Automated Burned Area Delineation Using IRS AWiFS satellite data

Abstract. India is endowed with a rich forest cover. Over 21% of country’s area is covered by forest of varied composition and structure. Out of 67.5 million ha of Indian forests, about 55% of the forest cover is being subjected to fires each year, causing an economic loss of over 440 crores of rupees apart from other ecological effects. Studies carried out by Forest Survey of India reveals that on an average 53% forest cover of the country is prone to fires and 6.17% of the forests are prone to severe fire damage. Forest Survey of India in a countrywide study in 1995 estimated that about 1.45 million hectares of forest are affected by fire annually. According to Forest Protection Division of the Ministry of Environment and Forest (GOI), 3.73 million ha of forests are affected by fire annually in India. Karnataka is one of the southern states of India extending in between latitude 110 30' and 180 25' and longitudes 740 10' and 780 35'. As per Forest Survey of India's State of Forest Report (SFR) 2009, of the total geographic area of 191791sq.km, the state harbors 38284 sq.km of recorded forest area. Major forest types occurring in the study area are tropical evergreen and semi-evergreen, tropical moist and dry deciduous forests along with tropical scrub and dry grasslands. Typical forest fire season in the study area is from February–May with a peak during March-April every year, though sporadic fire episodes occur in other parts of the year sq.km, the state harbors 38284 sq.km of recorded forest area. Major forest types occurring in the study area are tropical evergreen and semi-evergreen, tropical moist and dry deciduous forests along with tropical scrub and dry grasslands. Significant area of the deciduous forests, scrub and grasslands is prone to recurrent forest fires every year.

In this study we evaluate the feasibility of burned area mapping over a large area (Karnataka state, India) using a semi-automated detection algorithm applied to medium resolution multi spectral data from the IRS AWiFS sensor. The method is intended to be used by non-specialist users for diagnostic rapid burnt area mapping.