Ice aggregation in clouds plays a crucial role in cloud development and precipitation formation. Despite the significance of ice aggregation, direct in situ quantification of aggregation rates in natural clouds has been challenging due to the difficulty of tracking ice crystals. Here, we present in situ measurements of ice aggregation rates in persistent supercooled stratiform clouds. Using novel glaciogenic seeding experiments (CLOUDLAB), ice crystals are nucleated upwind and subsequently measured downwind after a known advection time in cloud, allowing us to estimate their age. A deep-learning-based detection algorithm (IceDetectNet) counts the individual monomers of aggregates to derive the initial ice crystal number concentration (ICNCt0). We considered several factors that may influence ice aggregation, including ICNCt0, temperature, ice crystal size, aspect ratio, and turbulence. Among these, ICNCt0 was found to be the dominant factor controlling aggregation rates by three independent approaches: causal inference, a physical equation, and machine learning models. We report, however, a subquadratic dependence of the aggregation rate on ICNCt0 (mean exponent similar to 0.92; 95 % CI: 0.88-0.97), in contrast to theoretical expectations (quadratic dependence). One possible explanation is that aggregation may also involve smaller ice crystals, but this remains hypothetical. To predict aggregation rates, we evaluated 11 machine learning models and a physically based formulation. CatBoost achieved the best statistical performance, while the physical model proved more robust in sensitivity tests. These findings provide new insights into the microphysical and environmental controls of ice aggregation and establish a robust methodological foundation for studying aggregation processes in natural clouds.

For decades, silver iodide (AgI) has been widely used for laboratory ice nucleation experiments and glaciogenic cloud-seeding operations due to its ability to nucleate ice at relatively warm temperatures (up to −3 °C). Despite being one of the most well-characterized ice-nucleating substances, gaps remain in the understanding of how its ice nucleation behavior in the laboratory translates to natural clouds. Here, we present, for the first time, measurements of the ice-nucleated fractions (INFs) of AgI-containing seeding particles, derived from in situ measurements of ice crystal number concentrations (ICNC) and seeding particle number concentrations during glaciogenic cloud-seeding experiments. The experiments were performed as part of the CLOUDLAB project, in which we used targeted cloud seeding with an uncrewed aerial vehicle to try to answer fundamental questions about ice-phase cloud microphysics. Data from 16 seeding experiments show strong linear correlations between ICNC and seeding particle concentration, indicating relatively constant INFs throughout each experiment. Median INFs (0.07 %–1.63 %) were found to weakly increase with decreasing cloud temperature at seeding height (range of −5.1 to −8.3 °C). We compare our results with previous key laboratory experiments and discuss the possible freezing mechanisms. This study can help to bridge the gap in understanding of AgI ice nucleation behavior between laboratory and field experiments which further helps to inform future cloud-seeding operations.

The ice phase is a major contributor to precipitation formation over continents due to its efficiency in growing hydrometeors to large enough sizes for sedimentation. One prominent growth mechanism is the vapor deposition onto ice crystals. However, its actual growth rates remain ambiguous. In the CLOUDLAB project, we conducted field experiments in supercooled clouds with the goal to infer ice crystal growth rates through local perturbations from cloud seeding. In this study, we combine a high‐resolution model setup of 65 m with Lagrangian trajectories to achieve a more straightforward comparison to the observations. We first show that the chosen field experiments can be reproduced in the model in terms of ice crystal number concentration. Second, we perform a series of sensitivity studies by perturbing two parameters in the vapor depositional growth equation. The goal is to understand what change is needed to achieve an agreement between simulated and observed ice crystal growth rates since the default model configuration fails to do so. Increasing the vapor deposition efficiency by a factor of up to three yields comparable growth rates to the observations. Last, we try to quantify the different contributions to the vertical motions within the seeding plume, such as the large‐scale forcing, the underlying topography, and latent heat release upon ice nucleation and growth. We show the different factors are superposed with the large‐scale forcing being a dominant factor. The Lagrangian trajectories proved to be crucial to bridge dynamics and cloud microphysical processes.

