Microscale Distribution of Environmental Proxies in Snow

Research Collection

Doctoral Thesis

Author(s): Trachsel, Jürg

Publication Date: 2019

Permanent Link: https://doi.org/10.3929/ethz-b-000387457

Rights / License: Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International

This page was generated automatically upon download from the ETH Zurich Research Collection. For more information please consult the Terms of use.

ETH Library DISS. ETH NO. 26360

Microscale Distribution of Environmental Proxies in Snow

A thesis submitted to attain the degree of

DOCTOR OF SCIENCES of ETH Zurich

(Dr. sc. ETH Zurich)

presented by

JÜRG CHRISTIAN TRACHSEL

MSc ETH in Civil Engineering

born on 06 January 1986

citizen of Brugg AG and Wattenwil BE

accepted on the recommendation of

Prof. Dr. Konrad Steffen Dr. Martin Schneebeli Dr. Franziska Scholder-Aemisegger Dr. Ulrich Krieger

2019

“People think focus means saying yes to the thing you've got to focus on. But that's not what it means at all. It means saying no to the hundred other good ideas that there are. You have to pick carefully.” Steve Jobs, 1997

Contents

Zusammenfassung...... v

Summary ...... ix

1 Introduction ...... 1

1.1 Project Overview ...... 2 1.2 Snow – an environmental proxy ...... 2 1.3 Instrument development and experimental ...... 4 1.3.1 Droplet Chopper (PSI) – Artificial snow ...... 4 1.3.2 Metamorphism chamber (SLF) – “Aging” snow ...... 5 1.3.3 Micro-computer tomography scanner (SLF) – Snow structure analysis ...... 6 1.3.4 Elution experiments (SLF) – Investigation of ion (re)distribution in the snow ...... 7 1.3.5 Ion chromatography (IC) and plasma mass spectrometry (ICP-SF-MS) (PSI) ...... 7 1.3.6 Picarro laser spectrometry (Weissfluhjoch) – Water vapor isotope measurement ...... 8 1.3.7 Sampling snow pit – Collecting impurity records ...... 9 1.4 Study site ...... 10 1.5 Goals and Outline ...... 11 References ...... 13 2 Microscale rearrangement of ammonium induced by snow metamorphism ...... 19

Abstract ...... 20 2.1 Introduction ...... 20 2.2 Methods ...... 23 2.2.1 Preparation of artificial and natural snow samples ...... 23 2.2.1.1 Preparation of artificial snow ...... 23 2.2.1.2 Sampling of natural snow at WFJ ...... 24 2.2.2 Laboratory-controlled and natural snow metamorphism ...... 24 2.2.2.1 Laboratory-controlled metamorphism experiment ...... 24 2.2.2.2 Natural metamorphism: Samples of natural snow from a WFJ snow pit ...... 26 2.2.3 Elution Experiments ...... 26 2.2.4 Vertical distribution of ions after temperature gradient storage...... 27 2.2.5 Ion chromatography (IC) ...... 28 2.2.6 X-ray microtomography (microCT) ...... 28

Contents

2.3 Results ...... 28 2.3.1 Structural development during metamorphism ...... 28 2.3.1.1 Temperature gradient metamorphism ...... 28 2.3.1.2 Isothermal metamorphism ...... 31 + 2.3.2 Redistribution of NH4 and other major ions during snow metamorphism ...... 31 2.3.2.1 Elution experiments ...... 31 2.3.2.2 Temporal changes of ion concentrations in first eluate and residual snow ...... 33 2.3.2.3 Temporal changes in ion concentrations under isothermal conditions...... 34 + 2.3.2.4 Enrichment of NH4 in the residual sample (gradient/isothermal) ...... 34 2.3.2.5 Vertical distribution of ions after temperature gradient storage ...... 35 2.4 Discussion ...... 35 2.4.1 Structural change under temperature gradient metamorphism ...... 36 2.4.2 Structural changes under isothermal conditions ...... 37 + 2.4.3 Rearrangement of NH4 and other major ions during snow metamorphism ...... 37 2.4.3.1 Initial chemical characterization of the snow ...... 38 2.4.3.2 First eluate vs residual sample: What we learn with respect to accessibility? ..... 39 2.4.3.3 Impact of structural changes during metamorphism on distribution of ions ...... 41 2.4.3.4 Environmental implications ...... 45 2.5 Conclusion ...... 46 Acknowledgements ...... 47 References ...... 48 3 Melt-induced fractionation of major ions and trace elements in an Alpine snowpack ...... 57

