　they　are　vital　to　our　lives　in　many　direct　and　indirect　ways.　However,　forest　fires　can　cause　severe　damage　to　the　land　and　many　other　resources　such　as　property,　human　life,　and　wildlife.　Previously,　metacellular　automata　have　been　used　to　simulate　the　spread　of　forest　fires,　and　the　traditional　metacellular　automata　model　outputs　the　spread　of　forest　fires　by　inputting　forest　fire-related　factors　such　as　topography,　meteorology,　and　vegetation　type,　but　these　inputs　are　highly　unstable　and　fires　can　have　an　effect　on　the　wind　direction　and　speed,　so　the　inputs　must　be　calibrated.　The　aim　of　this　study　is　to　explore　a　forest　fire　spread　simulation　model　based　on　Long　Short-Term　Memory(LSTM)　and　Cellular　Automaton　(CA),　and　to　improve　the　accuracy　of　the　simulation　results　by　calibrating　the　incoming　parameters　of　the　CA　in　order　to　improve　the　fire-fighting　action　during　forest　fires.　©　2024　IEEE.