The microstructure of liquid and mixed-phase clouds is characterized by the cloud droplet size distribution (CDSD), which influences the cloud evolution and its interaction with radiation. However, state-of-the-art cloud probes still face challenges because they require either platforms that move at constant speed or inlets that can directly alter the actual CDSD. Therefore, precise and accurate in situ measurements of CDSDs, especially of cloud droplets smaller than 6 mu m, are still lacking. This can lead to uncertainties in the microphysics and thus in weather and climate models, which are based on parameterizations often derived from these measurements.We present a new in situ instrument, the Small Holographic Imager for Microscopic Objects (SmHOLIMO), specifically designed to measure a broad spectrum of the CDSDs, i.e., from 3.7 to approximate to 100 mu m, with a sample volume rate of 2.5 cm3s-1. Thereby, SmHOLIMO pushes the resolution limit towards the limits seen with forward-scattering probes, while still maintaining the advantages of open-path holography, i.e., a well-defined sample volume (operation at variable wind speed); no need for an inlet; independence of particle size, phase, refractive index, and shape; and the potential of spatial analyses. After calibrating SmHOLIMO in the laboratory, the instrument was deployed in the field, on a tethered balloon system, probing a dissipating low stratus. We demonstrate its ability to measure the cloud microphysical properties at high spatio-temporal resolution. Furthermore, we compare the SmHOLIMO observations to those of another holographic imager (resolution: 6 mu m) and to co-located remote sensing measurements. We unequivocally show the importance of SmHOLIMO's skills to capture the lower tail of the CDSD, which significantly affects the derived quantities of cloud droplet mean diameter (up to 1.6 times smaller), number concentration (up to 4 times higher), and cloud optical depth (up to 2.7 times higher). SmHOLIMO's high-resolution in situ data of cloud droplets will help us to better interpret observations and to refine the representation of clouds in climate and weather models.

A comprehensive approach to study the seeder-feeder mechanism in unprecedented detail from a combined remote-sensing, in situ, and model perspective is shown. This publication aims at investigating the role of the interplay of a seeder-feeder cloud system and its influence on precipitation formation based on a case study from 8 January 2024 observed over the Swiss Plateau in Switzerland.This case study offers an ideal setup for applying several advanced remote-sensing techniques and retrieval algorithms, including fall streak tracking, radar Doppler peak separation, dual-wavelength radar applications, a liquid detection retrieval, a riming retrieval, and an ice crystals shape retrieval. Results indicate that a large portion of ice mass was rimed, which is attributed to persistent coexistence of falling ice crystals and supercooled water within low-level supercooled liquid water layers. Interaction of seeder and feeder clouds results in a significant precipitation enhancement. This has implications on the water cycle. From the anti-correlation between surface precipitation and liquid water path we estimated that 20 %-40 % of the precipitation stems from the feeder cloud. However, we have to note that the value of 20 %-40 % is strongly dependent on the assumed reproduction rate of liquid water in the feeder cloud. This study aims at giving an overview from a remote-sensing, in situ and model perspective on a seeder-feeder event in an unprecedented detail by exploiting a big set of retrievals applicable to remote-sensing and in situ data. Utilizing different retrievals gives a consistent view on the seeder-feeder case study which is an important basis for future studies. It is demonstrated how improved understanding of seeder-feeder interactions can contribute to enhancing weather forecast models, particularly in regions affected by persistent low-level supercooled stratus clouds.

Ice crystals are essential in the evolution of mixed-phase clouds, as ice crystals can quickly grow to large sizes by vapor diffusion and thereby trigger precipitation formation. Vapor diffusional growth rates of ice crystals were quantitatively studied in the laboratory for several decades, forming the basis of various ice crystal growth models. Since field measurements generally only provide snapshots that lack information on ice crystal age or changes induced by cloud processes, significant gaps remain in quantitative field observations, impeding the validation of laboratory experiments and models.Our study addresses this gap through innovative glaciogenic cloud seeding experiments in persistent low-level stratus clouds in the CLOUDLAB project. The controllability and repeatability of our seeding experiments facilitates the quantification of diffusional ice crystal growth rates in natural clouds via in situ measurements. We report growth rates of 0.17-0.81 mu ms-1 (major axis of pristine ice crystals) from 14 seeding experiments between -5.1 and -8.3 degrees C. We also observe how microphysical characteristics induce strong variations in the growth rates, e.g., reduced growth rates in seeding-induced regions of high ice crystal number concentrations. For better comparison to laboratory and non-seeded clouds, we developed two filtering methods to isolate growth conditions less affected by the experimental setup. The comparison shows that the temperature-dependent growth rate variations align with laboratory data, whereas absolute laboratory values are higher. Our findings provide valuable insights into the vapor diffusional growth of ice crystals in natural clouds and connect in situ observations with laboratory and modeling studies.

