/cumulonumba

Moist thermodynamic and convective parameters implemented with Numba

Primary LanguagePythonApache License 2.0Apache-2.0

Cumulonumba: moist thermodynamic and convective parameters implemented with Numba

This project is in a proof-of-concept/pre-alpha state, based on the immediate research needs of @jthielen for fast CAPE calculations compatible with Awkward Arrays. Caveat emptor!

More details to come, but suffice to say for now that this is a collection of MetPy calculations ported to Numba JIT routines with all the Pint- and xarray-specific handling stripped out, and top-level interfaces redesigned for "many-profile" style data structures.

Dev Notes

Overall Plan/Brainstorming

  • Goal is to have at least CAPE, CIN, and SRH implemented in numba and runnable on Awkward Array (as I would get out of a parquet store)
  • Should be easy though, once having that, to also have a version for grids (as a demo)
  • Not sure on API though, given that your "awkward array" version will likely be structured different from a gufunc/grid version?
  • there will be lots of copy-from-metpy, remove-unit-stuff (esp. tests!)