基于地理探测器下的川西高原地表温度空间格局及影响因子分析——以西昌市为例

以西昌市为例，选取2010年和2015年 Landsat系列遥感影像，在“3S”技术的支持下，采用监督分类中的最大似然法，并结合Google Earth高分辨率影像和GPS野外实地验证数据获取西昌市2010年和2015年土地利用信息；再利用大气校正法反演获得西昌市2010年和2015年地表温度，并对反演结果中的高温异常区进行实地调研验证；最后，采用地理探测器定量分析了坡度、总辐射、坡向、海拔、年均降水量、年均气温、植被类型、土壤类型以及土地利用类型9个影响因子对地表温度影响的解释力。结果表明：（1）地表温度空间分布存在显著差异。（2）影响因子对地表温度的影响存在差异，其中海拔和年均气温对地表温度的影响较大，而总辐射影响最小。（3）不存在单一因素或是单一性质的因素影响地表温度。不同影响因子之间存在交互作用，对地表温度的影响是相互增强或非线性增强的。（4）部分影响因子对地表温度具有显著性差异影响，并且地表温度均值最大时，对应着影响因子不同的范围或类型。

Geographical exploration of the spatial pattern of the surface-temperature and its influencing factors in western Sichuan-Plateau:A case of Xichang City

In this paper,the 2010 and 2015 Landsat series satellite images of Xichang City,Sichuan,China have been used.To obtain the 2010 and 2015 land use information of Xichang City,the maximum likelihood method of the supervision classification,the Google Earth high resolution images,and the GPS field verification data were utilized with the help of the “3S” technology.Next,the 2010 and 2015 surface temperature of Xichang City was inversely studied by the atmospheric correction method,and the high-temperature anomaly areas were verified by the field investigation.Finally,the geographical detector technique was adopted to quantitatively analyze the influence of nine factors,such as slope,total radiation,aspect,elevation,average annual precipitation,average annual temperature,vegetation type,soil type,and land use type on surface temperature.The results showed as follows:(1) the difference of the LST spatial distribution is remarkable.(2) the effects of the influencing factors on LST are different; the influences of altitude and annual temperature are more obvious while the total radiation has the least impact.(3) the factors that influence LST interact with each other.In general,LST is enhanced linearly or nonlinearly by the interaction of influencing factors,and (4) the influencing factors have significant differences based on surface temperature.For the highest mean value of surface temperature,the ranges or types of influencing factors will be different.