Fuzzy Classification of Mediterranean Type Forest Using Envisat Meris Data

The aim of this study was to classify Envisat MERIS and Landsat ETM satellite sensor imagery using fuzzy classification techniques such as, linear mixture modelling and artificial neural networks. The images were classified successfully using these two techniques. The fuzzy results were more accurate then hard classification. Landsat ETM imagery was classified using maximum likelihood classifier and the output was resampled to 300 m to produce test data. Land cover classes comprised agriculture, bare ground, Tukish pine (Pinus brutia), Crimean pine (Pinus nigra), Lebanese cedar (Cedrus libani), Taurus fir (Abies sp.), Juniper (Juniperus sp.) and water. The classification accuracy was poor as there was insufficient number of training pixels available for these classes. As a result of this, overall accuracy was considered to evaluate the potential of these techniques. Overall results of soft classification from linear mixture modeling and artificial neural network and hard classification were 80%, 78% and 57% respectively. It can be concluded that soft classifiers particularly ANN for classifying Mediterranean type forest cover has a great potential. Additionally, there is no significant difference between soft classification outputs for certain land cover classes from linear mixture modeling and artificial neural networks, however artificial neural networks tackled pixels with high degree of mixing more accurately than LMM.