Research On Air Temperature Extrapolation Based On The Classification Of Urban Underlying Surface In Xi’an

Near surface air temperature data is one of the essential parameters in urban studies,with the development of urbanization of city underlying surface change,the traditional research on different areas in a city using the unified method of air temperature data is no longer applicable,air temperature data from a single point of monitoring station data are urgently needed to travel to the surface domain data.The urban underlying surface is a factor that has a great direct influence on the air temperature except solar radiation.The study on the influence mechanism of the urban undersurface air temperature is an important premise to accurately obtain the urban near surface air temperature.Based on this research background,this study took Xi’an as an example,and referred to the classification of Local Climate Zone(LCZ)to classify the urban undersurface of the main urban area.Temperature-Vegetation Index method was used to retrieve the raster data of the air temperature near the ground in the main urban area,and the spatial distribution characteristics of the air temperature in different underlying surfaces were studied.And using multiple regression method to establish the air temperature Inference model for the different underlying surface.The main content is as follows.(1)Referring to the method of Local Climate Zone,the paper classifies the underlying surface of six administrative areas in the main urban area of Xi’an.The natural environment areas were classified using Landsat-8 remote sensing images,and the study areas were divided into six types,namely dense forest land,sparse forest land,low vegetation,natural bare land and water body,and hard paved surface,by the supervised classification method of WUDAPT(World Urban Database and Access Portal Tools).The overall classification accuracy was 72.56%,and the Kappa coefficient was 0.6479,showing good consistency.Buildings vector data was adopted and GIS(Geographic Information System)was used for calculation built-up area,According to the limit value of building density equal to 40%and the stipulations of building height in the uniform standard of civil building design,8 types of underlay surfaces in built-up areas were formed within the research scope.(2)The Land surface temperature and vegetation index data of MODIS(Moderate-Resolution Imaging Spectroradiometer)were adopted.The temperature-Vegetation index method was used to invert and obtain the raster data of near-surface air temperature in the main urban area of Xi’an in summer.A total of 14 sets of temperature data were obtained for 7 days in summer 2019,with two times of 10:30 and 13:30 each day.The mean values were calculated respectively to represent the raster data of the mean temperature of Xi’an city at two times in summer,among which the air temperature distribution in the classification of each underlying surface at afternoon time was greatly different.The air temperature of the built-up area is higher than that of the natural environment,and the temperature change is more stable.The air temperature is lowest in the dense woodland area under natural environment.(3)Through the method of sample selection analysis,the urban form parameters such as Mean Height of Buildings(MHB),Building Surface Fraction(BSF),and Ratio of Green Space(GSP)calculated in the undersurface of various built-up areas.The influence mechanism of urban morphology on air temperature was studied.The results showed that MHB and GSP were negatively correlated with the mean air temperature,while BSF was positively correlated with the mean air temperature,and the correlation was weak in this study.The correlation analysis between land surface temperature and air temperature shows that air temperature and surface temperature are highly positively correlated,and the correlation coefficient R of various underlying surfaces is above 0.649.Multiple regression models were established by taking three types of urban morphological parameters and land surface temperature as independent variables as air temperature.The correlation coefficient R~2 of each underlying surface model was between0.08126～0.35233.The model evaluation results show that the air temperature prediction model in the afternoon of low-density high-rise buildings is the worst,and the air temperature prediction model in the afternoon of low-density multistory buildings is the best.Based on the classification of urban undersurface,this study establishes an air temperature Inference model,which provides a new reference for obtaining air temperature data near the ground.