Abstract ...... 58 3.1 Introduction ...... 58 3.2 Materials and Methods ...... 60 3.2.1 Study site and meteorological setting ...... 60 3.2.2 Snow pit sampling ...... 61 3.2.3 Major ion, water stable isotope and trace element analysis ...... 62 3.2.4 Data Evaluation ...... 62 3.3 Results and Discussion ...... 63 3.3.1 Major ions (MIs) ...... 63 3.3.1.1 Comparison of the five MI concentration profiles ...... 63 3.3.1.2 Preferential elution of MIs: Elution sequence and discussion ...... 64 3.3.2 Trace elements (TEs) ...... 67 3.3.2.1 TE concentration profiles ...... 67 3.3.2.2 Different preservation of TEs under melt conditions ...... 69 3.4 Conclusion ...... 72

Contents

Acknowledgements ...... 73 References ...... 74 4 Recording frontal passages with the ẟ18O of Alpine snow ...... 81

Abstract ...... 82 4.1 Introduction ...... 83 4.2 Experimental methods ...... 87 4.2.1 Overview of measurements ...... 87 4.2.2 Stable water isotope measurements ...... 88 4.2.2.1 Snow sampling ...... 88 4.2.2.2 Analysis of δD and δ18O from snow samples in the lab...... 89 4.2.2.3 Measuring the stable water isotope composition of the vapor phase ...... 89 4.3 Winter evolution and association of precipitation events ...... 90 4.3.1 Evolution of the seasonal snow cover ...... 90 4.3.2 Weather system-based classification of precipitation events ...... 91 4.3.3 Snowpack model for associating snow layers to precipitation events ...... 92 4.4 Results and Discussion ...... 93 4.4.1 Physical properties of the snow cover ...... 93 4.4.2 Five profiles of ẟ18O in snow ...... 95 4.4.3 Variability in air and snow surface δ18O signal ...... 97 4.4.4 Archival of frontal passages in the seasonal snow cover ...... 99 4.4.4.1 Six examples of cold and warm precipitation events ...... 100 4.4.4.2 Event analysis winter storm Egon: 11-17 January 2017 ...... 103 4.4.4.3 Event analysis of a warm front on 18-19 March 2017 ...... 104 In the association of the different precipitation events ...... 104 4.4.5 Post-depositional Processes ...... 107 4.4.5.1 Preservation of the snow δ18O during winter despite dry metamorphism ...... 107 4.4.5.2 Smoothing and shifting of the δ18O record during spring ...... 108 4.5 Conclusion and Outlook ...... 111 Acknowledgements ...... 112 Appendix Chapter 4 ...... 113 References ...... 114 5 Conclusions and Outlook ...... 121

Appendix A White and wonderful? Microplastics prevail in snow from the Alps to the Arctic ... 131

Abstract ...... 132 A.1 Introduction ...... 132 A.2 Results ...... 133 A.2.1 Fiber and microplastic quantities at different locations ...... 133

iii

Contents

A.2.2 Size of microplastics and fibers ...... 135 A.2.3 Material composition...... 136 A.2.4 Other particles ...... 136 A.3 Discussion ...... 138 A.3.1 Abundance of microplastic particles and microfibers ...... 138 A.3.2 Comparison with microplastics levels in Arctic sea ice and deep-sea sediments ...... 140 A.3.3 Microplastics size ...... 140 A.3.4 Abundance of fibers ...... 141 A.3.5 Polymer composition ...... 141 A.3.6 Health implications ...... 143 A.4 Material and Methods ...... 144 A.4.1 Study sites ...... 144 A.4.2 Contamination prevention and procedural blanks ...... 146 A.4.3 Analytical procedure for the detection of (microplastic) plastics ...... 146 A.4.4 Detection of fibers ...... 147 A.4.5 Data analysis ...... 147 A.5 Supplementary materials ...... 148 Acknowledgements ...... 151 References ...... 152 Appendix B Collecting snow samples of environmental proxies – Best practice ...... 157

