城市功能区夏季昼夜热环境分异机制研究OA

Urban thermal environments are jointly shaped by urban form,land-surface conditions,functional organization,and human activity.Taking the main urban area of Hefei,China,as a case study,this study examines the daytime-nighttime differentiation in summer Land Surface Temperature(LST)across urban functional zones.Daytime and nighttime LST data were derived from the MODIS MOD11A1 product for the summer of 2024,and multi-source spatial data were used to construct 11 indicators representing building morphology,ecological and surface-cover characteristics,and activity functions.In total,1,217 road-bounded blocks were classified into different functional zones.Random forest models,permutation importance,kernel-smoothed response curves,and bootstrap confidence intervals were employed to identify dominant factors,nonlinear responses,and candidate thresholds.The study unveils pronounced temporal and functional differentiation in the urban thermal environment.Daytime hotspots were mainly concentrated in industrial and high-intensity mixed-use zones.Industrial zones recorded the highest mean daytime LST(36.7 ℃)and the largest day-night temperature difference(9.5 ℃).In contrast,nighttime hotspots shifted toward residential,mixed-use,and commercial zones,with residential zones exhibiting the highest mean nighttime LST(27.8 ℃).Agricultural zones maintained the lowest mean LST during both daytime and nighttime(at 32.7 ℃ and 26.1 ℃,respectively),indicating their cool-background and ecological-buffering functions.The nighttime random forest model showed substantially greater explanatory power than the daytime model(R2 values:0.725 and 0.415,respectively).The driving structures differed markedly between daytime and nighttime.At night,the activity-function variables accounted for 72.6％of the total feature importance,with Population Activity Intensity(PAI)alone contributing 55.6％,whereas building morphology and surface-cover characteristics contributed 16.6％and 10.8％,respectively.During daytime,the contributions were more balanced among building morphology,activity functions,and surface cover,which accounted for 38.8％,36.0％,and 25.2％,respectively.The PAI,impervious surface area ratio(ISA),and building height were the leading daytime predictors.These findings suggest that daytime heat is jointly associated with development intensity,surface hardening,and activity concentration,whereas nighttime heat is more strongly related to persistent activity,facility agglomeration,stored-heat release,and restricted radiative and ventilative cooling.PAI should be interpreted as an integrated proxy for human activity and correlated with built-environment intensity,not as an independent causal factor.Clear,non-linear,and function-specific thresholds were identified.Across all samples,daytime PAI and ISA changed the effect direction by nearly 7.46 and 33.14％,respectively;nighttime PAI changed near 10.29％.The daytime ISA threshold increased to 52.53％in industrial zones.Residential nighttime LST became more sensitive near a PAI of 11.20 and a point of interest density of 197.12 km-2.The nighttime response of commercial zones changed near a land-use-mix value of 0.31,whereas the daytime NDVI threshold in green zones was 0.49,the nighttime SVF threshold in green zones was approximately 0.97.The principal contribution of this study lies in its replacement of spatially uniform analytical units with functionally meaningful urban zones,integration of daytime and nighttime thermal processes within a unified framework,and translation of explainable machine learning results into quantitative and function-specific planning thresholds.Industrial and high-intensity mixed-use zones should prioritize reducing continuous impervious surfaces,applying cool roofs and high-albedo materials,and improving shade and ventilation.Residential and commercial zones should enhance nocturnal ventilation and permeable green spaces,as well as manage nighttime lighting and energy-intensive operations.The proposed framework provides a quantitative basis for time-sensitive and function-specific urban heat governance,although finer-resolution and multi-temporal data are required to test the stability and applicability of the identified mechanisms.