Analysis Of The Influencing Factors And Scale Effects Of Urban Surface Thermal Environment

In recent years,rapid urbanization has not only increased urban impermeability but also changed urban surface morphology,which in turn has disrupted the urban surface energy balance,caused the urban heat island(UHI)effect and deteriorated the urban thermal environment,with far-reaching impacts on the urban water environment,air quality,urban public health,energy consumption and vegetation soils.Therefore,exploring the distribution characteristics,influencing factors and scale effects of the urban surface thermal environment is of great practical significance for the regulation of the urban thermal environment at multiple scales.Based on Landsat8 remote sensing images,3D building vector data,OSM road network data,point of interest data and other data,this paper employs the single window algorithm to invert the surface temperature and analyze the seasonal distribution characteristics of land surface temperature(LST)and its scale effects in the central city of Jinan in four seasons.The spatial distribution characteristics of the influencing factors and their scale effects are also analyzed.Moreover,this paper explores the non-linear relationship between surface temperature and the influencing factors by using boosted regression tree model(BRT),and investigates the relative contribution and marginal effects of the influencing factors on the LST in all seasons at different scales.The main results are as follows.(1)There is a strong spatial heterogeneity in surface temperature in the study area.On the whole,the high temperature zones are mainly distributed in the old city,the eastern industrial zone and the western industrial zone;at the season level,the surface temperature shows a decreasing trend of high in the northeast and low in the southwest in spring and summer,a decreasing spatial distribution pattern from northwest to southeast in autumn,and a pattern of high in the northeast and southwest,and low in the south and middle in winter.(2)The impact of influencing factors on LST has non-linear characteristics and seasonal effects.In terms of relative contribution,in any season and at any scale,the two-dimensional surface cover index makes the largest contribution to LST and is much higher than other indicators,while the three-dimensional building form index is the second and the human activity intensity index is the smallest.Normalized difference vegetation index(NDVI),Normalized difference building index(NDBI)and modified normalized difference water index(MNDWI)are the dominant surface temperature factors in spring and summer seasons,NDVI,mean building height(MBH)and NDBI are the dominant surface temperature factors in autumn,and NDBI,NDVI,MNDWI and MBH are the dominant surface temperature factors in winter,from summer to transitional season to winter,the urban thermal environment influencing factors gradually tend to be complex structural processes.The marginal effect of NDVI on surface temperature in all seasons shows a shift from positive to negative influence,NDBI shows a shift from negative to positive contribution in most cases,MDNWI shows a shift from positive to negative influence in spring,summer and autumn,and a shift from negative to positive contribution in winter.The building coverage ratio(BCR)basically shows a shift from negative to positive contribution,while the building volume density(BVD)and MBH show a shift from positive to negative contribution,while the sum of point of interest(SPOI)and road network density(RND)curves are smoother.(3)LST and its influencing factors have an obvious scale effect.From the perspective of LST,with the change of scale,the LST in the central city of Jinan shows a trend of decreasing extreme difference and decreasing data dispersion in all seasons;from the perspective of relative contribution,the relative contribution of each indicator to LST changes with the scale,with different trends,and the sensitivity of each indicator to scale also varies.In terms of marginal effects,the shape of the marginal effect curve of each indicator on LST varies slightly at different scales,and the thresholds for positive and negative contribution conversion also vary with scale,with the two-dimensional surface coverage indicator having a small range and the threedimensional building form indicator having a large range,in addition,there are some points of highest and lowest contribution.The scale effects suggest that the implications and potential uncertainties associated with the choice of different spatial analysis scales should be considered in subsequent studies of the urban surface thermal environment.This study extends and deepens the scientific understanding of the complexity of surface temperature influencing factors and scale effects,and has important implications for the regulation of multi-scale urban surface thermal environments.