Large-scale atmospheric models, which typically describe secondary organic aerosol (SOA) formation based on chamber experiments, tend to systematically underestimate observed organic aerosol burdens. Since SOA constitutes a significant fraction of atmospheric aerosol, this discrepancy translates into an underestimation of SOA contribution to radiative forcing of atmospheric aerosol. Here we show that the underestimation of SOA yields can be partly explained by wall losses of SOA forming compounds during chamber experiments. We present a chamber experiment where α-pinene and ozone are injected into a Teflon chamber. When these two compounds react, we observe rapid formation and growth of new particles. Theoretical analysis of this formation and growth event indicates rapid formation of oxidized volatile organic compounds (OVOC) of very low volatility in the chamber. If these oxidized organic compounds form in the gas phase, their wall losses will have significant implications on their partitioning between the gas and particle phase. Although these OVOCs of very low volatility contribute to the growth of new particles, their mass will almost completely be depleted to the chamber walls during the experiment, while the depletion of OVOCs of higher volatilities is less efficient. According to our model simulations, the volatilities of OVOC contributing to the new particle formation event can be of the order of 10⁻⁵ μg m⁻³.

Boreal forests are a major source of climate-relevant biogenic secondary organic aerosols (SOAs) and will be greatly influenced by increasing temperature. Global warming is predicted to not only increase emissions of reactive biogenic volatile organic compounds (BVOCs) from vegetation directly but also induce large-scale insect outbreaks, which significantly increase emissions of reactive BVOCs. Thus, climate change factors could substantially accelerate the formation of biogenic SOAs in the troposphere. In this study, we have combined results from field and laboratory experiments, satellite observations and global-scale modelling in order to evaluate the effects of insect herbivory and large-scale outbreaks on SOA formation and the Earth's climate. Field measurements demonstrated 11-fold and 20-fold increases in monoterpene and sesquiterpene emissions respectively from damaged trees during a pine sawfly (Neodiprion sertifer) outbreak in eastern Finland. Laboratory chamber experiments showed that feeding by pine weevils (Hylobius abietis) increased VOC emissions from Scots pine and Norway spruce seedlings by 10–50 fold, resulting in 200–1000-fold increases in SOA masses formed via ozonolysis. The influence of insect damage on aerosol concentrations in boreal forests was studied with a global chemical transport model GLOMAP and MODIS satellite observations. Global-scale modelling was performed using a 10-fold increase in monoterpene emission rates and assuming 10 % of the boreal forest area was experiencing outbreak. Results showed a clear increase in total particulate mass (local max. 480 %) and cloud condensation nuclei concentrations (45 %). Satellite observations indicated a 2-fold increase in aerosol optical depth over western Canada's pine forests in August during a bark beetle outbreak. These results suggest that more frequent insect outbreaks in a warming climate could result in substantial increase in biogenic SOA formation in the boreal zone and, thus, affect both aerosol direct and indirect forcing of climate at regional scales. The effect of insect outbreaks on VOC emissions and SOA formation should be considered in future climate predictions.

