Evaluation of aerosol representation in TM5 using a one-at-a-time sensitivity analysis constrained by aerosol optical depth

Aerosol parametric uncertainty in global climate models can be as large as intermodal uncertainty, posing a significant challenge for reliable climate projections. This study investigates the magnitude and drivers of aerosol-related uncertainty in the TM5 chemical transport model (CTM), the atmospheric chemistry and transport component of the EC-Earth3 Earth System Model (ESM), and a contributor to the Coupled Model Intercomparison Project (CMIP).

Aerosol parameters in TM5 describe characteristics of emissions, removal, transformations, physical, chemical, and optical properties. To assess aerosol representation in TM5, we have performed one-at-a-time sensitivity studies, where individual aerosol parameters were perturbed to the minimum and maximum of their respective uncertainty ranges. The resulting impacts were evaluated using climate-relevant outputs, cloud condensation nuclei number (CCN) concentrations at supersaturations of 0.2% and 1% averaged over the lowest five model levels, and aerosol optical depth (AOD) at 550 nm. Model responses were analyzed on seasonal timescales over a two-year period (2017–2018).

The results indicate that sea-salt and dimethyl sulfide (DMS) emissions, particle size of biomass-burning emissions, and dry-deposition rates exert the strongest influence on CCN number concentrations and aerosol optical depth (AOD).  Simulated AOD from each sensitivity experiment was constrained using the AOD merged product for 2017 by Sogacheva et al. (2020). Seasonal comparisons were performed for 2017 across four regions: West Asia and North Africa (WANA), India and South Asia (ISA), Southern Africa (SA), and Mexico to Colombia (MC). Across all seasons, high AOD regions, TM5 exhibits minimal sensitivity to the perturbed simulations and accurately captures high AOD values. In contrast, the model shows more variation in low AOD values than the observations, where sensitivity to parameter perturbations, particularly emission-related parameters, is most pronounced. Increased dry deposition and reduced SO₂ emissions consistently improve low-AOD predictions with respect to the satellite product, especially in WANA and ISA, while the default setup performs best in South Asia and during JJA.

These findings identify the aerosol parameters that contribute most significantly to uncertainty in TM5 and highlight the key sensitivities that will inform future work involving emulated perturbed-parameter ensemble (PPE) experiments. The PPEs will vary these parameters over their uncertainty range simultaneously to study their combined effect on TM5.