Summary
This map shows the locations of 409 vegetation training sites used to provide signatures for supervised classification of satellite images. The map shows the vegetation category at each location. A photograph is attached to each location. This helps in looking at the site in that lab and reaffirming the vegetation category.
Description
The vegetation training sites were taken from all over the Central Himalayas and used for supervised classification of IRS 1C and 1D satellite images. The training sites were taken during extensive field work during the period 1994 to 2000. Some of the locations were used to create a training model and some of the training sites were kept for estimating the errors of classification.
We believe that collecting a large pool of such sites along with the photographs will help in using these sites for other studies as well.
Methodology
The training sites were collected during field surveys in the Central Himalayas during the period 1994 to 2000. During field visits, good representative training sites for different vegetation categories were identified and their locations were marked on a toposheet. Along with this, for every site a photograph was captured and linked to the training site.
In the lab, these training sites were digitized and the photography was examined to affirm the vegetation category. These training sites were randomized and a third of the sites for each category was kept aside. With two-thirds of the training, a supervised classification was performed and the remaining one-third sites were used to estimate the efficacy of the classification. The results of the classification were used to estimate the extent of degredation in the Central Himalayas, along with the errors of classification. The results have been published.
References
R. Prabhakar and B.S. Mehta, 2006. How Degraded are Himalayan Forests? with Current Science, 91(1): 61-67.