New particle formation (NPF) in the atmosphere is globally an important source of climate relevant aerosol particles. Occurrence of NPF events is typically analyzed by researchers manually from particle size distribution data day by day, which is time consuming and the classification of event types may be inconsistent. To get more reliable and consistent results, the NPF event analysis should be automatized. We have developed an automatic analysis method based on deep learning, a subarea of machine learning, for NPF event identification. To our knowledge, this is the first time that a deep learning method, i.e., transfer learning of a convolutional neural network (CNN), has successfully been used to automatically classify NPF events into different classes directly from particle size distribution images, similarly to how the researchers carry out the manual classification. The developed method is based on image analysis of particle size distributions using a pretrained deep CNN, named AlexNet, which was transfer learned to recognize NPF event classes (six different types). In transfer learning, a partial set of particle size distribution images was used in the training stage of the CNN and the rest of the images for testing the success of the training. The method was utilized for a 15-year-long dataset measured at San Pietro Capofiume (SPC) in Italy. We studied the performance of the training with different training and testing of image number ratios as well as with different regions of interest in the images. The results show that clear event (i.e., classes 1 and 2) and nonevent days can be identified with an accuracy of ca. 80 %, when the CNN classification is compared with that of an expert, which is a good first result for automatic NPF event analysis. In the event classification, the choice between different event classes is not an easy task even for trained researchers, and thus overlapping or confusion between different classes occurs. Hence, we cross-validated the learning results of CNN with the expert-made classification. The results show that the overlapping occurs, typically between the adjacent or similar type of classes, e.g., a manually classified Class 1 is categorized mainly into classes 1 and 2 by CNN, indicating that the manual and CNN classifications are very consistent for most of the days. The classification would be more consistent, by both human and CNN, if only two different classes are used for event days instead of three classes. Thus, we recommend that in the future analysis, event days should be categorized into classes of “quantifiable” (i.e., clear events, classes 1 and 2) and “nonquantifiable” (i.e., weak events, Class 3). This would better describe the difference of those classes: both formation and growth rates can be determined for quantifiable days but not both for nonquantifiable days. Furthermore, we investigated more deeply the days that are classified as clear events by experts and recognized as nonevents by the CNN and vice versa. Clear misclassifications seem to occur more commonly in manual analysis than in the CNN categorization, which is mostly due to the inconsistency in the human-made classification or errors in the booking of the event class. In general, the automatic CNN classifier has a better reliability and repeatability in NPF event classification than human-made classification and, thus, the transfer-learned pretrained CNNs are powerful tools to analyze long-term datasets. The developed NPF event classifier can be easily utilized to analyze any long-term datasets more accurately and consistently, which helps us to understand in detail aerosol–climate interactions and the long-term effects of climate change on NPF in the atmosphere. We encourage researchers to use the model in other sites. However, we suggest that the CNN should be transfer learned again for new site data with a minimum of ca. 150 figures per class to obtain good enough classification results, especially if the size distribution evolution differs from training data. In the future, we will utilize the method for data from other sites, develop it to analyze more parameters and evaluate how successfully CNN could be trained with synthetic NPF event data.

Atmospheric new particle formation (NPF) is an important source of climate-relevant aerosol particles which has been observed at many locations globally. To study this phenomenon, the first step is to identify whether an NPF event occurs or not on a given day. In practice, NPF event identification is performed visually by classifying the NPF event or non-event days from the particle number size distribution surface plots. Unfortunately, this day-by-day visual classification is time-consuming and labor-intensive, and the identification process renders subjective results. To detect NPF events automatically, we regard the visual signature (banana shape) which has been observed all over the world in NPF surface plots as a special kind of object, and a deep learning model called Mask R-CNN is applied to localize the spatial layouts of NPF events in their surface plots. Utilizing only 358 human-annotated masks on data from the Station for Measuring Ecosystem–Atmosphere Relations (SMEAR) II station (Hyytiälä, Finland), the Mask R-CNN model was successfully generalized for three SMEAR stations in Finland and the San Pietro Capofiume (SPC) station in Italy. In addition to the detection of NPF events (especially the strongest events), the presented method can determine the growth rates, start times, and end times for NPF events automatically. The automatically determined growth rates agree with the manually determined growth rates. The statistical results validate the potential of applying the proposed method to different sites, which will improve the automatic level for NPF event detection and analysis. Furthermore, the proposed automatic NPF event analysis method can minimize subjectivity compared with human-made analysis, especially when long-term data series are analyzed and statistical comparisons between different sites are needed for event characteristics such as the start and end times, thereby saving time and effort for scientists studying NPF events.

