Beschreibung:
The natural landscape is fast turning into impervious surfaces with the increase in urban density and the spatial extent of urbanized areas. Remote sensing data are crucial for mapping impervious surface area (ISA), and several methods for ISA extraction have been developed and implemented successfully. However, the heterogeneity of the ISA spectra and the high similarity of the ISA spectra to those of bare soil in dry climates were not adequately addressed. The objective of this study is to determine which spectral impervious surface index best represents impervious surfaces in arid climates using two seasonal Landsat-8 images. We attempted to compare the performance of various impervious surface spectral Index for ISA extraction in dry climates using two seasonal Landsat-8 data. Specifically, nine indices, i.e., band ratio for the built-up area (BRBA), built-up area extraction method (BAEM), visible red near infrared built-up index (VrNIR-BI), normalized ratio urban index (NRUI), enhanced normalized difference impervious surfaces index (ENDISI), dry built-up index (DBI), built-up land features extraction index (BLFEI), perpendicular impervious surface index (PISI), combinational biophysical composition index (CBCI), and two impervious surface binary methods (manual method and ISODATA unsupervised classification). According to the results, PISI and CBCI combined with the manual method had the best accuracy with 88.5% and 88.5% overall accuracy (OA) and 0.76 and 0.81 kappa coefficients, respectively, while DBI combined with the manual method had the lowest accuracy with 75.37% OA and 0.56 kappa coefficients. PISI is comparatively more stable than the other approaches in terms of seasonal sensitivity. The ability of PISI to discriminate ISA from soil and vegetation accounts for much of its good performance. In addition, spring is the ideal time of the year for mapping ISA from Landsat-8 images because the impervious surface is generally less likely to be confused with bare soil and sand at this time of year. Therefore, this study can be used to determine spectral indices for studying ISA extraction in drylands in conjunction with binary approaches and seasonal effects.