B.1 Introduction ...... 158 B.2 Sampling ...... 159 B.2.1 Basics...... 159 B.2.2 Materials ...... 159 B.2.2.1 Basic equipment ...... 159 B.2.2.2 Sampling containers ...... 159 B.2.3 Choice of field site ...... 160 B.2.4 Snow pit preparation ...... 161 B.2.5 Sample collection ...... 161 B.3 Conclusion ...... 163 Acknowledgements ...... 167

Zusammenfassung

Während eines Schneefalls lagern sich chemische Verbindungen aus der Atmosphäre auf der Erdoberfläche ab. Wenn der Schnee nicht vollständig schmilzt, wie dies auf polaren und alpinen Gletschern der Fall ist, bleiben die verunreinigten Lufteinschlüsse in der Schneedecke erhalten. Durch nachfolgende Schneefälle verdichtet, wandelt sich der Schnee nach und nach zu Firn und später zu Eis. Auf diese Weise im Gletschereis eingeschlossene Ablagerungen bilden unschätzbar wichtige Archive vergangener Klimata und der ursprünglichen Atmosphärenzusammensetzung. Durch Eisbohrkerne können diese Archive wieder an die Oberfläche geholt und analysiert werden. Um aber die darin enthaltenen Informationen zu erschliessen, müssen die Prozesse, die zur Einbettung geführt hatten, verstanden werden. Insbesondere die Rekristallisation der Schneedecke durch die Schneemetamorphose kann zu einer Umverteilung der eingebetteten Verbindungen führen. Weil Schnee an der Erdoberfläche physikalisch betrachtet relativ nahe am Schmelzpunkt liegt, ergibt sich ein hoher Dampfdruck. Analog zu glühendem Baustahl, der hochgradig duktil ist und seine Form laufend umwandelt, ändert sich auch die Schneestruktur. Der hohe Dampfdruck führt dazu, dass sich in der Schneedecke ständig Wassermoleküle aus Schneekristallen lösen (Sublimation) und sich an benachbarten Flächen anlagern (Re-Sublimation). Dieser Strukturumbau kann auch einen Einfluss auf die darin enthaltenen Fremdstoffe haben. Da die Umlagerung verschiedene Phasenübergänge beinhaltet und es zusätzlich zu Wechselwirkungen zwischen den enthaltenen Verbindungen kommen kann, drängt sich eine nähere Betrachtung der dabei ablaufenden Prozesse auf, im Kontext, dass die im Eis vorgefundene Zusammensetzung nicht mehr der ursprünglich durch die Schneekristalle eingebrachten entsprechen könnte. In dieser Arbeit untersuchten wir den Einfluss der Schneemetamorphosen auf die Umverteilung von Ionen, Spurenelementen und stabilen Wasserisotopen in der Schneedecke. Allesamt stellen sie wichtige Umweltindikatoren (Proxys) dar und spielen bei der Interpretation der Eisbohrkerne eine wichtige Rolle. Unsere erste Studie untersuchte mit einer Reihe von Auswasch-Experimenten (Elutionen) die

+ 2+ - - + 2- Umlagerung von Ammonium (NH4 ) und fünf weiteren Ionen (Ca , Cl , F , Na und SO4 ) während der Metamorphose. Ammonium in Eiskernen ist ein wichtiger Proxy für Waldbrände sowie für biogene und anthropogene Emissionen. Es bestimmt auch das Säure-Basen-Gleichgewicht und die Pufferkapazität im Schnee. Für die Elutionen wurden Schneeproben aus künstlichem und natürlichem Schnee im Labor während bis zu drei Monaten einer kontrollierten Metamorphose ausgesetzt. Mit fortschreitender

Zusammenfassung

Lagerungsdauer wurde deutlich, dass sich der Strukturumbau unterschiedlich auswirkt: Ionen, welche

+ - - über eine hohe Löslichkeit im Eis verfügen (NH4 , F und Cl ), wurden tiefer in die Eisstruktur der Schneepartikel eingebaut. Sie konnten bei den Elutionen immer weniger ausgewaschen werden. Ionen

