Accurate representations of clouds are required in large-scale weather and climate models to make detailed and precise predictions of the Earth's weather and climate. Representations of clouds within these models are limited by the present understanding of the role of aerosols in the microphysical processes responsible for cloud formation and development. As part of a NERC funded CASE studentship with the Met Office, this thesis aims to test new aerosol-dependent mixed-phase cloud parametrizations by obtaining extensive cloud microphysical measurements in-situ and comparing and contrasting them with model simulations. Cloud particle concentrations were measured during the Ice NUcleation Process Investigation And Quantification (INUPIAQ) field campaign at Jungfraujoch in Switzerland. A new probe was used to separate droplet and small ice concentrations by using depolarisation ratio and size thresholds. Whilst the new small ice crystal and droplet number concentrations compared favourably with other instruments, the size and depolarisation ratio thresholds were found to be subjective, and suggested to vary from cloud to cloud. An upwind site was chosen to measure out-of-cloud aerosol particle concentrations during INUPIAQ. During periods where the site was out-of-cloud and upwind of Jungfraujoch, several large-scale model simulations were run using the aerosol concentrations in an aerosol-dependent ice nucleation parametrization. The inclusion of the parametrization failed to increase the simulated ice crystal number concentrations, which were several orders of magnitude below those observed in-situ at Jungfraujoch. Several possible explanations for the high observed ice crystal number concentrations at Jungfraujoch are tested using further model simulations. Further primary ice nucleation was ruled out, as the inclusion of additional ice nucleating particles in the model simulations suppressed the liquid water content, preventing the simulation of the mixed-phase clouds observed during INUPIAQ. The addition of ice crystals produced via the Hallett-Mossop process upwind of Jungfraujoch into the model only infrequently provided enough ice crystals to match the observed concentrations. The inclusion of a simple surface flux of hoar crystals into the model simulations was found to produce ice crystal number concentrations of a similar magnitude to those observed at Jungfraujoch, without depleting the simulated liquid water content. By confronting models with in-situ observations of cloud microphysical process, this thesis highlights interactions between surface ice crystals and mixed-phase clouds, and their potential impact on large-scale models.

Arctic clouds are poorly represented in numerical models due to the complex, small-scale interactions which occur within them. Modelled cloud fractions are often significantly less than observed in this region; therefore, the radiative budget is not accurately simulated and forecasts of the melting cryosphere are fraught with uncertainty. Our ability to accurately model Arctic clouds can be improved through observational studies. Recent in situ airborne measurements from the springtime Aerosol-Cloud Coupling and Climate Interactions in the Arctic (ACCACIA) campaign are presented in this thesis to improve our understanding of the cloud microphysical interactions unique to this region. Aerosol-cloud interactions - where aerosol particles act as ice nucleating particles (INPs) or cloud condensation nuclei (CCN) - are integral to the understanding of clouds on a global scale. In the Arctic, uncertainties caused by our poor understanding of these interactions are enhanced by strong feedbacks between clouds, the boundary layer, and the sea ice. In the Arctic spring, aerosol-cloud interactions are affected by the Arctic haze, where a stable boundary layer allows aerosol particles to remain in the atmosphere for long periods of time. This leads to a heightened state of mixing in the aerosol population, which affects the ability of particles to act as INPs or CCN. Aerosol particle compositional data are presented to indicate which particles are present during the ACCACIA campaign, and infer how they may participate in aerosol-cloud interactions. Mineral dusts (known INPs) are identified in all flights considered, and the dominating particle classes in each case vary with changing air mass history. Mixed particles, and an enhanced aerosol loading, are identified in the final case. Evidence is presented which suggests these characteristics may be attributed to biomass burning activities in Siberia and Scandinavia. Additionally, in situ airborne observations are presented to investigate the relationship between the Arctic atmosphere and the mixed-phase clouds - containing both liquid cloud droplets and ice crystals - common to this region. Cloud microphysical structure responds strongly to changing surface conditions, as strong heat and moisture fluxes from the comparatively-warm ocean promote more turbulent motion in the boundary layer than the minimal heat fluxes from the frozen sea ice. Observations over the transition from sea ice to ocean show that the cloud liquid water content increases four-fold, whilst ice crystal number concentrations, N_ice, remain consistent at ~0.5/L. Following from this study, large eddy simulations are used to illustrate the sensitivity of cloud structure, evolution, and lifetime to N_ice. To accurately model mixed-phase conditions over sea ice, marginal ice, and ocean, ice nucleation must occur under water-saturated conditions. Ocean-based clouds are found to be particularly sensitive to N_ice, as small decreases in N_ice allow glaciating clouds to be sustained, with mixed-phase conditions, for longer. Modelled N_ice also influences precipitation development over the ocean, with either snow or rain depleting the liquid phase of the simulated cloud.

