Falwa

There were resolution-dependent procedures which worked with the (360x181) grid we used for NHN2022, GRL. This bugfix enables the new inversion routine to take data of other grid sizes.

Major update:

- Thanks Christopher Polster chpolste for developing an **xarray interface** for the QGField object using the SOR solver to compute reference state as documented in [Nakamura and Huang (2018, Science)](https://www.science.org/doi/abs/10.1126/science.aat0721).
- Sample code that use this interface: `examples/nh2018_science/demo_script_for_nh2018_with_xarray.ipynb`
- Please refer to the corresponding [xarray interface documentation](https://hn2016-falwa.readthedocs.io/en/latest/Xarray%20Interface.html) page for details.

Minor update:

- Added `.gitignore` for this repo
- Fix documentation display on readthedocs.io
- Fortran code files are now moved to `hn2016_falwa/f90_modules/` such that the compiled `.so` modules are located in `hn2016_falwa/`
- Cleaned up f2py modules and remove unused variables

Code change

Detailed change in the code can be viewed on this GitHub commit comparison page:

https://github.com/csyhuang/hn2016_falwa/compare/44fabe7..a656062

to fields at the equator, which improves the quality of the analysis.

The analysis code to reproduce results in Neal et al. "The 2021 Pacific Northwest heat wave and associated blocking: Meteorology and the role of an upstream cyclone as a diabatic source of wave activity" (submitted to GRL) can be found in the directory `scripts/nhn_grl2022/`.

Zenodo badge: [](https://zenodo.org/badge/latestdoi/63908662)

- Great thanks to chpolste for pull request 36 that optimizes the fortran code that computes LWA and fluxes. To incorporate the changes made, after pulling the changes, please re-compile the f2py modules. One way of doing this is:

$ python setup.py develop -u
$ git pull
$ python setup.py develop

- The module `hn2016_falwa/download_data.py`, which downloads ERA-Interim data, has been removed.
- The module `hn2016_falwa/beta_version.py` has been moved to the `legacy/` folder. This module is no longer maintained, and will be deprecated in the upcoming release.
- Added one unit test for the class `BarotropicField`.
- Instead of using Travis CI, now deployment test and code coverage test are implemented using *GitHub workflow*.

close 37

There are two fixes in this release:
- The bugs in the class BarotropicField is fixed. Thanks chpolste for the pull request.
- Now `QGField` can handle masked arrays without raising an error. It converts masked array to numpy array with the `fill_value` stored in the masked array.

Enhancement in functionality
- The computation of LWA, reference state and vertically averaged fluxes in the **Southern Hemisphere** is now available in the `QGField` class!
- The computed variables can be accessed via the QGField class without calling any methods, e.g. the 3D wave activity can be retrieved from the QGField object `qgfield` by `qgfield.lwa`. See documentation/example for all the attributes.
- Added more sanity checks for the inputs of `QGField`.

Other changes
- The documentation page is fixed and now available at [http://csyhuang.github.io/hn2016_falwa](http://csyhuang.github.io/hn2016_falwa).
- Added test data and increase unit test coverage for `QGField`.