IRS 1C LISS-III sensor data was used to generate a medium scale vegetation cover map. Four scenes with minimum cloud cover were acquired, pre-processed and geo-referenced to Survey of India (SOI) topomaps. The satellite images were then subjected to knowledge-based hybrid classification. A standard forest vegetation / land cover classification legend was used for this purpose. All the vegetation classes were visited on ground to collect information on their structure and composition, which was utilized in the classification exercise. Total land cover of over 20000 km2 of Subansiri Himalaya was classified into seventeen categories. The vegetation classes derived from digital classification were compared with the existing ground-based forest classification given by Champion and Seth. Area estimates were made for various land cover categories. Distribution of various forest vegetation types when compared with altitudinal zones of the area has shown good relationship. Correspondence using field-gathered GPS points for vegetation classes showed 89.25% overall accuracy. The methodology used here for classification exercise has contributed to improved classification accuracy. All the vegetation classes have been described with respect to their dominant species composition, spectral response on satellite images, occurrence zone with respect to altitude & climate and their correspondence with existing ground-based forest type classification given by Champion and Seth. This study envisages the use of satellite remote sensing and its kindred technologies like GIS and GPS supplemented by ground-based limited field survey for characterizing forest vegetation cover.
No takes yet. Share an insight, caveat, or question.
Behera et al. (2001) studied this question.