Sulphuric acid, ammonia, amines, and oxidised organics play a crucial role in nanoparticle formation in the atmosphere. In this study, we investigate the composition of nucleated nanoparticles formed from these compounds in the CLOUD (Cosmics Leaving Outdoor Droplets) chamber experiments at CERN (Centre européen pour la recherche nucléaire). The investigation was carried out via analysis of the particle hygroscopicity, ethanol affinity, oxidation state, and ion composition. Hygroscopicity was studied by a hygroscopic tandem differential mobility analyser and a cloud condensation nuclei counter, ethanol affinity by an organic differential mobility analyser and particle oxidation level by a high-resolution time-of-flight aerosol mass spectrometer. The ion composition was studied by an atmospheric pressure interface time-of-flight mass spectrometer. The volume fraction of the organics in the particles during their growth from sizes of a few nanometers to tens of nanometers was derived from measured hygroscopicity assuming the Zdanovskii–Stokes–Robinson relationship, and compared to values gained from the spectrometers. The ZSR-relationship was also applied to obtain the measured ethanol affinities during the particle growth, which were used to derive the volume fractions of sulphuric acid and the other inorganics (e.g. ammonium salts). In the presence of sulphuric acid and ammonia, particles with a mobility diameter of 150 nm were chemically neutralised to ammonium sulphate. In the presence of oxidation products of pinanediol, the organic volume fraction of freshly nucleated particles increased from 0.4 to ~0.9, with an increase in diameter from 2 to 63 nm. Conversely, the sulphuric acid volume fraction decreased from 0.6 to 0.1 when the particle diameter increased from 2 to 50 nm. The results provide information on the composition of nucleated aerosol particles during their growth in the presence of various combinations of sulphuric acid, ammonia, dimethylamine and organic oxidation products.

We carried out a closure study of aerosol–cloud interactions during stratocumulus formation using a large eddy simulation model UCLALES–SALSA (University of California Los Angeles large eddy simulation model–sectional aerosol module for large applications) and observations from the 2020 cloud sampling campaign at Puijo SMEAR IV (Station for Measuring Ecosystem–Atmosphere Relations) in Kuopio, Finland. The unique observational setup combining in situ and cloud remote sensing measurements allowed a closer look into the aerosol size–composition dependence of droplet activation and droplet growth in turbulent boundary layer driven by surface forcing and radiative cooling. UCLALES–SALSA uses spectral bin microphysics for aerosols and hydrometeors, and incorporates a full description of their interactions into the turbulent-convective radiation-dynamical model of stratocumulus. Based on our results, the model successfully described the probability distribution of updraught velocities and consequently the size dependency of aerosol activation into cloud droplets, and further recreated the size distributions for both interstitial aerosol and cloud droplets. This is the first time such a detailed closure is achieved not only accounting for activation of cloud droplets in different updraughts, but also accounting for processes evaporating droplets and drizzle production through coagulation–coalescence. We studied two cases of cloud formation, one diurnal (24 September 2020) and one nocturnal (31 October 2020), with high and low aerosol loadings, respectively. Aerosol number concentrations differ more than 1 order of magnitude between cases and therefore, lead to cloud droplet number concentration (CDNC) values which range from less than 100 cm⁻³ up to 1000 cm⁻³. Different aerosol loadings affected supersaturation at the cloud base, and thus the size of aerosol particles activating to cloud droplets. Due to higher CDNC, the mean size of cloud droplets in the diurnal high aerosol case was lower. Thus, droplet evaporation in downdraughts affected more the observed CDNC at Puijo altitude compared to the low aerosol case. In addition, in the low aerosol case, the presence of large aerosol particles in the accumulation mode played a significant role in the droplet spectrum evolution as it promoted the drizzle formation through collision and coalescence processes. Also, during the event, the formation of ice particles was observed due to subzero temperature at the cloud top. Although the modelled number concentration of ice hydrometeors was too low to be directly measured, the retrieval of hydrometeor sedimentation velocities with cloud radar allowed us to assess the realism of modelled ice particles. The studied cases are presented in detail and can be further used by the cloud modellers to test and validate their models in a well-characterized modelling setup. We also provide recommendations on how increasing amount of information on aerosol properties could improve the understanding of processes affecting cloud droplet number and liquid water content in stratiform clouds.