A Comparison of the Classification of Vegetation Characteristics by Spectral Mixture Analysis and Standard Classifiers on Remotely Sensed Imagery within the Siberia Region

Tan, S.-Y. (2003). A Comparison of the Classification of Vegetation Characteristics by Spectral Mixture Analysis and Standard Classifiers on Remotely Sensed Imagery within the Siberia Region. IIASA Interim Report. IIASA, Laxenburg, Austria: IR-03-020

Preview

Text IR-03-020.pdf Download (785kB) | Preview

Abstract

As an alternative to the traditional method of inferring vegetation cover characteristics from satellite data by classifying each pixel into a specific land cover type based on predefined classification schemes, the Spectral Mixture Analysis (SMA) method is applied to images of the Siberia region. A linear mixture model was applied to determine proportional estimates of land cover for, (a) agriculture and floodplain soils, (b) broadleaf, and (c) conifer classes, in pixels of 30 m resolution Landsat data. In order to evaluate the areal estimates, results were compared with ground truth data, as well as those estimates derived from more sophisticated method of image classification, providing improved estimates of endmember values and subpixel areal estimates of vegetation cover classes than the traditional approach of using predefined classification schemes with discrete numbers of cover types. This technique enables the estimation of proportional land cover type in a single pixel and could potentially serve as a tool for deriving improved estimates of vegetation parameters that are necessary for modeling carbon processes.

Actions (login required)

View Item