Re: how to improve the interpolation?

The problem occurs in regions where the data have been *extrapolated*.
Extrapolation is always problematical and, generally, should be avoided.

What to do to get "smooth" contours? Well, I guess you could use "smth9"
over the just the regions where the "saw tooth" pattern appears.

http://www.ncl.ucar.edu/Document/Functions/Built-in/smth9.shtml

Say, the subscript indices of the "sawtooth" region correspond to
jStrt, jLast, iStrt, iLast

EG:

 x(jStrt:jLast,iStrt:iLast) = smth9(x(jStrt:jLast,iStrt:iLast), 0.50,
0.25, True)

>>The attached is the plot, the black dot in land indicate the station
observation. Then I interpolate them to grid data for plot with the
function natgrids. >>The upper two plots have the serious problem in the
Northeat China obviously. The observation stations is less, the result
is not accurate enough, >>however, it should be smooth. How can I
improve this plot in NCL.

hjb wrote:
> Dear everyone,
>
> I have a problem with interpolation. I interpolated the station data to
> grid with the fuction natgrids.
> However, the "sawtooth phenomenon" is serious near the missing value
> region. Can anyone give some suggestion to improve this kind
> interpolation? Thanks in advance.
>
>
> ------------------------------------------------------------------------
> hjb
> 2008-08-16
>
>
> ------------------------------------------------------------------------
>
> _______________________________________________
> ncl-talk mailing list
> ncl-talk_at_ucar.edu
> http://mailman.ucar.edu/mailman/listinfo/ncl-talk
_______________________________________________
ncl-talk mailing list
ncl-talk_at_ucar.edu
http://mailman.ucar.edu/mailman/listinfo/ncl-talk
Received on Sat Aug 16 2008 - 08:55:42 MDT