Hello to all familiar with the software idrisi Andes or Taiga!
I am using idrisi for my master thesis in which I am trying to detect vegetation and land cover change along a river in Patagonia. I’m working with Landsat data from three different years and I am doing a hard classification of the scenes. As I am only interested in the vegetation between the different braids of the river I excluded the extensively used agricultural area and surrounding steppe vegetation since they showed to strongly influence the classification results. For exclusion, I used a mask with the pixel value zero, while at the same time defining a class “mask” so that this won´t influence the other, important land cover classes. This worked very well when using maximum likelihood for classification but I also did a classification tree analysis which showed the best results. Unfortunately, with this method, some clusters of pixels in my area of interest were also classified as “mask”, especially water bodies and forest. This happens, I assume, because when the classification tree is created, all pixels in channel 2 with a value lower than approximately 0.003 are put into one class with the mask pixels (value=0) as they do not reflect much radiation.
Does anyone have a solution for this problem or know someone who might be able to help? Any help would be very much appreciated.