[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

20050927: computing an overlay using multiple grids



Matt,

The current version of GEMPAK (5.8.3a) completed the transition of
all grid programs to support calculations involving grids of different
projections/domain. Gdbiint is no longer needed.

The MISS(S1,S2) function can be used to replace missing values in a grid
S1 with values in S2. You can create a new grid in a program, or through
gddiag and interpolate the high resolution to the new grid, and then
use the MISS function to set the remaining missing values to the
low resolution grid data values. 

Steve Chiswell
Unidata User Support

On Tue, 2005-09-27 at 11:53, address@hidden wrote:
> I have two SST grids of differing dimensions, one low-res grid which
> covers my entire model domain (GAREA=15;-110;50;-50), and one high-res
> which covers only a small subset of my model domain (GAREA=25;-90;40;-67).
>  Ultimately what I'd like to do is: overlay the high-res grid on the low
> res-grid, perform a smoothing function, save the overlay as a new grid,
> convert the overlay grid to grib so that it can be used as input into WRF.
> 
> I was thinking that I could just use gdbiint to interpolate the high-res
> grid to the same grid as the low-res SSTs, hoping that GEMPAK would just
> write missing values in for the area for which the high-res grid does not
> cover.  But when I do this my high-res grid has SST values ranging from
> negative infinity to infinity.  So either I've done something wrong...or
> GEMPAK expects the grid to cover the same GAREA as the grid it is being
> interpolated to.
> 
> I know I could use fortran to expand the high-res grid with blank
> columns/rows so that it covers my entire model domain.  I am really hoping
> I can avoid this though, as it would be a bit painstaking with my limited
> fortran background to ensure that I'm adding the correct number of grid
> points in the correct places.
> 
> I figure since I can do the overlay visually in GEMPAK, that there must be
> a way to write out the results to a new grid...as it seems like such a
> simple thing to do.  Is there a program I'm overlooking?  Can someone help
> me out with this?
> 
> Thanks,
> Matt