Index Development And Validation For Urban Impervious Surface Area Mapping

Accurate understanding of the spatial distribution and dynamics of impervious surfaces area(ISA)is critical for urban planning and management,urban ecological environments,urban hydrological cycles,and biodiversity.Remote sensing images have been widely used in urban ISA mapping due to their fast,wide-ranging and reproducible advantages.The methods of extracting ISA based on remote sensing image are continuously proposed,especially the remote sensing spectral index method has attracted much attention because of its clear physical meaning and ease of use.However,the current ISA index still does not effectively solve the inherent heterogeneity of the ISA spectra and the external miscibility between the ISA and the soil.In this paper,the spectral data of the US Geological Survey(USGS)and the Advanced Spaceborne Thermal Radiation Reflectance Detector(ASTER)are selected as the research samples,and the Landsat-8 OLI and WorldView-2 images are used as experimental data.Two new indices were constructed: The Perpendicular Impervious Surface Index(PISI)and the Enhanced Impervious Surface Index(EISI).The main work is as follows:(1)Construction of the ISA index.In this paper,V-I-S is used as a conceptual model to analyze the spectral of major components of the city according to the samples in the USGS and ASTER spectral libraries.The results show that the impervious surface,the spectra of soil and vegetation are significantly different in the near-infrared and blue bands,and the PISI index is constructed accordingly.In view of the fact that the PISI index is affected by the water content of the soil,EISI replaces the blue-NIR combination in the PISI by the blue-SWIR combination based on the fact that the blue band is least affected by the water content and the SWIR band is the most affected.The EISI index can effectively increase the difference between the ISA and the soil,and can greatly inhibit the influence of soil water content on the ISA extraction accuracy.(2)Evaluation of the ISA index.In this paper,the mixed pixel decomposition method is used to simulate the threshold of the new indices,and then the correlation between the index value and the proportions of each component in the pixel and the coincidence between the index threshold and the ISA proportion are analyzed.Finally,the two new indices are compared with the existing indices from the aspects of separability and extraction accuracy.A comprehensive evaluation method for the impervious surface extraction index is given.Using Landsat-8 and Worldview-2 images,four cities of Fuzhou,Xi'an,Xining and Wuhan were used as research areas to verify the accuracy and applicability of the newly constructed two indexes.The conclusions are as follows:(1)PISI has a significant positive correlation with the proportion of ISA in pixels.The recommended threshold [0.0098,0.1462] obtained in the simulation experiment can be used in actual urban scenes.Compared with NDBI and BCI,PISI has obvious advantages in the precision and separability of ISA in different geographical environments and different resolution images.It is more prominent in areas with more soil.(2)There is a significant negative correlation between the EISI and the proportion of ISA in the pixels.The threshold [0,0.46] obtained in the simulation experiment is consistent with the EISI value distribution on the Landsat-8 images.The extraction accuracy and separability of EISI in different geography environments are better than PISI.(3)The PISI and EISI proposed in this paper have excellent extraction ability for ISA,but the two indices have their own characteristics: PISI is characterized by its required blue and NIR bands is very common in sensors and has a wide range of applications,but it is necessary to remove the water in advance,and the result is susceptible to the moisture content of the soil.The characteristic of EISI is that it has stronger separability to the ISA,soil and vegetation,and has a certain inhibitory effect on the soil water content,so it has better extraction precision for the ISA and does not need remove the water in advance.However,the SWIR band is not common in sensors,thus its application range is limited.