Abstract. Transatlantic dust from North Africa is the largest intercontinental aerosol flux on Earth, affecting radiative forcing, cloud processes, and biogeochemical cycling, yet the inter-model spread of dust transport remains large in previous AeroCom experiments. With the progress in 17 AeroCom Phase III (P3) models, this study evaluates the transatlantic dust cycle for 2010 against AERONET inversions, satellite products (MODIS, CALIOP, MISR, IASI, POLDER, and MODIS-TIR), DustCOMM, CAMS, MERRA-2, in-situ records, and GPCP precipitation across subregions from North Africa to the Caribbean. We found that AeroCom P3 narrows mid-visible dust optical depth (DOD) diversity to 30–45 % across the transport domain relative to AeroCom P2 (38–59 %), yet this convergence masks compensatory biases, as models with lower (or higher) mass loading may have higher (or lower) mass extinction efficiency (MEE), yielding similar DOD despite a large spread of mass representations. Compared with observational estimates, model biases intensify with transport distance, with median DOD underestimated by ~50 % and ~70 % over mid- and long-range regions, and loss frequencies exceed satellite estimates by a factor of 2–3. Such biases remain unresolved even with super-coarse extensions, indicating that constraining mid-visible DOD alone is inadequate for the dust cycle. The excess removal implicates deficiencies in gravitational settling, wet scavenging, and vertical transport parameterizations. This systematic underestimation could propagate into dust radiative forcing, cloud-dust interactions, and nutrient delivery to Atlantic and Amazonian ecosystems. Progress requires size-resolved benchmarking (DOD at 550 nm and 10 µm with size-segregated mass), non-spherical settling treatments, and mechanisms sustaining coarse particles aloft.

Uncertainty in estimations of the net contribution of anthropogenic aerosol particles, particularly of aerosol-cloud interactions (ACIs) to the Earth's radiation budget, limits our ability to understand past and project future climate change. Earth System Models (ESMs) are among the key tools for assessing the magnitude and impacts of changes in various forcing agents on the global climate system. Hence, improving aerosol and cloud descriptions in ESMs is an important way forward to increase the confidence in estimates of climate impacts of aerosol perturbations in the past, present and future. In the framework of the FORCeS project, experimental and theoretical approaches were combined to bridge the current key gaps in the fundamental understanding of essential aerosol and cloud processes and their descriptions in selected European ESMs. Regarding aerosol types and processes, we focused on organic aerosol, particulate nitrate, absorbing aerosols, and ultrafine aerosol sources including new particle formation and growth. In terms of cloud processes, we targeted cloud droplet activation, hydrometeor growth and evaporation, ice formation and multiplication as well as aerosol processing and scavenging by clouds. The selection was made based on the identified knowledge gaps in the scientific understanding of these processes and/or their current representation in ESMs, as well as a novel perturbed parameter ensemble approach to detecting potential structural deficiencies in an ESM. Here, we review the state-of-the-art, outline our approach for arriving at recommendations for improving the representation of key aerosol and cloud processes within ESMs, and then provide such recommendations applicable in models operating at the Earth system scale. The limitations of the recommendations, applicability, as well as alternative approaches and future research directions are discussed. Overall, the findings highlight the need for continuous efforts towards smart ways for representing the aerosol number size distribution as well as consistent representations of key parameters (e.g., liquid water content and cloud droplet number concentration). Furthermore, we provide guidance for future ESM evaluation emphasising, in particular, the need for exploring the consistency of key parameters, process-based (as opposed to parameter-based), and the complementarity of in-situ and remote-sensed measurements for model evaluation.

The balance between liquid and ice in clouds remains a major challenge in climate modeling, largely due to uncertainties in ice-related processes. We investigate the relative importance of four microphysical processes—primary ice nucleation (PIN), secondary ice production (SIP), sedimentation, and transport of ice crystals—for the supercooled liquid fraction (SLF) in mixed-phase clouds using three global climate models: EC-Earth3-AerChem, NorESM2-MM, and ECHAM6.3-HAM2.3. All models identify PIN as the dominant influence on SLF at cold temperatures in high northern latitudes, but diverge elsewhere and for higher temperatures. Implementing a unified SIP parameterization produced varied model responses, revealing fundamental differences in how microphysical processes interact within each model framework. These discrepancies suggest that each model prioritizes different processes in shaping the cloud phase. Such divergence may limit the reliability of conclusions regarding microphysical processes drawn from any single model.

