Over the past several months I have been working to identify spectral recovery patterns for the 2000 Jasper Fire in the Black Hills region of South Dakota. This has been an enjoyable and challenging experience that has given me the opportunity to develop many new geospatial skills and gain considerable knowledge of fire ecology.
I went into this process thinking I would would extract data, apply analysis methods and get direct results but I have been surprised at how many different methods of analysis and definitions of recovery there are in remote sensing. As an example there is no widely accepted definition of "spectral recovery". Each study I found when researching utilized different spectral recovery metrics and different ways of representing "recovery". There were some similarities and general categories of spectral metrics and recovery variables that were used- many studies utilized some sort of percentage of the pre-fire spectral value to determine recovery patterns. Others used the magnitude of the disturbance when the fire occurred as a way of calculating a slope or a point of recovery.
With the help of my mentors I developed my own process of measuring the recovery of the burn scar. I took the pre-fire value and utilized the 80% of pre-fire spectral value as a "recovery" point, which I had seen used in several different studies. One flaw I found when analyzing the data was that this metric was influenced by climate patterns in a way that could give a "false" recovery reading- there was one year that was extremely wet and had a noticeable increase in "recovered" pixels. There were many pixels that "recovered" for that year and then never hit the recovery value again. To me this was not a true representation of recovery. To account for this discrepancy I chose to represent my data in terms of "Years at Recovery", which was the number of years an area hit the 80% of the pre-fire NBR. I have created a map to show this initial analysis. I will be using this recovery metric to try to identify factors that have influenced recovery, such as topography, climate, and proximity to unburned pixels.
