Estimating Infrared irradiance into surfaces?

[cesss]

Hi!

Can your sky model we used for estimating the infrared irradiance into a surface, in W*m-2? I'm asking this because I see that your papers seem to be limited to the visible spectrum, but at the same time you have researched a lot about the sky physics, so maybe you can provide some advice about this.

What I want to simulate is heating of surfaces due to the Sun and the sky dome. At the moment I was looking at the (luminance-only) Perez model, but then I learnt about your model and wondered if it could be more realistic than Perez for the infrared radiation.

[cesss]

Actually, I just realized that what I need is not just the infrared range, but the whole thermal radiation (ie: summing the irradiance for all wavelengths)... would it be possible to do this with your model, or is it limited to the visible spectrum only? I'd need the W*m-2 on surfaces as total thermal irradiance coming from the sun and from the sky dome.

[ebruneton]

This is not possible without extending the model. The atmosphere and the ground can emit infrared, and there are currently no emission terms in the model (only absorption, scattering and reflectance). Also there is not necessarily a radiative equilibrium over the whole spectrum (the temperature could be increasing or decreasing), unlike in the visible range, and the model currently assumes this.

[cesss]

Thanks a lot for your reply. I think I feel confused because the Perez model has been used both for solar energy simulation (the Insight Solar Analysis plugin for Revit and other solar energy software are based on it) and for realistic rendering and lighting simulation (notably Radiance was perhaps the first renderer to include the Perez model).

The common area between both grounds (energy simulation and lighting simulation) is the Stefan–Boltzmann law, which links temperature to radiation power. But however, it works for the overall power across all wavelengths (which might explain why the Perez model is monochromatic, as I believe it was developed thinking in energy simulation, and not in realistic rendering).

Of course, a monochromatic model is not accurate enough for physically-based rendering, so quite a few new models (including yours) have been developed, but their field of research is rendering, not energy. The thing is that all these new models compare themselves to the results of the Perez model, so if you are doing energy simulation, you cannot avoid the question of how these new models could be used for energy simulation, and whether they would be more accurate for the simulation or not.

The usual formulation for estimating the temperature increase in a surface due to solar radiation can be read for example in this thread, and you can see that it includes some values that are either not straightforward to achieve, or empirical, such as the "sky temperature" for example (estimating the temperature increase on a surface depends not only in the solar radiation, but on the "sky temperature" as well).

I firmly believe there should be some model that could be valid for both areas of research (energy and rendering), designed in a way flexible enough for avoiding the energy vs rendering confusion.