Atmospheric icing, the accumulation of ice on surfaces, is a severe concern for the aviation industry. Deicing and icing prediction tools are necessary for pilots to ensure flight safety, and while there is established technology for large aircraft icing, more research is needed for smaller uncrewed aerial vehicles (UAVs). Here, we present measurements from 59 flights of a multirotor UAV into wintertime low stratus clouds of temperatures between -3 and -10 degrees C. The UAV is equipped with rotor heating to allow flights up to 10 min in icing conditions. Icing severity was quantified by using the rate of increase in battery current during icing, and was then compared with simultaneous, co-located measurements of liquid water content (LWC). LWC measurements were (a) calculated from cloud droplets measured with an in situ holographic imager on a tethered balloon system and (b) retrieved from remote sensing observations (microwave radiometer, ceilometer, cloud radar). We show that, for these environmental conditions, icing was strongly positively correlated to LWC over an LWC range of 0.02 to 0.5 g m(-3), independent of temperature and mean droplet size, though droplets >50 mu m in diameter may contribute to increased icing severity. We also show that the LWC retrieved from remote sensing agrees well with the in situ measurements, indicating that remote sensing measurements can effectively be used to assess icing conditions. These are the first known measurements of multirotor UAV icing with co-located LWC measurements in natural clouds.

The ice phase in clouds is essential for precipitation formation over continents. The underlying processes for ice growth are still poorly understood, leading to large uncertainties in precipitation forecasts and climate simulations. One crucial aspect is the Wegener–Bergeron–Findeisen (WBF) process, which describes the growth of ice crystals at the expense of cloud droplets, leading to a partial or full glaciation of the cloud. In the CLOUDLAB project, we employ glaciogenic cloud seeding to initiate the ice phase in supercooled low-level clouds in Switzerland using uncrewed aerial vehicles with the goal of investigating the WBF process. An extensive setup of ground-based remote-sensing and balloon-borne in situ instrumentation allows us to observe the formation and subsequent growth of ice crystals in great detail. In this study, we compare the seeding signals observed in the field to those simulated using a numerical weather model in large-eddy mode (ICON-LEM). We first demonstrate the capability of the model to accurately simulate and reproduce the seeding experiments across different environmental conditions. Second, we investigate the WBF process in the model by comparing the simulated cloud droplet and ice crystal number concentration changes to in situ measurements. In the field experiments, simultaneous reductions in cloud droplet number concentrations with increased ice crystal number concentrations were observed, with periods showing a full depletion of cloud droplets. The model can reproduce the observed ice crystal number concentrations most of the time; however, it cannot reproduce the observed fast reductions in cloud droplet number concentrations. Our detailed analysis shows that the WBF process appears to be less efficient in the model than in the field. In the model, exaggerated ice crystal number concentrations are required to produce comparable changes in cloud droplet number concentrations, highlighting the inefficiency of the WBF process in the numerical weather model ICON.

