Re: Bilinear Interpolation

Hi Chris,

WRF output is usually *NOT* on a regular lat-lon grid.

You have to convert both higher and lower resolution grids to a regular
lat-lon grid using rcm2rgrid before the comparison

The function is documented at
http://ncl.ucar.edu/Document/Functions/Built-in/rcm2rgrid.shtml

; An example, assuming your file pointer is called "f"

    times = wrf_user_list_times(f) ; get times in the file
    ntimes = dimsizes(times) ; number of times in the file
    var = wrf_user_getvar(f,vname,ntimes-1)
    lat2d = f->XLAT(0,:,:) ; this depends on the kind of variable
    lon2d = f->XLONG(0,:,:) ; this depends on the kind of variable
    dimll = dimsizes(lat2d) ; get size of dimensions
    nlat = dimll(0)
    mlon = dimll(1)
    lat1=(lat2d(0,0))
    lon1=(lon2d(0,0))
    lat2=(lat2d(nlat-1,mlon-1))
    lon2=(lon2d(nlat-1,mlon-1))
    lat = fspan(lat1,lat2,nlat)
    lon = fspan(lon1,lon2,mlon)
    var_r = rcm2rgrid(lat2d,lon2d,var,lat,lon,0)

--
saji
On Wed, Oct 6, 2010 at 4:10 PM, Christopher Steele <
christopher.steele@uea.ac.uk> wrote:
> Hi,
>
> I've been attempting to do a comparison of WRF simulations with differing
> resolutions. To do this I have interpolated the lower resolution simulation
> onto a grid the same size as the higher one using the linint2 command. Then
> I simply find the difference between the two. The script itself works but it
> produces strange output. There is a band of data running along the SW-NE
> diagonal that is close to zero (or at least giving differences that are of a
> magnitude that you would perhaps expect), which is sandwiched between
> regions of unrealistically large values in the NW and SE corners. These
> areas are the polar opposite to one another, ie, the higher resolution run
> is unrealistically larger than the low in the NW and the lower is
> unrealistically larger than the high resolution in the SE. These patterns
> are apparent in the simulations at all times, suggesting that somehow one of
> the data sets have been flipped in both the horizontal and vertical
> dimensions though I cannot see how this has happened.
>
> Chances are I've made a glaring mistake somewhere in the script that I
> can't see (possibly from staring at it blankly for too long), but has anyone
> had experience with interpolating WRF data like this before and could offer
> help?
>
> Sorry if the script appears a little messy!
>
> cheers
>
> Chris
>
>
>
> _______________________________________________
> ncl-talk mailing list
> List instructions, subscriber options, unsubscribe:
> http://mailman.ucar.edu/mailman/listinfo/ncl-talk
>
>

_______________________________________________
ncl-talk mailing list
List instructions, subscriber options, unsubscribe:
http://mailman.ucar.edu/mailman/listinfo/ncl-talk
Received on Wed Oct 6 01:57:57 2010