Arctic climate is changing fast, but weather forecast and climate models have serious deficiencies in representing the Arctic atmosphere, because of the special conditions that occur in this region. The cold ice surface and the advection of warm air aloft from the south result in a semi-continuous presence of a temperature inversion, known as the “Arctic inversion”, which is governed by interacting large-scale and local processes, such as surface fluxes and cloud formation. In this thesis these poorly understood interactions are investigated using observations from field campaigns on the Swedish icebreaker Oden: The Arctic Summer Cloud Ocean Study (ASCOS) in 2008 and the Arctic Clouds in Summer Experiment (ACSE) in 2014. Two numerical models are also used to explore these data: the IFS global weather forecast model from the European Center for Medium-range Weather Forecasts and the MIMICA LES from Stockholm University. Arctic clouds can persist for a long time, days to weeks, and are usually mixed-phase; a difficult to model mixture of super-cooled cloud droplets and ice crystals. Their persistence has been attributed to several mechanisms, such as large-scale advection, surface evaporation and microphysical processes. ASCOS observations indicate that these clouds are most frequently decoupled from the surface; hence, surface evaporation plays a minor role. The determining factor for cloud-surface decoupling is the altitude of the clouds. Turbulent mixing is generated in the cloud layer, forced by cloud-top radiative cooling, but with a high cloud this cannot penetrate down to the surface mixed layer, which is forced primarily by mechanical turbulence. A special category of clouds is also found: optically thin liquid-only clouds with stable stratification, hence insignificant in-cloud mixing, which occur in low-aerosol conditions. IFS model fails to reproduce the cloud-surface decoupling observed during ASCOS. A new prognostic cloud physics scheme in IFS improves simulation of mixed-phase clouds, but does not improve the warm bias in the model, mostly because IFS fails to disperse low surface-warming clouds when observations indicate cloud-free conditions. With increasing summer open-water areas in a warming Arctic, there is a growing interest in processes related to the ice marginal zones and the summer-to-autumn seasonal transition. ACSE included measurements over both open-water and sea-ice surfaces, during melt and early freeze. The seasonal transition was abrupt, not gradual as would have been expected if it was primarily driven by the gradual changes in net solar radiation. After the transition, the ocean surface remained warmer than the atmosphere, enhancing surface cooling and facilitating sea-ice formation. Observations in melt season showed distinct differences in atmospheric structure between the two surface types; during freeze-up these largely disappear. In summer, large-scale advection of warm and moist air over melting sea ice had large impacts on atmospheric stability and the surface. This is explored with an LES; results indicate that while vertical structure of the lowest atmosphere is primarily sensitive to heat advection, cloud formation, which is of great importance to the surface energy budget, is primarily sensitive to moisture advection. / <p>At the time of the doctoral defense, the following paper was unpublished and had a status as follows: Paper 4: Manuscript.</p>

In the atmosphere, cloud droplets can remain in a supercooled liquid phase at temperatures as low as -40 °C. Above this temperature, cloud droplets freeze via heterogeneous ice nucleation whereby a rare and poorly understood subset of atmospheric particles catalyze the ice phase transition. As the phase state of clouds is critical in determining their radiative properties and lifetime, deficiencies in our understanding of heterogeneous ice nucleation poses a large uncertainty on our efforts to predict human induced global climate change. Experimental challenges in properly simulating particle-induced freezing processes under atmospherically relevant conditions have largely contributed to the absence of a well-established model and parameterizations that accurately predict heterogeneous ice nucleation. Conversely, the sparsity of reliable measurement techniques available struggle to be interpreted by a single consistent theoretical or empirical framework, which results in layers of uncertainty when attempting to extrapolate useful information regarding ice nucleation for use in atmospheric cloud models. In this dissertation a new framework for describing heterogeneous ice nucleation is developed. Starting from classical nucleation theory, the surface of an ice nucleating particle is treated as a continuum of heterogeneous ice nucleating activity and a particle specific distribution of this activity g is derived. It is hypothesized that an individual particle species exhibits a critical surface area. Above this critical area the ice nucleating activity of a particle species can be described by one g distribution, 𝑔, while below it 𝑔 expresses itself expresses externally resulting in particle to particle variability in ice nucleating activity. The framework is supported by cold plate droplet freezing measurements for dust and biological particles in which the total surface area of