Uncrewed aerial vehicles (UAVs) have become widely used in a range of atmospheric science research applications. Because of their small size, flexible range of motion, adaptability, and low cost, multirotor UAVs are especially well-suited for probing the lower atmosphere. However, their use so far has been limited to conditions outside of clouds, first because of the difficulty of flying beyond visual line of sight and second because of the challenge of flying in icing conditions in supercooled clouds. Here, we present two UAVs for cloud microphysical research: one UAV (the measurement UAV) equipped with a Portable Optical Particle Spectrometer (POPS) and meteorological sensors to probe the aerosol and meteorological properties in the boundary layer and one UAV (the seeding UAV) equipped with seeding flares to produce a plume of particles that can nucleate ice in supercooled clouds. A propeller heating mechanism on both UAVs allows for operating in supercooled clouds with icing conditions. These UAVs are an integral part of the CLOUDLAB project in which glaciogenic cloud seeding of supercooled low stratus clouds is utilized for studying aerosol–cloud interactions and ice crystal formation and growth. In this paper, we first show validations of the POPS on board the measurement UAV, demonstrating that the rotor turbulence has a small effect on measured particle number concentrations. We then exemplify the applicability for profiling the planetary boundary layer, as well as for sampling and characterizing aerosol plumes, in this case, the seeding plume. We also present a new method for filtering out high-concentration data to ensure good data quality of POPS. We explain the different flight patterns that are possible for both UAVs, namely horizontal or vertical leg patterns or hovering, with an extensive and flexible parameter space for designing the flight patterns according to our scientific goals. Finally, we show two examples of seeding experiments: first characterizing an out-of-cloud seeding plume with the measurement UAV flying horizontal transects through the plume and, second, characterizing an in-cloud seeding plume with downstream measurements from a POPS and a holographic imager mounted on a tethered balloon. Particle number concentrations and particle number size distributions of the seeding plume from the experiments reveal that we can successfully produce and measure the seeding plume, both in-cloud (with accompanying elevated ice crystal number concentrations) and out-of-cloud. The methods presented here will be useful for probing the lower atmosphere, for characterizing aerosol plumes, and for deepening our cloud microphysical understanding through cloud seeding experiments, all of which have the potential to benefit the atmospheric science community.

The representation of cloud processes in models is one of the largest sources of uncertainty in weather forecast and climate projections. While laboratory settings offer controlled conditions for studying cloud processes, they cannot reproduce the full range of conditions and interactions present in natural cloud systems. To bridge this gap, here we leverage weather modification, specifically glaciogenic cloud seeding, to investigate ice growth rates within natural clouds. Seeding experiments were conducted in supercooled stratus clouds (at -8 to -5 degrees C) using an uncrewed aerial vehicle, and the created ice crystals were measured 4-10 min downwind by in situ and ground-based remote sensing instrumentation. We observed substantial variability in ice crystal growth rates within natural clouds, attributed to variations in ice crystal number concentrations and in the supersaturation, which is difficult to reproduce in the laboratory and which implies faster precipitation initiation than previously thought. We found that for the experiments conducted at -5.2 degrees C, the ice crystal populations grew nearly linearly during the time interval from 6 to 10 min. Our results demonstrate that the targeted use of weather modification techniques can be employed for fundamental cloud research (e.g. ice growth processes, aerosol-cloud interactions), helping to advance cloud microphysics parameterizations and to improve weather forecasts and climate projections.

The shape of ice crystals affects their radiative properties, growth rate, fall speed, and collision efficiency; thus, it plays a significant role in cloud optical properties and precipitation formation. Ambient conditions, like temperature and humidity, determine the basic habit of ice crystals, while microphysical processes, such as riming and aggregation, further shape them, resulting in a diverse set of ice crystal shapes and effective densities. Current classification algorithms face two major challenges: (1) ice crystals are often classified as a whole (at the image scale), necessitating identification of the dominant component of aggregated ice crystals, and (2) single-label classifications lead to information loss because of the compromise between basic habit and microphysical process information. To address these limitations, we present a two-pronged solution here: (1) a rotated object detection algorithm (IceDetectNet) that classifies each component of an aggregated ice crystal individually and (2) a multi-label classification scheme that considers both basic habits and physical processes simultaneously. IceDetectNet was trained and tested on two independent datasets obtained by a holographic imager during the NASCENT campaign in Ny-& Aring;lesund, Svalbard, in November 2019 and April 2020. The algorithm correctly classified 92 % of the ice crystals as either aggregate or non-aggregate and achieved an overall accuracy of 86 % for basic habits and 82 % for microphysical process classification. At the component scale, IceDetectNet demonstrated high detection and classification accuracy across all sizes, indicating its ability to effectively classify individual components of aggregated ice crystals. Furthermore, the algorithm demonstrated a good generalization ability by classifying ice crystals from an independent generalization dataset with overall accuracies above 70 %. IceDetectNet can provide a deeper understanding of ice crystal shapes, leading to better estimates of ice crystal mass, fall velocity, and radiative properties; therefore, it has the potential to improve precipitation forecasts and climate projections.

