how to clearly define step 1 point shape input
Just in the first step (step 1 load reference data) I stuck because the user is asked to load a point shape for reference data or reference data extraction (as I can read in the documentation). Here we need to clearly distinguish between two cases A)the user wants to extract reference data on a current image time series using current point shape (this is the utils step) OR B) the user already have a complete point shape with reference time series saved as columns in the *.dbf and wants to apply this to another scene in the past or the future. We need to realize that a fresh user needs to be pointed to that decision: Do you want to mark some points and see the class distribution of them in a map OR do you want to see the distribution in the past, then please first provide or extract a current point shape, somehow (this somehow could be a barrier for a fresh user).
By the way the point shape does not extract reference data, it provides the spectra to compare with the target classes in the respective time series stack. So in order to have a good comparison of shapefile spectra with other scenes we need a good initial time series for spectra extraction. In this regard I do not understand which time series the utils function provides? This is confusing and leads me to the first question (A or B).
To understand the user's benefit: If the user once produced the point shape or reference table (with good temporal resolution), it can be applied to any other time step for which no reference data exist. When I start step1 this connection is not obvious for me and the documentation is a bit misleading.