The practitioner must investigate carefully the various parts of the image. Various situations carry some specific radiometric information such as:
wind swept areas,
very dense vegetation,
very dark bareland,
inland pools and lakes,
flow of foul waters into a clear water body (or reverse as is the case here inside the green polygon),
shadowed slopes on land,
Know your image:
the practitioner must investigate carefully the various parts his image.
In this small image subset, it is not possible to get a clear picture of the BPL until a certain central portion is excluded from the area investigated.
It turns out that the area excluded exhibits much brighter bottoms ( the "bump" in the left scatter plot of linearized data for the whole image) than anywhere else in this small image.
Trying to estimate a value for the ratio K 1/K 2from the scatter plot of linearized data for the whole image (left) causes dire embarrassment!
By just excluding the central part of the image through a mask, the "bump" disappears and the K 1/K 2ratio is now easily estimated from the scatter plot of linearized data for the restricted image (right).
Such very bright bottoms are present at all depthsin the whole image of Caicos Bank, but they are restricted to a very limited area inside this image subset