The Yarnell Hill wildfire, ignited by lightning lasted from June 28 to July 10, 2013. It is a high-risk, dynamic environment in Arizona. The total death toll of 19 crew members of the Prescott firefighters was reported. It was one of the deadliest wildfires in US history after the Los Angeles Griffith Park fire in 1933 (National Weather Service).
In this study, we addressed the burn severity by using the pre, and post-fire remotely sensed data. Various indices based on NBR, NDVI, spectral response, and slope are evaluated for the 2013 Yarnell Hill Fire. The time series analysis of vegetation regeneration over the 4 years is also carried out in Google Earth Engine (GEE). The study shows a burned area of approximately 31.96 km2. Near-short infrared data (Band 7) show an increased reflectivity during the fire event. However, the landscape recovery is indicated by strongly increasing NDVI values over the time span of 4 years. Moreover, areas with combined high burn severities and high slopes are identified, leading to debris flow potentials. Therefore, this assessment confirms the effectiveness of using NBR, NDVI, and dNDVI methods for assessing fire events.
Using the Landsat 8 bands that are most sensitive to post-fire reflectance changes: band 4 (0.64 0.67 mm), band 5 (0.85-0.88 mm), and band 7 (2.11-2.29 mm). After a fire, reflectance in the visible (band 4) and near-short infrared (bands 5 and 7) region increases. To capture this information, The NDVI combines Red (band 4) and NIR (band 5) data. Whereas, the NBR combines the NIR (band 5) band with a SWIR2 (band 7). The NBR has become the standardized spectral index for determining burn severity, particularly in the United States. In band 7, the fire-induced reflectance increase is expected to be more obvious than in band 5. As a result, an index with band 7 instead of band 5 can capture a wide range of post-fire effects variation.
The band 7 reflectance is also plotted in the GEE environment by using Landsat 8 imagery from 2013 to 2014. The B7 reflectance graph illustrates the drastic increase in reflectance during a fire in July.
The long-term time series analysis of NDVI is carried out in GEE by using MODIS imageries from 2013 to 2017. The graph shows lower NDVI during the fire event and an increased post-fire NDVI. The NDVI graph shows an overall increasing trend from July 2013 to 2017.
In conclusion, overall results show a moderate to high severity of the 31.96 km2 affected area. In addition, the moderate to high severity and slope areas that are at increased risk of mass movements. Based on the drainage network and the “Burn Severity – Slope” map, identifies potential flow pathways for mud and debris flows. According to the stream network in the area, the flow pathways indicate a potential threat to a settlement 1.6 km south of the Granite Mountain Hotshots Memorial. Above all, by distinguishing between unburned and burned pixels based on their NBR pre/postfire difference values, the dNBR appears to be a good approach to quantifying burn severity (dNBR). (Escuin et al., 2008) found that the NBR was better than the NDVI at distinguishing burns and different degrees of severity.