Case Study of Camp Fire Employing Novel Metric for Time Series Analysis
of Vegetation Recovery
Abstract
Abstract—Wildfires are a major global issue, costing the United States
71.1 to 347.8 billion dollars annually (Graham, 2020). Rising global
temperatures have increased the frequency and intensity of wildfires
(Cuevas-Gonzales, 2009). Climate change has thus created a need for new
methods to examine the effects of wildfires. In this research, we
evaluated how the 2018 Camp Fire environmentally impacted land cover in
California, and the extent to which the area’s land cover suffered from
long-term damage. Our hypothesis was that the affected land was damaged
significantly, but recovered partially by the end of the investigated
period. In order to assess the healing of a patch of land burnt in the
Camp Fire, corrected reflectance images of the patch collected from NASA
Worldview were analyzed using Python to return each image’s Pixel
Greenness Value (PGI)–an original metric developed by our team that
analyzes an image’s color content to return a numerical value
corresponding to vegetation health. These values were then plotted. Over
the course of 26 weeks after the Camp Fire, the patch of land partially
regained its PGI. Major recovery occurred between weeks 7 and 15. We
concluded that the area burnt in the Camp Fire only partially recovered,
as the moving average of the PGI value only reached 81% of the baseline
value by the end of the investigated period. Our findings demonstrate
the environmental damage that wildfires can cause and the potential of
PGI as a useful metric for assessing the impact of wildfires. Future
studies could compare PGI results with those of the Normalized
Difference Vegetation Index (NDVI), known for its usage in such
vegetation recovery analyses. The case study could also repeat the
experiment with Landsat data taken from the United States Geological
Survey website, to account for the lack of atmospheric correction in
NASA Worldview data. Key words: extreme event, wildfire, land cover,
environmental science, programming