Re: Grid to grid interpolation (Correia, James)

From: Seth McGinnis <mcginnis_at_nyahnyahspammersnyahnyah>
Date: Sun, 17 May 2009 17:59:40 -0600

I have found it tolerably fast to use rcm2points in a loop over the x & y dimensions, calculating values for one gridcell at a time:

do j = 0,ny-1
  do i = 0,nx-1
    lon = wrflon2d(j,i)
    lat = wrflat2d(j,i)
    temp = rcm2points(narrlat2d,narrlon2d,indata,lat,lon,1)
    outdata(:,j,i) = (/temp/)
    end do
end do

(I can't remember whether I decided to try looping to avoid memory problems with certain big files, or because it seemed easier than massaging the output back into the shape I wanted, or both, but it works.)

I *think* rcm2points uses the same inverse distance weighting as linint2; I have no idea how it performs with regard to position errors.


--Seth McGinnis

>Does anyone have advice on interpolating from the NARR grid to a WRF grid
>(both have 2d lat lon arrays)? What method would be pretty fast?
>Triple2grid won't do because I am concerned about even minor position errors
>as I plan to do some validating of the wrf model dataset.
>Linint2 does interpolation but not grid to grid.
>James Correia Jr., PhD
>Climate Physics Group
>Post. Doc.
>Pacific Northwest National Lab

ncl-talk mailing list
List instructions, subscriber options, unsubscribe:
Received on Sun May 17 2009 - 17:59:40 MDT

This archive was generated by hypermail 2.2.0 : Tue May 19 2009 - 09:17:30 MDT