Cloud radar Doppler spectra are of particular interest for investigating cloud microphysical processes, such as ice formation, riming and ice multiplication. When hydrometeor types within a cloud radar observation volume have different terminal fall velocities, they can produce individual Doppler spectrum peaks. The peaks of different particle types can overlap and be further broadened and blended by turbulence and other dynamical effects. If these (sub-)peaks can be separated, properties of the underlying hydrometeor populations can potentially be estimated, such as their fall velocity, number, size and to some extent their shape. However, this task is complex and dependent on the operation settings of the specific cloud radar, as well as atmospheric dynamics and hydrometeor characteristics. As a consequence, there is a need for adjustable tools that are able to detect peaks in cloud radar Doppler spectra to extract the valuable information contained in them. This paper presents the synergistic use of two algorithms used for analyzing the peaks in Doppler spectra: PEAKO and peakTree. PEAKO is a supervised machine learning tool that can be trained to obtain the optimal parameters for detecting peaks in Doppler spectra for specific cloud radar instrument settings. The learned parameters can then be applied by peakTree, which is used to detect, organize and interpret Doppler spectrum peaks.The application of the improved PEAKO-peakTree toolkit is demonstrated in two case studies. The interpretation is supported by forward-simulated cloud radar Doppler spectra by the Passive and Active Microwave TRAnsfer tool (PAMTRA), which are also used to explore the limitations of the algorithm toolkit posed by turbulence and the number of spectral averages chosen in the radar settings.From the PAMTRA simulations, we can conclude that a minimum number of n = 20-40 spectral averages is desirable for Doppler spectrum peak discrimination. Furthermore, small liquid peaks can only be reliably separated for eddy dissipation rate values up to approximately 0.0002 m2s-3 in the simulation setup which we tested here.The first case study demonstrates that the methods work for different radar systems and settings by comparing the results for two cloud radar systems which were operated simultaneously at a site in Punta Arenas, Chile. Detected peaks which can be attributed to liquid droplets agree well between the two systems, as well as with an independent liquid-predicting neural network. The second case study compares PEAKO-peakTree-detected cloud radar Doppler spectrum peaks to in situ observations collected by a balloon-based holographic imager during a campaign in Ny-& Aring;lesund, Svalbard. This case demonstrates the algorithm toolkit's ability to identify different hydrometeor types but also reveals its limitations posed by strong turbulence and a low n.Despite these challenges, the algorithm toolkit offers a powerful means of extracting comprehensive information from cloud radar observations. In the future, we envision PEAKO-peakTree applications on the one hand for interpreting cloud microphysics in case studies. The identification of liquid cloud peaks emerges as a valuable asset, e.g., in studies on cloud radiative effects, in seeder-feeder processes, or for tracing vertical air motions. Furthermore, the computation of the moments for each subpeak enables the tracking of hydrometeor populations and the observation of growth processes along fallstreaks. On the other hand, PEAKO-peakTree applications could be extended to statistical evaluations of longer data sets. Both algorithms are openly available on GitHub, offering accessibility for the scientific community.

