Effectiveness of ALOS AVNIR-2 imagery for vegetation mapping

Abstract

This paper examines the effectiveness of ALOS AVNIR-2 imagery for the application of vegetation mapping. Two classification methods were applied; the maximum likelihood method as a pixel-based analysis and the object-based classification. The imagery taken by Landsat ETM was also analyzed for comparisons. The vegetation map produced by the object-based classification method using ALOS AVNIR-2 was most similar to the existing vegetation map by the aero photo interpretation. The accuracy assessment also showed that the vegetation map produced by the above method had the highest accuracy (0.606) in terms of Kappa coefficient. It can be considered that the higher spatial resolution (10m) of AVNIR-2 contributes to the more accurate detection of land cover objects, and the map can be used as a base map for vegetation survey.