多尺度城市地表温度降尺度方法

Downscaling urban land surface temperature based on multi-scale factor

针对目前星载热红外传感器的空间分辨率低,无法满足城市尺度的生态环境研究需求的现状,该文选择地表覆盖类型复杂的区域,根据研究区土地覆盖类型,选取归一化植被指数(NDVI)、城市不透水面指数(ISA)、改进的归一化差异水体指数(MNDWI)等因子加入DisTrad模型,采用移动窗口逐步回归统计地表温度和因子的线性关系,利用半方差曲线函数和均方根误差综合确定最优移动窗口的大小,以提高地表温度降尺度精度。研究结果表明:改进的DisTrad模型在地表覆盖类型复杂区域,具有良好的降尺度目视效果,且具有较高的降尺度精度,尤其在低植被覆盖的建筑区、水体区域具有更高的精度。

The spatial resolution of spaceborne thermal infrared sensor is low, which cannot meet the demand of urban scale ecological environment research. In this paper, aiming at the complex land cover type area, according to the type of land cover in the study area, the normalized difference vegetation index （NDVI）, the city impervious surface area （ISA）index and the modified normalized difference water index （MNDWI）were added into the DisTrad model; the size of the optimal mobile window was determined by using the semi-variance curve function and the root mean square error to improve the accuracy of the sur- face temperature drop scale. The results showed that the improved DisTrad model had good visual effect in the complex area of surface coverage type and higher downscaling accuracy, especially in the low vegeta- tion cover area and the water area.