建设强度的变化会对城市热岛效应产生不同方向或强度的影响。选用Landsat-8遥感影像，以地表温度热岛强度（简称为热岛强度）作为城市热岛效应的量化标准，以片区和连续序列空间尺度作为基本分析单元，使用空间自相关方法和空间自回归模型代替经典统计学方法，对热岛强度空间分布特性进行研究，探究热岛强度与建设强度的空间响应机制；并基于分析结果对研究区域进行分区，探究合理的城市热岛效应缓解措施。结果表明：(1)研究区域热岛强度具有显著的空间自相关性，且随着分析尺度的精确自相关性愈发明显；“热点”区域主要分布在西南部的工厂密集区和中部的城市中心区，“冷点”区域则以五象岭和青秀山风景保护区为主。(2)建设强度直接或间接影响了城市热岛效应的形成和演化；100 m规则格网尺度的空间特性表现最显著，可作为常用分析尺度；作为回归模型的解释变量，各建设强度指标的显著性在不同尺度呈现较大差异。(3)空间误差模型的拟合效果优于普通最小二乘法模型，空间尺度越精确、模型的拟合效果越好。(4)基于热岛强度“热点”和“冷点”区域进行进一步研究，得出基于不同分区的缓解思路和连接破碎化“冷点”区域，打造城市“冷廊”等热岛缓解措施。

The change of construction intensity can have impact on urban heat island effect direction or intensity. This paper selected Landsat 8remote sensing images, took the surface urban heat island intensity(referred to as heat island intensity) as urban heat island effect quantitative standard, and area and continuous sequence scale as basic analysis unit. This paper used spatial autocorrelation methods and spatial autoregressive models instead of classical statistical methods to study the spatial distribution characteristics of urban heat island intensity, explored the spatial response mechanism of urban heat island intensity and construction intensity, and partitioned the study area based on the analysis results to explore reasonable measures to mitigate urban heat island effect. The results show that:(1) the surface urban heat island intensity of the study area has significant spatial autocorrelation. With more precise autocorrelation analysis scale, the autocorrelation becomes more obvious. The "hot spots" areas are mainly distributed in the factory-intensive areas in the southeast and the central urban center district, and the "cold spot" areas are mainly in Wuxiangling and Qingxiu Mountain Scenic Reserve.(2) The construction intensity directly or indirectly affects the urban heat island effect formation and function. As the regression model explanatory variables, the construction intensity indicator significance and correlation vary greatly at different scales.(3) The fitting effect of the spatial error model is better than that of the ordinary least squares model. The more accurate the spatial scale, the better the model fitting effect.(4) Based on further research based on the urban heat island intensity in the "hot spots" and "cold spots" areas, the mitigation ideas based on different city districts, construction of urban "cold corridors" by connecting fragmented "cold spots" areas and other heat island mitigation measures were proposed.