Simulating isoprene-NOx interactions in deep convective events using large-eddy simulations with online chemistry

Isoprene is a biogenic volatile organic compound and the most significant non-methane hydrocarbon by total emission flux (400-600 Tg C/year). Once in the atmosphere, isoprene is oxidized by the hydroxyl (OH) radical on the order of hours, with its atmospheric lifetime depending on local OH concentrations. Although 80% of global isoprene emissions occur in the tropics, chemistry and emissions over remote tropical atmospheres remain highly uncertain due to limited in-situ observations and high cloud cover. These uncertainties over the tropics may have significant implications on new particle formation, especially during deep convective events where isoprene emissions interact with high lightning NO_x. Furthermore, recent studies show that soil NO_x emissions in Amazonia may be underestimated by an order of magnitude, which can subsequently alter local [OH] and isoprene chemistry. Here we address these uncertainties in the remote tropics by explicitly simulating isoprene-NO_x-OH chemistry and emissions in a large-eddy simulation (SAM-Chem). SAM-Chem, which uses the dynamical and microphysical core of the System for Atmospheric Modeling, includes gas-phase and heterogeneous chemistry; dry and wet deposition; and interactive emission fluxes for isoprene and soil/lightning NO_x at a spatial resolution of ~50 meters and a temporal resolution of ~1 second. We constrain SAM-Chem with in-situ observations from the goAMAZON campaign (2014-2015) and simulate a week during Amazonia's wet season, which contained some of the highest convective rainfall events during the campaign. We find that both soil and lightning NO_x emissions, and the uncertainties in those emissions, have significant impacts on the resulting isoprene column and vertical profiles, which can subsequently impact the fate of isoprene oxidation products in tropical regimes. We also highlight the potential for SAM-Chem for investigating chemistry and dynamics at spatiotemporal scales too fine for traditional chemical transport models.