基于HJ-1B的城市热岛季节变化研究——以北京市为例

Study on Seasonal Variation of Urban Heat Island with HJ-1B Satellite Imagery:A Case Study of Beijing

基于中国的HJ-1B卫星反演地表温度,通过建立一种2.5维高斯表面模型来对北京城市热岛建模,定量描述热岛的位置、形状和大小,并用这种模型来描述城市热环境及其季节变化。研究发现,北京市热岛效应季节变化明显,夏季市中心形成热岛且强度和范围最大,冬季城区中心变成冷岛。在2009年夏季,北京市热岛区域东西横跨33.92 km,南北横跨40.92 km,面积约为1 090.14 km2。相对的,在冬季,北京市区温度比郊区低,从而形成一个冷岛。研究表明HJ-1B卫星是城市热环境监测中的一个良好的数据源。

This article presents an urban heat island retrieval and analysis method with HJ-1B satellite imagery. The study area is Beijing, one of the largest cities in the world, which has experienced a rapid urban expansion over the last two decades. Accelerated urbanization has created serious impacts on its thermal environment. Retrieval and quantitative analysis of UHI patterns in Beijing is significant to understand the worsening urban climate problem in the area. More than 200 HJ-1B images of the study area were acquired and analyzed. Then 48 least-clouded scenes were selected to generate LST maps. The digital number （DN） values of the HJ-1B TIR band were converted from their recorded DN to spectral radiance. In order to perform the extraction, temperature images are created by masking out all areas of clouds and open water. To eliminate atmospheric interference on HJ-1B images, atmospheric correction of the red and near infrared reflectance band of HJ-1 image was carried out using ENVI FLAASH atmospheric correction module. The wavelength of the TIR band of HJ-1B is similar to the thermal band of TM5, and the spectral response curves of the two TIR bands matched well. Because HJ-1B has only one suitable thermal band to retrieval LST, so we adopted single-channel algorithm to retrieve LST from TIR band of the HJ-1B satellite. Land surface emissivity （ε） can effectively transmit thermal energy across the surface into the atmosphere, which was obtained through the NDVI thresholds method. A 2.5-dimensional Gaussian surface approximation was modeled to quantitatively describe the location, shape and size of the heat island, and to describe the urban heat environment and its seasonal variation. The algorithmic procedure for UHI analysis was performed in IDL （Interactive Data Language）. The input data of UHI extraction were the HJ-1B LST imagery, as described above. The average suburban LST plus a threshold value （1℃） were regarded as potential heat island pixels. It was found that the daytime UHI demonstrated distinctive seasonal variation, with the maximum during summer and the minimum during winter. A series of 2.5-D model were created to analyze the UHI and its seasonal variation. Through monitoring and analyzing the UHI effect, the results illustrated that the spread of the UHI in Beijing extended for 33.92 km in longitude direction, 40.92 km in latitude direction and 1 090.14 km2 in footprint area in summer, respectively. The urban area surface temperature was lower than the rural area temperature in winter, which has become an urban cooling island. On the basis of the accurately retrieval and analysis of Beijing UHI with HJ-1B satellite imagery, it was concluded that the HJ-1B TIR image is an ideal source for UHI study.