Classification of wetland vegetation by texture analysis methods using ERS-1 and JERS-1 images

Images obtained by the C-band (5.3 GHz, 5.7 cm) VV-polarized ERS-1 SAR and the L-band (1.275 GHz, 23.5 cm) HH-polarized JERS-1 SAR were analyzed for the classification of wetland vegetation types. Both scenes were obtained when the wetland vegetation was at the maximum biomass stage. The authors applied texture analysis method using the cooccurrence matrix to classify the vegetation of the Kushiro mire. The result of the classification was evaluated comparing with the result using the SPOT data. As a result, it is shown that several texture features of SAR images are useful for the wetland vegetation classification. The authors especially demonstrate that the bog and the fen vegetation in the wetland can be segregated using JERS-1 image, and also the swamp forest can be delineated from the fen vegetation using ERS-1 image.<<ETX>>