[ncl-talk] Explicit levels (using WRF output)

[Arianna Valmassoi]

Hi Mary,
thank you very much, now there is no problem with the filling!

Have you notice (and it happened to me before) that in the light blue you
can see the world coast-line? (In the figure you sent me there is in the
bottom right corner a part of Africa and Madagascar).. I wasn't able to do
anything to remove them (without removing the coordinates..)

Thank you very much again

Cheers

Arianna

Il giorno gio 19 gen 2017 alle ore 22:20 Mary Haley <haley at ucar.edu> ha
scritto:

> Hi Arianna,
>
> NCL does its contouring by filling the contours based on what values fall
> above or below a contour level, and not exactly equal to a contour level.
> This causes confusion when your data is all integers.
>
> If you specify contour levels of 1,2,3,4,5, then NCL is going to create a
> contour plot with *6* colors, where the colors represent the following:
>
>   color 1 - all values < 1
>   color 2 - all values >= 1 and < 2
>   color 3 - all values >= 2 and < 3
>   color 4 - all values >= 3 and < 4
>   color 5 - all values >= 4 and < 5
>   color 6 - all values >= 5
>
> This is why you see the extra "Label_1", because NCL only expects (n-1)
> labels if you have n colors, because it's assuming you are labelling
> *between* values.  When you tell it to center the labels, then suddenly you
> are one label short, and hence NCL gives you the "Label_1" label.
>
> Back to your contour levels:
>
> In your code, you set the first contour level to 1, so this means that the
> first color (darkgreen) is going to represent all values < 1, which means
> any value equal to 0, since you only have integer values.
>
> Then, then next color, "forestgreen" is going to represent all values >=1
> and < 2, which, again, in your case, since you just have integers, is going
> to be all colors equal to 1.
>
> If you want the first color to represent all values equal to 1, and the
> second color all values equal to 2, and you want twenty colors total, where
> the last color represents values equal to 20, then you need to create 19
> contour levels that start just slightly higher than 1, 2, 3, 4, etc:
>
> This is one way of doing it, by adding 0.1 to the levels:
>
>   res at cnLevels  = (/1, 2, 3, 4, 5, 6, 7, 8, 9,10, 11,12,13,14,
> 15,16,17,18,19/)+0.1
>
> I see you are using cnFillPalette, which is okay, but it's better to use
> cnFillColors when you have a specific color for each contour level.  This
> way you don't need to mess with gsnSpreadColors:
>
>   res at cnFillColors     = colorscheme        ; distinct colors for
> categories
>  ; res at gsnSpreadColors = False           ; comment this or remove it
>
> If you are trying to plot this data over a map, you generally want to use
> wrf_contour *and* wrf_map_overlays.  But, your XLAT and XLONG on the file
> do not have a time dimension, which is going to cause problems for
> wrf_map_overlays.
>
> So, I modified your code to use gsn_csm_contour_map instead, but I used
> the map projection defined on the file.
>
> I've attached the modified script and a new PNG.
>
>
> --Mary
>
>
> On Thu, Jan 19, 2017 at 10:57 AM, Arianna Valmassoi <
> arianna.valmassoi at gmail.com> wrote:
>
> Hi all,
>
> I'm working with WRF and it's landuse representation. In the plot output I
> have the following problems:
>
>    - label: there is one more tic in the label with the first color, and
>    has a name I did not put.
>    - assign color to discrete numbers: I had to impose to the landcat to
>    start with zero, not as one as it is, otherwise the first level will not
>    match
>    - Domain visualization: I can't use the model coordinates (even using
>    other plot option)
>
>
> I attach here both the code and the .nc file, in order to help the debug
> category 13 should be red and category 17 light blue.
>
> Thank a lot for any help and sorry if it's a duplicate problem!
>
> Cheers,
> Arianna Valmassoi
> --
> Arianna Valmassoi
>
> _______________________________________________
> ncl-talk mailing list
> ncl-talk at ucar.edu
> List instructions, subscriber options, unsubscribe:
> http://mailman.ucar.edu/mailman/listinfo/ncl-talk
>
>
> --
Arianna Valmassoi
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mailman.ucar.edu/pipermail/ncl-talk/attachments/20170119/8469e7d7/attachment.html