particle material available is varied. Freezing spectra above a certain surface area are shown to be successfully fitted with 𝑔 while a process of random sampling from 𝑔 can predict the freezing behavior below the identified critical surface area threshold. The framework is then extended to account for droplets composed of multiple particle species and successfully applied to predict the freezing spectra of a mixed proxy for an atmospheric dust-biological particle system. The contact freezing mode of ice nucleation, whereby a particle induces freezing upon collision with a droplet, is thought to be more efficient than particle initiated immersion freezing from within the droplet bulk. However, it has been a decades’ long challenge to accurately measure this ice nucleation mode, since it necessitates reliably measuring the rate at which particles hit a droplet surface combined with direct determination of freezing onset. In an effort to remedy this longstanding deficiency a temperature controlled chilled aerosol optical tweezers capable of stably isolating water droplets in air at subzero temperatures has been designed and implemented. The new temperature controlled system retains the powerful capabilities of traditional aerosol optical tweezers: retrieval of a cavity enhanced Raman spectrum which could be used to accurately determine the size and refractive index of a trapped droplet. With these capabilities, it is estimated that the design can achieve ice supersaturation conditions at the droplet surface. It was also found that a KCl aqueous droplet simultaneously cooling and evaporating exhibited a significantly higher measured refractive index at its surface than when it was held at a steady state temperature. This implies the potential of a “salting out” process. Sensitivity of the cavity enhanced Raman spectrum as well as the visual image of a trapped droplet to dust particle collisions is shown, an important step in measuring collision frequencies of dust particles with a trapped droplet. These results may pave the way for future experiments of the exceptionally poorly understood contact freezing mode of ice nucleation.

Due to the unique environmental conditions and different feedback mechanisms, the Arctic region is especially sensitive to climate changes. The influence of clouds on the radiation budget is substantial, but difficult to quantify and parameterize in models. In the framework of the PhD, elastic backscatter and depolarization lidar observations of Arctic clouds were performed during the international Arctic Study of Tropospheric Aerosol, Clouds and Radiation (ASTAR) from Svalbard in March and April 2007. Clouds were probed above the inaccessible Arctic Ocean with a combination of airborne instruments: The Airborne Mobile Aerosol Lidar (AMALi) of the Alfred Wegener Institute for Polar and Marine Research provided information on the vertical and horizontal extent of clouds along the flight track, optical properties (backscatter coefficient), and cloud thermodynamic phase. From the data obtained by the spectral albedometer (University of Mainz), the cloud phase and cloud optical thickness was deduced. Furthermore, in situ observations with the Polar Nephelometer, Cloud Particle Imager and Forward Scattering Spectrometer Probe (Laboratoire de Météorologie Physique, France) provided information on the microphysical properties, cloud particle size and shape, concentration, extinction, liquid and ice water content. In the thesis, a data set of four flights is analyzed and interpreted. The lidar observations served to detect atmospheric structures of interest, which were then probed by in situ technique. With this method, an optically subvisible ice cloud was characterized by the ensemble of instruments (10 April 2007). Radiative transfer simulations based on the lidar, radiation and in situ measurements allowed the calculation of the cloud forcing, amounting to -0.4 W m-2. This slight surface cooling is negligible on a local scale. However, thin Arctic clouds have been reported more frequently in winter time, when the clouds' effect on longwave radiation (a surface warming of 2.8 W m-2) is not balanced by the reduced shortwave radiation (surface cooling). Boundary layer mixed-phase clouds were analyzed for two days (8 and 9 April 2007). The typical structure consisting of a predominantly liquid water layer on cloud top and ice crystals below were confirmed by all instruments. The lidar observations were compared to European Centre for Medium-Range Weather Forecasts (ECMWF) meteorological analyses. A change of air masses along the flight track was evidenced in the airborne data by a small completely glaciated cloud part within the mixed-phase cloud system. This indicates that the updraft necessary for the formation of new cloud droplets at cloud top is disturbed by the mixing processes. The measurements served to quantify the shortcomings of the ECMWF model to describe mixed-phase clouds. As the partitioning of cloud condensate into liquid and ice water is done by a diagnostic equation based on temperature, the cloud structures consisting of a liquid cloud top layer and ice below could not be reproduced correctly. A small amount of liquid water was calculated for the lowest (and warmest) part of the cloud only. Further, the liquid water content was underestimated