Explore the vast range of titles available.

Over Boreal regions, monoterpenes emitted from the forest are the main precursors for secondary organic aerosol (SOA) formation and the primary driver of the growth of new aerosol particles to climatically important cloud condensation nuclei (CCN). Autoxidation of monoterpenes leads to rapid formation of Highly Oxygenated organic Molecules (HOM). We have developed the first model with near-explicit representation of atmospheric new particle formation (NPF) and HOM formation. The model can reproduce the observed NPF, HOM gas-phase composition and SOA formation over the Boreal forest. During the spring, HOM SOA formation increases the CCN concentration by ~10 % and causes a direct aerosol radiative forcing of −0.10 W/m 2 . In contrast, NPF reduces the number of CCN at updraft velocities < 0.2 m/s, and causes a direct aerosol radiative forcing of +0.15 W/m 2 . Hence, while HOM SOA contributes to climate cooling, NPF can result in climate warming over the Boreal forest. Forests emit compounds into the atmosphere that are oxidized into highly oxygenated molecules that serve as precursors for cloud condensation nuclei–a process that impacts the climate, but is poorly represented in models. Here the authors create a new model that accurately depicts highly oxygenated molecule and climate dynamics over Boreal forests.

The new particle formation (NPF) process is a significant source of atmospheric secondary particles, which has remarkable impacts on the regional air quality and global radiative forcing. Most NPF studies conduct their measurements at a single site, which can hardly provide information about the regionality of NPF events at large scales (>100 km). During the 2022 Winter Olympic and Paralympic Games, simultaneous measurements of particle number size distributions and NPF-associated precursors were conducted at a mountainous site close to the Winter Olympic Village in Chongli (CL), Zhangjiakou, and an urban site in Beijing (BJ) located 150 km southeast of the CL site. High NPF frequencies were observed at the CL (50%) and BJ (52%) sites; however, the fraction of concurrent NPF events was smaller than the results in other regions. In addition, the wind distributions exhibited distinct air mass origins at the two sites during the concurrent NPF events. Compared with the BJ site, the NPF growth rates were higher at the CL site due to higher levels of volatile organic compounds (VOCs) and radiation. Surprisingly, the formation rates at the CL site were lower than at the BJ site, even with a higher sulfuric acid concentration and lower CS, which may be attributed to lower dimethylamine concentrations in the mountainous area. This study reveals that, although NPF events are commonly thought to occur on regional scales, their intensity and mechanisms may have significant spatial inhomogeneity. Further studies are required to reduce the uncertainty when expanding the mechanisms based on the urban conditions to regional or global scales in the models.

A multitude of biogeochemical feedback mechanisms govern the climate sensitivity of Earth in response to radiation balance perturbations. One feedback mechanism, which remained missing from most current Earth System Models applied to predict future climate change in IPCC AR6, is the impact of higher temperatures on the emissions of biogenic volatile organic compounds (BVOCs), and their subsequent effects on the hydroxyl radical (OH) concentrations. OH, in turn, is the main sink term for many gaseous compounds including methane, which is the second most important human-influenced greenhouse gas in terms of climate forcing. In this study, we investigate the impact of this feedback mechanism by applying two models, a one-dimensional chemistry-transport model, and a global chemistry-transport model. The results indicate that in a 6 K temperature increase scenario, the BVOC-OH-CH 4 feedback increases the lifetime of methane by 11.4% locally over the boreal region when the temperature rise only affects chemical reaction rates, and not both, chemistry and BVOC emissions. This would lead to a local increase in radiative forcing through methane (ΔRF CH4 ) of approximately 0.013 Wm −2 per year, which is 2.1% of the current ΔRF CH4 . In the whole Northern hemisphere, we predict an increase in the concentration of methane by 0.024% per year comparing simulations with temperature increase only in the chemistry or temperature increase in chemistry and BVOC emissions. This equals approximately 7% of the annual growth rate of methane during the years 2008–2017 (6.6 ± 0.3 ppb yr−1) and leads to an ΔRF CH4 of 1.9 mWm −2 per year.