Anthropogenic aerosol emissions have historically exerted a net cooling effect which has masked some of the simultaneous warming from greenhouse gases (roughly -0.5°C since pre-industrial times). This mean effect is the result of heterogenous climate forcing through aerosol-radiation and aerosol-cloud interactions both locally close to emission sources and remotely via teleconnections. Future reductions and shifts in aerosol emission patterns due to regional clean air policies and shifting industrial production could therefore unmask additional warming and induce spatially complex climate impacts. Therefore, there is a need for computationally efficient tools to assess the climate impacts of possible future aerosol policy decisions.We have developed a machine-learning emulator using Gaussian Processes (GP), trained on output from the Norwegian Earth System Model (NorESM), to predict the global spatially resolved surface temperature response to regional aerosol emission perturbations. We use a novel design for our GP model which considers the joint spatial covariance of the outputs. We show the efficacy of the emulator is comparable to that of the parent model NorESM for a fraction of the computational cost, and then use it to assess potential future aerosol emission scenarios that might be relevant to European policy decisions.

Aerosol particles from both natural and anthropogenic sources play a critical role in the Earth's climate by interacting with solar radiation and clouds. Anthropogenic aerosol and precursor emissions have historically exerted a global cooling effect, which has partially offset the warming from concurrent greenhouse gas emissions. Recent reductions and shifts in aerosol and precursor emission patterns may reduce this offset and introduce spatially and temporarily varying climate impacts. Investigating aerosol-climate effects is typically done with computationally expensive Earth System Models, which include complex representations of physical, chemical, biological, and geological processes and their coupled interactions for the entire global climate system. In this study, we develop a machine-learning climate emulator using Gaussian processes, called AeroGP , that can be used to quickly assess, for example, the impact of different policy decisions on future climate mitigation strategies. The emulator is trained on a unique data set from the Norwegian Earth System Model (NorESM), analyzed as an ensemble here for the first time. AeroGP accounts for the joint spatial covariance of the output variables and captures the complex, heterogeneous impacts of aerosols on surface temperature using coregionalization. We believe this is the first time this method has been used to account for the spatial correlation of such climate data. We show that AeroGP retains the spatial complexity of NorESM at a fraction of the computational cost and demonstrate its usefulness to assess the sensitivity of temperature to idealized future aerosol emission scenarios.

While Arctic amplification is a robust feature of both observed and projected climate change, projections of Arctic climate change are characterized by substantial uncertainty. To better understand the drivers of this uncertainty, we performed a coordinated set of fully coupled experiments with the second version of the Norwegian Earth System Model (NorESM2), in which selected processes of key importance for the Arctic climate have been modified. They include improved representation of (1) mixed-phase clouds, (2) eddy processes in the upper ocean, (3) Greenland ice sheet coupling, (4) snow on sea ice processes, and (5) ozone chemistry. For each modification, we carried out sensitivity experiments following the protocols for the historical simulation of the sixth phase of the Coupled Model Intercomparison Project (CMIP6) and a future high-emissions scenario (ssp585). This results in an ensemble of modified historical and ssp585 experiments.The sensitivity experiments all demonstrate enhanced future Arctic warming compared to the unmodified historical and ssp585 experiments. Moreover, the amplitude of the additional warming varies considerably, with the difference between the experiment with the strongest and weakest Arctic mean warming reaching similar to 9 K during the winter season by the end of the 21st century. The warming signal is dominated by a relatively uniform Arctic warming, which, according to the CMIP6 ssp585 long-term extension, starts to equilibrate during the 22nd century. Surface temperature decomposition shows that winter warming is primarily driven by an enhanced greenhouse effect due to increased cloud cover, near-surface humidity, and the resulting increase in downwelling longwave radiation. The temperature response is most pronounced in the sea ice retreat regions, with the greatest variability between experiments occurring on the Atlantic side. We also identify an emergent constraint, linking changes in Arctic surface temperatures to changes in ocean heat fluxes and sea ice area. This highlights the importance of correctly representing (contemporary) Northern Hemisphere (NH) sea ice when assessing future projected Arctic warming.

