Comparison of AVIRIS and Landsat ETM+ detection capabilities for burn severity Jan W. Van Wagtendonk, Ralph R. Root and Carl H. Key Presenter Alpana Khairom

Overview: Introduction Study area Methods Results Discussion Conclusion

Introduction Wildfire is one of the most significant forms of natural disturbance impacting a wide range of ecosystems ranging from boreal forests to grasslands Burned forested areas have patterns of varying burn severity as a consequence of various topographic, vegetation, and meteorological factors

Need for information on forest fires “Throughout the past century the main form of forest management in regard to fire prevention and control has been suppression of fire” (Foley) To extend knowledge and to quantify the extent and severity of fires

Previous study Advanced Very High Resolution Radiometer (AVHRR) – South America by Hlavka et al. (1996) to map fire scars in the Brazilian Cerrodo SPOT-VEGETATION – used to detect burned areas in Northern Australia (Stroppiana et al., 2002)

Definitions Severity – the degree to which a site has been altered by fire (National Wildfire Coordinating Group, 1996) NBR (Normalized Burn Ratio) - is a difference ratio, similar to NDVI, but uses Reflectance of TM Band 4 and Band 7. NBR = (R 4 - R 7 ) / (R 4 + R 7 ) dNBR (differenced Normalized Burn Ratio) - The temporal difference between the pre- and post-fire NBR values CBI (Composite Burn Index) – is a field of measure of burn severity that improves quantification and comparative study of fire effects

Objective To determine whether the dNBR calculated from Landsat ETM+ data provides optimum signals of burn severity, independently validated by AVIRIS

Study area

Methods EROS USGS developed dNBR data from pre- and post-fire Landsat TM images NBR ETM+ =1000[(R 4 -R 7 )/(R 4 +R 7 )] dNBR ETM+ =NBR pre – NBR post Yosemite Fire Effects Crew performed burn severity validation the field using the CBI method of measuring the magnitude of change in the vegetative community from pre-fire conditions 63 CBI plots were sampled inside the burn perimeter

Plot locations High severity – complete combustion of litter, duff and small logs, the mortality of small sized trees and combustion of the crowns of large trees Moderate severity – mortality of small trees and the consumption of the crowns of medium and large trees Low severity – burning of fine fuels and some understory trees Unburned – accumulations of woody and litter fuels, thick understory trees and low live branches

2 AVIRIS images were recorded on August 17, 2001 and July 22, 2002 Data collected by the ETM+ sensor and by the AVIRIS sensor were used to compare burn severity NBR AVIRIS =1000[(R 47 -R 210 )/(R 47 +R 210 )] dNBR AVIRIS =NBR pre – NBR post

Results

High severity areaModerate severity area

Low severity areaUnburned area

Discussion Both ETM+ and AVIRIS datasets showed changes in reflectance The two AVIRIS channels were very close to the ETM+ sensor suggested that ETM+ bands 4 and 7 were capturing very dynamic bandwidths reacting to fire AVIRIS 17m resolution is 3.1 times more improved over Landsat 30m

Conclusion Similarities in burn severity detection capabilities exist between AVIRIS and ETM+ Due to high cost of data acquisition and complexity of the mission, use of AVIRIS may be limited AVIRIS data acquisitions are eventually archived for access by users other than initial principal investigators