A Comparative Study of Four Spectral Mixture Analysis Methods for Land Surface Composition in a Hilly Coastal City

Abstract

Spectral mixture analysis has been used to monitor and analyze urban surface compositions in recent years.Many of those urban researches select inland or coastal cities on flat plain as study areas,meanwhile water is masked off before spectral mixture analysis,to acquire acceptable results and conform to the conceptual model of urban composition named V-I-S(Vegetation-Impervious Surface-Soil).This paper took Xiamen as an example of hilly coastal cities,compared four spectral mixture analysis methods using Landsat TM image acquired on January 8,2007,and assessed their accuracy through visual interpretation of SPOT 5 image acquired on December 25,2006.Results show that 3-endmember methods masking water off would lead to spectral confusion between dark vegetation and impervious surface in such hilly coastal cites,while 4-endmember methods including water got better results.Spectral normalization could not improve the results remarkably.More tests should be carried out to find the optimal spectral mixture analysis methods for hilly coastal cities like Xiamen.