耦合夜光遥感的粤港澳不透水面提取及其热岛效应研究

Extraction of Impervious Surface in Guangdong,Hong Kong and Macao Bay Coupled with Nighttime Remote Sensing and Analysis of its Heat Island Effect

不透水面作为反映城市化进程及城市生态环境的重要指标,也是影响城市热岛效应的重要因素,是城市化研究中的重要数据源。当前通过卫星遥感反演不透水面是一种主要的方法,数据源及反演机理的不同,也给不透水面提取带来了差异。因此,本文利用VIIRS夜光数据与Landsat8等多源遥感数据,基于LISI指数法反演了2015年粤港澳湾区四个城市不透水面,并与线性光谱混合法(LSMA)进行了对比,同时结合高分不透水面产品进行了不透水面地表类型、空间结构等方面的深入对比分析;再结合反演的地表温度,分析城市热岛与不透水面的关系。主要结论如下:①基于LISI指数表征的不透水面,与线性光谱混合法反演的不透水面相关性R2约0.62,但与耦合大数据及机器学习模型的不透水面产品相关性较高,R2=0.8882。LISI指数法能较高精度挖掘不透水面密度及其时空特征。②不透水面、地表温度空间特征均与下垫面性质密切相关,强不透水面与强热岛均分布在建筑地表,而弱不透水面与冷岛则往往分布在植被地类附近。且不透水面与地表温度呈明显正比例关系。研究表明,LISI指数表征的粤港澳湾区不透水面提取具有较高的精度,可高效耦合多源数据,为进一步提高不透水面提取精度及相关应用提供科学依据。

As an important index reflecting the process of urbanization and urban ecological environment,impervious surface is an important factor affecting urban heat island effect and also an important data source in urbanization research.The inversion of impervious surface through satellite remote sensing is a main method.The different data sources and inversion mechanism also bring differences to the extraction of impervious surface.In this paper,the impervious surface of four cities in Guangdong,Hong Kong and Macao Bay Area in 2015 is retrieved by using VIIRS nighttime light data,Landsat 8 and other multi-source remote sensing data based on Large-scale Impervious Surface Index(LISI),and compared with linear spectral mixing method(LSMA).Meanwhile,we make an in-depth comparative analysis on the surface type and spatial structure of impervious surface based on high-resolution impervious surface products.In the end,the relationship between urban heat island and impervious surface is analyzed combined with the retrieved surface temperature.The main conclusions are as follows.Firstly,The impervious surface represented by LISI method has a correlation of R2=0.62 with the result by LSMA method,but has a higher correlation with the impervious surface products coupled with big data and machine learning model,R2=0.8882.LISI method can mine impervious surface density and its temporal and spatial characteristics with high accuracy.Secondly,The spatial characteristics of impervious surface and surface temperature are closely related to the properties of underlying surface.Strong impervious surface and strong heat island are distributed on the building surface,while weak impervious surface and cold island are often distributed near vegetation types.And the impervious surface has an obvious positive proportional relationship with the surface temperature.The research shows that the extraction of impervious surface in Guangdong,Hong Kong and Macao Bay area characterized by LISI model has higher accuracy and can efficiently couple multi-source data,which provides a scientific basis for further improving the extraction accuracy of impervious surface and related applications.