Dear all
my goal is to create a map showing the occurrence of shifting cultivation (a type of farming). The final product should be binary (i.e. 1 = Shifting Cultivation Yes, 0 = No).
I have training data (in the same area where my raster data is located) that already has the binary information (1 or 0) and different explanatory variables (raster-datasets).
So far I have been using the forest based classification and regression tool but I am stuck on an error (not enough variability within an explanatory raster).
Is there a solution that the rasters (which seem to have too little variability for the function) can be adjusted so that this function can still be performed?
If not, is there an alternative classification method?
FYI: I am not working with spectral information, but with raster data that has been calculated based on spectral information, then morphed into binary data and finally Landscape Metrics (shannon div. index, etc.) have been calculated on top of it. I need these rasters (my independent variables) to calculate the final map.
Do you have any idea how to classify this?
Thank you