by an order of magnitude compared to in situ observations. The airborne lidar observations of 9 April 2007 were compared to space borne lidar data on board of the satellite Cloud-Aerosol Lidar and Infrared Pathfinder Satellite Observations (CALIPSO). The systems agreed about the increase of cloud top height along the same flight track. However, during the time delay of 1 h between the lidar measurements, advection and cloud processing took place, and a detailed comparison of small-scale cloud structures was not possible. A double layer cloud at an altitude of 4 km was observed with lidar at the West coast in the direct vicinity of Svalbard (14 April 2007). The cloud system consisted of two geometrically thin liquid cloud layers (each 150 m thick) with ice below each layer. While the upper one was possibly formed by orographic lifting under the influence of westerly winds, or by the vertical wind shear shown by ECMWF analyses, the lower one might be the result of evaporating precipitation out of the upper layer. The existence of ice precipitation between the two layers supports the hypothesis that humidity released from evaporating precipitation was cooled and consequently condensed as it experienced the radiative cooling from the upper layer. In summary, a unique data set characterizing tropospheric Arctic clouds was collected with lidar, in situ and radiation instruments. The joint evaluation with meteorological analyses allowed a detailed insight in cloud properties, cloud evolution processes and radiative effects. / Die Arktis mit ihren speziellen Umweltbedingungen ist besonders empfindlich gegenüber Klimaveränderungen. Dabei spielen Wolken eine große Rolle im Strahlungsgleichgewicht, die aber nur schwer genau bestimmt und in Klimamodellen dargestellt werden kann. Die Daten für die Promotionsarbeit wurden im Frühjahr 2007 bei Flugzeug-Messungen von Wolken über dem Arktischen Ozean von Spitzbergen aus erhoben. Das dafür verwendete Lidar (Licht-Radar) des Alfred-Wegener-Instituts lieferte ein höhenaufgelöstes Bild der Wolkenstrukturen und ihrer Streu-Eigenschaften, andere Messgeräte ergänzten optische sowie mikrophysikalische Eigenschaften der Wolkenteilchen (Extinktion, Größenverteilung, Form, Konzentration, Flüssigwasser- und Eisgehalt, Messgeräte vom Laboratoire de Météorologie Physique, France) und Strahlungsmessungen (Uni Mainz). Während der Messkampagne herrschte Nordwind vor. Die untersuchten Luftmassen mit Ursprung fern von menschlichen Verschmutzungsquellen war daher sehr sauber. Beim Überströmen der kalten Luft über den offenen warmen Arktischen Ozean bildeten sich in der Grenzschicht (ca. 0-1500 m Höhe) Mischphasenwolken, die aus unterkühlten Wassertröpfchen im oberen Bereich und Eis im unteren Bereich der Wolken bestehen. Mit den Flugzeug-Messungen und numerischen Simulationen des Strahlungstransports wurde der Effekt einer dünnen Eiswolke auf den Strahlungshaushalt bestimmt. Die Wolke hatte lokal eine geringe Abkühlung der Erdoberfläche zur Folge. Ähnliche Wolken würden jedoch im Winter, wenn keine Sonnenstrahlung die Arktis erreicht, durch den Treibhauseffekt eine nicht vernachlässigbare Erwärmung der Oberfläche verursachen. Die Messungen der Mischphasenwolken wurden mit einem Wettervorhersagemodell (ECMWF) verglichen. Für die ständig neue Bildung von flüssigen Wassertropfen im oberen Teil der Wolke ist das Aufsteigen von feuchten Luftpaketen nötig. Während einer Messung wurden entlang der Flugstrecke verschiedene Luftmassen durchflogen. An der Luftmassengrenze wurde eine reine Eiswolke inmitten eines Mischphasen-Systems beobachtet. Die Messungen zeigen, dass das Mischen von Luftmassen den Nachschub an feuchter Luft blockiert, was unmittelbare Auswirkungen auf die thermodynamische Phase des Wolkenwassers hat. Weiterhin wurde bestimmt, wie groß die Abweichungen der Modellrechnungen von den Messungen bezüglich Wassergehalt und der Verteilung von Flüssigwasser und Eis waren. Durch die vereinfachte Wolken-Parameterisierung wurde die typische vertikale Struktur von Mischphasenwolken im Modell nicht wiedergegeben. Die flugzeuggetragenen Lidar-Messungen vom 9. April 2007 wurden mit Lidar-Messungen an Bord des Satelliten CALIPSO (Cloud-Aerosol Lidar and Infrared Pathfinder Satellite Observations) verglichen. Die Messungen zeigten beide eine ansteigende Wolkenobergrenze entlang desselben Flugwegs. Da die Messungen jedoch nicht genau gleichzeitig durchgeführt wurden, war wegen Advektion und Prozessen in den Wolken kein genauer Vergleich der kleinskaligen Wolkenstrukturen möglich. Außerdem wurde eine doppelte Wolkenschicht in der freien Troposphäre (4 km Höhe) analysiert. Die Wolke bestand aus zwei separaten dünnen Schichten aus flüssigem Wasser (je 150 m dick) mit jeweils Eis darunter. Die untere Schicht entstand wahrscheinlich aus verdunstetem Eis-Niederschlag. Diese feuchte Schicht wurde durch die Abstrahlung der oberen Wolkenschicht gekühlt, so dass sie wieder kondensierte. Solche Wolkenformationen sind in der Arktis bisher vor allem in der Grenzschicht bekannt. Ein einzigartiger Datensatz von arktischen Wolken wurde mit einer Kombination verschiedener Flugzeug-Messgeräte erhoben. Zusammen mit meteorologischen Analysen konnten für verschiedene Fallstudien Wolkeneigenschaften, Entwicklungsprozesse und Auswirkungen auf den Strahlungshaushalt bestimmt werden.

