Project

General

Profile

Regridding and intepolation NCEP to HadCM3 grid

Added by Rashid Mahmood almost 12 years ago

hi,
I want to use the NCEP data for SDSM model for downscaling. So I want to regrid and interpolate NCEP grid data (2.5 by 2.5 deg) to HadCM3 grids (2.5 by 3.75 deg)
How can i use the cdo to do this job
Thanks


Replies (2)

RE: Regridding and intepolation NCEP to HadCM3 grid - Added by Uwe Schulzweida almost 12 years ago

Horizontal interpolation can be done with the CDO operators:
  • remapbil - Bilinear interpolation
  • remapbic - Bicubic interpolation
  • remapdis - Distance-weighted average remapping
  • remapnn - Nearest neighbor remapping
  • remapcon - First order conservative remapping

Parameter for these operators is a description of the target grid. There are several different ways to define the target grid (see CDO Users Guide https://code.zmaw.de/embedded/cdo/1.4.7/cdo.html#x1-110001.3). Here is an example for a bilinar interpolation:

cdo  remapbil,targetgrid  ifile  ofile
Before the interpolation you should check whether CDO reads the source grid correctly. You can use the CDO operator sinfo or griddes to check the grid information of the source grid. If the result is not correct or there is no grid information available (gridtype: generic) then you also have to define the source grid with setgrid:
cdo  remapbil,targetgrid  -setgrid,souregrid  ifile  ofile
Here is an example template of a grid description for a regular lon/lat grid:
gridtype = lonlat
xsize    =
ysize    =
xfirst   =
xinc     =
yfirst   =
yinc     =

Regards,
Uwe

RE: Regridding and intepolation Daymet to lat/lat grid - Added by Ping Yang about 10 years ago

Hi Uwe,

I am following on a previous top on the regriding and interpolation using cdo. now I have a file which has the following information:
cdo sinfo 11754_2010_prcp.nc |more
cdo sinfo: Processed 2 variables over 365 timesteps. ( 0.00s )
File format: netCDF
-1 : Institut Source Param Time Typ Grid Size Num Levels Num
1 : unknown unknown -1 var I16 1 1 1 1
2 : unknown unknown -2 var F32 43105 2 1 1
Horizontal grids :
1 : generic > size : dim = 1
2 : curvilinear > size : dim = 43105 nx = 185 ny = 233
lon : min = -74.6866982 max = -71.5919006 degrees_east
lat : min = 39.8215531 max = 42.4474037 degrees_north
available : xvals yvals
Vertical grids :
1 : surface : 0
Time axis : 365 steps
RefTime = 1980-01-01 00:00:00 Units = days Calendar = STANDARD Bounds = true
YYYY-MM-DD hh:mm:ss YYYY-MM-DD hh:mm:ss YYYY-MM-DD hh:mm:ss YYYY-MM-DD hh:mm:ss
2010-01-01 12:00:00 2010-01-02 12:00:00 2010-01-03 12:00:00 2010-01-04 12:00:00
2010-01-05 12:00:00 2010-01-06 12:00:00 2010-01-07 12:00:00 2010-01-08 12:00:00
2010-01-09 12:00:00 2010-01-10 12:00:00 2010-01-11 12:00:00 2010-01-12 12:00:00
....

What I plan to do is this: regriding(aggregating) this 1 km resolution into 4 km resolution and finally change the projection from "lamber_conic"(here showed curvilinear) into lat/lon.
My questions:
1) actually I have a bunch of this files (which is tiled in the 2 degree*2 degree) for my study area, I am asking which I should do first,
A: do the regriding(interpolation) for each file and then merge them spatially to one grid
B: merge all the files into one grid and do the regriding(interpolation)
2) for the regriding, if I need regrid it to 4 km, I plan to use different methods for the interpolation, I am asking how to define the target grid (like the r360x180):
cdo remapbil,r360x180 -selname,thetao ifile ofile

3)should I do the project(from lambert to lat/lon) first or the regridding first?

Looking forward to hearing from you.

Best Regards,

Ping

    (1-2/2)