The susceptibility of cloud droplet number to cloud condensation nuclei number is one of the major factors controlling the highly uncertain change in the amount of solar radiation reflected by clouds when aerosol emissions are perturbed (the radiative forcing due to aerosol-cloud interactions). We investigate this susceptibility in low-level stratiform clouds using long-term (3-10-yr) in situ observations of aerosols and clouds at three high-latitude locations. The in situ observations show higher susceptibility for low-level stratiform clouds than values reported for satellite data. We estimate -1.16 W m-2 for the aerosol indirect radiative forcing on the basis of our observations, which is at the higher end of satellite-derived forcing estimates and the uncertainty range of the most recent Intergovernmental Panel on Climate Change report. We evaluate four Earth system models against the observations and find large inter-model variability in the susceptibility. Our results demonstrate that, even if the susceptibility in some of the models is relatively close to observations, the underlying physics in the models is unrealistic when compared with observations. We show that the inter-model variability is driven by differences in sub-grid-scale updraught velocities and aerosol size distributions, raising a need to improve these aspects in models.

Terrestrial isoprene, a biogenic volatile organic compound emitted by many plants, indirectly influences Earth's radiative balance through its interactions with atmospheric oxidants, affecting ozone formation, methane lifetime, and secondary aerosol production. Elucidating its historical changes is therefore important for predicting climate change and air quality. Isoprene emissions can respond to climate (e.g. temperature, shortwave radiation, precipitation), land use and land cover change (LULCC), and atmospheric CO2 concentrations. However, historical trends of isoprene emissions and the relative influences of the respective drivers of those trends remain highly uncertain. This study addresses uncertainty in historical isoprene emission trends and their influential factors, particularly the roles of climate, LULCC, and atmospheric CO2 (via fertilization and inhibition effects). The findings are expected to reconcile discrepancies among different modelling approaches and to improve predictions of isoprene emissions and their climate change effects. To investigate isoprene emission trends, controlling factors, and discrepancies among models, we analysed long-term (1850–2014) global isoprene emissions from online simulations of CMIP6 Earth system models and offline simulations using the Vegetation Integrative SImulator for Trace gases (VISIT) dynamic vegetation model driven by climate reanalysis data. Mean annual global present-day isoprene emissions agree well among models (434–510 TgC yr−1) with a 5 % inter-model spread (24 TgC yr−1), but regional emissions differ greatly (9 %–212 % spread). All models show an increasing trend in global isoprene emissions in recent decades (1980–2014), but their magnitudes vary (+1.27 ± 0.49 TgC yr−2, 0.28 ± 0.11 % yr−1). Long-term trends of 1850–2014 show high uncertainty among models (−0.92 to +0.31 TgC yr−2). Results of emulated sensitivity experiments indicate meteorological variations as the main factor of year-to-year fluctuations, but the main drivers of long-term isoprene emission trends differ among models. Models without CO2 effects implicate climate change as the driver, but other models with CO2 effects (fertilization only/fertilization and inhibition) indicate CO2 and LULCC as the primary drivers. The discrepancies arise from how models account for CO2 and LULCC alongside climate effects on isoprene emissions. Aside from LULCC-induced reductions, differences in CO2 inhibition representation (strength and presence or absence of thresholds) were able to mitigate or reverse increasing trends because of rising temperatures or in combination with CO2 fertilization. Net CO2 effects on global isoprene emissions show the highest inter-model variation (σ=0.43 TgC yr−2), followed by LULCC effects (σ=0.17 TgC yr−2), with climate change effects exhibiting more or less variation (σ=0.06 TgC yr−2). The critical drivers of isoprene emission trends depend on a model's emission scheme complexity. This dependence emphasizes the need for models with accurate representation of CO2 and LULCC effects alongside climate change influences for robust long-term predictions. Important uncertainties remain in understanding the interplay between CO2, LULCC, and climate effects on isoprene emissions, mainly for CO2. More long-term observations of isoprene emissions across various biomes are necessary, along with improved models with varied CO2 responses. Moreover, instead of reliance on the current models, additional emission schemes can better capture isoprene emissions complexities and their effects on climate.

