Hi Kris -
I'm replicating your results, and this looks to be related to
Inquiry 1252, where Aggregate Grids By Time data type doesn't work if the navigation/dimensions aren't the same between all of the files. As an FYI, this appears to be a 'limitation' with NetCDF itself. Looking at Unidata's
NcML Aggregation page, there's a note under JoinExisting (which is what we are doing here, aggregating the files around an existing time dimension):
...All the other variables, dimensions, and attributes are taken from the first file that declares them.
I wrote up a NcML wrapper file using joinNew and the behavior matches the Aggregate Grids By Time data type. Doing a ncdump on the NH files, there are 4 different sizes:
nlines x pixels.....times
2452 x 3263.......1, 2, 3, 4, 5, 7, 10, 11, 13, 14, 15, 16
2714 x 3600.......6, 12
1606 x 3267.......8
892 x 3160.........9
When Aggregate Grids By Time is used, the dimensions from time 1 are applied to all of the timesteps. If you just create a loop of the aggregated NH times, there are a few things you will notice:
- Time steps 1, 2, 3, 4, 5, 7, 10, 11, 13, 14, 15, and 16 will be correct.
- Time steps 6 and 12 will display, but their nav will be shifted.
- Time steps 8 and 9 won't display at all
This is all related to the dimensions of time step 1 being applied across the board to all time steps. One workaround I've found thus far is to load each file individually (or select all of them and load with the 'Grid files (netCDF/GRIB/OPeNDAP/GEMPAK)' data type). This will create individual data sources for each time step. At this point, you can use the 'Miscellaneous > Make a time sequence from single time grids/images'. This formula does require you to individually select the field you want to display in each file so it is a bit cumbersome, but it does create a good loop where everything displays and there's no incorrect navigation shift.
Thanks -
Bob Carp