Abstract
Impervious surface is a key indicator for monitoring urban land cover changes and human-environment interaction. Although the normalized difference impervious surface index (NDISI) has shown the potential to extract impervious surface areas (ISA) from multi-spectral imagery, it may lack robustness due to the spectral heterogeneity within urban impervious materials and confusion between other land covers. In this letter, a multi-source composition index is proposed to overcome the limitations of the original method. Three data sources: night-time light luminosity, land surface temperature and multi-spectral reflectance are integrated to create a modified NDISI (MNDISI), which aims to enhance impervious surfaces and suppress other unwanted land covers. Experimental results reveal that the MNDISI offers a stable and close relationship with ISA and is shown to be an effective index for mapping and estimating impervious surfaces in heterogeneous urban land cover environment.
Acknowledgements
The authors would like to thank Prof. Tim Warner from West Virginia University and three anonymous referees for their constructive comments.