论文部分内容阅读
以厦门岛为研究区,以CBERS-02B的CCD影像为数据源,采用基于可变端元的线性光谱混合模型估算了城市不透水面组分含量,并探讨了该方法的实现过程与优势。通过端元评估确定了研究区的4个典型端元,即高反射不透水面、低反射不透水面、高反射土壤和植被。在此基础上,以高、低反射不透水面端元的组分含量对城市不透水面含量进行估算。精度评价结果显示:基于可变端元的方法要优于一般带全约束法;而在混合像元分解过程中加入全色波段(band5)有助于提高模型估算精度,使得在像元尺度的精度与采用Landsat的已有报道相近,而在土地利用单元尺度实现了对城市不透水面的无偏估计。研究实例也表明,尽管目前CBERS-02B数据在辐射定标和地理定位等方面还有待改进,通过采用适当的处理过程和技术手段,依然能利用该数据对城市不透水面进行有效估算。
Taking Xiamen Island as a research area and CCD image of CBERS-02B as data source, the city water-permeable component content was estimated by using a linear mixed spectral model with variable terminal elements. The realization process and advantages of this method were also discussed. Four typical end-points of the study area were identified by endmember assessment, that is, highly reflective impervious surfaces, low-reflective impervious surfaces, highly reflective soils and vegetation. On this basis, the content of impervious surface of city is estimated by the content of the components of high and low reflectance end-elements. The results of the accuracy evaluation show that the method based on the variable endmember is better than the generalized all-constraint method, while the addition of the full-color band (band5) in the mixed pixel decomposition helps to improve the model estimation accuracy, The accuracy is similar to what has been reported with Landsat, and an unbiased estimate of the impervious surface of the city is achieved at the land-use unit scale. The research also shows that although CBERS-02B data still needs to be improved in radiometric calibration and geolocation, the data can still be used to effectively estimate the urban impervious surface by adopting appropriate processing and technical means.