Re: contouring within a domain

From: Dave Brown (dbrown AT XXXXXX)
Date: Thu Aug 08 2002 - 17:34:37 MDT


If, as you say, the original data are actually gridded -- that is,
representable as a 2d array with indexes (i,j), and you can build
associated 2d lat and lon coordinate arrays with matching dimension sizes,
then you should be able to contour the data directly using the newest
version of NCL. The contour boundaries will exactly match the data
boundaries.

I'm not sure if this version has been released yet, but, if not, it will
be soon.
 -dave
 dbrown@ucar.edu

On Thu, 8 Aug 2002, David Reusch wrote:

> Q: Given an irregular set of points, in a polar setting,
> is there a way to do contouring only within the domain
> covered by the points?
>
> My data looks like this:
> a vector of actual data
> a vector of latitudes
> a vector of longitudes
> The lat/lon values are specific to each data point and
> are definitely *not* monotonic. (For what it's worth,
> the lat/long data are from a 250 km polar EASE grid.)
> The boundaries of the data are straight making it a
> polygon of sorts but also removing any chance of using
> a latitude line as some kind of cutoff.
>
> So far I've been using idsfft to convert to a rectangular
> grid and gsn_csm_contour_map_polar to do the contours.
> Unfortunately, the latter is contouring outside the
> bounds of my data. Since my mpMaxLatF is set to -60,
> it fills in the whole region up to -60 (while doing at
> least the min/max longitude correctly). Some of my
> data does go to -60 but only a small fraction.
>
> In short, I guess these are my questions:
> - do I need to use idsfft in the first place? I think
> "yes" since everything else I tried failed... If
> not, I'd love to hear the alternative. I'd guess
> this may be causing problems since it introduces a
> rectangular grid with no real memory of the original
> data bounds.
> - is there a way to use a mask to keep the contours
> within my domain? if so, how? I've seen the sea
> ice example but don't quite see how to translate
> it to my data (at least w/o a nice continental
> mask as in that example).
>
> This is a rather urgent problem for me to have solved
> since visualization is key to doing any analysis of
> my PhD research results.
>
> thanks!
> dave reusch
> dbr@geosc.psu.edu
> _______________________________________________
> ncl-talk mailing list
> ncl-talk@ucar.edu
> http://mailman.ucar.edu/mailman/listinfo/ncl-talk
>

_______________________________________________
ncl-talk mailing list
ncl-talk@ucar.edu
http://mailman.ucar.edu/mailman/listinfo/ncl-talk



This archive was generated by hypermail 2b29 : Thu Aug 08 2002 - 22:02:44 MDT