| Surface temperature is one of the important indicators of urban ecological environment.By studying its differentiation characteristics and formation mechanism,it can provide decision-making basis for urban planning and construction.This paper selects the downtown area of Chengdu as the research area,uses the Radiative Transfer Equation model,Mono-window Algorithm model and General Single-channel model to retrieve the land surface temperature of remote sensing data in different time phases,and uses Exploratory Regression Analysis model,Ordinary Least Square model and Geographically weighted regression model to analyze the relationship between factors and surface temperature.At the same time,combined with frequency ratio method and variable rate method,the influence and driving effect of factors are discussed on the formation of land surface temperature,and the following main understandings are obtained:(1)The results of different inversion algorithms show that the Radiative Transfer Equation method is more suitable for Landsat TM data.The inversion of Landsat OLI/TIRS data was suitable for Mono-window Algorithm method.Although the general adaptive single channel method is simple in calculation,it has a large error.Compared with the temperature measurements,the Radiative Transfer Equation method has an inversion error of 1.34 °C,and the inversion effect is the best.The difference between the Radiative Transfer Equation method with the General Single-channel method? Radiative Transfer Equation method shows that the Radiative Transfer Equation method showed small plaque distribution and difference between the General Single-channel,the difference is the smallest in the urban construction area.Compared with the Radiative Transfer Equation method and Mono-window Algorithm method,there are larger patches and denser areas in different regions,which are mostly distributed in other areas besides urban building areas.(2)Analysis of multi-correlation model shows that there are different degrees of significance between geography factors(independent variables)and land surface temperature(dependent variables).There is a very significant correlation between land surface temperature with Temperature? Rainfall? Normalized Vegetation Index(NDVI)? Normalized Building Index(NDBI)? Gross Social Product(GDP);land surface temperature is a general significant correlations wth wind speed and a weak correlations with soil moisture.The Ordinary Least Square regression model show that the geographic factors can explain 59.7% of the geothermal variation in the region.Compared with Ordinary Least Square regression model,the Geographically Weighted Regression model improves the goodness of fit of each geographic factor,and the goodness of fit among factors can be increased to 0.616.(3)The maximum similarity method is used to divide the pixel values of each factor from low to high into 1-5 intensity intervals,and then the FR index of each interval is obtained and compared by combining the frequency ratio method.The results show that: the maximum contribution intensity interval of population density and GDP density to ground temperature migrates from the maximum grade intensity interval to the low grade intensity interval,and its intensity value also shows a decreasing trend;Normalized vegetation index and normalized building index change little every year,normalized vegetation index is negatively correlated with land surface temperature,normalized building index is positively correlated with land surface temperature;soil moisture and land surface temperature contribute the greatest contribution at 5 grades in each year;Rainfall contributes the greatest contribution at 3 grades in each year,and increases and decreases significantly before and after 3 grades;Wind speed decreases the ground temperature most weakly when it is in grade 1.The phenomenon of land surface temperature decreases obviously during the increase of wind speed(4)The driving mechanism of the formation of abnormal high temperature area is analyzed by the correlation of explanatory variables,FR index and variability.it is found that the increase of land surface temperature in the Shuangliu District,Wenjiang District,Qidu District,Xindu District,Qingjiangbai District and Longquanyi District are mainly caused by the increase of building density and the decrease of vegetation.The geothermal phenomena in Jinniu District,Chenghua District,Jinjiang District,Wuhou District and Qingyang District are mostly distributed between the Second Ring Road and the First Circumcity Expressway.The formation of geothermal differentiation pattern is mainly related to road,building density and vegetation coverage;wind speed and rainfall can alleviate the increase of land surface temperature,but it can reduce the land surface temperature when the rainfall reach a certain intensity.(5)Combining with the differential pattern and driving mechanism of urban land surface temperature,it is suggested that in terms of spatial optimization of urban land use,control measures should be taken to reduce the growth rate of land use in new urban areas,strictly control the spread of urban land,and Protect and maintain the continuity and integrity of natural landscape pattern in Chengdu.In the aspect of ecological space optimization,urban layout should be rationally planned and Urban Ventilation corridor should be constructed.Form the urban spatial structure which is easy to ventilate and cool down,effectively alleviate the heat island effect of Chengdu urban area;take reasonable and effective measures to optimize the layout of urban land use function,update the spatial structure of old urban area and build ventilation and cooling system. |
PDF Full Text Request