A Study of ERS-1 SAR and Landsat-4 TM Synergism for Forest Cover Studies
要旨Merging SAR data with imagery from different parts of the electromagnetic spectrum provides an almost infinite range of possible combinations to be explored. We examined the synergism between radar and optical dataset using various digital image processing techniques in a test site of Western Ghats, India. This study evaluated the advantages of combining traditional spaceborne optical data (Landsat4 TM) from the visible and infrared wavelengths with the longer wavelengths of radar (ERS1 SAR). The analysis involved speckle suppression of SAR (C-VV) data using Gamma MAP filter, optimum band selection for extraction of forest cover information, supervised classification, principal component analysis and IHS transformation. SAR data, when subjected to 9x9 Gamma MAP filters provided more meaningful joint distribution of grey level values. The triplet combination of blue, NIR and SAR possessed highest Optimum Index Factor of 31.8. Most of the forest type classes could be distinguished on TM images and the addition of SAR image ameliorated the accuracy. Improved classification accuracy for degraded forest (1%) and tropical semi-evergreen forest (0.7%) was observed could be that could be attributed to the incorporation of the structural details into the classification exercise. PC1 and PC2 were found to be highly correlated with optical and radar channels respectively. Physiographic and hydrographic features were significantly highlighted when SAR data was transformed as intensity and saturation components; supporting the fact that SAR provides more terrain details. It is apparent from the preliminary study that the effect of operations in terms of feature identification and interpretation would depend on the combinations of radar and optical systems.