Exploratory MPAS Sensitivity Experiments on Rainforest Biogenic Salt Aerosols, Tropical Rainfall, and Poleward Moisture Transport

Brian Lue

Rainforest ecosystems emit biologically influenced aerosol particles, including potassium-rich and other hygroscopic components that may affect warm-rain microphysics. While tropical biogenic aerosols have been studied extensively, their sensitivity within coarse-resolution global models remains incompletely characterized, particularly under differing background aerosol states. Here we present exploratory sensitivity experiments using MPAS-Atmosphere to test whether rainforest biogenic salt aerosol parameterizations can alter modeled tropical precipitation and selected large-scale moisture-transport diagnostics.

Using 30-day single-member simulations at 240 km and 120 km resolution, we examined ten alternative implementations ranging from simple autoconversion perturbations to a more explicit giant-cloud-condensation-nuclei (GCCN) lifecycle treatment with activation, condensational growth, coalescence, and wet scavenging. Across these implementations, diagnosed latent heat transport at 30°N varied substantially (range −211 to +153 TW; standard deviation ~126 TW — comparable in magnitude to some published transport estimates), indicating strong sensitivity of some circulation metrics to aerosol microphysical assumptions within these experiments.

v4 introduces the prescribed-CCN methodology of Heikenfeld et al. (2019, ACP, doi:10.5194/acp-19-2601-2019) into the MPAS-Atmosphere experimental framework, in which water-friendly aerosol concentrations are held at observed pristine values (Pöhlker et al. 2012) throughout each simulation. With a 5-pair ensemble (Jan 2022–2026) at 120 km on the Pöhlker-Dg-matched activation configuration (Dg = 160 nm, κ = 0.8), we find a robust reduction in northward latent heat transport at 30°N (mean −80 ± 22 TW, 5/5 pairs negative, t = −7.98, p = 0.0013). This is consistent with the hypothesized mechanism: K-salt enhances equatorial convective rainout, retaining latent heat in the tropical region rather than exporting it through the Hadley cell upper branch. A complementary trend at 70°N (mean +15 ± 20 TW, 4/5 pairs positive, p = 0.16) is consistent with the predicted secondary effect of strengthened mid-latitude baroclinic eddies driven by the increased equator-to-pole temperature gradient, but does not reach statistical significance in this ensemble size. The Amazon-mean precipitation response itself is dominated by natural meteorological variability and is not statistically resolved (+1.5 ± 9.5 mm, p = 0.75). v4 retires v3's polluted-baseline sign-flip narrative because the Thompson aerosol-aware single-species framework cannot represent the chemistry of real anthropogenic pollution at the levels v3's "polluted" baseline implied; multi-species K-salt-versus-smoke separation is deferred to a planned follow-up using MPAS-CMAQ.

These simulations are not intended as detection or attribution evidence. v3 phases (1–7) are limited to single 30-day realizations; the v4 ensemble (Phase 9 in §6.9) advances on this with a 5-pair design but remains coarse-resolution and parameterized for tropical convection. Rather than producing detection-grade numbers, the experiments demonstrate that rainforest bioaerosol assumptions can materially influence modeled tropical rainfall responses and some transport diagnostics, and that the 30°N northward-latent-heat-transport reduction is statistically resolved within the v4 5-pair ensemble at the Pöhlker-Dg-matched configuration. The results motivate targeted follow-up using larger ensembles, additional seasons (a July ensemble is planned for v5), convection-permitting resolution, observational constraints on rainforest aerosol emissions, and higher-fidelity cloud microphysics. We release all code, bug-fix patches, Docker builds, and analysis scripts at github.com/bluesaltbarrier/blue-salt-barrier.