Step 1. Field Data

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.                                      

                                                                                                                                        >>>> Step 2. Image Analysis

The first steps to calculate the fractional cover of green vegetation and char within a Landsat or other similar image is to (1) measure the cover on the ground and (2) record what the reflectance of the green and char surface are like with a detailed field measurement device. An example of how we go about measuring (1) is shown to the upper right and the output of (2) is shown to the lower right.

To measure fractional cover in the field, we use ocular estimation of the % cover of material within a quadrat. These quadrats are sampled across the variability of pre and post-0fire conditions.

To measure the reflectance of the surfaces we use a device called a spectroradiometer. It sampled the reflectance of the surfaces every 1 -3 nm and produced curves as is shown here on the right.

The image analysis method we then apply is very simple. It relies on the assumption that the reflectance of a pixel is a (linear) mixture of the reflectances of the components that make up that pixel. e.g., for a pixel made up of exactly half green and half soil we would have:

pixel reflectance = green reflectance * 0.5

                            + soil reflectance * 0.5

Or simply "one plus the other and divided by two"

In a similar manner to unmix the pixel reflectance into % green, % brown, and % black we assume that:

pixel reflectance = green reflectance * unknown %

                       + brown reflectance * unknown %

                       + char reflectance * unknown %

                       + error terms 

This method has been widely applied in the fire-remote sensing literature and for more information please look at:

    Cochrane MA, Souza CM (1998) Linear mixture model classification of burned forests in the Eastern Amazon. International Journal of Remote Sensing 19, 17, 3433-3440.

    Smith AMS, Lentile LB, Hudak AT, Morgan P (2007), Evaluation of linear spectral unmixing and ∆NBR for predicting post-fire recovery in a N. American ponderosa pine forest, International Journal of Remote Sensing, in press PDF Link

    Wessman, C.A., Bateson, T.L., and Benning, T.L. 1997. Detecting fire and grazing patterns in tallgrass prairie using spectral mixture analysis. Ecological Applications 7(2): 493-511.