Study of various types of vegetation cover using high resolution multispectral and microwave remote sensing data
Student name: Mr Anshuman Bhardwaj
Guide: Mr V.S.P.Sinha
Year of completion: 2011
Host Organisation: Maharashtra Remote Sensing Applications Centre (MRSAC)
Supervisor (Host Organisation): Mrs Tutu Sengupta
Abstract: The use of sub-pixel level classification as a tool for spectral library generation for different land
features can be very effective specially when there is availability of high resolution multi-spectral
data. Also, digital classification of high resolution multi-spectral data in itself is very challenging. In
this study, a high resolution land use-land cover map was prepared using digital NDVI based
classification on LISS-4, 5.8 m resolution data as the study area mainly consisted of various types of
vegetation cover (crops, hydrophytes and forest). This map was used for generation of statistics and
the results were more informative and in accordance with the earlier LULC studies for the same area
using LISS-3imagery. Sub-pixel classification was done on the patches identified as crop fields and
hydrophytes (distinguished using texture analysis of ENVISAT ASAR backscatter image) using end
member signature files. Extensive ground truthing was done to develop those signature files as
spectral libraries for different crops and hydrophytes. The spectral libraries can be very useful for
future projects related to crop identification and wetland mapping.
Due to volumetric interaction of Advanced synthetic aperture radar (ASAR) with vegetation and its
sensitivity to the variation in vegetation density, it was used to obtain plant density. However, the
sensitivity of ASAR backscatter to the vegetation density depends upon the frequency, polarization
and angle of incidence at which the ASAR is operated. This study provides comparative evaluation of
the sensitivity of multi-polarized ASAR backscatter to the plant density of Tectona grandis (Teak).
Forest density mapping was performed using ENVISAT ASAR, 12.5 m pixel spacing and 30 m
resolution data and extensive ground truth. The output density images were obtained for both VV and
VH polarizations. The results clearly showed the higher sensitivity of cross-polarized VH backscatter
image towards stem density in forest. The results can be very helpful for the forest department to
implement conservation strategies as the areas surrounding the forest are getting converted in to
scrub and wasteland very rapidly.
Key words: Sub-pixel classification, spectral library, backscatter, polarization, angle of incidence