Earth system models (ESMs) participating in the latest Coupled Model Intercomparison Project Phase 6 (CMIP6) simulate various components of fine particulate matter (PM2.5) as major climate forcers. Yet the model performance for PM2.5 components remains little evaluated due in part to lack of observational data. Here, we evaluate near-surface concentrations of PM2.5 and its five main components over China as simulated by fourteen CMIP6 models, including organic carbon (OC, available in 14 models), black carbon (BC, 14 models), sulfate (14 models), nitrate (4 models), and ammonium (5 models). For this purpose, we collect observational data between 2000 and 2014 from a satellite-based dataset for total PM2.5 and from 2469 measurement records in the literature for PM2.5 components. Seven models output total PM2.5 concentrations, and they all underestimate the observed total PM2.5 over eastern China, with GFDL-ESM4 (–1.5 %) and MPI-ESM-1-2-HAM (–1.1 %) exhibiting the smallest biases averaged over the whole country. The other seven models, for which we recalculate total PM2.5 from the available components output, underestimate the total PM2.5 concentrations, partly because of the missing model representations of nitrate and ammonium. Concentrations of the five individual components are underestimated in almost all models, except that sulfate is overestimated in MPI-ESM-1-2-HAM by 12.6 % and in MRI-ESM2-0 by 24.5 %. The underestimation is the largest for OC (by –71.2 % to –37.8 % across the 14 models) and the smallest for BC (–47.9 % to –12.1 %). The multi-model mean (MMM) reproduces fairly well the observed spatial pattern for OC (R = 0.51), sulfate (R = 0.57), nitrate (R = 0.70) and ammonium (R = 0.75), yet the agreement is poorer for BC (R = 0.39). The varying performances of ESMs on total PM2.5 and its components have important implications for the modeled magnitude and spatial pattern of aerosol radiative forcing.

Uncertainty in cloud feedbacks in climate models is a major limitation in projections of future climate. Therefore, evaluation and improvement of cloud simulation are essential to ensure the accuracy of climate models. We analyse cloud biases and cloud change with respect to global mean near-surface temperature (GMST) in climate models relative to satellite observations and relate them to equilibrium climate sensitivity, transient climate response and cloud feedback. For this purpose, we develop a supervised deep convolutional artificial neural network for determination of cloud types from low-resolution (2.5∘×2.5∘) daily mean top-of-atmosphere shortwave and longwave radiation fields, corresponding to the World Meteorological Organization (WMO) cloud genera recorded by human observers in the Global Telecommunication System (GTS). We train this network on top-of-atmosphere radiation retrieved by the Clouds and the Earth’s Radiant Energy System (CERES) and GTS and apply it to the Coupled Model Intercomparison Project Phase 5 and 6 (CMIP5 and CMIP6) model output and the European Centre for Medium-Range Weather Forecasts (ECMWF) Reanalysis version 5 (ERA5) and the Modern-Era Retrospective Analysis for Research and Applications version 2 (MERRA-2) reanalyses. We compare the cloud types between models and satellite observations. We link biases to climate sensitivity and identify a negative linear relationship between the root mean square error of cloud type occurrence derived from the neural network and model equilibrium climate sensitivity (ECS), transient climate response (TCR) and cloud feedback. This statistical relationship in the model ensemble favours models with higher ECS, TCR and cloud feedback. However, this relationship could be due to the relatively small size of the ensemble used or decoupling between present-day biases and future projected cloud change. Using the abrupt-4×CO2 CMIP5 and CMIP6 experiments, we show that models simulating decreasing stratiform and increasing cumuliform clouds tend to have higher ECS than models simulating increasing stratiform and decreasing cumuliform clouds, and this could also partially explain the association between the model cloud type occurrence error and model ECS.

Here we present a pilot study of the sensitivity of summer monsoon precipitation in the Yangtze River Valley (YRV; 110-122 circle E and 27-33 circle N, eastern China) to climatic boundary conditions from the Last Glacial Maximum (LGM), pre-industrial conditions, and the Representative Concentration Pathway 6 emission scenario from two different climate models. Using a quantitative Lagrangian moisture source diagnostic based on backward trajectories, we are able to interpret changes in precipitation amount and seasonality in terms of processes at the source regions and during transport that contribute to YRV precipitation. Thereby, we gain insight into influential processes and characteristics related to precipitation variability and the sensitivity of the summer monsoon hydroclimate in East Asia to boundary-condition changes in models. Comparing 10-year time slices similar to present-day conditions from the NorESM1-M and CAM5.1 models to reanalysis data reveals overall similar moisture source regions, albeit with a tendency for a more local precipitation origin in the climate models. The general characteristics of the moisture sources and moisture transport in the YRV are relatively stable across different climate forcings, both concerning the mean location of source regions, transport distance, and the relative contributions of moisture from land and ocean areas. Changes regarding regional precipitation contributions from the East Asian continent indicate that precipitation recycling responds to different climate forcings. We interpret these findings such that models to first order respond with a scaling rather than reorganisation of the hydroclimate to climatic forcing, while land-atmosphere interactions play an important, but secondary, role. If the model simulations are accurate, the moisture source regions and thus the general processes of precipitation in the YRV could remain relatively stable across different climates. However, some differences in moisture source conditions are larger between the different climate models than between different climatic boundary conditions in the same model. It may therefore be possible that current climate models underestimate the potential for non-linear responses to changing boundary conditions, for example due to precipitation recycling. Although limited by the relatively short analysis period, our findings demonstrate that the diagnosis of moisture sources provides a useful additional perspective for understanding and quantifying precipitation mechanisms and the hydroclimate simulated by models and enables more detailed evaluation of model simulations, for example using paleoclimate records.

