Hi - Thanks for the report. I don't see any reason that what Patrick wants to do couldn't be implemented, so we'll look into doing that. As Marco says an axis with more than one point does save with the correct bounds as expected. Ansley Marco Steinacher wrote: Hi Patrick, It seems that ferret does not use the bounds variable because your depth axis has only one point. When I add another one (eg. at z=15 with bounds at z=10 and z=20) the bounds are used correctly. With only one grid cell Ferret apparently assumes a regular grid with boundaries at z=4.5 and z=5.5 (probably in order to have weight=1). When you save your data Ferret is using these bounds values. You can check which boundaries Ferret is using with list zboxlo[g=SOSALINE_GLOBAL],zboxhi[g=SOSALINE_GLOBAL] As a workaround you could extend the z-axis with define axis/z/bounds/unit=meters `SOSALINE_GLOBAL,r=zaxis` = \ {5,15},{-0.000244140625,10.000244140625},{10.000244140625,20} Then, the boundaries are treated as you are expecting. Marco Brockmann Patrick wrote:Hi all, I have noticed that when a boundarie variable is present in a netcdf file, ferret detects this "associate" coordinate variable as a "classic" variable. Consequence in my scripts, the use of yes? shade `..varnames[i=1]` to produce a shade with the first variable found in a file becomes not usable since ferret takes the boundaries variable. I have also saved a file with /bounds qualifier and noticed that boundaries have been changed Initially I have: DEPTH1_1 = 5 ; DEPTH1_1_bnds = -0.000244140625, 10.000244140625 ; After a yes? save/clobber/bounds/file=new.nc SOSALINE_GLOBAL I get a DEPTH1_1 = 5 ; DEPTH1_1_bnds = 4.5, 5.5 ; I do not understand why ferret has changed those values. Note that when I use /bounds qualifier, I also save bounds even when axis is regular. Is there a way to control the write of boundarie variables for each axis (ie qualifiers XBOUNDS, YBOUNDS, ZBOUNDS, TBOUNDS). Download, http://dods.ipsl.jussieu.fr/brocksce/ferret_bugs/OCE_sosaline_global_ave.nc and try http://dods.ipsl.jussieu.fr/brocksce/ferret_bugs/err620_bnds.jnl Tests made with ferret v620 (Linux 32bits) Thanks gor any help on this topic. Patrick |