中国地表城市热岛驱动因素及其空间异质性

基于MODIS卫星遥感数据，计算了中国284个地级市2018年的年平均地表城市热岛强度，分析了中国地表城市热岛的空间分布规律和空间集聚模式.结合多元遥感数据、气象数据和社会经济统计数据，利用地理加权回归模型分析了日间和夜间地表城市热岛强度主要驱动因素的空间异质性.结果表明，中国地表城市热岛强度的空间分布存在明显的空间自相关性；地理加权模型相比传统的普通最小二乘模型，极大地提高了驱动因素的解释程度，日间和夜间的决定系数分别由最小二乘模型的0.659和0.189提高到了0.876和0.651并且具有更低的残差平方和以及赤池信息准则值，从驱动因素来看，除日间的植被因素对地表城市热岛强度的影响显著为负外，其余因素的影响方向均会随着地理位置的改变存在结构性的差异.总体来讲，日间的地表城市热岛强度受城乡植被差异的影响最大，而在夜间则更容易受社会经济因素的影响.

China's surface urban heat island drivers and its spatial heterogeneity

Based on satellite remote sensing data acquired through Moderate Resolution Imaging Spectrometer (MODIS), not only was the annual mean surface urban heat island intensity of 284prefecture-level cities in 2018 figured out, but spatial distribution patterns and spatial agglomeration models of surface urban heat islands in China were analyzed. Combining multivariate remote sensing data, meteorological data and socioeconomic statistics, a geographically weighted regression model was utilized to analyze spatial heterogeneity in main drivers for surface urban heat island intensity during daytime and nighttime. As demonstrated by relevant results, an obvious spatial autocorrelation existed in spatial distribution of China's surface urban heat island intensity. Compared with the traditional global ordinary least squares (OLS) model, interpretation of the drivers was significantly improved according to the geographically weighted regression model. Moreover, determination coefficients for daytime and nighttime increased from 0.651 and 0.189 in the OLS model to 0.876 and 0.659 respectively. In addition, both the residual sum of squares and the Akaike information criterion were calculated to be lower by the geographically weighted regression model. In terms of the drivers, vegetation placed a significantly negative influence on surface urban heat island intensity during the daytime, while structural differences were proved to exist in directions of influence that was applied by other factors along with geographic position changes. On the whole, surface urban heat island intensity was most significantly affected by differences in urban and rural vegetation in daytime; but at night, it was susceptible to socio-economic factors.