CF compliance

[arango]

ROMS is now following the NetCDF CF compliance more strictly. However, many users are unaware or ignoring it. I modified get_cycle.F to check if the attribute cycle_length is not defined as the same external data type as the time variable, which is always a floating-point (single or double precision) variable. The code now checks it the user defined this variable as an integer. Many thanks to Jerome Fiechter for bringing this to my attention.

I also changed nf_fread2d.F to allow negative _FillValue data which is possible when interpolating input fields. The check now is over absolute values:

          DO i=1,MyNpts
            IF (ABS(wrk(i)).ge.ABS(Aspval)) THEN
              wrk(i)=0.0_r8             ! land/sea maked with _FillValue
            ELSE
              ...
            END IF
          END DO

The station data in def_station.F is now defined as independent access variables during parallel I/O. Many thanks to Kate Hedstrom for reporting this problem.

[kate]

The comparison to _FillValue also happens at lines 411 and 752 in nf_fread2d.F. These should also be changed to account for negative _FillValues.

[arango]

Yes, I missed those. Thank you. I also corrected a typo in checkdefs.F.