conservative interpolation

General scientific issues regarding ROMS

Moderators: arango, robertson

Post Reply
Message
Author
turuncu
Posts: 128
Joined: Tue Feb 01, 2005 8:21 pm
Location: Istanbul Technical University (ITU)
Contact:

conservative interpolation

#1 Unread post by turuncu »

Hi,

I try to interpolate heat fluxes from atmospheric model to ROMS and i want to use conservative type interpolation for heat and rain fluxes. To do that i need to calculate area of the each grid cell in ROMS model. I was using following code to calculate rho centered grid area,

Code: Select all

          area(i,j) = (1.0d0/pm(i,j))*(1.0d0/pn(i,j))
As it can see from code, the pm and pn can be used to calculate grid area but in this case the grid area is in projected coordinate and area of all grid cells are in same value. Otherwise, the area of the grid cell is different in spherical coordinate (not projected one). Which one must be used to interpolate data into ROMS grid? Projected area or non-projected one? The non projected grid areas can be calculated using following formula,

Code: Select all

          area(i,j) = (6.3712d6*6.3712d6)*
     &                     abs(grid_corner_lon(1,i,j)*deg2rad-
     &                         grid_corner_lon(4,i,j)*deg2rad)*
     &                     abs(dsin(grid_corner_lat(3,i,j)*deg2rad)-
     &                         dsin(grid_corner_lat(4,i,j)*deg2rad))
Any suggestion are welcome :)
Thanks,

User avatar
kate
Posts: 4091
Joined: Wed Jul 02, 2003 5:29 pm
Location: CFOS/UAF, USA

Re: conservative interpolation

#2 Unread post by kate »

I assume you want to be consistent with what ROMS considers the area, or 1/(pm*pn). That could be the projected area or the unprojected area, depending on how you set up your pm,pn. What I do is make the grid in a flat coordinate system, then recompute the pm,pn based on distance on a sphere (or ellipsoid) once I know the lat,lon of the gridpoints.

Post Reply