To start consolidating the work that @sadamov and @joeloskarsson have done to train neural-lam on COSMO and DANRA reanalysis datasets I thought it would be a good exercise for me to work through their research branch () and identify specific fixes and features that could be upstreamed:
- add support for supplying lead-time to
top-of-atmosphere derived variable calculation: sadamov@b1901a1
- fixes for how target dimensions for derived variables are defined: 8040378
To start consolidating the work that @sadamov and @joeloskarsson have done to train neural-lam on COSMO and DANRA reanalysis datasets I thought it would be a good exercise for me to work through their research branch () and identify specific fixes and features that could be upstreamed:
top-of-atmospherederived variable calculation: sadamov@b1901a1