Abstract:
Information regarding spatial distribution of different crops in a region
of multi-cropping system is required for management and planning. In the present
study, multi dated LISS-III and AWiFS data were used for crop identification. The
cultivable land area extracted from the landuse classification of LISS-III image
was used to generate spectral-temporal profile of crops according to their growth
stages with Normalised Difference Vegetation Index (NDVI) method. The
reflectance from the crops on 9 different dates identified separate spectral behavior.
This combined NDVI image was then classified by Fuzzy C-Mean (FCM) method
again to get 5 crop types for around 12,000 km2 area on Narmada river basin of
Madhya Pradesh. The accuracy assessment of the classification showed overall
accuracy of 88 % and overall Kappa of 0.83. The crop identification was done for
one entire Ravi season from 23 October 2011 to 10 March 2012.
