Forums » Open discussion »
Regridding MERRA data -- What grid type to use?
Added by Laura Hinkelman about 10 years ago
Hi,
I want to conservatively regrid MERRA data, which is on a 2/3 deg lon, 1/2 deg lat grid, to a basic 1 deg x 1 deg grid. Fine, just use remapcon or remapcon2, right? The problem is that I don't see a definition of the right type of grid for the MERRA data: There are not 360 latitudes but 361, i.e., there is a point at each pole. This means that there are 540 "overlapping" points at each pole. I'm told that this means that, while the areas represented by the other points are +/-0.25 and +/-0/333 degrees about the point, the areas represented by the points with latitude index 1 or 361 represent triangles running from, say 89.75 to 90 degrees. I can only find one definition of a CDO grid that sounds like this, but of course it assumes equal spacing of the points: rNXxNY, where the longitudes start at 0 with an increment of (360/<NX>
and the latitudes go from south to north with an increment of (180/<NY> . I'm guessing that I need to make some kind of grid description data file? As I mentioned, within the MERRA file, the locations don't all mean the same thing. Surely there's a quick solution to the problem because people must regrid similar files all the time.
Thanks for your help and for maintaining the CDO system,
Laura
Hi Laura,
Could you please upload a small example file? This will help a lot to find a solution for this issue.
Cheers, Uwe
OK, I cut most of the variables out of this one. I'm not sure it will tell you much, though, unless you recognize the format. It's not obvious that, if you consider the points to be grid box centers, you have to treat the poles differently, unless you think about it.
You do realize that those smileys above were unintentional, right?
You don't need to take care about the MERRA grid, it is handled correctly in CDO. The file contains all information needed for a conservative interpolation. Here is an example to interpolate the data to a global 1 degree grid:
cdo remapcon,global_1 merradata result
If you don't like the definition of the predefined grid
global_1 then you have to define you own grid by writing a grid description file.
Hi Uwe,
OK, I'm looking at the documentation again now. I do see that the grid specifier is for the OUTPUT grid, not the grid of the input file. So you're saying that CDO just knows how to handle this data format? Obviously the reason I'm confused is that none of the standard horizontal grid types described in the documentation (regular, Gaussian, or hexagonal) fits the MERRA data, and I would've expected reanalysis data to be on a grid that meets some king of standard.
Laura