PhD Defence - Rahul Ranjan

From clouds over the boreal forest to fog events in polluted plains
Abstract
Aerosol–cloud interactions (ACI) remain the largest source of uncertainty in the estimates of anthropogenic radiative forcing from Earth system models (ESMs). In this thesis, we combine theoretical tools including inverse modeling, sensitivity analysis, and cloud parcel model (CPM) simulations with observations from clean and polluted environments (SMEAR II station in Finnish boreal forest, and San Pietro Capofiume station in the Italian Po Valley) to yield new insights on ACI. For the boreal forest, we showed that if bulk chemical composition of submicron aerosol particles is used for cloud condensation nuclei (CCN) prediction, it would lead to significant errors (for e.g. a bias of ~34% at 1.0% supersaturation). Therefore, inverse CCN closures were performed in which optimized size-segregated chemical composition and thus the hygroscopicity parameter (κ) were obtained, which reduced the bias to below 21%. The results further confirmed the lower hygroscopicity of the ultrafine (smaller than about 100 nm in diameter) Aitken mode particles (median κ ≈ 0.1) as compared with the larger accumulation mode (median κ ≈ 0.3) – except in cases with very numerous Aitken mode particles. We then performed multiple CPM simulations of warm and non-precipitating clouds using a decade-long aerosol and cloud base updraft data from the same location. The median of modelled maximum supersaturation (SSmax) and cloud droplet number concentration at SSmax height level (CDNCmax) were ~0.2-0.3% and ~200 cm⁻³, respectively, increasing to ~0.32% and ~500 cm⁻³ in summer cumulus clouds due to higher updrafts and aerosol number concentrations. Observed CCN showed a moderate correlation with modeled CDNCmax (Spearman correlation coefficient r ≈ 0.5), highlighting the importance of other parameters (particularly SS and κ).
Furthermore, in case of the cumulus clouds, the observed peak liquid water path showed low correlation (Spearman correlation coefficient r = 0.26) with the ground observation of total aerosol number concentration. Co-condensation of organic vapors on growing droplets was found to enhance the cloud base CDNC by 16 to 22%, depending on updraft and the type of size distribution. In the Po Valley fog, in contrast to the boreal forest clouds, high aerosol loading and slower cooling rate resulted in low SS (0.01% to 0.03%) and lower CDNC ~12 cm⁻³ leaving, most (~87%) as hydrated haze particles.
In the boreal forest, the most important factors driving the CDNC varied depending primarily on the size distribution shape. For cases with a relatively large number of accumulation mode particles present, their number and updraft velocities were the most important determining factors. For cases with numerous ultrafine particles, however, the Aitken mode properties (hygroscopicity, concentrations and size) became important as well. In contrast to the boreal forest, in the Po Valley, the SS driven by the radiative cooling and aerosol number concentration were always the most important parameters for accurate prediction of fog microphysics – instead of the chemical composition.
This thesis highlights that accurate representation of updraft velocity and aerosol number concentration is consistently needed for reliable CDNC estimates, while chemical composition becomes especially critical when the size distribution includes a large number of ultrafine particles. We recommend continuing current long-term measurements, which can be complemented by campaign-based vertical profiling of aerosol size distributions beneath low stratus and cumulus clouds to help close remaining knowledge gaps. For fog microphysics studies in polluted environments, hydrated particles must be accounted for, as activated and non-activated droplets behave very differently in terms of processes such as evaporation and sedimentation.
Last updated: 2026-05-26
Source: Department of Environmental Science