Status of geodesy packages in Julia (DGGS experiments)

[allixender]

Hi all,

I would like to instigate a short review discussion on geodesy-related packages in Julia. My background is GIS, geospatial, and recently Discrete Global Grid Systems (DGGS) in particular. If you have heard of Uber H3, Google S2, DGGRID, or A5 systems, this is the direction. I had some experiences with the Pangeo folks picking up DGGS [0]. But still it feels inconvenient in Python. The reason is, most DGGS-related software is currently implemented in C/C++ and increasingly Rust, with Python-bindings at different levels of maturity and speed. I am increasingly enticed by the “not 2 languages” promise of Julia and had some encouraging success in dabbling with JuliaGeo packages. So I would like to try to do more with DGGS in Julia.

One aspect in underlying geo logic is that most DGGS are implemented on the sphere and using them with ellpsoidal/WGS84 coordinates/geometries requires classic geodesic calculations to make an adjustment of the latitude values (authalic conversion) to be correct.
I am aware of two geodesy packages in Julia (Geodesy.jl and [1] and Geodesics.jl [2]). Geodesy.jl is apparently in maintenance mode, so I reached out to Geodesics.jl. I managed to port an extract of some of the routines for authalic latitude conversion to Julia ([3]), based from a Python package [4], which is based, like most packages, one way or another on Geographiclib by M. Karney [5].

But my question is also if there is interest in general to explore/compose native Julia implementations. There are DGGS-related activities like HEALPIX.jl [6] and DGGS.jl [7].

Thanks for consideration, happy to discuss more.
Alex

Some references:

• [0] Discrete Global Grid Systems (DGGS) use with Pangeo - Data - Pangeo
• [1] GitHub - JuliaGeo/Geodesy.jl: Work with points defined in various coordinate systems.
• [2] GitHub - anowacki/Geodesics.jl: Calculate great circles on a flattened sphere in Julia
• [3] Auxiliary latitudes feature? · Issue #15 · anowacki/Geodesics.jl · GitHub
• [4] GitHub - mrJean1/PyGeodesy: Pure Python geodesy tools
• [5] https://geographiclib.sourceforge.io/
• [6] GitHub - JuliaAstro/Healpix.jl: Healpix library written in Julia
• [7] GitHub - danlooo/DGGS.jl: Discrete Global Grid System for Julia

[juliohm]

Hi @allixender ,

Take a look into CoordRefSystems.jl and Meshes.jl on top of it. It is a strongly typed system with no equivalent in other languages.

Please reach out if you have questions or would like to contribute.

[allixender]

Thanks, yes and no Preferably in DGGS I’d like to go as long as possible without geometries, DGGS are also quite well-structured, so a mesh is a bit of a step backwards. But the package looks very powerful.
And CoordRefSystems look good, too, but if I see correclty it also does not yet implement spherical to ellpisoid. Would you see a way forward to add this (or use my draft implementation as provided above) ?

[joa-quim]

Packages like ArchGDAL and GMT.jl or others that depend on GDAL have access to the PROJ library, which AFAIK is what almost everybody uses in the outside world.

[juliohm]

@allixender

I’d like to go as long as possible without geometries, DGGS are also quite well-structured, so a mesh is a bit of a step backwards. But the package looks very powerful.

You might be surprised if you study the Meshes.jl source code. Dont be fooled by the name. Grids are structured and fully supported with lazy data structures.

@allixender

And CoordRefSystems look good, too, but if I see correclty it also does not yet implement spherical to ellpisoid.

It does. Study the source code in detail and feel free to ask questions. If it doesnt have a specific conversion yet, it should be straightforward to add. We already worked out the design to acommodate all use cases.