The underlying logic of this approach is based on the 
high reciprocality of bare soil status and vegetation 
status. Therefore, by combining both B1 and AVI in the 
analysis, it becomes possible to estimate the ground 
status on a continuum ranging from vegetatively rich 
conditions to exposed soil conditions. The BI is 
formulated with medium infrared information. In 
addition to vegetation type and status, the unique 
characteristic of forest is its three dimensional structure. 
To extract information on this shadow index (SI) is used 
through extraction of the low radiance of visible light. 
This approach isolates vegetation feature space 
using advanced vegetation index (AVI) and bare soil 
index (BI). The vegetation feature space is further 
stratified using shadow index (SI) on the basis of texture 
variation introduced by the canopy shadow of the forest 
stand. 
The Landsat TM bands (except band 6) are 
normalised using linear transformation. The temperature 
calibration using coefficients for Landsat 5 was done to 
estimate relative ground temperature. The temperature 
data has only been used to separate soil and non-tree 
shadow. The spectral dataset is subjected to physical 
transformation using enhancement techniques (Anon. 
1993). The vegetation feature space data was stratified 
based on the ‘texture’ of the data as influenced by the 
canopy shadow. Finally a rule-based logic is 
implemented to achieve land cover and forest density 
classification. The results can be improved by using 
water mask from Landsat TM band 7. The entire 
approach is packaged in the form of semi-expert system 
for wider application under ITTO project (Fig. 2). 
3.2 Vegetation Type Mapping and Monitoring 
Changes 
Forest vegetation types are classified based on 
physiognomy, structure, function and composition and 
also on height classes and distinction of woody tissue. 
Champion and Seth (1968) classified Indian forest based 
on physiognomy, climate, succession and ecological 
status. Satellite remote sensing data presents above 
information on integrated manner and has been used for 
vegetation stratification by several authors (Houghton 
and Woodwell, 1981; Botkin et al, 1984; Roy, 1993). 
The different studies carried out in India demonstrates 
that satellite remote sensing can stratify forest based on 
following criteria: 
• Phenological types as a function of leaf duration 
(e.g. Evergreen, Semi-evergreen, Moist deciduous 
and Dry deciduous) 
• Major communities and gregarious species (e.g. 
Sal, Dipterocarpus, Pine, Teak, Bamboo. Oak, 
Deodar, etc.) 
• Vegetation types of unique environmental setup 
(e.g. Mangroves, Sholas, Riverine, Alpine pastures, 
etc.) 
• Canopy closure expressed as forest density (e.g. 
encroachments, shifting cultivation and different 
density levels) 
Since vegetation is the indicator of environmental 
condition, its continuous monitoring acts as the 
watchdog for the environment. Remote sensing provides 
information on any positive and negative change in 
vegetation cover unambiguously, being above all the 
levels of socio-economic or political bias. Hence remote 
sensing has the potential to act as a legal tool to 
overcome all the problems/difficulties & socio-economic 
conflicts. 
The basic premise of the change detection through 
remote sensing is that the spectral signatures change 
commensurate with the change in the land cover (Roy et. 
al., 1996). Superimposition of two period maps to find 
the change is an established procedure. Digital change 
detection methods involve more time on computer, for 
the response and changes in ground limit its accuracy. 
Though it is possible to identify classes on both images, 
scene dependent spectral change detection viz. image 
differencing, image rationing, principal component 
analysis, image regression etc. 
The medium resolution Wide Field Sensor (WiFS) 
is available in IRS-1C/1D and IRS -P3 satellites. WiFS 
has also demonstrated its ability in classifying forest 
type, land use/land cover, and estimating agricultural 
production in Indian subcontinent. It is well tuned for 
monitoring vegetation status and dynamics. 
Improvement over existing accuracy of 80% using 
careful analysis and combination of multidate data has 
been suggested (Roy et al., 1995). The approach to 
stratify forests using phenology as discriminant using 
WiFS holds a great promise. 
3.3 Deforestation Monitoring 
The amount of vegetation loss/deforestation due to 
encroachment can be estimated by the use of remote 
sensing technique. The impact of slash and burn during 
and after the ‘jhumming’ operations is clearly visible 
from remote sensing imagery. The representative 
relationship between the population density and the 
percent of forest cover provides information about the 
rate of deforestation and thereby helps in formulating the 
mitigation plan. 
42