Plugin for More Accurate Light Pollution Simulation

[smln]

Hi all,

I've been mulling over an idea I'd like to work on in the coming months.

Basically, I want to build a plugin for Stellarium which will generate a light dome for light pollution at any arbitrary GPS coordinate.

The utility of this plugin would exceed that of Stellarium's existing light pollution tool in that for each pixel above the horizon a distinct magnitude/arcsec^2 value can be displayed. These values would be generated using publicly available nightlight data (i.e. Black Marble), elevation data, and some fancy math.

Example use case: A Bortle 4 location with a large city 50 miles to the west and a mid-size town 20 miles to the south. The plugin would generate a light dome with a diffuse bright area on the western horizon and a more localized bright area to the south, and darker skies to the north and east.

What I need are resources and tips for getting started. I have some familiarity with C++ coding, although I haven't touched it for some time. I've used Stellarium for years, but this is my first time looking at its source code.

• Does anyone have any suggestions on where to start looking in the source code to understand Stellarium's internal structure, how to render the light dome, etc.
• Is there a forum or chat room the devs use to communicate / collaborate other than Github?
• Any thoughts on whether this idea is suitable for a plugin, or would it make more sense to fork Stellarium and add it as an option in the Sky and Viewing Options window?

The idea for this plugin is inspired by many trips to dark sky sites, and the inadequacy of the Bortle scale to fully capture the nuances of light pollution in a given location. I also do astrophotography and am a little tired of showing up to a Bortle 1 or 2 site, only to find an unexpected light dome on the horizon interfering with a shot I had planned 😅

Any help and guidance is very much appreciated! And if anyone is interested in collaborating on this with me, please reach out and let's chat 😊

[10110111]

Do you have any model for this?

I was planning to implement a similar thing as part of CalcMySky simulation, but still haven't found a good way of saving the data that would yield a small enough dataset (because this would include dependence of luminance on observer's altitude, distance to the city, as well as azimuth and altitude of view direction).

[smln]

My rough plan is to use ray tracing to calculate each patch of sky based on nearby light sources and the atmospheric thickness at that point. To make it computationally feasible, my plan is to use a low granularity with bicubic interpolation.

To be honest though, I'm not sure if this is the best or even a practical approach 😅. I still need to do a lot of research and experimenting.

Does Stellarium use CalcMySky for its atmospheric modelling? I'll need to read more about it to understand what you are saying about needing to yield a smaller dataset. Are you able to elaborate on that a little more for me?

[10110111]

Does Stellarium use CalcMySky for its atmospheric modelling?

Yes, and you can see some useful references about the methods used in its README.

[10110111]

P.S. you need to choose the ShowMySky model in the View settings to actually use it, otherwise a legacy model will be used by default.

[gzotti]

I only know about the modelling papers by Garstang (e.g. https://iopscience.iop.org/article/10.1086/132933) and Cinzano https://arxiv.org/abs/1209.2031

[smln]

Thank you, I will look into these!

[roman-dvorak]

I really like this idea. The suggested solution is maybe too heavy but it would be useful

[oneredpanther]

Potentially of interest: I built a light pollution simulator at www.darkskysites.com that uses VIIRS input and a custom volumetric atmospherics shader to compute global horizon glow for any location on Earth. Select a location of interest and choose > Virtual 360 Horizon.

[10110111]

I'm getting "Error: WebGPU adapter not available" when trying to get a 360° All-Sky simulation on Chromium Version 137.0.7151.103 (Official Build) for Linux Mint (64-bit). As a sanity check, WebGL Water demo works perfectly in the same browser.

[oneredpanther]

Thanks for this. I added a fallback and warning. Try chrome://flags/#enable-unsafe-webgpu and reload, let me know.

[gzotti]

Now that's impressive! A QWebEngine-based plugin could access your website and probably get an all-sky pano to be used as light texture in all Landscapes. Either as permanent addition to whatever is the current landscape, or as dynamic addition that injects the light pollution layer only into the current session, or @10110111 wants to integrate the information to his skylight model. (Or both? SMS drains more power from field laptops, so I rather use the old model there.) Do I see right you also make horizon calculation from DEM? Then maybe even as complete Landscape with separate terrain and light layers? Maybe @smln can work with you on this?

[oneredpanther]

Yeah I think that's definitely possible. I have a (gated, on request) public API that can generate all sorts of sky data for any lat/long, and could be modified to compute and output hi-res panoramas for ingestion to Stellarium. I'll think about it. And yes, I baked a DEM into the platform to calculate terrain masks for the horizon. It's all layered and modular.