The PolarCAP project aims to decode the interplay between aerosol and cloud-microphysical processes by investigating the evolution of the ice phase under slightly supercooled conditions (T > -10°C) within a thermodynamically and aerosol-controlled environment, employing radar polarimetry and spectral-bin modelling. In collaboration with the CLOUDLAB project at the ETH Zurich, PolarCAP investigates the development of an artificially initiated ice phase within supercooled stratus clouds. Utilizing cloud seeding with silver iodide, CLOUDLAB initiates the freezing process of cloud droplets. The subsequent evolution is monitored using in-situ measurements and conventional ground-based cloud remote sensing tools. The collaboration has yielded a unique dataset, incorporating polarimetric radar and lidar observations from the Leipzig Aerosol and Cloud Remote Observing System (LACROS) in tandem with data from the cloud-resolving spectral bin model COSMO-SPECS. We will present a comparative evaluation between observational data from CLOUDLAB and LACROS, and the innovative COSMO-SPECS simulations featuring an artificial INP source, termed COSMO-SPECS-FLARE (CSF). The inclusion of this source within the COSMO-SPECS framework is pivotal for precipitating ice nucleation, thereby significantly improving the accuracy of cloud microphysical process simulations. A notable advancement of this research is the coupling of CFS output with the radar-forward-simulator PAMTRA. This integration is instrumental in translating the model output into observational parameters, enabling intricate analysis of radar reflectivity, mean Doppler velocity, and other vital atmospheric metrics. Preliminary results are centered on the alignment of radar-forward-simulation predictions based on CSF output with standard remote sensing assessments, including ice and liquid water content retrievals. The analysis highlights the enhanced capacity of the CSF model to replicate observed cloud responses, thus providing comprehensive insights into cloud microphysical processes, particularly in the context of induced ice nucleation.

Ice formation and growth processes play a crucial role in the evolution of cloud systems and the formation of precipitation. However, the initial formation and growth of ice crystals are challenging to study in the real atmosphere resulting in uncertainties in weather forecasts and climate projections. The CLOUDLAB project tackles this problem by using supercooled stratus clouds as a natural laboratory for targeted glaciogenic cloud seeding to advance the understanding of ice processes: Ice nucleating particles are injected from an uncrewed aerial vehicle (UAV) into supercooled stratus clouds to induce ice crystal formation and subsequent growth processes. Microphysical changes induced by seeding are measured 3-15 min downstream of the seeding location using in situ and ground-based remote sensing instrumentation. The novel application of seeding with a multirotor UAV combined with the persistent nature of stratus clouds enables repeated seeding experiments under similar and well-constrained initial conditions. This article describes the scientific goals, experimental design, and first results of CLOUDLAB. First, the seeding plume is characterized by using measurements of a UAV equipped with an optical particle counter. Second, the seeding-induced microphysical changes observed by cloud radars and a tethered balloon system are presented. The seeding signatures were detected by regions of increased radar reflectivity (>-20 dBZ), which were 10-20 dBZ higher than the natural background. Simultaneously, high concentrations of seeding particles and ice crystals (up to 2,000 L-1) were observed. A cloud seeding case was simulated with the numerical weather model ICON to contextualize the findings.

<p>Wintertime stratus clouds over the Swiss Plateau can last for days.  They dissipate either due to airmass changes, absorption of solar radiation during the day, or after glaciation when a sufficiently large number of cloud droplets freezes. After formation, the ice crystals grow in an ice-supersaturated environment via vapor deposition until they are large enough to sediment from the cloud as drizzle or freezing drizzle.</p> <p>To better understand how quickly ice crystals of various habits grow in real clouds with turbulence, we conduct glaciogenic seeding experiments in wintertime stratus clouds over the Swiss Plateau in our project CLOUDLAB[1]. During these experiments, a drone releases silver iodide (AgI) particles into the cloud, upwind of our measurement site, and we use various ground-based remote sensing and in-situ cloud and aerosol instruments to detect the microphysical changes induced by seeding. Preliminary results from the first CLOUDLAB field campaign proved that our method successfully allows us to detect the seeding signal in the cloud radar. In addition to our field measurements, we conduct numerical model simulations with ICON at different horizontal resolutions and different seeding particle concentrations to understand which seeding AgI concentration is theoretically needed for partial or full glaciation of the cloud, i.e. how fast the ice crystals grow at the expense of the evaporating cloud droplets due to the Wegener-Bergeron-Findeisen process.</p> <p>First results will be presented in this talk.</p> <div><br /> <div> <p>[1] https://cloudlab.ethz.ch/</p> </div> </div>

