利用多源遥感数据进行城市不透水面覆盖度估算

Estimating Urban Impervious Surface Percentage with Multi-source Remote Sensing Data

利用深圳实验区4种不同传感器获取的遥感数据,通过CART算法进行城市ISP估算。讨论了多光谱遥感数据的不同波段在ISP估算中的重要性,比较了针对三种不同中分辨率影像建立的ISP估算模型在性能上的差异。实验结果表明,近红外波段对ISP估算结果的贡献最大,具有较高空间分辨率和成像辐射质量的遥感影像得到的估算结果精度较高,所有的估算结果均在实际ISP分布范围的两端分别存在着高估和低估的现象。

In this article large-area impervious surface mapping is carried out within the Shenzhen urban area by employing the CART（classification and regression tree） method with remote sensing data acquired by 4 satellites.The experimental results show that near-infrared（NIR） band is of most importance for ISP estimation.Another conclusion is that higher estimation accuracy can be obtained with multispectral imagery of higher spatial resolution and better radiometric quality.However,an unfavorable phenomenon can be consistently observed from all results that over-estimation and under-estimation do exist around the lower and upper bound of actual ISP value range respectively.