Pycontrails

Fixes

- Fix bug in which `Cocip._process_rad` dropped radiation dataset attributes introduced in v0.49.4.

Breaking changes

- Remove the `CocipGridParams.met_slice_dt` parameter. Now met downselection is handled automatically during contrail evolution. When the `met` and `rad` data passed into `CocipGrid` are not already loaded into memory, this update may make `CocipGrid` slightly more performant.
- No longer explicitly load `met` and `rad` time slices into memory in `CocipGrid`. This now only occurs downstream when interpolation is performed. This change better aligns `CocipGrid` with other pycontrails models.
- Remove the `cocipgrid.cocip_time_handling` module. Any useful tooling has been moved directly to the `cocipgrid.cocip_grid` module.
- Remove the `CocipGrid.timedict` attribute. Add a `CocipGrid.timesteps` attribute. This is now applied in the same manner that the `Cocip` model uses its `timesteps` attribute.
- Simplify the runtime estimate used in constructing the `CocipGrid` `tqdm` progress bar. The new estimate is less precise than the previous estimate and should not be trusted for long-running simulations.
- Deprecate `MetBase.variables` in favor of `MetBase.indexes`.

Features

- Add support for 9 additional aircraft types in the [Poll-Schumann](https://py.contrails.org/notebooks/AircraftPerformance.html) (PS) aircraft performance model. The new aircraft types are:
- A338
- A339
- A35K
- B37M
- B38M
- B39M
- B78X
- BCS1
- BCS3
- Modify PS coefficients for B788, B789, and A359.
- Support running `CocipGrid` on meteorology data without a uniformly-spaced time dimension. The `CocipGrid` implementation now no longer assumes `met["time"].diff()` is constant.
- Add a `MetDataset.downselect_met` method. This performs a met downselection in analogy with `GeoVectorDataset.downselect_met`.

Fixes

- Improve clarity of warnings produced when meteorology data doesn't cover the time range required by a gridded CoCiP model.
- No longer emit `pandas` warning when `Flight.resample_and_fill(..., drop=True, ...)` is called with non-float data.
- Correctly handle `CocipGrid` `rad` data with non-uniform time steps.

Features

- Re-organize notebooks in documentation.
- Add new [model level](https://py.contrails.org/examples/model-levels.html) tutorial notebook.
- Add new high-level `Flight.clean_and_resample` method. This method parallels the `Flight.resample_and_fill` method but performs additional altitude filtering. In essence, this method is a combination of `Flight.filter_altitude` and `Flight.resample_and_fill`.

Breaking changes

- Remove `Flight.fit_altitude` method in favor of `Flight.filter_altitude`. The new method now only applies a median filter during cruise flight phase.

Fixes

- Remove opaque warning issued when all tau_contrail values are nan in `Cocip` evolution.
- Emit warning in `Cocip.eval` if the advected contrail is blown outside of the domain of the met data.
- Remove empty flights in `Fleet.from_seq`. Issue warning if an empty flight is encountered.
- Emit warning when `Flight.resample_and_fill` returns an empty flight.

Internals

- Modify test workflow to use Makefile recipes and ensure early failures are detected in CI.
- Pin `black` and `ruff` versions for consistency between local and CI/CD environments.
- Improve development documentation.
- Improve handling of missing credentials in tests (`make nb-test`, `make doctest`).
- Update time frequency aliases for `pandas` 2.2 compatibility.
- Update cython annotations for `scipy` 1.12 compatibility.
- Improve notebook output testing capabilities (`make nb-test`).
- Add new convenience Make recipe to execute all notebooks in the docs (`make nb-execute`).
- Add new Make recipe to cleanup notebooks (`make nb-clean`).
- Add pre-commit hook to check if notebooks are *clean*.
- Re-organize notebooks in documentation.
- Clean up contributing and develop documentation.
- Automatically parse `np.timedelta64`-like model params in `Model.update_params`.

Features

- Support [pandas Copy-on-Write](https://pandas.pydata.org/docs/user_guide/copy_on_write.html). This can be enabled with `pd.set_option("mode.copy_on_write", True)` or by setting the `PANDAS_COPY_ON_WRITE` environment variable.

Fixes

- Ensure the `Flight.fuel` attribute is preserved for the `Flight.filter` method.
- Ensure the `Fleet.fl_attrs` attribute is preserved for the `Fleet.filter` method.
- Raise `ValueError` when `Flight.sort` or `Fleet.sort` is called. Both of these subclasses assume a custom sorting order that is enforced in their constructors.
- Always correct intermediate thrust coefficients computed in the `PSGrid` model. This correction is already enabled by default in the `PSFlight` model.
- Include `attr` fields in the ValueError message raised in `CocipGrid` when not all aircraft performance variables are present on the `source` parameter.
- Allow `mach_number` as a replacement for `true_airspeed` in `CocipGrid` aircraft performance determination.

Internals

- Make `Fuel` and its subclasses `JetA`, `SAFBlend`, and `HydrogenFuel` frozen.
- No longer copy `met` when `Models.downselect_met` is called. In some modes of operation, this reduces the memory footprint of the model.
- Update codebase for more harmony with [PDEP 8](https://jorisvandenbossche.github.io/pandas-website-preview/pdeps/0008-inplace-methods-in-pandas.html) and Copy-on-Write semantics.
- Add `default` parameter to the `VectorDataset.get_data_or_attr` method.

Fixes

- Fix memory bottleneck in `CocipGrid` simulation by avoiding expensive call to `pd.concat`.
- Require `oldest-supported-numpy` for python 3.12. Remove logic for `numpy` 1.26.0rc1 in the `pyproject.toml` build system.

This release updates the Poll-Schumann (PS) aircraft performance model to version 2.0. It also includes a number of bug fixes and internal improvements.

Features

- Add convenience `Fleet.resample_and_fill`.
- Update the PS model aircraft-engine parameters.
- Improve PS model accuracy in fuel consumption estimates.
- Improve PS model accuracy in overall propulsive efficiency estimates.
- Include additional guardrails in the PS model to constrain the operational limits of different aircraft-engine types, i.e., the maximum permitted Mach number by altitude, maximum available thrust coefficient, maximum lift coefficient, and maximum allowable aircraft mass.

Fixes

- Update polygon algorithm to use `shapely.Polygon` instead of `shapely.LinearRing` for contours with at least 4 vertices.
- Fix `Fleet.to_flight_list` to avoid duplicating global attributes on the child `Flight` instances.
- Add `__slots__` to `GeoVectorDataset`, `Flight`, and `Fleet`. The base `VectorDataset` class already uses `__slots__`.
- Add `Fleet.copy` method.
- Improve `Fleet.__init__` implementation.
- Ensure `source` parameter is mutated in `CocipGrid.eval` when the model parameter `copy_source=False`.