This thesis evaluates the limitations of passive airborne remote sensing methods to retrieve optical and microphysical properties of Arctic mixed-phase clouds. These limitations are circumvented using a synergy of passive and active remote sensing techniques, and large eddy simulations. Using this synergetic approach, the three-dimensional spatial distribution of the thermodynamic phase of two cloud case studies is characterized. The findings are subsequently applied to a statistical analysis of the cloud properties measured during the Arctic Cloud Observations Using airborne measurements during polar Day (ACLOUD) campaign.

Im Rahmen der Arbeit wurden Lidar- und Wolkenradarmessungen von troposphärischen Schichtwolken durchgeführt und ausgewertet, um den Zusammenhang zwischen Vertikalwinden und Eisbildung in diesen Wolken zu untersuchen. Der Eis- und Flüssigwassergehalt von Schichtwolken wurde mit einer Kombination aus Raman-Lidar und Wolkenradar untersucht. Die vertikalen Windbewegungen an der Wolkenunterkante wurden mit einem Doppler-Lidar aufgezeichnet. Durch die Auswertung vorangegangener Messkampagnen konnte die Vertikalwindstatistik in mittelhohen Schichtwolken zwischen den Standorten Leipzig und Praia (Kap Verde) verglichen werden. Messverfahren für die Vertikalwindmessung mit Doppler-Lidar wurden im Rahmen dieser Arbeit weiterentwickelt. In Zusammenarbeit mit dem Deutschen Wetterdienst wurde außerdem die Kombination von Doppler-Lidar, Wolkenradar und Wind-Profiler getestet. Die Eisbildungseffizienz in der Troposphäre wurde im Temperaturbereich zwischen 0 und -40°C für den Standort Leipzig untersucht und sowohl mit vorangegangenen Lidarmessungen, als auch mit aktuellen Satellitenmessungen verglichen. Zum ersten Mal wurde außerdem die statistische Verteilung von Vertikalwinden an der Basis von Mischphasenwolken dargestellt. Es wurde festgestellt, dass sich bei einer Temperatur von (-9 +/- 3)°C bereits in 50% der Schichtwolken über Leipzig Eis bildet. Zwischen -15 und 0°C wurden Verhältnisse zwischen Eis- und Flüssigwasserpfad zwischen 0,1 und 0,0001 abgeschätzt. Im Rahmen der Messgenauigkeit wurden zwischen den Standorten Leipzig und Praia keine Unterschiede in der Vertikalwindstatistik festgestellt.

In dieser Arbeit wird der auf maschinellem Lernen basierende Algorithmus zur Erkennung von unterkühlten Flüssigwasserschichten in Mischphasenwolken (MPCs) jenseits der Lidarattenuation VOODOO (reVealing supercOOled liquiD beyOnd lidar attenuatiOn) vorgestellt. Beobachtungen von zwei Langzeitmesskampagnen bilden die Datengrundlage. Für die nördliche Hemisphäre wurden die Daten mittels der mobilen bodengebundenen Fernerkundungsanlage des Leipziger Instituts für Meteorologie (LIM) in Leipzig (Deutschland, 51.3°N, 12. 4°E) erhoben. Für die südliche Hemisphäre wurde ein 9-monatiger Teil der dreijährigen Feldkampagne DACAPO-PESO (Dynamics Aerosol Clouds And Precipitation Observation in the Pristine Environment of the Southern Ocean; Beobachtung von Dynamik, Aerosol, Wolken und Niederschlag in der unverschmutzten Umgebung des Südozeans) aus Punta Arenas (Chile, 53.1°S, 70.9°W) verwendet. Dieser Datensatz wurde mittels des 94GHz Wolkenradars des LIM in Kombination mit LACROS (Leipzig Aerosol and Cloud Remote Observations System; Leipziger Aerosol- und Wolken-Fernerkundungssystem)- Instrumenten erhoben. Datensätze von vertikal ausgerichteten Doppler-Wolkenradaren und Lidarsystemen liefern Erkenntnisse über Wolkeneigenschaften mit hoher zeitlicher und räumlicher Auflösung. Die Identifizierung von flüssigen Wolkentröpfchen ist jedoch aufgrund der Abschwächung des Lidarsignals oft eingeschränkt. Wolkenradare sind andererseits in der Lage, mehrere Flüssigwassersschichten zu durchdringen, und können potenziell eingesetzt werden, um die Identifizierung der thermodynamischen Wolkenphase auf die gesamte vertikale Säule jenseits der Lidar-Signalabschwächungshöhe auszudehnen. Dazu werden morphologische Merkmale in Wolkenradar-Doppler-Spektren extrahiert, um auf das Vorhandensein von Flüssigwasser zu schließen. Das wichtigste Ergebnis dieser Arbeit ist die Implementierung des Open-Source-Retrievals VOODOO. Für beide Langzeitdatensätze zeigt die Wolken- und Flüssigwasseridentifikation von VOODOO hervorragende zeitliche und räumliche