Using Virtual Datasets in ERMapper

When you are comparing images of different epochs, or seeking to run classification algorithms on ratios and principal components in place of or in addition to raw data, these difference surfaces of data must be combined under one .ers file to be used. The best way to do this is to make a virtual dataset instead of a new dataset. A virtual dataset uses an algorithm to process data from the original files to arrive at the ratios, principal components, etc and uses these data in place of a new set of "bands."

  1. Build the algorithm containing multiple surfaces and layers within surfaces.
    1. each layer should be a "pseudolayer"
    2. a transform can be applied or not depending on your plans for the data analysis later
    3. if you want to see the individual layers or surface, you'll have to move them, or turn them off and on in sequence.
  2. Save the algorithm with a good name ("Building ratios virtual dataset for 1999 image.alg") so you can change it if you need to.
  3. Save As... choosing virtual dataset option. Make sure that "virtual" is part of the name. You'll get this box.

    if you're combining raw bands, you should choose yes (so that the same data that went in, comes out "unstretched"), but for ratios and PC's you can choose either. You can always stretch the data in the new algorithm that makes use of the virtual dataset.

  4. When you open the virtual dataset, you'll see the same layers as above now combined into one dataset on which calculations can be made (differences etc).
    In this case from the file "pc1-3_1999_1984_rec_lin_nn_virtual.ers"