SMPS and HTDMA dataset for analysis of particle hygroscopicity

Hygroscopic properties of aerosols are an essential parameter to predict the roles of aerosols in cloud droplet formation and to estimate particle lung deposition. The objective of this study was to introduce a new approach to classify and predict the hygroscopic growth factors (Gf) of specific atmospheric sub-micrometre particle types in a mixed aerosol based on measurement of the ensemble hygroscopic growth factors and particle number size distribution (PNSD). PNSDs were measured in two intensive field campaigns in an urban background area of London in 2012. Based on a non-linear regression model between aerosol source contributions from PMF applied to the PNSD dataset and the measured Gf values (at 90% relative humidity) of ambient aerosols, the estimated mean Gf values for secondary inorganic, mixed secondary, nucleation, urban background, fresh and aged traffic generated particle classes at a diameter of 110 nm were found to be 1.51, 1.44, 1.23, 1.44, 1.09 and 1.10, respectively. The probability density function (Gf-PDF) of Gf for particles from each source was also determined by a PMF model using the combined data sets of PNSD and hygroscopic particle counts as inputs. This study also investigated a new approach based on a Random Forest algorithm to estimate hygroscopic properties of particles from the data sets of Gf, PNSD, chemical composition and meteorological conditions. It is found possible to impute missing TDMA datasets using a Random Forest regression on PNSD and meteorological conditions