Study On The Scale Effects Of The Relationships Between Urban Heat Islands And Land Surface Parameters

In order to explore the basic characteristics of the spatial distribution of the Urban Heat Island(UHI)and the variation laws of the relationships between urban heat islands and land surface parameters at different spatial resolutions,this thesis takes Chongqing as an example and focuses on the systematic and in-depth research in the scale effects of the relationships between urban heat islands and land surface parameters,and the conclusions of this thesis are as follows:1.For purpose of improving the accuracy of the built-up quick mapping,this thesis proposed an Improved Double-window Flexible Pace Search(IDFPS)approach.Experimental results showed that the IDFPS approach was more intuitive and comprehensively superior to the Double-window Flexible Pace Search(DFPS)approach.The overall accuracies and Kappa coefficients of the IDFPS approach were improved by at least 4.30%and 0.12,respectively,and the commission errors and omission errors of the IDFPS approach were reduced by at least 8.74%and 2.26%,respectively.2.The spatial distributions of urban heat islands are consistent with the spatial distributions of the buildings.The urban heat island of the study area mainly consists of three relatively smaller urban heat islands,which is mainly distributed in the built-up areas between Zhongliang mountain and Tongluo mountain;the study area has a plurality of heat island centers,which are mainly distributed in some areas with high energy consumption and dense population,such as industrial areas and airport.And the heat island effect is not clear in densely built areas which close to the Yangtze River and Jialing River.3.This thesis had focused on the in-depth research in the scale effects of the relationships between urban heat islands and land surface parameters.Firstly,the Soil-adjusted Vegetation Index(SAVI),the Index-based Built-up Index(IBI),the Soil Brightness Index(NDSI)and Land Surface Temperature(LST)were aggregated from30 m to 60 m,120 m,240 m,480 m,960 m,and the relationships between UHI(LST)and land surface parameters(SAVI,IBI and NDSI)were established by the Geographically Weighted Regression(GWR)model.Then,the advantages of the GWR models over the Ordinary Least Squares(OLS)models were analyzed and compared.Last,the scale effects of the relationships between UHI and land surface parameters by using GWR model were analyzed further.Experimental results showed as follows:(1)The multi-variation model had a better fitting result than single-variation models.In single-variation analysis,both NDSI and IBI had positive correlations with LST,and SAVI had a negative correlation with LST.This indicated that the built-up area and the bare land had the function of aggravating the urban heat island effect,while the vegetation had the function of alleviating the urban heat island effect.(2)Both the local coefficients and the localR~2 of the GWR models were changing with the spatial location,and this indicated that the relationships between urban heat islands and its land surface parameters in Chongqing exhibited significant spatial non-stationarity.(3)The multi-variation GWR models of urban heat island at different scales were analyzed and found that with the increase of the spatial resolution,the local differences of the spatial distributions of the local R~2 and the residuals of the multi-variation GWR model were gradually reduced,the degree of the spatial autocorrelation of the local residuals of the multi-variation GWR model was gradually enhanced,and the performance differences between the multi-variation GWR model and its corresponding OLS model were gradually reduced.This indicated the GWR model became increasingly global which revealing the relationships with more generalized geographical patterns,and then spatial non-stationarity in the relationship tended to be neglected with the increase of the spatial resolution.