Re: weird contour lines [message #68478 is a reply to message #68470] |
Sat, 31 October 2009 20:08   |
Libo Wang
Messages: 6 Registered: October 2009
|
Junior Member |
|
|
On Oct 31, 1:16 pm, David Fanning <n...@dfanning.com> wrote:
> Libo Wang writes:
>> Thanks, David. I've read helpful tips in your website. It is actually
>> a bit of both aesthetic problem and a science problem! A meteorologist
>> would probably tell me that my map is not what a standard pressure
>> field plot should look like!
>
>> I tried to expand my data from 144*37 to 1440*370 using cubic=-0.5 in
>> Congrid,the contour map did improved in some parts, but some parts got
>> worse:
>
>> http://picasaweb.google.com/lh/photo/pbT1Ftfbx0nqS9O3uw0tGg? feat=dire...
>
>> The NCEP data is low in resolution: 2.5 degree spacing. What else
>> could I try?
>
> Well, this is a pain-in-the-kiester solution, but what you
> *could* do is obtain each closed contour from the CONTOUR
> command (PATH_XY keyword, etc.) and "resample" it with the
> ArcSample program from my web page:
>
> http://www.dfanning.com/programs/arcsample.pro
>
> The purpose of ArcSample is to re-sample the contour
> at approximately equally spaced intervals, and then to
> use those points to interpolate the contour. This has
> the effect of smoothing the contour. I use this, for
> example, when I am getting the starting points for
> an active contour or snake algorithm.
>
> I wouldn't go to this kind of trouble unless your
> meteorologist friend is EXTREMELY anal.
>
> Cheers,
>
> David
>
> P.S. I assume you tried using the MIN_CURVE_SURF program
> on your data and gave up because you don't have that many
> hours in the day.
>
> --
> David Fanning, Ph.D.
> Fanning Software Consulting, Inc.
> Coyote's Guide to IDL Programming:http://www.dfanning.com/
> Sepore ma de ni thui. ("Perhaps thou speakest truth.")
Thanks, David. I'll probably try the arcsample approach some day (I'm
pretty sure I'll make more contours in IDL!). For this "simple" task,
I'll see if I could do it in other softwares :(
I did try the MIN_CURVE_SURF fuction, but killed the run after 5
minutes.
|
|
|