The impact of aerosol size-dependent hygroscopicity and mixing state on the cloud condensation nuclei potential over the north-east Atlantic

We present an aerosol cloud condensation nuclei (CCN) closure study over the north-east Atlantic Ocean using six approximating methods. The CCN number concentrations (NCCN) were measured at four discrete supersaturations (SSs; 0.25 %, 0.5 %, 0.75 % and 1.0 %). Concurrently, aerosol number size distribution, sub-saturation hygroscopic growth factor and bulk PM1 chemical composition were obtained at matching time resolution and after a careful data validation exercise. Method A used a constant bulk hygroscopicity parameter κ of 0.3; method B used bulk PM1 chemical composition measured by an aerosol mass spectrometer (AMS); method C utilised a single growth factor (GF) size (165 nm) measured by a humidified tandem differential mobility analyser (HTDMA); method D utilised size-dependent GFs measured at 35, 50, 75, 110 and 165 nm; method E divided the aerosol population into three hygroscopicity modes (near-hydrophobic, more-hygroscopic and sea-salt modes), and the total CCN number in each mode was cumulatively added up; method F used the full-size-scale GF probability density function (GF–PDF) in the most complex approach. The studied periods included high-biological-activity and low-biological-activity seasons in clean marine and polluted continental air masses to represent and discuss the most contrasting aerosol populations. Overall, a good agreement was found between estimated and measured NCCN with linear regression slopes ranging from 0.64 to 1.6. The temporal variability was captured very well, with Pearson's R value ranging from 0.76 to 0.98 depending on the method and air mass type. We further compared the results of using different methods to quantify the impact of size-dependent hygroscopicity and mixing state and found that ignoring size-dependent hygroscopicity induced overestimation of NCCN by up to 12 %, and ignoring a mixing state induced overestimation of NCCN by up to 15 %. The error induced by assuming an internal mixing in highly polluted cases was largely eliminated by dividing the full GF–PDF into three conventional hygroscopic modes, while assuming an internal mixing in clean marine aerosol did not induce significant error.