Re: contour smoothing

From: David B. Reusch <dbr120_at_nyahnyahspammersnyahnyah>
Date: Fri Oct 30 2009 - 14:05:31 MDT

Yes, that does seem to be helping, though I have to apply it three times
to get where I want to be (with p=0.5, q=0.25, so fairly "heavy"
smoothing). Is there any more guidance out there for the smoothing
parameters? For example, is it worth trying different values of p or
should I just stick with the web page advice and do multiple passes? I
tried finding the source code so I could evaluate this myself, but
didn't have any luck (and I do a binary install anyway so I might not
even have it!).

For what it's worth, I'm still curious about what I wasn't seeing with
cnSmoothingTension -- but smth9 seems to meet my needs.


Adam Phillips said the following on 10/30/09 3:20 PM:
> Hi Dave,
> I would highly recommend using smth9_Wrap to smooth your data before
> plotting. This allows you to have complete control over the amount of
> smoothing (heavy, light, pass the data through the function a few
> times), as opposed to trying to mess with the cnSmoothingTension resource.
> Adam
> David B. Reusch wrote:
>> Hello,
>> I have a long-time-working script that uses gsn_csm_contour_map_polar to
>> produce panelled plots of Antarctic meteo data and I've been wanting to
>> experiment with tweaking the contouring so that basically I get a
>> "smoother" look (for example, get rid of any really small areas that
>> aren't particularly representative of the larger picture). Rather than
>> approaching this by changing the cmin/cmax/cint values (which might also
>> work, I think), I've been trying to use the cnSmoothingOn and
>> cnSmoothingTensionF resources. The reason I'm writing is that these
>> don't appear to be doing anything! I've tried a wide range of tension
>> values with no resulting plot changes. All I've done to the script is
>> add these two resources -- am I missing something else I need to get
>> this to work? I didn't see any examples that address these particular
>> resources (though I know there can't be examples for everything).
>> I can provide the script/data if needed (I didn't attach the script
>> because it's *long* and ugly). Or if someone has an example, that would
>> be great. Also, if the prevailing wisdom is to use cmin/cmax/cint to
>> accomplish such smoothing, that would be really nice to know too!
>> This is NCL 5.1.1 on Mac 10.5.8.
>> Thanks,
>> Dave Reusch

517 Deike Building
Earth & Environmental Systems Institute
Penn State University
University Park, PA 16802

ncl-talk mailing list
List instructions, subscriber options, unsubscribe:
Received on Fri Oct 30 14:05:12 2009

This archive was generated by hypermail 2.1.8 : Mon Nov 02 2009 - 08:49:47 MST