The radiative properties of clouds depend partially on the cloud droplet number concentration, which is determined by the concentration of cloud condensation nuclei (CCN) when the clouds are formed. In turn, CCN concentrations are determined by the atmospheric particle size distribution and their chemical composition. We use a novel Lagrangian modelling framework to examine the origins and history of gas and aerosol components observed at the boreal forest measurement site SMEAR II, and their potential to act as CCN. This framework combines: (a) global emission datasets, (b) backward trajectories from the FLEXible PARTicle dispersion model (FLEXPART) airmass dispersion model, (c) a detailed description of atmospheric chemistry and aerosol dynamics from the Model to Simulate the Concentration of Organic Vapours, Sulphuric Acid and Aerosol Particles (SOSAA). We apply this SOSAA-FP (FP standing for FLEXPART) framework to simulate a period from March to October 2018 with 1 h time resolution, focusing on the concentrations of CCN between 0.1 % and 1.2 % maximum supersaturation as calculated by the κ-Köhler theory (with respective dry particle diameter of activation ca. 175–35 nm). We find that the model PM1 fraction of primary particles, sulfates and secondary organic aerosol correlate well with the observed organic aerosol and sulfate trends and explain most of the observed organic aerosol and sulfate PM1 mass. Our results show that primary particle emissions play a considerable role in CCN concentrations even at a rural site such as SMEAR II. Changes in atmospheric cluster formation rates had a relatively weak impact on the CCN concentrations in the sensitivity runs. Enhanced cluster formation increased (decreased) the CCN concentrations for the highest (lowest) maximum supersaturation. Without any cluster formation our modelled median CCN concentrations changed by −48 % and +23 % for supersaturations of 1.2 % and 0.1 %, respectively, whereas omitting primary particle emissions had a decreasing effect in all calculated CCN supersaturation classes (−82 % and −33 % decrease in median CCN of 1.2 % and 0.1 % supersaturation, respectively). While the enhancing effect of cluster formation to high supersaturation (i.e., small diameter) CCN concentrations is unsurprising, the weak sensitivity to cluster formation rates and the decreasing effect to lowest supersaturation CCN was unexpected, as was the strong influence of anthropogenic primary emissions. The Lagrangian model framework showed its power, as it was possible to trace down the causes behind the unexpected outcomes by comparing how the particle population evolved along the trajectories in different sensitivity tests.

Urban air consists of a complex mixture of gaseous and particulate species from anthropogenic and biogenic sources that are further processed in the atmosphere. This study investigated the characteristics and sources of volatile organic compounds (VOCs) and submicron organic aerosol (OA) in a traffic environment in Helsinki, Finland, in late summer. The anthropogenic VOCs (aVOCs; aromatic hydrocarbons) and biogenic VOCs (bVOCs; terpenoids) relevant for secondary-organic-aerosol formation were analyzed with an online gas chromatograph mass spectrometer, whereas the composition and size distribution of submicron particles was measured with a soot particle aerosol mass spectrometer. This study showed that aVOC concentrations were significantly higher than bVOC concentrations in the traffic environment. The largest aVOC concentrations were measured for toluene (campaign average of 1630 ng m−3) and p/m xylene (campaign average of 1070 ng m−3), while the dominating bVOC was α-pinene (campaign average of 200 ng m−3). For particle-phase organics, the campaign-average OA concentration was 2.4 µg m−3. The source apportionment analysis extracted six factors for OA. Three OA factors were related to primary OA sources – traffic (24 % of OA, two OA types) and a coffee roastery (7 % of OA) – whereas the largest fraction of OA (69 %) consisted of oxygenated OA (OOA). OOA was divided into less oxidized semi-volatile OA (SV-OOA; 40 % of OA) and two types of low-volatility OA (LV-OOA; 30 %). The focus of this research was also on the oxidation potential of the measured VOCs and the association between VOCs and OA in ambient air. Production rates of the oxidized compounds (OxPR) from the VOC reactions revealed that the main local sources of the oxidation products were O3 oxidation of bVOCs (66 % of total OxPR) and OH radical oxidation of aVOCs and bVOCs (25 % of total OxPR). Overall, aVOCs produced a much smaller portion of the oxidation products (18 %) than bVOCs (82 %). In terms of OA factors, SV-OOA was likely to originate from biogenic sources since it correlated with an oxidation product of monoterpene, nopinone. LV-OOA consisted of highly oxygenated long-range or regionally transported OA that had no correlation with local oxidant concentrations as it had already spent several days in the atmosphere before reaching the measurement site. In general, the main sources were different for VOCs and OA in the traffic environment. Vehicle emissions impacted both VOC and OA concentrations. Due to the specific VOCs attributed to biogenic emissions, the influence of biogenic emissions was more clearly detected in the VOC concentrations than in OA. In contrast, the emissions from the local coffee roastery had a distinctive mass spectrum for OA, but they could not be seen in the VOC measurements due to the measurement limitations for the large VOC compounds. Long-range transport increased the OA concentration and oxidation state considerably, while its effect was observed less clearly in the VOC measurements due to the oxidation of most VOC in the atmosphere during the transport. Overall, this study revealed that in order to properly characterize the impact of different emission sources on air quality, health, and climate, it is of importance to describe both gaseous and particulate emissions and understand how they interact as well as their phase transfers in the atmosphere during the aging process.

