Fire Severity and Vegetaon Recovery in Yellowstone Naonal Park, Wyoming, USA Introducon In August 1988, Yellowstone Naonal Park had a series of devastang wild fires that were considered the largest forest fire burn event in the recorded history of the naonal park. (Scullery, 1989). The fire started as a series of smaller fires which were exacerbated by increasing winds and drought which burned for several months. Thousands of firefighters and military personal fought the fire but it was not unl early October that the fire was brought under control by cool and moist weather condions. This project aims to evaluate burn severity following the forest fire along with the vegetaon re- covery response 23 years later. This process was done through the use of re- mote sensing technology which has been proven to be an accurate tool to evaluate forest fires and ecological recovery. Natural-like Rendion of Yellowstone Naonal park Aſter the Fire on October 10, 1988 Figure 1. This image was created with the band combinaon of 7,4,2 which provides a natural-like ren- dion while also penetrang smoke and atmospheric parcles. Red indicates a recent forest fire, bright green areas indicate healthy vegetaon, pink represents barren soil, and blue areas represent water. Burn Severity Assessment Using Normalized Burn Rao in 1988 NDVI Change No Change Increase (Moderate) Increase (High) NDVI Change and Vegetaon Recovery of Severe Burned areas as of 2011 No Change Increase (Moderate) Increase (High) Percent of High severity Area 0.24% 53.60% 46.16% Table 2. NDVI Change between 2011 and the days aſter the fire in 1988 Methodology Three images for Yellowstone Naonal Park were acquired from Landsat 5. The first image was acquired for September 22, 1987, which was a year before the wild fire started. The second image was acquired for October 10, 1988 which was eight days directly aſter the fire. The third image was ac- quired for September 24, 2011 which was 23 years aſter the fire. The dates were chosen as close to the day and month of each other as possible to avoid any spectral variances due to the me of year. To analyze the severity of the forest fire, Normalized Burn Rao (NBR) was used. NBR ulizes the short-wave infrared bands which are not affected by dust, smoke and atmospheric parcles (Avery & Berlin, 1998; Eva & Lamb- in, 1998). Cocke et al. (2005) states that several studies have concluded that using this band combinaon provides the highest accuracy for burn severity analysis. The NBR algorithm was applied on the pre-fire and post-fire images and then a change in NBR was calculated through an image differencing al- gorithm. The unsupervised classificaon was then combined with the change in the NBR map in a GIS soſtware to create a fire severity index seen in Figure 3. To analyze the restoraon of the damaged forest, change detecon using normalized vegetaon index (NDVI) was used. NDVI gives an indicaon of the amount of green vegetaon and is effecve in detecng vegetaon re- covery aſter a fire event (Delgado et al., 2003). Change in NDVI values were examined in high severity burned areas since those areas experienced com- plete vegetaon loss. Results and discussion Burned area was classified based on how much damage the forest canopy sustained and was characterized by four categories: low severity, moderate- low severity, moderate-high severity and high severity (Figure 3). With a rela- vely even distribuon of percent burned areas in each category (Table 1), it becomes clear that certain areas of forest were affected by the fire to a different degree. Further ground based research could indicate what tree species or environmental factors contribute to each severity category. Within the high severity burned areas, NDVI change detecon revealed that 53.6% of this area experienced a moderate increase in vegetaon while 46.16% experienced a high increase in vegetaon. This indicates, without a doubt, that the forest within high severity category is recovering to some de- gree. Low Severity Moderate- Low Severity Moderate- High Severity High Severity Percent of Burned Area 21.61% 22.46% 32.77% 23.16% Table 1. Fire Severity Established Through Normalized Burn Rao References Avery T. E, Berlin G. L .(1992) ‘Fundamentals of remote sensing and air photo interpretaon.’ (Prence Hall: Upper Saddle River, NJ) 472 pp. Cocke, A. E, Fule, P. Z, & Crouse, J. E. (2005). Comparison of burn severity assessments using differenced Normalized Burn Rao and ground data. Internaonal Journal of Wildland Fire Vol 14, 198-198. Delgado, D, Lloret, F, & Pons, X. (2003). Influence of fire severity on plant regeneraon by means of remote sensing imagery. Internaonal Journal of Remote Sensing Vol 24, No. 8, 1751-1763. Eva H, Lambin E. F. (1998) Burnt area mapping in Central Africa using ATSR data. Internaonal Journal of Remote Sensing 19, 3473–3497. Schullery, P. 1989. The fires and fire policy. Bioscience 39: 686-695. 1988-2011 NDVI Composite Figure 2. This image combines both NDVI images from October 10, 1988 and September 24, 2011 to a show change in NDVI values . Bright cyan indicates an increase in vegetaon, bright red indicates a decrease in vegetaon, black represents water, grey indicates barren soil, light red indicates forests that have no change and light cyan indicates grassland that have not increased in vegetaon values. Fire Severity High Severity Moderate-high Severity Moderate-low Severity Low Severity 0 30 60 15 Km 0 30 60 15 Km 0 30 60 15 Km 0 30 60 15 Km Figure 3. Figure 4. More Informaon Data source: Landsat 5 Adriano Nicolucci -Poster presented for paral fulfillment of The Professional Geographer (GEO871).