Ethan Gutmann

This seems more critical before substantial additional development ramps up so that we can easily test a variety of configurations and confirm updates don't break anything... will work on this...

On selection: I agree, that zero-precip days should be left in when selecting a collection of analog days. If a single analog day is being sampled from that collection, then...

After #66 files are nearly (might be?) CF-compliant. Should be run through a CF-checker

Almost trivial I suspect, but it still depends on NR and I don’t want that dependency in GARD. Once that is ironed out I would prefer to have it as...

agreed

This could incorporate David Pierce's EOF Fortran library http://meteora.ucsd.edu/~pierce/eof/eofs.html

locality option should be able to specify how many days on either side of the current day should be included (at least for analog options, that gets really difficult with...