[ncl-talk] unexpected white contours in color filled plot
Mark Branson
mark at atmos.colostate.edu
Mon Aug 22 12:02:29 MDT 2016
Hi Mary.
Thanks for the update. I’ll keep up my debugging efforts on this end and let you know if I discover any work-arounds or fixes.
Mark
> On Aug 21, 2016, at 10:48 PM, Mary Haley <haley at ucar.edu> wrote:
>
> Hi Mark,
>
> [I'm posting back to ncl-talk here.]
>
> Thanks for providing the data file. I was able to reproduce the problem, and unfortunately I'm not sure about a work-around at this time.
>
> That's not entirely true: if I don't reverse the Y axis, then the bug goes away. But, with a pressure/height plot, you want to reverse the Y axis. I tried a few things, like reversing the leftmost dimension of your data (that didn't work), and reversing the Y axis myself and then just using gsn_csm_contour (that didn't work either because then a different contour showed up white).
>
> I have filed a bug report using your small data file and a modified version of your script. The bug report is NCL-2478 for your reference. I'm sorry I couldn't find a work-around. Hopefully this isn't a difficult bug to fix.
>
> --Mary
>
>
> On Fri, Aug 19, 2016 at 10:49 AM, Mark Branson <mark at atmos.colostate.edu <mailto:mark at atmos.colostate.edu>> wrote:
> Hi Mary.
>
> Wow, what a cool debugging tool. Attached is the resulting mary_debug.nc <http://mary_debug.nc/> file. If need be, I can definitely upload the original data file (uandps.nc <http://uandps.nc/>). That file size is around 1GB.
>
> Thanks!
> Mark
>
>
>
>> On Aug 19, 2016, at 10:41 AM, Mary Haley <haley at ucar.edu <mailto:haley at ucar.edu>> wrote:
>>
>> [Offline for a bit.]
>>
>> Yeah, this must point to the coordinate somehow, but this is a little odd.
>>
>> Would you be able to provide the data? I think if you set this unadvertised resource before you call gsn_csm_contour:
>> res at gsnDebugWriteFileName = "mary_debug"
>>
>> Then it should create a "mary_debug.nc <http://mary_debug.nc/>" NetCDF file, with just the "uwnd" data you're plotting, along with the coordinate arrays. This way you don't have to provide me with the full data file if you don't want.
>>
>> The "mary_debug,nc" file should probably be small enough to just attach in an email. I think a "mary_debug.ncl" script will also get created, and you can email that to me, although I don't need it since I have your original script.
>>
>> Thanks,
>>
>> --Mary
>>
>>
>>
>>
>>
>> On Fri, Aug 19, 2016 at 10:34 AM, Mark Branson <mark at atmos.colostate.edu <mailto:mark at atmos.colostate.edu>> wrote:
>> Hi Mary.
>>
>> Here is that. Interestingly, the problem has gone away. So perhaps the coordinate arrays are indeed the source of the problem?
>>
>> Mark
>>
>>
>>
>>> On Aug 19, 2016, at 10:26 AM, Mary Haley <haley at ucar.edu <mailto:haley at ucar.edu>> wrote:
>>>
>>> Hi Mark,
>>>
>>> That is a little odd. You did good work debugging the problem, so missing values and nans are ruled out.
>>>
>>> To further debug this, can you try plotting a very basic contour plot, without the pressure /height axes, and without any coordinate information and send the image:
>>>
>>> plot = gsn_contour (wks,uwind,res)
>>>
>>> The gsn_contour script ignores any coordinate arrays, which I want to make sure are not the source of the problem.
>>>
>>> Thanks,
>>>
>>> --Mary
>>>
>>>
>>>
>>> On Fri, Aug 19, 2016 at 10:07 AM, Mark Branson <mark at atmos.colostate.edu <mailto:mark at atmos.colostate.edu>> wrote:
>>> Hello.
>>>
>>> I'm running ncl v6.2.1 and here is what I'm doing in the attached script:
>>>
>>> 1) read in a long time series of zonal wind and surface pressure data from CAM output.
>>> 2) interpolate them to constant pressure levels using vinth2p.
>>> 3) compute long-term monthly means with clmMonTLLL.
>>> 4) compute the annual mean of the monthly means.
>>> 5) subtract off the annual mean from the monthly means.
>>> 6) select one latitude (the equator), and then take the mean of all longitudes.
>>> 6) make a color-filled height versus time cross section plot.
>>>
>>> I get the plot (see attached), but some of the contours near the top are white. I thought maybe it was because there were some missing or bogus values, so I inserted a check for those with the ismissing and isnan_ieee functions and it seems the data is fine.
>>>
>>> Any ideas?