The sizes and shapes of ice crystals influence the radiative properties of clouds, as well as precipitation initiation and aerosol scavenging. However, ice crystal growth mechanisms remain only partially characterized. We present the growth processes of two complex ice crystal habits observed in Arctic mixed-phase clouds during the Ny-angstrom lesund AeroSol Cloud ExperimeNT campaign. First, are capped-columns with multiple columns growing out of the plates' corners that we define as columns on capped-columns. These ice crystals originated from cycling through the columnar and plate temperature growth regimes, during their vertical transport by in-cloud circulation. Second, is aged rime on the surface of ice crystals having grown into faceted columns or plates depending on the environmental conditions. Despite their complexity, the shapes of these ice crystals allow to infer their growth history and provide information about the in-cloud conditions. Additionally, these ice crystals exhibit complex shapes and could enhance aggregation and secondary ice production.

The Arctic is very susceptible to climate change and thus is warming much faster than the rest of the world. Clouds influence terrestrial and solar radiative fluxes and thereby impact the amplified Arctic warming. The partitioning of thermodynamic phases (i.e., ice crystals and water droplets) within mixed-phase clouds (MPCs) especially influences their radiative properties. However, the processes responsible for ice crystal formation remain only partially characterized. In particular, so-called secondary ice production (SIP) processes, which create supplementary ice crystals from primary ice crystals and the environmental conditions that they occur in, are poorly understood. The microphysical properties of Arctic MPCs were measured during the Ny-Ålesund AeroSol Cloud ExperimENT (NASCENT) campaign to obtain a better understanding of the atmospheric conditions favorable for the occurrence of SIP processes. To this aim, the in situ cloud microphysical properties retrieved by a holographic cloud imager mounted on a tethered balloon system were complemented by ground-based remote sensing and ice-nucleating particle measurements. During the 6 d investigated in this study, SIP occurred during about 40 % of the in-cloud measurements, and high SIP events with number concentrations larger than 10 L−1 of small pristine ice crystals occurred in 4 % of the in-cloud measurements. This demonstrates the role of SIP for Arctic MPCs. The highest concentrations of small pristine ice crystals were produced at temperatures between −5 and −3 ∘C and were related to the occurrence of supercooled large droplets freezing upon collision with ice crystals. This suggests that a large fraction of ice crystals in Arctic MPCs are produced via the droplet-shattering mechanism. From evaluating the ice crystal images, we could identify ice–ice collision as a second SIP mechanism that dominated when fragile ice crystals were observed. Moreover, SIP occurred over a large temperature range and was observed in up to 80 % of the measurements down to −24 ∘C due to the occurrence of ice–ice collisions. This emphasizes the importance of SIP at temperatures below −8 ∘C, which are currently not accounted for in most numerical weather models. Although ice-nucleating particles may be necessary for the initial freezing of water droplets, the ice crystal number concentration is frequently determined by secondary production mechanisms.

Ice crystals are an important component of clouds due to their strong impact on cloud radiative properties and precipitation formation. The shape of an ice crystal impacts its radiative effect, riming efficiency, and fall speed. The ice crystal shapes are dependent on (i) the environment (temperature and humidity) that they grow in and (ii) the microphysical processes (i.e. riming, aggregation) that they have experienced. These connections offer a great opportunity to trace back the previous in-cloud conditions and the microphysical processes in clouds. Thus, ice crystal shape classification is crucial to better understand radiation properties and precipitation formation of clouds. Scientists have explored and developed various algorithms to automatically classify ice crystal shapes in the past decades. Among those, the machine learning algorithm Convolutional Neural Network (CNN), shows a good performance due to its ability to catch the main features that describe ice crystal habits and recognize patterns between images. However, the existing classification methods all show an overlap between physical process categories (i.e. rimed) and basic habit categories (i.e. column) of ice crystals due to the existence of compound ice (i.e. column-rimed), especially in situations conducive to light riming. A CNN was trained using over 10’000 images of pristine and complex ice crystals recorded by a holographic imager during the NASCENT campaign (Pasquier et al., BAMS, in revision) in Fall 2019, in Ny-Ålesund, Svalbard. To avoid the overlap between physical process and basic habit categories of ice crystals, each ice label contains two properties; one of 7 basic habits property (i.e. column) and up to 3 physical process properties (aggregate, rimed, aged). The trained model gives us both the basic habit and the physical processes information, which helps us to better understand the microphysical processes in clouds.

