(2) Supervised classification: A total of 102 training sites were created
for various landscape elements using information from
204 ground truthing plots and 517 ground truthing points collected
during field survey. In addition, the vegetation maps from
French Institute (Pascal et al. 1982) were referenced for certain
areas. Based on signatures files created from these training sites,
the remaining image pixels were classified using Maximum
Likelihood Classifier algorithm (Lillesand and Kiefer 2000). This
classification delineated different LSEs for the entire range.
Their areas were then calculated separately for each forest division.