Übereinstimmungen mit der weit verbreitenten Cloudnet-Klassifizierung. Vergleiche des vom Mikrowellenradiometers (MWR) gemessenen Flüssigwasserpfads (LWP) mit modelliertem adiabatischen LWP zeigen jedoch die Überlegenheit von VOODOO in der Detektion hochreichender und mehrschichtiger MPCs im Vergleich zu Cloudnet. Der Einfluss von Turbulenzen auf die Vorhersageleistung von VOODOO wurde analysiert und als gering eingestuft. Darüber hinaus bestätigen weltraumgestützte Lidar-Beobachtungen die VOODOO-basierten Vorhersagen von Oberkanten von Flüssigwasserschichten für ausgewählte Satellitenüberflüge über Punta Arenas. Das Endresultat zeigt für ein Fallbeispiel die resultierende Reduktion der Fehler zwischen kurzwelliger solarer Strahlung am Erdboden, sowie des Strahlungseffekts von Wolken zwischen Beobachtungen und Strahlungstransfersimulationen um den Faktor 2, bei der Verwendung der VOODOO-Flüssigwasseridentifikationen.:i Mixed-Phase Clouds 1 Introduction 3 2 Remote-sensing of mixed-phase clouds 7 2.1 Definition, occurrence, and impact . . . . . . . . . . . . . . . . . . . . . . . 7 2.2 Instrumentation and operating principles . . . . . . . . . . . . . . . . . . . 9 2.2.1 Doppler cloud radar . . . . . . . . . . . . . . . . . . . . . . . . . . . 11 2.2.2 Lidar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15 2.2.3 Microwave radiometer . . . . . . . . . . . . . . . . . . . . . . . . . . 17 2.2.4 Numerical weather forecast model . . . . . . . . . . . . . . . . . . . 18 2.2.5 Additional data sources . . . . . . . . . . . . . . . . . . . . . . . . . 18 3 Datasets 19 3.1 Punta Arenas, Chile . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19 3.2 Leipzig, Germany . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21 ii Methodology 4 Identifying the thermodynamic phase of hydrometeors 25 4.1 Multisensor-based approaches . . . . . . . . . . . . . . . . . . . . . . . . . . 25 4.1.1 Cloudnet: Illingworth et al., 2007 . . . . . . . . . . . . . . . . . . . . 25 4.1.2 Shupe, 2007 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26 4.2 Radar-moment-based approaches . . . . . . . . . . . . . . . . . . . . . . . . 27 4.2.1 Silber et al., 2020 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27 4.2.2 Kalogeras et al., 2021 . . . . . . . . . . . . . . . . . . . . . . . . . . . 28 4.3 Doppler-Spectrum-based approaches . . . . . . . . . . . . . . . . . . . . . . 28 4.3.1 Yu et al., 2014 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28 4.3.2 PEAKO + peakTree: Kalesse et al., 2019; Radenz et al., 2019 . . . . 28 4.3.3 Luke et al., 2010 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29 4.4 Challenges in cloud-phase classification . . . . . . . . . . . . . . . . . . . . 30 5 Machine learning model 33 5.1 Mathematical basics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34 5.2 Pre-processing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35 5.3 Training and validation dataset . . . . . . . . . . . . . . . . . . . . . . . . . 36 5.4 Architecture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38 5.4.1 Convolution layer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39 5.4.2 Perceptron layer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40 5.4.3 Output layer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42 5.5 Training process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42 5.6 Post-processing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43 5.7 Evaluation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43 5.7.1 Confusion matrix and binary classification metrics . . . . . . . . . 43 5.7.2 Correlation with independent measurements . . . . . . . . . . . . . 45 5.7.3 Influence of LWP and turbulence on the performance . . . . . . . . 45 5.7.4 Probability density functions . . . . . . . . . . . . . . . . . . . . . . 46 5.7.5 Validation via space-borne lidar satellite CALIPSO . . . . . . . . . 46 5.7.6 Radiative closure study . . . . . . . . . . . . . . . . . . . . . . . . . 46 iii Results 6 Results 51 6.1 Training results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51 6.2 Case study: 1. August 2019, Punta Arenas, Chile . . . . . . . . . . . . . . . 53 6.3 Case study: 30 December 2020, Leipzig, Germany . . . . . . . . . . . . . . 