Re: How to interpolate low resolution data to high resolution?

I had replied to "Lawrence" but forgot to include ncl-talk.
He sent a clarification to me offline. See below

=======
I am not sure what you mean by 'quite different from the original'
Generally, interpolation will not maintain the high/low values
from the high resolution grid.

In the case you mention, linint2 searches for the 4 nearest
grid points of the high res grid that surrounds a low res
grid point. It then computes a weighted average of the 4
high res points. The weights are a function of the distance
of the distance from the low res point.

Averaging is a smoother. Also, if the low res grid points
do not match where the high res grid points have max/min
values, you will not see those in the low res grid.

When going from high->low res the grid will *always* look smoother
mo matter what interpolator you use. There is no way to preserve the
detailed
features.

D

 Fri, 24 Aug 2007, Lawrence wrote:

> Hi Dennis,
>
> I am so sorry to type the wrong mail title. It should be "*How to
> interpolate high resolution data to low resolution*". When I
interpolate low
> resolution data to high resolution, linint2 does work fine, but when
> interpolate high resolution to low, the results often quite differs
from the
> original, for maximum values are greatly smoothed. So I wonder is
there any
> other function in NCL that can make this type of interpolation
results more
> close to the original?
>

_______________________________________________
ncl-talk mailing list
ncl-talk_at_ucar.edu
http://mailman.ucar.edu/mailman/listinfo/ncl-talk
Received on Fri Aug 24 2007 - 10:59:53 MDT