Step 2. Image Analysis

The purpose of this section is to briefly instruct users who are given a Landsat scene acquired immediately post-fire (that is in reflectance) how to calculate the fraction cover of green vegetation and char.

 <<<< Part 1. Field Data                                                                     

Several common GIS and Image analysis packages now have the capability to produce the fraction cover maps. Their implementation is all very similar:

1. On the image select areas of interest (AOIs) that represent the main surface components that you collected ground spectra of. Therefore in our example areas that represent green, brown, and char.

2. Next in the program interface match-up the ground spectra with those areas of interest.

3. Next, run the algorithm. When doing this you typically have two options: These are 'setting a sum constraint to 1' and 'no negative or >1 fractions'. Although counter-intuitive most people only use the first constraint - this ensures that all fractions within a pixel will sum to 1, however it does remain possible for fractions of individual components to either exceed 1 or be less than zero. The reason for this is due to the fact that the error in the measurement is not well quantified.

Once run, you will produce an image like that shown on the far right. The left image is the false color composite of the burned extent of the Jasper fire and the image in the middle is the dNBR image produced for this fire. This image actually shows the fractional char within each pixel. As you can see like the dNBR it captures the variability within the burned area. Importantly, the fractional measure can directly be compared to the field measures and is transferable across different regions.