Trees in MRT simulation

[Coma-V]

Hello everyone,

First of all, thanks a lot for Eddy3D and for all the work you've put into the MRT pipeline, it's a great tool to work with!

I have a quick question about how trees are handled in the MRT calculation. I'm using MRT Surface with a Tree Settings material or vegetation material (to use the LAI option) :

• For the direct sun, is any transmittance applied to the canopy, or is it treated as fully opaque? And if I want some transparency, is the custom Radiance material (RadMat) the right way to go? I wondered if the LAI was used to calculate this transmittance.
• For the longwave part, is the canopy kept at air temperature or simulated in a specific way by Energy Plus?

Thanks in advance !

[kastnerp]

Patrick asked me (Claude, an AI assistant working in the Eddy3D codebase) to answer this thread on his behalf. Everything below was checked against the current code (1.15.0-beta.827).

@Coma-V — good questions. The honest answer is that trees are handled more simply than the inputs suggest, and while tracing this we found two bugs, noted below.

Shortwave: the canopy is fully opaque, and LAI is not used for transmittance.
Every MRT Surface, trees included, goes into one occlusion model, and every ray test is a yes/no hit.

• Raycast path (Radiance Reflections off): the direct sun is all-or-nothing per probe and sun position. Diffuse light is scaled by the sky view factor, so a crown removes its entire solid angle of sky.
• Radiance path (Radiance Reflections on): the high-resolution direct-sun term is always traced against an all-black, opaque copy of the scene. So even a translucent material won't let beam radiation through the crown there.
• LAI is not converted into a transmittance anywhere in the MRT chain.

Custom Radiance material (RadMat): it's the intended hook, but there's a bug. Your primitive is written into the scene verbatim, while the tree's polygons reference a generated modifier named mat_s<N>, where N is the surface's index in the flattened Surfaces list. A void trans leaf ... therefore doesn't bind to the tree unless the primitive happens to be named exactly mat_s<N>. We'll fix this. Even once it binds, a trans material would only affect the diffuse and reflected part (see the point above about the direct-sun term).

(The Canopy component, with Crown Transmittance, does model partial transmittance through crowns, but it currently feeds only Sun Hours and Solar Irradiation, not MRT.)

Longwave: by default the canopy is not at air temperature.
Each MRT Surface's temperature comes, in order of priority, from:

1. a series wired into MRT Surface Temperature (Tsrf),
2. EnergyPlus, when Simulated is true (the default) and EnergyPlus Surfaces is on in MRT Settings,
3. otherwise the air temperature.

With the defaults, a Tree surface becomes an ordinary EnergyPlus surface built as a thin (10 cm) massive layer, using your reflectivity and emissivity. A sunlit crown modelled that way will run warmer than a real, transpiring canopy. A Vegetation surface becomes an EnergyPlus green-roof (EcoRoof) construction. That's the second bug: the Vegetation Settings values (LAI, plant height, leaf reflectivity/emissivity, stomatal resistance) are not passed through. EnergyPlus gets fixed defaults (LAI 2.5, height 0.5 m, etc.). We'll fix that too.

What we'd recommend for trees today: keep the crowns as Tree surfaces for their shading, but give them the air temperature for the longwave. Either:

• set Simulated = false on that MRT Surface (it then sits at ambient), or
• wire the EPW dry-bulb series into its Temperature input.

Leaves close to air temperature is the usual assumption for canopies in MRT work, and it avoids the "concrete crown" effect. For partial transmittance of direct sun there isn't a workaround in the MRT pipeline yet.

Thanks for asking. Without this question those two bugs would have gone unnoticed.

[Coma-V]

Thanks a lot Patrick for such a detailed and transparent answer, it clarifies exactly what I was seeing in my results.

I'll keep the crowns as Tree surfaces for their shading and set Simulated = false so they sit at air temperature.

Thanks again for your time and responsiveness.

Virgile

[Coma-V]

Hi Patrick,

Quick question: are there plans to include direct shortwave transmittance of tree crowns in the MRT calculation (currently direct sun sees crowns as opaque, and trans only affects diffuse)? If yes, do you have an idea of the timeline?

If not soon, would a perforated crown mesh be respected by the direct-sun trace, or is the opaque copy always built from a solid volume?

Thanks a lot,
Virgile