Abstract. Here we study the sensitivity of monsoon season precipitation in the Yangtse River Valley (YRV, 110–122° E and 27–33° N, East China) to climatic boundary conditions from the last glacial maximum (LGM), pre-industrial conditions, and with the RCP6.0 emission scenario. Using a quantitative Lagrangian moisture source diagnostic, we interpret changes in precipitation amount and seasonality in terms of processes at the source regions that contribute to YRV precipitation. Thereby, we gain insight into influential processes and characteristics related to precipitation variability, and the sensitivity of the summer monsoon hydroclimate in East Asia to boundary condition changes in models. Comparing 10-year time slices similar to present- day conditions from the NorESM1-M and CAM5.1 models to ERA Interim reanalysis data reveals overall very similar moisture source regions, albeit with a tendency to more local precipitation origin in the climate models. Also across different climate forcings, the general characteristics of the moisture sources and moisture transport to the YRV are relatively stable, both concerning the location of source regions, their magnitudes, and the relative contributions of moisture from land and ocean areas. Differences in moisture source conditions are larger between the different climate models, than between different climatic boundary conditions using the same model. Overall, these findings imply that the moisture source regions, and thus the general processes of precipitation in the YRV could remain relatively stable across different climatic periods. However, the results may also indicate that current climate models underestimate the potential for non-linear responses to changing boundary conditions. The plausibility of moisture source changes simulated by the different models could in be evaluated in the future using paleoclimatic records, such as the stable isotope composition in cave sediments. Overall, our findings underline that the diagnosis of moisture sources provides a useful additional perspective for understanding and quantifying precipitation mechanisms and the hydroclimate simulated by models.

Many different emission pathways exist that are compatible with the Paris climate agreement, and many more are possible that miss that target. While some of the most complex Earth System Models have simulated a small selection of Shared Socioeconomic Pathways, it is impractical to use these expensive models to fully explore the space of possibilities. Such explorations therefore mostly rely on one‐dimensional impulse response models, or simple pattern scaling approaches to approximate the physical climate response to a given scenario. Here we present ClimateBench—the first benchmarking framework based on a suite of Coupled Model Intercomparison Project, AerChemMIP and Detection‐Attribution Model Intercomparison Project simulations performed by a full complexity Earth System Model, and a set of baseline machine learning models that emulate its response to a variety of forcers. These emulators can predict annual mean global distributions of temperature, diurnal temperature range and precipitation (including extreme precipitation) given a wide range of emissions and concentrations of carbon dioxide, methane and aerosols, allowing them to efficiently probe previously unexplored scenarios. We discuss the accuracy and interpretability of these emulators and consider their robustness to physical constraints such as total energy conservation. Future opportunities incorporating such physical constraints directly in the machine learning models and using the emulators for detection and attribution studies are also discussed. This opens a wide range of opportunities to improve prediction, robustness and mathematical tractability. We hope that by laying out the principles of climate model emulation with clear examples and metrics we encourage engagement from statisticians and machine learning specialists keen to tackle this important and demanding challenge.

Qirui Zhong1, Nick Schutgens1, Guido van der Werf 1, Twan van Noije2, Kostas Tsigaridis3,4, Susanne E. Bauer4,3, Tero Mielonen5, Alf Kirkevåg6, Øyvind Seland6, Harri Kokkola5, Ramiro Checa-Garcia7, David Neubauer8, Zak Kipling9, Hitoshi Matsui10, Paul Ginoux11, Toshihiko Takemura12, Philippe Le 5 Sager2, Samuel Rémy13, Huisheng Bian14,15, Mian Chin15, Kai Zhang16, Jialei Zhu17, Svetlana G. Tsyro6, Gabriele Curci18,19, Anna Protonotariou20, Ben Johnson21, Joyce E. Penner22, Nicolas Bellouin23, Ragnhild B. Skeie24, and Gunnar Myhre24