>>>
>>> Thanks,
>>> Mark Branson
>>>
>>>
>>> begin
>>> a = addfile("/pool/mark/spcam_nlev51/spcam_nlev51.cam.h0.0003-01.nc <http://spcam_nlev51.cam.h0.0003-01.nc/>","r")
>>> hyam = a->hyam
>>> hybm = a->hybm
>>>
>>> b = addfile("/pool/mark/spcam_nlev51/uandps.nc <http://uandps.nc/>","r")
>>> u = b->U ;u = [time | 338] x [lev | 53] x [lat | 96] x [lon | 144]
>>> ps = b->PS ;ps = [time | 338] x [lat | 96] x [lon | 144]
>>> u at _FillValue = default_fillvalue("float")
>>>
>>> if (any(ismissing(u))) then
>>> print("### original u array contains some missing data ###")
>>> end if
>>>
>>> ; interpolation variables
>>> plvlM = (/ 30.,50.,70.,100.,150.,200.,250.,300.,350.,400.,450.,500.,550., \
>>> 600.,650.,700.,750.,800.,850./)
>>> plvlM at units = "mb"
>>> nplvlM = dimsizes(plvlM)
>>>
>>> p0 = 1000. ; mb required by vinth2p
>>> interp = 2 ; log interpolation
>>> extrap = False ; no extrapolation past psfc.
>>>
>>> tmp = vinth2p(u,hyam,hybm,plvlM,ps,interp,p0,1,extrap)
>>> if (typeof(tmp).eq."double") then
>>> uwind1 = dble2flt(tmp)
>>> else
>>> uwind1 = tmp
>>> end if
>>>
>>> ; compute long-term monthly means
>>> uclm = clmMonTLLL(uwind1) ;uclm = [month | 12] x [lev_p | 33] x [lat | 96] x [lon | 144]
>>>
>>> ; annual mean of monthly means
>>> uann = dim_avg_n_Wrap(uclm,0) ;uann = [lev_p | 33] x [lat | 96] x [lon | 144]
>>>
>>> ; subtract off annual mean from monthly means
>>> do i = 0,11
>>> uclm(i,:,:,:) = (/ uclm(i,:,:,:) - uann(:,:,:) /)
>>> end do
>>>
>>> uwind2 = uclm(:,:,47,:) ; select the equator for the latitude
>>> uwind3 = dim_avg_Wrap(uwind2) ; mean of all longitudes
>>> uwind = uwind3(lev_p|:,month|:) ; swap the array dimensions: month,level --> level,month
>>>
>>> printMinMax(uwind,False)
>>>
>>> if (any(ismissing(uwind))) then
>>> print("### uwind array contains some missing data ###")
>>> end if
>>> if (any(isnan_ieee(uwind))) then
>>> print("### uwind array contains some NaNs ###")
>>> end if
>>>
>>> uwind at long_name = "SP-CAM, 51 levels, Monthly Means"
>>>
>>> wks = gsn_open_wks ("eps", "test1") ; send graphics to EPS file
>>> gsn_define_colormap(wks,"BlRe")
>>>
>>> res = True
>>> res at cnFillOn = True
>>> res at cnLinesOn = True
>>> res at lbLabelBarOn = True
>>> res at cnInfoLabelOn = False
>>>
>>> ; change labels for monthly mean plot
>>> res at tmXBMode = "Explicit"
>>> res at tmXBLabelFontHeightF = 0.015
>>> res at tmXBLabels=(/"Jan","Feb","Mar","Apr","May","Jun","Jul","Aug","Sep", \
>>> "Oct","Nov","Dec","Jan"/)
>>> res at tmXBValues = ispan(0,12,1)
>>>
>>> plot = gsn_csm_pres_hgt (wks,uwind,res)
>>> end
>>>
>>> ** OUTPUT FROM THE SCRIPT **
>>>
>>> /Users/mark/mmf/SPCAM5/stratosphere_gwdrag/spcam_nlev51> ncl hvstime.ncl
>>> Copyright (C) 1995-2014 - All Rights Reserved
>>> University Corporation for Atmospheric Research
>>> NCAR Command Language Version 6.2.1
>>> The use of this software is governed by a License Agreement.
>>> See http://www.ncl.ucar.edu/ <http://www.ncl.ucar.edu/> for more details.
>>> (0) min=-9.835 max=6.77056
>>>
>>>
>>>
>>> _______________________________________________
>>> ncl-talk mailing list
>>> ncl-talk at ucar.edu <mailto:ncl-talk at ucar.edu>
>>> List instructions, subscriber options, unsubscribe:
>>> http://mailman.ucar.edu/mailman/listinfo/ncl-talk <http://mailman.ucar.edu/mailman/listinfo/ncl-talk>
>>>
>>>
>>> _______________________________________________
>>> ncl-talk mailing list
>>> ncl-talk at ucar.edu <mailto:ncl-talk at ucar.edu>
>>> List instructions, subscriber options, unsubscribe:
>>> http://mailman.ucar.edu/mailman/listinfo/ncl-talk <http://mailman.ucar.edu/mailman/listinfo/ncl-talk>
>>
>>
>>
>
>
>
> _______________________________________________
> ncl-talk mailing list
> ncl-talk at ucar.edu
> List instructions, subscriber options, unsubscribe:
> http://mailman.ucar.edu/mailman/listinfo/ncl-talk
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mailman.ucar.edu/pipermail/ncl-talk/attachments/20160822/16ae7206/attachment.html
More information about the ncl-talk
mailing list