[ncl-talk] empty plots in panel-plot
Dennis Shea
shea at ucar.edu
Tue Apr 11 09:53:19 MDT 2017
mnmxlvl = nice_mnmxintvl(min(data),max(data),16,False)
Is a nice function, in particular for normally distributed data.
An alternative for non-normally distributed data is to use:
https://www.ncl.ucar.edu/Document/Functions/Contributed/stat_dispersion.
shtml
opt = True
;opt at PrintStat = True
statd = *stat_dispersion*(data, opt )
res at cnLevelSelectionMode = "ManualLevels"
; arbitrary
res at cnMinLevelValF = statd(5) ; lower sextile
res at cnMaxLevelValF = statd(11) ; upper sextile
res at cnLevelSpacingF = ...whatever...
https://www.ncl.ucar.edu/Applications/ESMF.shtml
Example 35
By looking at the output from stat_dispersion, one could manually set
contour levels that maximize the
information. The data are clusterd at the 'low' end but have significantly
larger values which are not artificial.
res at cnLevelSelectionMode = "ExplicitLevels"
res at cnLevels = (/0.001, 0.050, 0.060, 0.070, 0.080,
0.090, 0.100, 0.125, 0.150\
,0.200, 0.250, 0.50, 1, 5, 10, 15,
20, 25, 50 /)
On Tue, Apr 11, 2017 at 7:48 AM, Mary Haley <haley at ucar.edu> wrote:
> I usually end up increasing the levels too, simply by taking the "spacing"
> value and dividing it by a number, like 2 or 3 (for 2x or 3x as many
> contours).
>
> res at cnLevelSelectionMode = "ManualLevels"
> res at cnMinLevelValF = mnmxlvl(0)
> res at cnMaxLevelValF = mnmxlvl(1)
> res at cnLevelSpacingF = mnmxlvl(2) / 2. ; make the contour
> interval smaller for more contours
>
> On Tue, Apr 11, 2017 at 1:43 AM, Noelia otero <noeli1680 at gmail.com> wrote:
>
>> Hi Mary,
>>
>> Thanks a lot. I was trying to set min and max values, but I wasn't taking
>> a good value for levelSpacing ..,I follow your suggestion, although I had
>> to increase considerably the number of levels. Now it seems to work.
>>
>> Many thanks again,
>>
>> Best
>>
>> Noelia
>>
>> 2017-04-11 7:47 GMT+02:00 Mary Haley <haley at ucar.edu>:
>>
>>> Dear Noelia,
>>>
>>> When you are paneling a series of color contour plots and then paneling
>>> them later, you first want to be sure you use a fixed set of contour levels
>>> across all the plots, so you can then use a single labelbar in the final
>>> panel plot.
>>>
>>> The function nice_mnmxintvl
>>> is what you can use to fix your contour levels
>>>
>>>
>>> mnmxlvl = nice_mnmxintvl(min(data),max(data),16,False)
>>>
>>> res at cnLevelSelectionMode = "ManualLevels"
>>> res at cnMinLevelValF = mnmxlvl(0)
>>> res at cnMaxLevelValF = mnmxlvl(1)
>>> res at cnLevelSpacingF = mnmxlvl(2)
>>>
>>> You will need to get the min and max of your data across all plots, and pass that to
>>>
>>> The "min(data)" and "max(data)" are from another example. You will need
>>> to plug in the min/max of your own data across all plots.
>>>
>>> You now want to turn off the individual labelbar you're getting for each
>>> plot:
>>>
>>> res at lbLabelBarOn = False
>>>
>>> and then turn on the panel labelbar when you call gsn_panel:
>>>
>>> pres = True
>>> pres at gsnPanelLabelBar = True
>>> ...
>>> gsn_panel(...)
>>>
>>> You may want to see our "panel" examples page:
>>>
>>> http://www.ncl.ucar.edu/Applications/panel.shtml
>>>
>>> I think panel_35.ncl uses the nice_mnmxintvl function.
>>>
>>> --Mary
>>>
>>> On Mon, Apr 10, 2017 at 11:37 AM, Noelia otero <noeli1680 at gmail.com>
>>> wrote:
>>>
>>>> Hi!
>>>>
>>>> I am plotting in a panel plots seasonal means from different models,
>>>> and I got some "empty" plots for some specific models and seasons. My guess
>>>> is that it is due to the large spread of values (or some outliers). Maybe,
>>>> it's a simple question..but I don't know how to set the resources to get a
>>>> good label bar. I attach the final plot that I got with my code (see a part
>>>> of the code below):
>>>>
>>>> ; i= model ,s =season
>>>>
>>>> model = addfile(fils(i), "r")
>>>> c=str_split(fils(i), "/")
>>>> nmod=str_split(c(6), "#")
>>>> mmod=nmod(0)
>>>> print(nmod)
>>>> sfile=systemfunc("ls " + dir2 + mmod +"*"+ season(s)+".nc")
>>>> smodel=addfile(sfile, "r")
>>>>
>>>> svar = smodel ->$var$
>>>> svar!1="lat"
>>>> svar!2="lon"
>>>>
>>>> ;------Getting statistics for each grid point-------
>>>>
>>>> stats_seas_var = new((/nlat,mlon/), typeof(svar),-9e+33)
>>>>
>>>> ;-------get 95th possition 25 of statistics---------
>>>>
>>>>
>>>> opt=False
>>>> do nl=0,nlat-1
>>>> do ml=0,mlon-1
>>>>
>>>> seas_st_var = stat_dispersion( svar(:,nl,ml), opt)
>>>> stats_seas_var(nl,ml) = seas_st_var(0)
>>>>
>>>> ;stat_dispersion: (0) mean, (1) std, (2), min, (14),max
>>>> end do
>>>> end do
>>>>
>>>>
>>>> stats_seas_var at long_name = "mean"
>>>> stats_seas_var!0="lat"
>>>> stats_seas_var!1="lon"
>>>> stats_seas_var&lat = lat
>>>> stats_seas_var&lon = lon
>>>>
>>>> res = True
>>>> res at gsnMaximize = True
>>>> res at cnFillOn = True
>>>> res at cnLinesOn = False
>>>> res at gsnAddCyclic = False
>>>> res at gsnDraw = False
>>>> res at gsnFrame = False
>>>> res at mpMinLonF = min(lon)
>>>> res at mpMinLatF = min(lat)
>>>> res at mpMaxLonF = max(lon)
>>>> res at mpMaxLatF = max(lat)
>>>> res at lbLabelAutoStride=True
>>>> res at cnLevelSelectionMode = "ManualLevels"
>>>> res at cnMinLevelValF = 0.
>>>> res at cnMaxLevelValF = 10.
>>>> res at cnLevelSpacingF = 5.
>>>> res at lbLabelBarOn = False
>>>> res at gsnCenterString =nmod(0)
>>>> res at gsnRightString =season(s)
>>>> plotseas = gsn_csm_contour_map(wks,stats_seas_var,res)
>>>>
>>>> I really appreciate any suggestion to improve it,
>>>>
>>>> Many thanks in advance,
>>>>
>>>> Noelia
>>>>
>>>>
>>>> _______________________________________________
>>>> ncl-talk mailing list
>>>> ncl-talk at ucar.edu
>>>> List instructions, subscriber options, unsubscribe:
>>>> 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/20170411/bc39d18a/attachment.html
More information about the ncl-talk
mailing list