[Thread Prev][Thread Next][Index]

Re: [ferret_users] fill leaves corners of the plot blank

      This has to do with the way Ferret does the "contouring" with respect
to the given region and given contouring intervals.

      I faced the same problem recently and this was the work around
i have found:

  1. sometimes fine tuning the levels qualifier (increase the number of
         intervals or decrease the delta value) will improve the situation.

  2.  - decide on the values/options for hlim, vlim, levels, palette
      - first make a SHADE plot
      - then make an overlay FILL plot (don't forget to use /over with
          FILL commad and ppl fill command)

Basically, the blank region appears where the data is not changing with
respect to the specified contouring (ie fill) level. Since SHADE command
does not care about interpolations/contouring, plots will be just fine.
To handle this kind of situations, much sophisticated contouring routines
are required.

A similar issue is with small graphics in Ferret. The same plot
which looks perfectly fine in a bigger viewport, will have blank spaces
at edges/corners when the viewport size is reduced. This is particularly
true, if you are using a high resolution data (for example TMI SST).

Hope this helps,


> Dear Ferret users / developers,
> I've run into an old problem with the fill command.  It seems to leave
> random corners of the plot blank, as if it's forgetting to include the
> data point at the corner of the plot for the interpolation.  There is
> an old post in the archives about this where Ansley suggested setting
> a bigger data region than the plot window, but this doesn't help.
> See http://www.icce.rug.nl/~hein/ferret/ for some samples of the
> problem, these were all plotted with the same grid and plot options,
> yet the "missing corner" in the right top of the chart is not identical.
> The data grid used was that of etopo20.  What I've determined so far:
> - the data region is (much) bigger than the plot window, no
>   difference.  I tried limiting the window both with region and
>   vlimit/hlimit, and also setting smaller limits than the region.
> - the effect is "random": for certain specific values of hlimits and
>   vlimits it goes away.
> - Other plot options don't seem to matter (palette, set_up, nolabel, etc)
> - my data was saved in a netcdf file first, so it shouldn't be an
>   issue with variable definitions.
> I can reproduce the issue (left top, right top and right bottom corners) with
> just the following commands:
> use etopo20
> fill /hlim=83w:63w /vlim=28n:38n /lev=20c rose
> Kind regards,
>      Hein Zelle
> --
>  Unix is user friendly. It's just very particular about who
>  it's friends are.
>  Hein Zelle                     hein@xxxxxxxxxxx
> 	                        http://www.icce.rug.nl/~hein

[Thread Prev][Thread Next][Index]

Contact Us
Dept of Commerce / NOAA / OAR / PMEL / TMAP

Privacy Policy | Disclaimer | Accessibility Statement