57 6.4 Performance analysis for larger data sets . . . . . . . . . . . . . . . . . . . . 60 6.5 Probability density functions . . . . . . . . . . . . . . . . . . . . . . . . . . . 64 6.6 Case study for satellite-based cloud phase validation . . . . . . . . . . . . . 66 6.7 Radiative closure study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68 iv Outcome 7 Summary and Conclusion 77 8 Outlook 79 Publication record 83 List of Figures 85 List of Tables 88 List of Symbols 93 Bibliography 99 / This work presents a data driven retrieval algorithm for reVealing supercOOled liquiD beyOnd lidar attenuatiOn (VOODOO) in mixed-phase clouds (MPCs), which is based on deep convolutional neural networks (CNNs). Observations of two long-term field campaigns of mobile ground-based remote-sensing instrument deployments from both hemispheres are utilized. For the northern hemisphere, the data set was obtained by the mobile ground-based remote-sensing suite of the Leipzig Institute for Meteorology (LIM) in Leipzig (Germany, 51.3°N, 12.4°E) and for the southern hemisphere, 9-months of the three-year-long data set of the Dynamics Aerosol Clouds And Precipitation Observation in the Pristine Environment of the Southern Ocean (DACAPO-PESO) field campaign in Punta Arenas (Chile, 53.1°S, 70.9°W), collected by the supersite LACROS (Leipzig Aerosol and Cloud Remote Observations System). Data sets of vertically pointing Doppler cloud radars and lidars provide insights into cloud properties at high temporal and spatial resolution. However, the identification of liquid cloud droplets is often limited due to the attenuation of the lidar signal. On the contrary, cloud radars are able to penetrate multiple liquid layers and can potentially be used to expand the identification of cloud phase to the entire vertical column beyond the lidar signal attenuation height, by exploiting morphological features in cloud radar Doppler spectra that relate to the existence of supercooled liquid. The most important result of this work is the the open-source implementation of the VOODOO retrieval, predicting excellent temporal and spatial agreement in cloud-droplet bearing pixels detected by the widely-used Cloudnet atmospheric target classification. Comparisons of the liquid water path (LWP) measured by the microwave radiometer (MWR), with modeled adiabatic LWP show the superiority of VOODOO in detecting liquid in deep and multilayer MPCs compared to Cloudnet. The influence of turbulence on VOODOO’s predictive performance was analyzed and found to be minor. Additionally, space-borne lidar observations confirm liquid cloud top predictions of MPCs tops for selected satellite overpasses over Punta Arenas, Chile. The final results of this work is the demonstration of the ability to reduce the shortwave downward radiation bias and the bias in cloud radiative effect between ground-based pyranometer observations and radiative transfer simulations by a factor of 2 for a case study. This highlights the fact that from a measurement perspective, advanced cloud thermodynamic phase retrievals based on sophisticated remote-sensing observations can be a way to reduce the Southern Ocean radiation bias.:i Mixed-Phase Clouds 1 Introduction 3 2 Remote-sensing of mixed-phase clouds 7 2.1 Definition, occurrence, and impact . . . . . . . . . . . . . . . . . . . . . . . 7 2.2 Instrumentation and operating principles . . . . . . . . . . . . . . . . . . . 9 2.2.1 Doppler cloud radar . . . . . . . . . . . . . . . . . . . . . . . . . . . 11 2.2.2 Lidar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15 2.2.3 Microwave radiometer . . . . . . . . . . . . . . . . . . . . . . . . . . 17 2.2.4 Numerical weather forecast model . . . . . . . . . . . . . . . . . . . 18 2.2.5 Additional data sources . . . . . . . . . . . . . . . . . . . . . . . . . 18 3 Datasets 19 3.1 Punta Arenas, Chile . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19 3.2 Leipzig, Germany . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21 ii Methodology 4 Identifying the thermodynamic phase of hydrometeors 25 4.1 Multisensor-based approaches . . . . . . . . . . . . . . . . . . . . . . . . . . 25 4.1.1 Cloudnet: Illingworth et al., 2007 . . . . . . . . . . . . . . . . . . . . 25 4.1.2 Shupe, 2007 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26 4.2 Radar-moment-based approaches . . . . . . . . . . . . . . . . . . . . . . . . 