| Abstract | With the emergence of importance on wetland value, there are continuous efforts to conserve wetlands and to use & manage them wisely. Remote sensing is an efficient tool for the assessment and monitoring of natural resources. Wetlands are lands with mixture of water, vegetation and soil. Individual wetland plant species and surface materials occur as sub-pixel components. The low accuracy of landcover classification in wetland is largely attributed to the mixed pixel problems. Mixed pixels have the potential to be spectrally resolved and classified using sub-pixel processing techniques that can be distinguish surface materials smaller than the spatial resolution of the sensor. Linear spectral Mixture Analysis (LSMA) is one of the most often used methods in handling the spectral mixture problem. The main objective of this study was to evaluate the usefulness of the Linear Spectral Mixture Analysis for wetland land cover classification. Linear Spectral Mixture Analysis is an image processing method, which assumes that spectrum measured by a sensor is a linear combination of the spectra of all components within the pixel. In the Linear Spectral Mixture Analysis, spectral endmemers were driven from pure features in the imagery, using the scatter-plotters. A series of endmember models were developed based on green vegetation, soil and water endmembers which are the main indicators of wetlands. In Spectral Mixture Analysis, endmember selection is a crucial step. The selection of suitable endmemers often evolves an iterative process, i.e., selecting initial endmembers, refining these endmembers, evaluation fraction images, and further refining endmembers. Finally, selected endmembers should be independent of each other. The Linear Spectral Mixture Analysis decomposes the image spectra into a series of endmember fractions and calculates the proportion of different endmembers for each pixel, which is defined by user, and provides as a result so-called fraction images. In this study, fraction values from the fall image have very distinctive characteristics, which make it easy to discriminate land-use classes. In spring time, wetland condition being in initial growth and development, make it difficult to identify exact wetland areas from similar phonological cycle of forests. For the type classification of wetland in class level, finer spatial and spectral resolution imagery should be considered. The availably of high spatial resolution imagery to Spectral Mixture Analysis technique make it easy to extract wetland information in the mapping level of plant community. Despite the limitations, this study indicates that Linear Spectral Mixture Analysis adopting sub-pixel classification algorithm improved wetland classification and this routine will be helped to detect wetlands in regional scale. Phenologically, a fall image is appropriate to identify wetlands. from various land-covers |