[ncl-talk] Drawing several polygons on a xy plot

[Dave Allured - NOAA Affiliate]

That looks great in panel #1.  I am sure you can easily extend this to the
other five panels, if that is needed.


On Wed, Mar 10, 2021 at 9:44 AM Giorgio Graffino <g.graffino at tim.it> wrote:

> Hi Dave,
>
> Thanks for the hint. I haven't thought about looping gsn_add_polygon over
> years, and setting five (!) vertices. I've managed to find a good setting
> (see code snippet below and attached plot).
>
> do j = 0, nyear-1
> if (.not.ismissing(temp_amip2_tpi_year_filter(j))) then
> if (temp_amip2_tpi_year_filter(j).lt.0.) then
>  plot1x(j) =
> gsn_add_polygon(wks,plot1g(i),(/years(j)-0.5,years(j)+0.5,years(j)+0.5,years(j)-0.5,years(j)-0.5/),(/temp_amip2_tpi_year_filter(j)-1.e2,temp_amip2_tpi_year_filter(j)-1.e2,temp_amip2_tpi_year_filter(j)+1.e2,temp_amip2_tpi_year_filter(j)+1.e2,temp_amip2_tpi_year_filter(j)-1.e2/),pg_res)
> end if
> end if
> end do
>
> Cheers,
> Giorgio
>
> ----Messaggio originale----
> Da: dave.allured at noaa.gov
> Data: 10-mar-2021 14.04
> A: "Giorgio Graffino"<g.graffino at tim.it>
> Cc: <ncl-talk at ucar.edu>
> Ogg: Re: [ncl-talk] Drawing several polygons on a xy plot
>
> P.S.  I just remembered you are making a panel plot.  You will probably
> need to increase the size of the graphics array to hold all polygons in all
> six panels, due to that memory requirement.  Something like this:
>
>     poly = new ( (/ nyears, npanels /), graphic)
>
> There is more than one way to do this.  The array can be 1-D, 2-D, or 3-D,
> whatever you think is most convenient for the looping and indexing
> structure of your program.
>
>
> On Wed, Mar 10, 2021 at 5:54 AM Dave Allured - NOAA Affiliate <
> dave.allured at noaa.gov> wrote:
>
>> Gsn_add_polygon does not work like that.  It draws only a single polygon
>> per call, and you must specify all vertices of the polygon.  Missing values
>> can not be used to create discontinuities or unfilled areas of a polygon.
>>
>> The simplest approach is to draw multiple polygons, one narrow rectangle
>> for each year on the X axis.  Loop over years, and skip (i.e. do not draw)
>> polygons unless the IPO index is negative.  Within the loop, remember to
>> assign each polygon to a different element of a type graphics array, to
>> satisfy the memory requirement of the function.
>>
>>     poly = new (nyear, graphic)
>>     do i = 0, nyear-1
>>         if (temp_amip2_tpi_year_filter(i) .lt. 0) then
>>             poly(i) = gsn_add_polygon (...)
>>         end if
>>     end do
>>
>> I leave it up to you to calculate the correct polygon vertices for each
>> year.  Your initial code version is almost doing this already.  I recommend
>> using a fully closed polygon for each year.  That is 5 vertices, not 4, for
>> each rectangle.  Function behavior for non-closed polygons is undefined.
>>
>>
>> On Wed, Mar 10, 2021 at 3:20 AM Giorgio Graffino via ncl-talk <
>> ncl-talk at mailman.ucar.edu> wrote:
>>
>>> Short addition to this thread: I'm running NCL 6.6.2 on  Linux
>>> 5.4.101-1-MANJARO.
>>>
>>> Best,
>>> Giorgio
>>>
>>> ----Messaggio originale----
>>> Da: ncl-talk at mailman.ucar.edu
>>> Data: 10-mar-2021 0.11
>>> A: < ncl-talk at ucar.edu>
>>> Ogg: [ncl-talk] Drawing several polygons on a xy plot
>>>
>>> Dear NCL people,
>>>
>>> I'm trying to draw several polygons on a time series plot, to obtain
>>> something like this (
>>> https://www.nature.com/articles/nclimate2106/figures/1). Basically, I
>>> need separate filled boxes to be drawn anywhere the IPO index is negative.
>>> I'd like to not draw individual polygons myself, because I should set the x
>>> and y ranges explicitly. (That is, at all times when the IPO index changes
>>> sign). Rather, I'd like the boxes to be drawn automatically, without
>>> knowing where the index is negative.
>>>
>>> So far, my best attempt at the problem is to create a 1-D mask, setting
>>> all positive value of the index to missing values. Then, I'm creating one
>>> big box over the whole domain, relying on the missing values of the masked
>>> time series. The relevant snippet of my code is shown here.
>>>
>>> ;; Create polygons where AMIP II TPI/IPO index is negative
>>>  temp_amip2_tpi_year_filter_neg =
>>> where(temp_amip2_tpi_year_filter.lt.0.,temp_amip2_tpi_year_filter,temp_amip2_tpi_year_filter@
>>> _FillValue)
>>>  temp_amip2_tpi_year_filter_neg_xp =
>>> new((/2*nyear/),typeof(temp_amip2_tpi_year_filter_neg),temp_amip2_tpi_year_filter_neg@
>>> _FillValue)
>>>  temp_amip2_tpi_year_filter_neg_yp =
>>> new((/2*nyear/),typeof(temp_amip2_tpi_year_filter_neg),temp_amip2_tpi_year_filter_neg@
>>> _FillValue)
>>> do i = 0, nyear-1
>>>  temp_amip2_tpi_year_filter_neg_yp(i)          =
>>> temp_amip2_tpi_year_filter_neg(i)+1.e2
>>>  temp_amip2_tpi_year_filter_neg_xp(i)          = years(i)
>>>  temp_amip2_tpi_year_filter_neg_xp(2*nyear-1-i) = years(i)
>>>  temp_amip2_tpi_year_filter_neg_yp(2*nyear-1-i) =
>>> temp_amip2_tpi_year_filter_neg(i)-1.e2
>>> end do
>>>
>>> After, I'm calling gsn_add_polygon to draw the boxes over the time
>>> series plot, but I'm getting only one big box (see attached). It seems that
>>> the function doesn't care about missing values, and in the description
>>> that's clearly stated ("If a missing value is encountered in *x* and/or
>>> *y*, then this pair is ignored, but the polygon will still be closed";
>>> https://www.ncl.ucar.edu/Document/Graphics/Interfaces/gsn_add_polygon.shtml).
>>>
>>>
>>> My question: is there a way to make gsn_add_polygon not closing the
>>> polygon? Is there a better way to draw those polygons with NCL?
>>>
>>> Best,
>>> Giorgio
>>>
>>>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mailman.ucar.edu/pipermail/ncl-talk/attachments/20210310/76889ca0/attachment.html>