LunarLeaf-CFD · Astrobotany / AIRI

Gravity, geometry and enclosure control the gas-exchange boundary layer of Arabidopsis

A validated, in-browser computational fluid-dynamics model of how the loss of buoyant convection in microgravity thickens the leaf boundary layer and steepens O2/CO2/H2O gradients — from a single leaf to a canopy to real spaceflight growth hardware — and what that does to carbon gain.

Lattice-Boltzmann CFD 4 numerical benchmarks Anchored to measured gas exchange BRIC · CARA · VEGGIE npj Microgravity-style draft

01Overview

Plants grown in spaceflight show stress phenotypes whose mechanistic basis is debated. A long-suspected contributor is the loss of buoyancy-driven convection in microgravity, which should thicken the unstirred boundary layer around plant organs and steepen the gas gradients that govern exchange — but no whole-chamber sensor resolves the sub-millimetre boundary layer. We test the idea with a browser-based lattice-Boltzmann model that solves incompressible flow, multi-species (CO2/O2/H2O) advection–diffusion and Boussinesq buoyancy under an adjustable gravity vector. The solver is validated against four textbook benchmarks and anchored to measured Arabidopsis whole-plant-chamber gas exchange (net assimilation 3.85 µmol CO2 m−2 s−1). We separate what the data validate (flux magnitude, closed-chamber transport) from what the model predicts (the spatial, gravity- and scale-dependent gradients).

02Key findings

1.5–1.8×Steeper surface gas gaps from Earth to microgravity, at every scale (convection ∝ √g → 0).
2.6 → 11 → 21 cm/sForced airflow needed to null the µg penalty for a leaf → rosette → canopy — denser stands need far more ventilation.
~1% vs ~100%12-h carbon fixed in a sealed BRIC dish vs a ventilated VEGGIE system, once photosynthesis is CO2-limited.
~7 min / ~6.5 dSealed-dish CO2 exhausted in light / O2 to hypoxia in the dark (analytic mass balance).

03A trustworthy solver, anchored to real data

Before any biology, the solver reproduces four independent benchmarks; setting g = 0 then genuinely collapses convection to the diffusion limit.

GateBenchmarkReferenceModel
1Lid-driven cavity, Re 100Ghia et al. (1982)L2 = 0.011 ✓
2Cylinder wake, Re 100Kármán sheddingSt = 0.192 ✓
3Natural convection, Ra 10⁴de Vahl Davis (1983), Nu = 2.238Nu = 2.242 ✓
4Pure diffusion½·erfc analyticL2 ≈ 0 ✓
Closed-chamber mass conservation and the sustained surface-to-bulk concentration gap.
Closed-chamber validation. Sealed-chamber gas matches the imposed flux exactly (mass conserved to 2×10⁻⁵), and a persistent near-leaf vs far-wall gap develops — the boundary layer at chamber scale.

04Gravity thickens the boundary layer — geometry amplifies it

As gravity falls, the buoyant plume that ventilates the leaf on Earth disappears and a thicker, symmetric diffusive halo forms; the surface gas gaps steepen. Denser architectures trap air and make it worse: the rosette crown shows the steepest local gradients, a microgreen canopy the highest bulk gradient.

Water-vapour boundary layer: buoyant plume on Earth, stagnant halo in microgravity, trapped canopy air.
Water-vapour boundary layer. Buoyant plume (leaf, Earth) → symmetric stagnant halo (leaf, microgravity) → trapped within-canopy air (microgreen canopy, microgravity).
Single-leaf gravity sweep and the three-scale amplification.
Gravity and scale. Left: surface gaps rise and convection falls as gravity decreases. Right: the microgravity penalty amplifies from leaf to rosette to canopy.

05Forced airflow reverses it — but the requirement scales with density

Because the penalty is transport-limited, it is engineerable: a fan substitutes for the missing buoyant convection. An isolated leaf needs only ≈ 2.6 cm/s, but a rosette needs ≈ 11 and a dense canopy ≈ 21 cm/s — an ~8× increase. A single ventilation setting tuned on sparse plants under-serves a canopy.

Surface gap versus fan speed for a single leaf in microgravity.
Ventilation nulls the µg penalty (single leaf). Surface gap vs fan speed, with the ≈ 2.6 cm/s Earth-equivalent point.
Earth-equivalent ventilation speed rises steeply with planting density.
Ventilation requirement scales with density. Earth-equivalent airflow ≈ 2.6 / 11 / 21 cm/s for leaf / rosette / canopy.

06Spaceflight hardware as boundary conditions

The three most-used ISS plant systems differ, in transport terms, only in the boundary condition they impose on the Petri dish. BRIC (hermetically sealed) drifts toward hypoxia and gives the steepest surface gradient; CARA (gas-permeable micropore tape) vents the enclosure near ambient but leaves the µg boundary layer intact; VEGGIE (forced airflow) fixes both. Across all three scales, one VEGGIE fan speed that restores a leaf barely helps a rosette or canopy.

Enclosure CO2 drift over time and the leaf-surface gradient by hardware.
Hardware as dish boundary conditions. BRIC's enclosure atmosphere drifts without bound; the taped dish holds near ambient. Right: leaf-surface gradient by hardware.
BRIC/CARA/VEGGIE leaf-surface gradient across leaf, rosette and canopy.
Hardware × plant scale. BRIC ≈ CARA at the leaf surface at every scale; one VEGGIE fan speed under-serves denser stands.

07From transport physics to carbon gain

Closing the loop with CO2-limited photosynthesis, the boundary-layer depletion self-suppresses assimilation (most in the trapped rosette crown). Over a 12-h photoperiod the consequence is stark: a sealed BRIC dish exhausts its CO2 within minutes and fixes only ≈ 1% of the Earth carbon, while a taped CARA dish sustains ≈ 90% and a ventilated VEGGIE system ≈ 100%. A plant in a sealed, microgravity enclosure is not just stressed at its surface — it is starved of carbon substrate.

Net assimilation over a photoperiod and 12-hour carbon fixed by hardware.
Closed-loop CO2-limited photosynthesis. The sealed BRIC dish's photosynthesis collapses within minutes; 12-h carbon fixed is ≈ 1% (BRIC) vs ≈ 90% (CARA) vs ≈ 100% (VEGGIE) of Earth.

This dovetails with a transcriptome meta-analysis of 15 Arabidopsis spaceflight experiments (Barker et al., npj Microgravity 2023), which found flight hardware and lighting among the largest confounding effects and hypoxia/oxidative-stress signatures among the conserved responses — the model supplies a physical mechanism for that hardware dependence.

08Measured gas exchange

The model's stomatal flux is fixed by measurement, not assumed: a whole-plant-chamber dataset and an independent Vernier chamber trace over 4.8 days that shows the expected diel O2/humidity cycling.

Vernier whole-chamber trace: O2, temperature, relative and absolute humidity over 4.8 days.
Measured whole-chamber gas exchange. Diel O2 cycling and humidity swings driven by photosynthesis and respiration, over 4.8 days.

09Read, run & reuse

10Cite

Barker, R. J. et al. LunarLeaf-CFD: a validated in-browser CFD model of gravity-dependent gas-exchange boundary layers around Arabidopsis. (2026). github.com/dr-richard-barker/LunarLeaf-CFD — see CITATION.cff.
Manuscript author list, affiliations and a Zenodo DOI are being finalised.