The cloud radiative feedbacks are particularly complex and uncertain in the Arctic, which is a region of amplified warming. Within mixed-phase clouds (MPCs), the radiation fluxes are influenced by the thermodynamic phase and cloud particle concentrations. However, the processes responsible for ice crystal formation, particularly secondary ice production (SIP) processes, are still poorly understood. We conducted in-situ cloud microphysical measurements in Ny-Ålesund, Svalbard, as part of the Ny-Ålesund AeroSol Cloud ExperimeNT campaign (NASCENT, Pasquier et al., BAMS, in revision). The main instrument used was a holographic cloud probe mounted on a tethered balloon system to image cloud particles. Additionally, ambient ice nucleating particles (INPs) and cloud condensation nuclei (CCN) were measured at ground level, remote sensing instruments (e.g. cloud radar) profiled the entire troposphere, and radiosondes were launched to determine the in-cloud temperature profiles. Here we discuss the SIP occurrence in Arctic MPCs measured in autumn 2019 and spring 2020. We defined local SIP as occurring when the concentration of pristine ice crystals smaller than 100 µm in diameter is larger than the INP concentration. During the six days of measurements in MPCs, regions with local SIP were observed in 40% of the in-cloud measurements. Regions with high concentrations of small pristine ice crystals (>10 L-1) coincided with the presence of large frozen and broken drops, providing evidence for SIP during the freezing of drizzle drops. We suggest that the formation of drizzle drops initiating high SIP upon freezing was determined by the low CCN concentration in the pristine Arctic environment. Furthermore, SIP occurred at all observed temperatures (-24 °C to -2 °C). The frequency of SIP occurrence was highest in the temperature range between -24 °C and -18 °C (up to 96%), whereas the concentration of small pristine ice crystal peaked between -5 °C and -3 °C (up to 95 L-1). Our observations demonstrate the high importance of SIP for ice crystal formation in Arctic MPCs over a larger temperature range. Thereby, SIP influences the radiative properties of Arctic MPCs and hence the surface radiative energy budget.

Abstract. Understanding the evolution of the ice phase within mixed-phase clouds (MPCs) is necessary to reduce uncertainties related to the cloud radiative feedback in climate projections and precipitation initiation. Both primary ice formation via ice nucleating particles (INPs) and secondary ice production (SIP) within MPCs are unconstrained, not least because of the lack of atmospheric observations. In the past decades, advanced remote sensing methods have emerged which provide high resolution data of aerosol and cloud properties and could be key in understanding microphysical processes on a global scale. In this study, we retrieved INP concentrations, and ice multiplication factors (IMFs) in wintertime orographic clouds using active remote sensing and in situ observations obtained during the RACLETS campaign in the Swiss Alps. INP concentrations in air masses dominated by Saharan dust and continental aerosol were retrieved from a polarization Raman lidar and validated with aerosol and INP in situ observations on a mountaintop. A calibration factor of 0.0204 for the global INP parameterization by DeMott et al. (2010) is derived by comparing in situ aerosol and INP measurements improving the INP concentration retrieval for continental aerosols. Based on combined lidar and radar measurements, the ice crystal number concentration and ice water content were retrieved and validated with balloon-borne in situ observations, which agreed with the balloon-borne in situ observations within an order of magnitude. For seven cloud cases the ice multiplication factors (IMFs), defined as the quotient of the ice crystal number concentration to the INP concentration, were calculated. The median IMF was around 80 and SIP was active (defined as IMFs > 1) nearly 85 % of the time. SIP was found to be active at all observed temperatures (−30 °C to −5 °C) with highest IMFs between −20 °C and −5 °C. The introduced methodology could be extended to larger datasets to better understand the impact of SIP not only over the Alps but also at other locations and for other cloud types.