Optimizing CCN predictions through inferred modal aerosol composition – a boreal forest case study

Abstract. The contribution of natural aerosol particles from boreal forests to total aerosol loadings may increase with reduction in anthropogenic emissions. Aitken and accumulation mode particles in boreal regions differ significantly in hygroscopicity, and ignoring this size dependence can cause large uncertainty in Cloud Condensation Nuclei (CCN) prediction. We applied κ-Köhler theory to a multi-year dataset (2016–2020) from Hyytiälä, Finland, to evaluate different representations of aerosol chemical composition for CCN prediction. Overpredictions by forward closures using either bulk chemical composition from an Aerosol Chemical Speciation Monitor (ACSM) or a constant κ= 0.18 were mitigated to a great extent by optimizing size-resolved composition using two inverse modeling approaches: (1) Nelder–Mead method with the size distribution fixed to its median during each 2 h CCN measurement cycle, and (2) MCMC (Markov Chain Monte Carlo) accounting also for the variability in the size distribution during each cycle. Both methods improved closure at SS = 0.2 %–1.0 % (with Geometric Mean Bias GMB values 1.12–1.20 and 0.95–1.05, respectively), with moderate improvement at 0.1 % (GMBs of 1.53 and 1.32, respectively). The Aitken mode was enriched in organics in 77 % of cases using method (1) and 46 % using method (2) – with typical κ values of ∼ 0.1 for Aitken and ∼ 0.3 for accumulation modes. The results generally align with known size-dependent chemical composition in Hyytiälä and indicate that variability in CCN hygroscopicity is largely driven by Aitken mode composition. Our results demonstrate the potential of inverse CCN closure methods for obtaining valuable information of the size-dependent chemical composition.