Modern fire suppression has the potential to alter the natural distribution of forest cover types and age classes, which has consequences for resource management. A 21,000 ha prescribed burn to enhance wildlife habitat and secondarily reduce continuity of coniferous fuels was conducted in 1998 in the boreal forest of eastern interior Alaska. Our objectives within the burned area were >50% mortality of trees and spatially-varying burn severity to promote regeneration of deciduous trees and shrubs by root or crown sprouting and by seed germination on mineral soil.

After no precipitation the previous 8 days, aerial firing with incendiary plastic spheres was accomplished within prescription parameters on 21 July (observations: relative humidity 33%, wind 18 km/hr, Duff Moisture Code 49.5, Drought Code 246; calculations: Initial Spread Index 9.7, Buildup Index 67.3, Fire Weather Index 25.2). Most of burning occurred within a few hours of ignition, and 14.7 mm of precipitation fell the next day. Planning and implementation costs for the fire specialists, excluding wildlife staff time for planning, was $1.10/ha for the burn.

We have begun conducting a change detection of cover types at the landscape scale to evaluate the burn treatment. DigitalGlobe^TM Quickbird imagery (2.6 m multispectral / 0.6 m panchromatic) from September 2002 was ortho-rectified to create a base map for ortho-rectifying other digital images used in the study. Color-infrared aerial photos were digitally scanned for both pre-burn conditions (1:63,360-scale, summer 1981 or 1983) and post-burn conditions (1:12,670-scale, July 2002).

PCI Geomatics software is being used for the ortho-rectification of the imagery, and eCognition^© software will be used for classification of the Quickbird satellite imagery and aerial photos. This project will compare the accuracy and the time involved in classifying imagery vs. hand-type mapping of digitized aerial photos. It will also compare classification of woody broadleaf types between Quickbird imagery and digitized aerial photographs. Ground verification of the classified cover types is planned for summer 2003. Once the vegetation classification is completed, change detection will be preformed to analyze vegetation type changes. We will then incorporate a digital elevation model to examine the effect of slope, aspect, and elevation on vegetation changes caused by the fire.