Prior-knowledge-based spectral mixture analysis for impervious surface mapping |
| |
| Institution: | 1. Department of Geological and Atmospheric Sciences, Iowa State University, Ames, IA 50011, USA;2. Joint Global Change Research Institute, Pacific Northwest National Laboratory, College Park, MD 20740, USA;3. Maryland Instutute for Applied Environmental Health, School of Public Health, University of Maryland, College Park, MD 20740, USA;4. Aerobiology Research Laboratories, Nepean, ON K2E 7Y5, Canada |
| |
| Abstract: | In this study, we developed a prior-knowledge-based spectral mixture analysis (PKSMA) to map impervious surfaces by using endmembers derived separately for high- and low-density urban regions. First, an urban area was categorized into high- and low-density urban areas, using a multi-step classification method. Next, in high-density urban areas that were assumed to have only vegetation and impervious surfaces (ISs), the vegetation–impervious model (V–I) was used in a spectral mixture analysis (SMA) with three endmembers: vegetation, high albedo, and low albedo. In low-density urban areas, the vegetation–impervious–soil model (V–I–S) was used in an SMA analysis with four endmembers: high albedo, low albedo, soil, and vegetation. The fraction of IS with high and low albedo in each pixel was combined to produce the final IS map. The root mean-square error (RMSE) of the IS map produced using PKSMA was about 11.0%, compared to 14.52% only using four-endmember SMA. Particularly in high-density urban areas, PKSMA (RMSE = 6.47%) showed better performance than four-endmember (15.91%). The results indicate that PKSMA can improve IS mapping compared to traditional SMA by using appropriately selected endmembers and is particularly strong in high-density urban areas. |
| |
| Keywords: | Impervious surface V–I–S Spectral mixture analysis Prior-knowledge |
| 本文献已被 ScienceDirect 等数据库收录！ |
|
