Chapter this study can be used and tested to

Chapter 1:       
summary and conclusion

1.1       Introduction

We Will Write a Custom Essay Specifically
For You For Only $13.90/page!

order now

In the era of climate change and shifting of global
attention towards cheap and deep mitigation technologies, the forest carbon
holds a big promise. With digital format, coverage over vast
areas, synoptic view, repetitivity, stabilised and predictive data as
attributes of remote sensing data and vast geographical extent as an
essential attribute of forests, there is no alternative to remote sensing and
GIS technologies in forest carbon assessment. Remote sensing
and GIS can surmount the issues of inaccessibility and assist in scaling up
spatial distribution.

 But as the present research has
shown, the remote sensing based forest carbon assessment models developed for
one place cannot be simply transferred to other locations. This is going to be much
more challenging in case of high biodiversity mountainous terrain which
introduces huge geometric, spectral and radiometric distortions. The
model presented in this study can be used and tested to assess aboveground
biomass and carbon by other states, national or international agencies who are
interested in knowing the standing forest resource and carbon stock. It would at the same time be very useful for
forest managers when they plan to tap the emerging carbon finance opportunities.

1.2       Summary

The total above ground
forest carbon in Sikkim forests is to the tune of 29.46 Million tons. Montane  Wet Temperate Forest  contain the highest quantity of carbon to the
tune of 8.46 Million tonnes. lowest average carbon density of 12.03 tons/ha
is found in Moist Alpine Scrub while the least carbon (1.03 Million tonnes) is
stored in moist alpine scrub.

Spectral ratio
(Vegetation indices) and Gray Level Coherence Matrix based texture metrics
models do not yield satisfactory results, at least with moderate resolution
optical sensors, in high biodiversity mountainous forests. There is need for
research in finding other remote sensing variables which when combined with
these models, can give acceptable results.


The initial sampling plan
was made as per the NDVI values but later on Forest Density and Forest Type
based modeling had to be used for assessment of carbon. This sampling has caused
an asymmetric distribution of count of plots among forest type and density
classes. Even though four density classes were available, they had to be
aggregated to two classes due to insufficient data in all density classes.

For Landsat,  the data availability for the relevant period
was that of Landsat 7 but   all the data
were Landsat 7 ETM+ SLC-off data when the Scan
Line Corrector (SLC) had failed. These products have data gaps.  Landsat 7 ETM+ SLC-off inputs acquired after
May 31, 2003 are not gap-filled in spectral indices production.  For
Landsat  the Landsat 8 data had to be

Spectral indices
products may have increased uncertainties, inherited from the Surface
Reflectance source data, in areas where atmospheric correction is affected by
adverse conditions like in  snow-covered
regions and  areas with extensive cloud


forest types and forest density maps referred to different years and therefore
there may not always one to one pixel correspondence.

All uncertainties and errors associated with the ground samples viz.  height– diameter allometry, stem volume
equations and wood-density estimates
are attached to the model.

could be inherent locational inaccuracies in the GPS.

Future Pathways:

is a need to quantify high biodiversity induced pixel noise/contamination .
These can be researched upon and can later be integrated into remote sensing
based carbon model. 

biomass is a three-dimensional metric, precise estimation requires biophysical
measures addressing horizontal (e.g., canopy density/cover) and vertical (e.g.,
canopy height) structural characteristics of the vegetation. New technology
such as airborne laser scanning (LiDAR) has recently been introduced. There is
a need to carry out hierarchical modeling in multi-sensor environment which
includes optical, lidar, hyper spectral and microwave remote sensing, either
as standalone or in combination depending upon their usability, availability
and accuracy.

Understanding the major
determinants of uncertainty can also be a powerful tool for improving
methodology and the accuracy of the resulting estimates.