We introduce the Atmospherically Relevant Chemistry and Aerosol box model ARCA box (v.1.2.2). It is a zero-dimensional process model with a focus on atmospheric chemistry and submicron aerosol processes, including cluster formation. A novel feature in the model is its comprehensive graphical user interface, allowing for detailed configuration and documentation of the simulation settings, flexible model input, and output visualization. Additionally, the graphical interface contains tools for module customization and input data acquisition. These properties – customizability, ease of implementation and repeatability – make ARCA an invaluable tool for any atmospheric scientist who needs a view on the complex atmospheric aerosol processes. ARCA is based on previous models (MALTE-BOX, ADiC and ADCHEM), but the code has been fully rewritten and reviewed. The gas-phase chemistry module incorporates the Master Chemical Mechanism (MCMv3.3.1) and Peroxy Radical Autoxidation Mechanism (PRAM) but can use any compatible chemistry scheme. ARCA's aerosol module couples the ACDC (Atmospheric Cluster Dynamics Code) in its particle formation module, and the discrete particle size representation includes the fully stationary and fixed-grid moving average methods. ARCA calculates the gas-particle partitioning of low-volatility organic vapours for any number of compounds included in the chemistry, as well as the Brownian coagulation of the particles. The model has parametrizations for vapour and particle wall losses but accepts user-supplied time- and size-resolved input. ARCA is written in Fortran and Python (user interface and supplementary tools), can be installed on any of the three major operating systems and is licensed under GPLv3.

In this study, we modeled the aerosol particle formation along air mass trajectories arriving at the remote Arctic research stations Gruvebadet (67 m a.s.l.) and Zeppelin (474 m a.s.l.), Ny-Ålesund, during May 2018. The aim of this study was to improve our understanding of processes governing secondary aerosol formation in remote Arctic marine environments. We run the Lagrangian chemistry transport model ADCHEM, along air mass trajectories generated with FLEXPART v10.4. The air masses arriving at Ny-Ålesund spent most of their time over the open ice-free ocean. In order to capture the secondary aerosol formation from the DMS emitted by phytoplankton from the ocean surface, we implemented a recently developed comprehensive DMS and halogen multi-phase oxidation chemistry scheme, coupled with the widely used Master Chemical Mechanism (MCM). The modeled median particle number size distributions are in close agreement with the observations in the marine-influenced boundary layer near-sea-surface Gruvebadet site. However, while the model reproduces the accumulation mode particle number concentrations at Zeppelin, it overestimates the Aitken mode particle number concentrations by a factor of ∼5.5. We attribute this to the deficiency of the model to capture the complex orographic effects on the boundary layer dynamics at Ny-Ålesund. However, the model reproduces the average vertical particle number concentration profiles within the boundary layer (0–600 m a.s.l.) above Gruvebadet, as measured with condensation particle counters (CPCs) on board an unmanned aircraft system (UAS). The model successfully reproduces the observed Hoppel minima, often seen in particle number size distributions at Ny-Ålesund. The model also supports the previous experimental findings that ion-mediated H2SO4–NH3 nucleation can explain the observed new particle formation in the marine Arctic boundary layer in the vicinity of Ny-Ålesund. Precursors resulting from gas- and aqueous-phase DMS chemistry contribute to the subsequent growth of the secondary aerosols. The growth of particles is primarily driven via H2SO4 condensation and formation of methane sulfonic acid (MSA) through the aqueous-phase ozonolysis of methane sulfinic acid (MSIA) in cloud and deliquescent droplets.

