Hi Steve, You're right, of course. The time axis is indeed computed using (S_END - S_START+1)/S_DELTA But the point is that the descriptor file wants the first and last step, in your case 1 and 240 for a total of 20 years of 12 months each. The user does not have to worry about the +1; that is taken care of internally. You have 240 time steps (middle-of-the-month values) and you were thinking of the boundaries between time steps (end-of-the-month). Simple example -- if we had 10 days's data; s_start = 1, s_end = 10, s_delta = 1. There are 10 = (10-1+1)/1 steps, and you would give the descriptor file S_START=1, S_END =10, S_DELTA=1. Anyway, we got you off and going. Ansley Steve Cousins wrote: On Fri, 14 May 2004, Ansley Manke wrote:Your time unit of 2629746 corresponds to monthly data. (1/12 of 365.2425 days). You have S_START = 1, S_END = 239, S_DELTA = 1, I assume this corresponds to 20 years of data. But the time axis is computed using (S_END-S_START-1)/S_DELTA = (239-1) months. This is less than 20 years; and that looks to me like it's why you're seeing the error message, WARNING - NON-INTEGER NUMBER OF STEPS PER STEPFILE INDICATED IN DESCRIPTOR - TRYING TO CONTINUE ANYWAY Following the suggestion of using a DEFINE AXIS command to make the axis that you want, and using its steps to define your axis, if you want: yes? defaxis/t=15-jan-1980:15-dec-1999:1/units=months/t0="15-dec-1979:00:00" taxyes? list t[gt=tax] Then this suggests thatyou want S_END to be 240.Hi Ansley, I gave it a try using 240 and it works! It is a bit non-intuitive (for me anyway) though. Rather than using (S_END-S_START-1)/S_DELTA I would have thought it would be (S_END-S_START+1)/S_DELTA. I have 239 time slices so I would (and did) guess that S_END would be 1 and S_END would be 239. At least it works and it is a very easy fix. Thanks very much, SteveAnsley Manke Steve Cousins wrote:I have a series of 20 netCDF files, each with 12 time slices (except the last file only has 11 time slices). I have a descriptor file to connect these: &FORMAT_RECORD D_TYPE = ' MC', D_FORMAT = ' 1C', / &BACKGROUND_RECORD D_TITLE = 'BIOLOGY', D_T0TIME = '15-dec-1979 00:00:00', D_TIME_UNIT = 2629746.0, D_TIME_MODULO = .false., / &MESSAGE_RECORD D_MESSAGE = ' ', D_ALERT_ON_OPEN = F, D_ALERT_ON_OUTPUT = F, / &EXTRA_RECORD / &STEPFILE_RECORD S_FILENAME = '/neptune/usr2/pacific-g-46-99/bio.cdf.001', S_AUX_SET_NUM = 0, S_START = 1, S_END = 239, S_DELTA = 1, S_NUM_OF_FILES = 20, S_REGVARFLAG = ' ', S_INDEX = ' ' / &STEPFILE_RECORD S_FILENAME = '**END OF STEPFILES**' / In ferret I periodically (every 12 slices) get: ** unknown netCDF error code: -31 Here is an example: [ferret@rocky server]# ferret NOAA/PMEL TMAP FERRET v5.53 Linux(g77) 2.4.x - 09/26/03 14-May-04 13:02 yes? set data "/usr2/fer_dsets/descr/bio-g-80-99.des" WARNING - NON-INTEGER NUMBER OF STEPS PER STEPFILE INDICATED IN DESCRIPTOR - TRYING TO CONTINUE ANYWAY yes? shade/L=1/Z=5/levels=30 co2flx yes? shade/L=12/Z=5/levels=30 co2flx ** unknown netCDF error code: -31 yes? shade/L=13/Z=5/levels=30 co2flx yes? shade/L=11/Z=5/levels=30 co2flx yes? shade/L=24/Z=5/levels=30 co2flx ** unknown netCDF error code: -31 yes? shade/L=36/Z=5/levels=30 co2flx ** unknown netCDF error code: -31 yes? shade/L=48/Z=5/levels=30 co2flx ** unknown netCDF error code: -31 yes? shade/L=47/Z=5/levels=30 co2flx yes? shade/L=49/Z=5/levels=30 co2flx If I access the file directly (not using the Descriptor file) that corresponds to slice 13 I can plot that slice fine. I can't find what error -31 is. Anybody have any ideas? Thanks, Steve ______________________________________________________________________ Steve Cousins, Ocean Modeling Group Email: cousins@umit.maine.edu Marine Sciences, 208 Libby Hall http://rocky.umeoce.maine.edu Univ. of Maine, Orono, ME 04469 Phone: (207) 581-4302 |