For short-lived climate forcers such as black carbon (BC), the atmospheric concentrations, radiative forcing (RF), and, ultimately, the subsequent effects on climate, depend on the location and timing of the emissions. Here, we employ the NorESM1-Happi version of the Norwegian Earth System Model to systematically study how the RF associated with BC emissions depends on the latitude, longitude, and seasonality of the emissions. The model aerosol scheme is run in an offline mode to allow for an essentially noise-free evaluation of the RF associated with even minor changes in emissions. A total of 960 experiments were run to evaluate the BC direct RF (dirRF) and the RF associated with BC in snow/ice (snowRF) for emissions in 192 latitude–longitude boxes covering the globe, both for seasonally uniform emissions and for emissions in each of the four seasons separately. We also calculate a rough estimate of the global temperature response to regional emissions and provide a Fortran-based tool to facilitate the further use of our results. Overall, the results demonstrate that the BC RFs strongly depend on the latitude, longitude, and season of the emissions. In particular, the global mean dirRF normalized by emissions (direct specific forcing; dirSF) depends much more strongly on the emission location than suggested by previous studies that have considered emissions from continental-/subcontinental-scale regions. Even for seasonally uniform emissions, dirSF varies by more than a factor of 10, depending on the emission location. These variations correlate strongly with BC lifetime, which varies from less than 2 to 11 d. BC dirSF is largest for emissions in tropical convective regions and in subtropical and midlatitude continents in summer, both due to the abundant solar radiation and strong convective transport, which increases BC lifetime and the amount of BC above clouds. The dirSF is also relatively large for emissions in high-albedo, high-latitude regions such as Antarctica and Greenland. The dependence of snow specific forcing (snowSF) on the emission location is even larger. While BC emissions originating from most low-latitude regions result in negligible snowSF, the maxima of snowSF for emissions in polar regions greatly exceed the largest values of dirSF for low-latitude emissions. The large magnitude of snowSF for high-latitude BC emissions suggests that, for a given mass of BC emitted, the climate impacts are also largest for high-latitude emissions. The additivity of the RFs resulting from BC emissions in different regions and seasons is also investigated. It is found that dirRF is almost additive for current-day emissions, so that summing the RFs computed for individual regions/seasons without considering BC emissions from elsewhere overestimates dirRF by less than 10 %. For snowRF, the overestimate is somewhat larger, at ∼ 20 %.

The susceptibility of cloud droplet number concentration (CDNC) to cloud condensation nuclei (CCN) number concentration is one of the major factors controlling the aerosol indirect forcing. In this study we investigate the sensitivity of CDNC to CCN concentrations using long term in-situ observations from three stations (Puijo, Pallas, Zeppelin) locating in Finland and Arctic. These stations represent semiurban, remote and Arctic remote environments with differences in typical updraft velocity conditions as well as in aerosol number concentrations. We compare the in-situ observations with three large scale models (ECHAM-M7, ECHAM-SALSA and NorESM) having differences in aerosol presentation while the activation parametrization is the same in all three model setups. In the comparison we use CDNC and CCN model outputs of the gridbox corresponding to the location and the height for each station. In addition, we compare the updraft velocities from the models and stations when they are available. Our current observational results show very high susceptibility of CDNC and CCN in all investigated stations. The agreement between the large scale models and observations was very good for Puijo and Pallas stations, but for the Arctic station (Zeppelin) the modelled CDNC susceptibility to CCN was much lower than the observed. This might be related to the recent results demonstrating that Aitken mode particles can active to cloud droplets at Zeppelin station (Bulatovic et al., 2021; Karlsson et al., 2021). In addition, at Zeppelin CDNC exhibits very low values which are below the lower bound imposed by ECHAM. References: Bulatovic, I., Igel, A. L., Leck, C., Heintzenberg, J., Riipinen, I., and Ekman, A. M. L.: The importance of Aitken mode aerosol particles for cloud sustenance in the summertime high Arctic – a simulation study supported by observational data, Atmos. Chem. Phys., 21, 3871–3897, https://doi.org/10.5194/acp-21-3871-2021, 2021. Karlsson, L., Krejci, R., Koike, M., Ebell, K., and Zieger, P.: A long-term study of cloud residuals from low-level Arctic clouds, Atmos. Chem. Phys., 21, 8933–8959, 2021.