Knowledge of chemistry and meteorology controls on methane oxidation is needed to predict environmental impacts from atmospheric methane removal (AMR; [1-2]). The decadal lifetime of methane is set by reaction with the hydroxyl radical (OH), which is locally in chemical steady state (sources = sinks; OH lifetime is about a second [3-5]). Airborne sampling shows that methane loss rates (LCH₄) vary on synoptic scales resolved in global models [6], which simulated different non- linear chemical feedbacks [7-11]. Global mean OH integrates over regionally distinct chemical processes which differ in their responses to emissions of OH source and sink gases and to climate change (Figure 1). Climate variability affects OH (Figure 2) and may confound attributions of changing LCH₄ [12-15]. Estimates of global OH abundances and trends are method-dependent [16-18] and process-level insights to drivers of LCH₄ are limited [19-25].

Grounding AMR in sound science requires identifying robust cause-and-effect signals in a complex chemical system amidst climate variability [23; 26-27]; Figure 2). By characterizing spatially resolved sensitivities of LCH₄ to multiple drivers, including natural climate variability and human-caused changes, we will identify atmospheric regions where anthropogenic signals are most detectable. With a simplified 3D chemistry-climate model, we will perturb emissions of trace gases affecting LCH₄ (NO_x, CO, NMVOC, H₂) from different sources under present and possible future climates to assess their imprints on LCH₄. Major uncertainties exist, including those from emerging understanding of atmospheric chemistry; we will evaluate LCH₄ sensitivity to selected processes (e.g., halogens [28-29] and some kinetic rate coefficients [30-31]). A successful project will rank the impacts of specific processes and uncertainties on LCH₄ to guide assessment of AMR techniques.

By advancing knowledge of LCH₄ sensitivity to emissions of multiple chemical species in a changing climate, we can inform AMR impact assessments under different air pollution and climate scenarios. We aim to demonstrate how a “first-look” model framework can build fundamental, quantitative understanding by interrogating the impacts of specific processes and their uncertainty on LCH₄ (or other variables of interest). These insights can then guide targeted exploration of AMR scenarios with more comprehensive models. Our proposed approach permits exploration of chemical sensitivities under climate or atmospheric chemistry conditions not yet experienced, which could be used to generate training datasets for even faster machine-learning models.

Figure 1. Schematic illustrating sources (red arrows) and sinks (blue arrows) of the key trace gases that together with physical climate variables (e.g., solar radiation, water vapor) determine the global mean abundance of hydroxyl radical (OH) in the troposphere, and thus the methane lifetime (controlled by reaction of methane with OH): nitrogen oxides (NO_x = NO+NO₂); carbon monoxide (CO); non-methane volatile organic compounds (NMVOC); and methane (CH₄). Also shown is the anticipated response of weather-sensitive emissions of these key trace gases to climate change: positive (pink shading) for biogenic emissions (wetland CH₄; soil NO_x; NMVOC from vegetation) and biomass burning, or uncertain (pink-to-blue shading) for lightning NO_x. The strongest future impacts from anthropogenic emissions (e.g., agricultural activities and fossil fuel and biofuel combustion; gray shading) likely depend more on human choices than climate change (though some emissions from these sectors vary with weather). We note that hydrogen gas (H₂), which may increase in the future (as we propose to examine), will compete with CO, NMVOC, and CH₄ for reaction with OH. Reproduced from Figure 5 of Fiore et al. [32].

‍

Figure 2. Spatial distribution of the 13-member initial-condition ensemble (a) mean and (b) relative standard deviation of tropospheric column air mass-weighted methane loss rates (LCH₄) generated with the Community Earth System Model version 2 - Whole Atmosphere Community Climate Model version 6 (CESM2-WACCM6) chemistry-climate model [33-35] for 1950-2014. Each ensemble member is driven by the same historical anthropogenic emissions and greenhouse gas concentrations [23; 36]. Also shown are the zonal mean LCH₄ distributions for the (c) ensemble mean and (d) standard deviation across the ensemble, which show that most methane oxidation occurs in the lower tropical troposphere, consistent with earlier work (e.g., [37-39]). Note that global mean LCH₄ changes of only a few percent are sufficient to explain multi-year changes in the observed atmospheric methane growth rates during recent decades, implying that climate variability (at least as simulated in this model, see panels b and d) may confound interpretation of observed methane variations.