Closed platipodium closed 11 months ago
So I guess while your claim ".. useful for oceanographers" must be toned down to explicitly refer to structured (logically rectangular?) grids.
Hmm. That error message isn't quite right, as it's the data structure of the variable that CDO dislikes not the grid necessarily. I need to tweak some code so that it provides a better error message in this case.
Strictly speaking, half of the time CDO does work with unstructured grids. So it's tricky to provide clear guidance on this in the nctoolkit website.
Which claim are you referring to? The word "oceanographer" does not appear in the paper. Is this claim on the website?
You're right, not "oceanographer" but quoting from your statement of need:
"... oceanic model output .. irregular vertical grids"
Having the irregular vertical capability is of great help and sets your package apart from other software solutions. But in oceanography we not only have irregular vertical grids but also irregular horizontal grids. This has been (more or less) tackled by the psyplot package (also a JOSS paper, https://www.theoj.org/joss-papers/joss.00363/10.21105.joss.00363.pdf), but that one can't deal with irregular horizontal AFAIK
I've now added a "Data supported" section to the website that hopefully clarifies these issues.
Thanks for addressing this. The statement clarifies this, indeed, sufficiently. If you have the chance, I'd suggest you add some simple unstructured grids from typical ocean model outputs to your test suite and list support models.
Good suggestion. We use nctoolkit for fvcom analysis locally, so that should be added to the test suite. I've just raised an issue: https://github.com/pmlmodelling/nctoolkit/issues/93
I tried unstructured CF-compliant data with "Conventions = "CF-1.7, UGRID-1.0".
Nctoolkit failed on loading the dataset, ultimately resulting in
after lots of CDO warnings: