[ncl-talk] taylor_diagram question
Tao Zhang
tao.zhang at colorado.edu
Tue Apr 4 12:30:30 MDT 2023
Hi Adam,
Thanks for your great recommendations.
1) Dennis Shea mentioned that the Taylor diagram listed in NCL website
is hard-coded for normalized standard deviation, not for standard deviation
as shown in previous link I provided, right?
2) How to remove the "0.75" "1" labels close to correlation value? I
just want to keep one. Is possible to
add "0.25"? see attached plot.
Thanks
Tao
On 4/3/23 4:01 PM, Adam Phillips wrote:
> Hi Tao,
> I believe you corresponded with someone offline about this. I have
> some experience modifying the taylor_diagram function, but I have not
> done what you are looking for so I can only offer recommendations. You
> will likely need to make several attempts getting all the resources
> set the way you want; hopefully your test script runs quickly!
>
> To answer your 3 questions:
> 1 - Yes.
> 2 - I believe this is possible. Steps required:
> - Copy the taylor_diagram.ncl
> <https://www.ncl.ucar.edu/Applications/Scripts/taylor_diagram.ncl> code,
> and saving it as (say) taylor_diagram_2.ncl.
> - Alter the function name at the top to taylor_diagram_2 (at line
> one of taylor_diagram_2.ncl).
> - In your new function, go to the section entitled "Part 3". You
> will see a resource list that is initiated 5 lines beneath "Part 3"
> named respl. I would try adding the following resource settings
> *beneath* the current respl resource settings:
> respl at gsLineDashSegLenF = 0.30. ; Controls spacing between line
> labels; this setting will require some trial and error.
> respl at gsLineLabelFontColor = respl at gsLineColor
> respl at gsLineLabelFontHeightF = 0.0125. ; this font height
> setting might need adjustment
>
> - Then, within the "do n-1,ncon" do loop, you will need to tell NCL
> how to label the lines, so try this:
> do n=1,ncon
> rr = n*dx
> res at gsLineLabelString = rr. ; add this line
>
> That should do it. As I stated in the comments above, you will likely
> need to adjust the setting for gsLineDashSegLenF and
> gsLineLabelFontHeightF.
>
> 3 - Yes. Search within the function for the resource "gsLineColor".
> You can adjust these as needed.
>
> Hope that helps!
> Adam
>
> On Sat, Mar 25, 2023 at 3:00 PM Tao Zhang via ncl-talk
> <ncl-talk at mailman.ucar.edu> wrote:
>
> Hi,
>
> I can reproduce the taylor_4 figure using the sample script in
>
> https://www.ncl.ucar.edu/Applications/Images/taylor_4_1_lg.png
>
> https://www.ncl.ucar.edu/Applications/Scripts/taylor_4.ncl
>
> But I want Figure 1 in
>
> https://en.wikipedia.org/wiki/Taylor_diagram#:~:text=Taylor%20diagrams%20are%20mathematical%20diagrams,comparative%20assessment%20of%20different%20models
>
> Which shows the standard deviation (0~4 mm/day), not the standardized
> deviations (normalized) (0-1.5).
>
> 1) May I only change the taylor_4.ncl for my purpose, but keep the
> loaded "taylor_diagram.ncl" unchanged?
>
> 2) How to add "observed value" in the Taylor diagram as shown in
> Figure
> 1 in above link?
>
> 3) How to change the color in 3 lines of standard deviation,
> correlation, and RMS?
>
> Thanks
>
> Tao
>
>
> _______________________________________________
> ncl-talk mailing list
> ncl-talk at mailman.ucar.edu
> List instructions, subscriber options, unsubscribe:
> https://mailman.ucar.edu/mailman/listinfo/ncl-talk
>
>
>
> --
> Adam Phillips
> Associate Scientist IV, Climate Analysis Section
> Climate and Global Dynamics Laboratory
> National Center for Atmospheric Research
> www.cgd.ucar.edu/staff/asphilli/
> <http://www.cgd.ucar.edu/staff/asphilli/>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mailman.ucar.edu/pipermail/ncl-talk/attachments/20230404/87f68bbc/attachment.htm>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: taylor_GFSv2amip.GFSFV3amip.t2m.precip.cli.1991-2020.ann.globe.tropics.Contiguous_US.png
Type: image/png
Size: 49915 bytes
Desc: not available
URL: <https://mailman.ucar.edu/pipermail/ncl-talk/attachments/20230404/87f68bbc/attachment.png>
More information about the ncl-talk
mailing list