CDO

Hi,

From CDO docs: " ... interpolation is based on an adapted SCRIP library version. For a detailed description of the remapping methods see SCRIP" here http://climate.lanl.gov/Software/SCRIP/

You can find a detailed description of the remapping methods in the above link.

Also from CDO docs: [just in case, if you haven't noticed ...] "*Note:* For this module the author has converted the original Fortran 90 SCRIP software to ANSI C99. If there are any problems send a bug report to CDO and not to SCRIP!"

I'm not an expert on the subject but I think, the remapping method you want to use depends on the distribution and type of the variable. For example, I wouldn't use bilinear/bicubic to interpolate precipitation data but the same can be used to interpolate smooth [model] fields like Temperature..

Hope this helps,

Cheers,
J.

Hi,

This is very helpful, Thanks.

Can you suggest a remapping method for precipitation, Could you list some reason on why didn't use bilinear/bicubic to interpolate precipitation? I tried all the 3 three methods(Bilinear, Conservative and IDW), it seems that bilinear performs better than other two(from the monthly, seasonally and annually statistics).

I will be appreciated if there is an expert here to give suggestions(comments) on choosing those remapping methods for interpolation(regridding) from coarse resolution (e.g. 30 arcmin) to higher resolution (3 arcmin) for the atmospheric variables (precipitation, temperature, radiation, wind speed,etc)

Big thanks.

Ping

Could you list some reason on why didn't use bilinear/bicubic to interpolate precipitation?

You may end up in negative precipitation (ie the "fldmin") - although it's just a number, it doesn't have any physical meaning. Most cases, I use remapcon.

Cheers,
J.

Hi,

I am asking here why there are no other interpolation methods include Kriging and Spline in CDO? or Are they will be included in the new version of CDO?

Fortunately, Bilinear interpolation for precipitation works well with me (I didn't have negative values at all), but introduced small bias.

Best Regards,

Ping