JOSS paper

[milankl]

Hey Team! I finally got around writing our efforts here up into a paper to be submitted to JOSS 🎉 With v0.6 released two weeks ago I feel the model is now in a usable state and with a hopefully relatively stable user interface, with its interactivity and extensibility definitely something new for atmospheric models. You can read the current draft here (and it will be recompiled if you submit changes to paper.md to this pull request)

https://github.com/SpeedyWeather/SpeedyWeather.jl/blob/mk/josspaper/docs/joss/paper.pdf

I've decided to invite you all to be co-authors. Some of you contributed code, others contributed ideas that crucially changed where SpeedyWeather is today but you all believed in this project. Thank you for that 💙

As many of you don't know each other, I want to be more specific why I'd like to have you as co-author

• @maximilian-gelbrecht for many fundamental ideas about the model structure, user interface, contributions to the LowerTriangularMatrix implementation, and initial efforts to support GPUs (even though that hasn't been priority lately)
• @hottad for many ideas around the abstraction of our grids, thanks to which we have a grid-flexible spectral transform, the implementation of the OctaHEALPixGrid and contributions to our interpolation library in RingGrids
• @jmert for algorithms towards a grid-flexible spherical harmonic transform and teaching me through your blog series how to code this up in the first place. I essentially copied many parts of your code and only tweaked it to our needs 🙈 I'm sure you'd still find some of your handwriting hidden in our SpeedyTransforms module
• @simone-silvestri for fixing our vertical advection scheme, completely abstracting this component of our primitive models and even implementing a fancy WENO scheme
• @glwagner for many suggestions around the user interface and the urgent call to abandon a monolithic user interface. The words "interactivity, extensibility and composability" would probably not be in the title if it wasn't for your ideas although that's what I always wanted SpeedyWeather to be!
• @white-alistair for your implementations of our convection and large-scale condensation schemes. I already updated one of these to our new logic, the other one will follow soon!
• @samhatfield for first attemps to translate SPEEDY to Julia, much insight on how SPEEDY works and conceptual and analytical contributions to our spectral transform and its details
• @dmey for initial translations of SPEEDY's short and long-wave radiation schemes to Julia, I haven't gotten around it yet to overhaul them to our new logic but will try to do soon!
• @tomkimpson for your first translations of the primitive equation tendenices and the time stepping to Julia back in Oxford days! It took a while to patch it alltogether and get it going 🙈
• @navidcy for all your contributions to the documentation. I'm glad I had your inputs on how to write reproducible model setup examples and how to run it sections
• @christophernhill for all your guidance along the way when I suffered through the implementation of the semi-implicit scheme and the adiabatic term and many other things that took waaaay longer than anticipated! Anyway it's also a better model now!

Could you all read through the draft paper and commit changes directly to this pull request? (Or alternatively just post them in here). Also I believe many of you haven't actually used the model lately, feel free to follow some of the example model setups that are now documented. While you are busy doing that there are still a few things I want to do before hitting the submit button

• Document the Extending SpeedyWeather with some examples on how to get new schemes in
• Document the few parameterizations we have
• Add some surface fluxes so that Fig 1 doesn't actually represent an ocean of concrete with no fluxes and to close the hydrological cycle with a resupply of moisture to the atmosphere!

Anyway thanks for all your work and I'd love to hear your opinion on the current state of SpeedyWeather, its documentation and the paper draft 🚀

[navidcy]

a few comments after a first read

[hottad]

Thank you Milan!
I have nothing to say about this excellent manuscript, except that I have never been affiliated with Uni of Reading.
Also, I spotted some typos.

[milankl]

Thank you Milan! I have nothing to say about this excellent manuscript, except that I have never been affiliated with Uni of Reading. Also, I spotted some typos.

Wonderful, thanks Daisuke. Apologies for the Uni Reading, I don't know how this happened 🙈 Probably copied it from somewhere?

[samhatfield]

Lovely work Milan! Nothing to add but some very minor editorial comments.

[samhatfield]

Though on second thought I do have some questions:

• Is this intended as a very brief kind of announcement that SpeedyWeather.jl exists, rather than a long model description paper? I guess you're not in a position yet to do climatology verification like in the original 2003 SPEEDY paper.
• How about adding some simulated-things-per-thing numbers? Typical speed obtained on a regular laptop running on 4 threads etc.?
• Perhaps you should add a sentence about data formats -> NetCDF I/O supported.

[milankl]

@maximilian-gelbrecht @hottad @samhatfield @simone-silvestri @tomkimpson @white-alistair @jmert @christophernhill @glwagner ☝🏼 Our JOSS paper is accepted and published now!! 🎉

https://joss.theoj.org/papers/10.21105/joss.06323

[simone-silvestri]

Great work @milankl !!!

[samhatfield]

Excellent!

[tomkimpson]

Fantastic. Well done @milankl