Currently, both the model and abundance are read from a MARCS model. We should separate the two, so that they are not necessarily determined by the same file (though you should be able to feed both of them a MARCS model and have it work).
See #33 and #35 as they are relevant here.