Assessment of Post-Fire Vegetation Recovery in Washington State Using Landsat and Geographical Data

Abstract

Consequences of wildfires often result in the loss or change of vegetation, causing a reduction of biodiversity and an increase in soil erosion. Studies aiming to understand the potential dynamics in vegetation regeneration after a fire can benefit restoration programs by defining probable contributing factors. This report considered environmental variables and their impact on fire recovery for six fires in Washington State over a five-year period. Variables included the differenced normalized burn ratio (dNBR), the normalized difference vegetation index (NDVI), land cover type, and topological variables. Regression modeling was performed using both Ordinary Least Squares (OLS) and Geographic Weighted Regression (GWR) to determine the best-fit model. Over the five-year period, mixed vegetation showed the highest recovery rate with varied rates for both forested and desert vegetation. OLS regression demonstrated that land cover had high multicollinearity with other variables and land cover factors, thus it was excluded from GWR calculations. The best-fit models revealed a positive relationship with pre-fire NDVI and burn severity for most fire locations, indicating an increase in revegetation based on an increase in burn severity. Topological variable slope had both positive and negative relationships with NDVI. R2 values calculated through GWR were between 0.85 and 0.98. As Washington State is a diverse, widespread area, this study serves as an initial step to understand the potential relationships between fire recovery and the contributing factors. Additional steps should be taken to focus on specific vegetation type and assessing longer recovery time.