Numerous　studies　have　explored　multiple　factors　that　affect　the　urban　thermal　environment.　Most　current　studies　focus　on　establishing　the　holistic　quantitative　relationships　between　the　characteristics　of　urban　thermal　environmental　factors　and　land　surface　temperature　(LST),　but　less　attention　has　been　given　to　exploring　urban　subdivisions　in　detail.　The　local　climate　zone　(LCZ)　system　has　been　proven　effective　in　establishing　the　quantitative　relationships　between　urban　climate　and　spatial　factors,　thereby　revealing　the　characteristics　of　LST　spatial　variations.　This　study　utilized　the　random　forest　regression　model　to　analyze　the　relative　importance　and　marginal　effects　of　the　urban　thermal　environmental　factors　on　seasonal　LSTs　based　on　the　whole　study　area　and　the　LCZ-derived　built　type　area　of　Fuzhou　city,　known　as　a　“furnace　city”　in　southeastern　China.　The　results　show　that:　(1)　Building　morphology　and　spatial　distribution　exert　a　significant　impact　on　LST,　with　the　LST　levels　of　different　building　forms　ranked　as　open　＜　compact　and　high-rise　＜　mid-rise　＜　low-rise.　(2)　Landscape　composition　exhibits　the　greatest　influence　on　seasonal　LSTs　in　the　whole　study　city.　(3)　Building　morphology　emerges　as　the　most　influential　category　for　LST　in　the　LCZ　built　type　area,　with　the　standard　deviation　of　building　height　having　the　highest　relative　importance　at　approximately　34%.　(4)　Concentrated　compact　low-rise　self-built　houses　experience　higher　temperatures　and　should　be　a　focus　for　future　urban　renewal　efforts.　We　believe　that　these　findings　provide　quantitative　insights　for　urban　heat　island　analysis　and　have　substantial　implications　for　climate-friendly　urban　planning.　©　2023　Elsevier　Ltd