v1.1.0

Refactored the code and made it available on PyPI. This release also folds in the interim 1.0.1 milestone — the workflow's rewrite from bash scripts to pure Python — which was never separately tagged.

Workflow rewrite (1.0.1 milestone)

• Fully YAML-file-based configuration; no more Linux environment variables
• Models are now Model classes with user-provided methods (read_var, etc.) instead of shell-script modules
• Ensemble forecast runs in either batch mode or via a job scheduler
• Workflow control moved to Python, with scripts/run_exp.py as the top-level entry point and assimilate.py/ensemble_forecast.py as the two main steps
• Adaptive inflation algorithms added

assim_tools refactor

• Refactored into composable component classes: Assimilators (ETKF, TopazDEnKF, EAKF), Updators (Additive, Alignment), and transform functions (null, scale_bandpass)

New models & datasets

• Models: lorenz96, nextsim/v1, nextsim/dg, noresm, qg (+ emulator), topaz/v4, topaz/v5, vort2d, wrf
• Datasets: era5, argo (ifremer), rgps, osisaf (ice_conc, ice_drift), qg, lorenz96, vort2d

Docs & examples

• readthedocs documentation site
• QG-model benchmark example comparing filter/DA algorithm performance