Interactions in urban environment were investigated using a multidisciplinary model combination, with focus on traffic, emissions and atmospheric particles. An agent-based model was applied to simulate the evolution of unsustainable human behavior (usage of combustion-based personal vehicles) as a function of pro-environmental affordances (opportunities for sustainable choices). Scenarios regarding changes in multi-pollutant emissions were derived, and the non-linear implications to atmospheric particles were simulated with a box model. Based on the results for a Nordic city, increasing pro-environmental affordances by 10%, 50% or 100% leads to emission reductions of 15%, 30% and 40% within 2 years. To reduce ambient particle mass, emissions from traffic should decrease by > 15%, while the lung deposited surface area decreases in all scenarios (-23%, -32% and -36%, correspondingly). The presented case is representative of one season, but the approach is generic and applicable to simulating a full year, given meteorological and pollution data that reflects seasonal variation. This work emphasizes the necessity to consider feedback mechanisms and non-linearities in both human behavior and atmospheric processes, when predicting the outcomes of changes in an urban system.

Modeling atmospheric chemistry is complex and computationally intense. Given the recent success of Deep neural networks in digital signal processing, we propose a Neural Network Emulator for fast chemical concentration modeling. We consider atmospheric chemistry as a time-dependent Ordinary Differential Equation. To extract the hidden correlations between initial states and future time evolution, we propose ChemNNE, an Attention based Neural Network Emulator (NNE) that can model the atmospheric chemistry as a neural ODE process. To efficiently simulate the chemical changes, we propose the sinusoidal time embedding to estimate the oscillating tendency over time. More importantly, we use the Fourier neural operator to model the ODE process for efficient computation. We also propose three physical-informed losses to supervise the training optimization. To evaluate our model, we propose a large-scale chemical dataset that can be used for neural network training and evaluation. The extensive experiments show that our approach achieves state-of-the-art performance in modeling accuracy and computational speed.

Applying machine learning to increasingly high-dimensional problems with sparse or biased training data increases the risk that a model is used on inputs outside its training domain. For such out-of-distribution (OOD) inputs, the model can no longer make valid predictions, and its error is potentially unbounded. Since testing OOD detection methods on real-world datasets is complicated, we design a benchmark for OOD detection, which includes three novel and easily-visualisable toy examples. These simple examples provide direct and intuitive insight into whether the detector is able to detect (1) linear and (2) non-linear concepts and (3) identify thin in-distribution (ID) subspaces (needles) within high-dimensional spaces (haystacks). We use our benchmark to evaluate the performance of various methods from the literature. Since tactile examples of OOD inputs may benefit OOD detection, we also review several simple methods to synthesise OOD inputs for supervised training. We introduce two improvements, \\(t\\)-poking and OOD sample weighting, to make supervised detectors more precise at the ID-OOD boundary. This is especially important when conflicts between real ID and synthetic OOD sample blur the decision boundary. Finally, we provide recommendations for constructing and applying OOD detectors in machine learning.