CDO

Hi everyone,

I have a proposed feature addition for cdo, based on my experiences working with CMIP data, primarily the ocean(and ice) model data which is on curvilinear or unstructured grids. The goal is to make cdo capable of generating interpolation and rotation weights for vector valued fields from any one geophysical grid to another.

Last year I found the work of Grady B. Wright [https://gradywright.github.io/research.html]. Some of his work involves solutions to the general problem of interpolation and rotation of multivariate data at arbitrary spatial locations. His work started with global data but progressed towards solutions on finite domains. His work is general enough that it could take into account the u, v and mass grid locations of the Arakawa family of grids, as well as unstructured grids. What I do not know is how computationally expensive the numerical solutions are. However, I think adding this capability to cdo would be extrememly useful. I say this not only because of my own work, but in response to the number of questions asked on the cdo forums about interpolating/rotating model output.

I did communicate with Dr.Wright last year, and he responded that he was having difficulties getting funding for this specific work.

In a perfect world, the final piece of the puzzle would be a standardized description of geophysical model grids and the inclusion of this grid description metadata in the CF-Metadata standard. I asked about this on the CF-Metadata Github but no one shared my enthusiasm for such an endeavour. There has been prior work on standardization geophysical model grids [https://arxiv.org/abs/1911.08638], but I suspect such work, as with Dr.Wright's work, is simply not seen as "sexy" enough to get funding.

I am pretty sure that at the moment, all your development energies are principally directed towards updating cdo to a more recent version of the CF standard including data groups. However, please let me know if this idea holds any interest to you or if you think it might be feasible. Although the math and programming involved are way beyond my understanding, I am very good at communications and bringing people with a shared vision together. To that end I end, I am emailing a copy of this post to Dr.Wright to let him know that I am talking about his work.

All the best,
Brendan