27 4.2.1 Silber et al., 2020 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27 4.2.2 Kalogeras et al., 2021 . . . . . . . . . . . . . . . . . . . . . . . . . . . 28 4.3 Doppler-Spectrum-based approaches . . . . . . . . . . . . . . . . . . . . . . 28 4.3.1 Yu et al., 2014 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28 4.3.2 PEAKO + peakTree: Kalesse et al., 2019; Radenz et al., 2019 . . . . 28 4.3.3 Luke et al., 2010 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29 4.4 Challenges in cloud-phase classification . . . . . . . . . . . . . . . . . . . . 30 5 Machine learning model 33 5.1 Mathematical basics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34 5.2 Pre-processing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35 5.3 Training and validation dataset . . . . . . . . . . . . . . . . . . . . . . . . . 36 5.4 Architecture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38 5.4.1 Convolution layer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39 5.4.2 Perceptron layer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40 5.4.3 Output layer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42 5.5 Training process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42 5.6 Post-processing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43 5.7 Evaluation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43 5.7.1 Confusion matrix and binary classification metrics . . . . . . . . . 43 5.7.2 Correlation with independent measurements . . . . . . . . . . . . . 45 5.7.3 Influence of LWP and turbulence on the performance . . . . . . . . 45 5.7.4 Probability density functions . . . . . . . . . . . . . . . . . . . . . . 46 5.7.5 Validation via space-borne lidar satellite CALIPSO . . . . . . . . . 46 5.7.6 Radiative closure study . . . . . . . . . . . . . . . . . . . . . . . . . 46 iii Results 6 Results 51 6.1 Training results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51 6.2 Case study: 1. August 2019, Punta Arenas, Chile . . . . . . . . . . . . . . . 53 6.3 Case study: 30 December 2020, Leipzig, Germany . . . . . . . . . . . . . . 57 6.4 Performance analysis for larger data sets . . . . . . . . . . . . . . . . . . . . 60 6.5 Probability density functions . . . . . . . . . . . . . . . . . . . . . . . . . . . 64 6.6 Case study for satellite-based cloud phase validation . . . . . . . . . . . . . 66 6.7 Radiative closure study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68 iv Outcome 7 Summary and Conclusion 77 8 Outlook 79 Publication record 83 List of Figures 85 List of Tables 88 List of Symbols 93 Bibliography 99

Im Rahmen der Arbeit wurden Lidar- und Wolkenradarmessungen von troposphärischen Schichtwolken durchgeführt und ausgewertet, um den Zusammenhang zwischen Vertikalwinden und Eisbildung in diesen Wolken zu untersuchen. Der Eis- und Flüssigwassergehalt von Schichtwolken wurde mit einer Kombination aus Raman-Lidar und Wolkenradar untersucht. Die vertikalen Windbewegungen an der Wolkenunterkante wurden mit einem Doppler-Lidar aufgezeichnet. Durch die Auswertung vorangegangener Messkampagnen konnte die Vertikalwindstatistik in mittelhohen Schichtwolken zwischen den Standorten Leipzig und Praia (Kap Verde) verglichen werden. Messverfahren für die Vertikalwindmessung mit Doppler-Lidar wurden im Rahmen dieser Arbeit weiterentwickelt. In Zusammenarbeit mit dem Deutschen Wetterdienst wurde außerdem die Kombination von Doppler-Lidar, Wolkenradar und Wind-Profiler getestet. Die Eisbildungseffizienz in der Troposphäre wurde im Temperaturbereich zwischen 0 und -40°C für den Standort Leipzig untersucht und sowohl mit vorangegangenen Lidarmessungen, als auch mit aktuellen Satellitenmessungen verglichen. Zum ersten Mal wurde außerdem die statistische Verteilung von Vertikalwinden an der Basis von Mischphasenwolken dargestellt. Es wurde festgestellt, dass sich bei einer Temperatur von (-9 +/- 3)°C bereits in 50% der Schichtwolken über Leipzig Eis bildet. Zwischen -15 und 0°C wurden Verhältnisse zwischen Eis- und Flüssigwasserpfad zwischen 0,1 und 0,0001 abgeschätzt. Im Rahmen der Messgenauigkeit wurden zwischen den Standorten Leipzig und Praia keine Unterschiede in der Vertikalwindstatistik festgestellt.