2+ 2- + mit geringerer Löslichkeit (Ca , SO4 und Na ) hingegen wurden aus der Eis-Struktur ausgelagert und waren damit dem Elutionswasser stärker ausgesetzt. Diese Erkenntnis aus dem Labor auf die Natur

+ übertragen bedeutet, dass NH4 bei zunehmender Alterung einer Schneedecke möglichen Schmelzeinflüssen besser entzogen ist. Gleichzeitig ist es für chemische Reaktionen an der Oberfläche weniger verfügbar. Um Schmelzeinflüsse ging es auch in unserer zweiten Studie, in der wir den Erhalt von Spurenelementen im Schnee während der Metamorphose (Winter) und der Schneeschmelze (Frühling) untersuchten. Spurenelemente stellen wie die Ionen wichtige Umweltindikatoren dar. Anhand von Spurenelement-Analysen aus Eisbohrkernen konnte beispielsweise schon früher gezeigt werden, wie sich die Zusammensetzung der Atmosphäre mit Beginn der Industrialisierung verändert hat und dass somit ein direkter Zusammenhang zwischen Luftverschmutzung und menschlichen Aktivitäten besteht. Im Winter 2017 gruben wir im Abstand von je einem Monat auf dem Versuchsfeld Weissfluhjoch oberhalb Davos (2536 m ü.M., 46°49′47″ N 9°48′33″ E) fünf Schneeprofile und analysierten die im Schnee enthaltenen Spurenelemente. Der Vergleich der Profile zeigte, dass abgelagerte Elemente während des Hochwinters stabil in der Schneedecke eingeschlossen sind. Mit Einsetzen der Schneeschmelze kommt es zu einem Auswaschen der Elemente. Dabei beobachten wir unterschiedliche Konzentrationsänderungen: Während einige Elemente stark reduziert werden, scheinen andere resistent gegen Schmelzwassereinflüsse zu sein. Diese präferentielle Verminderung ist ein wichtiger Faktor im Hinblick auf die Interpretation von Eisbohrkernen, deren Eis Schmelzperioden ausgesetzt war. In unserer dritten Studie untersuchten wir anhand viermonatiger Wasserdampf- Isotopenmessungen in Kombination mit Schneeproben, wie das atmosphärische Isotopen-Signal in der Schneedecke abgelagert, respektive verändert wird. Wasser-Isotope in Eisbohrkernen können für die

18 Temperaturrekonstruktion verwendet werden, weil sich das Verhältnis von schwereren (H2O ) zu

16 leichteren Wasserisotopen (H2O ) je nach Atmosphärentemperatur verändert. Das im Schnee abgelagerte Signal ist eine komplexe Funktion des atmosphärischen Transports, der Wolkenbildung, des Niederschlags sowie der Veränderungen, die nach der Ablagerung auftreten - u.a. durch die Metamorphose. Unsere Isotopen-Untersuchungen fand zwischen Januar und Juni 2017 auf dem Weissfluhjoch statt. Auch hier wurden in monatlichen Abständen fünf Schneeprofile gegraben und die Proben analysiert. Das in der Schneedecke abgelagerte Isotopensignal konnte nachträglich einzelnen Niederschlagsereignissen zugeordnet werden. Die Klassifizierung der Ereignisse in Niederschläge aus warmen oder kalten Luftmassen zeigte eine klare Charakteristik: Die Schneeschichten, die durch

Zusammenfassung

18 Schneefall aus warmen Luftmassen entstanden, sind angereicherter (höherer Anteil H2O ) als die Schneeschichten, die bei Niederschlag aus kalten Luftmassen abgelagert wurden. Der Vergleich der fünf Schneeprofile zeigte weiter, dass das abgelagerte Isotopensignal während der kalten Wintermonate (Januar bis März) sehr stabil erhalten bleibt. Es war eine geringfügige Glättung des Signals zu erkennen, welche auf die Schneemetamorphose zurückgeführt werden konnte. Das Abschmelzen im Frühsommer (Juni) verursachte dann eine deutliche Verschiebung des ursprünglichen Signals. Unsere Ergebnisse zeigen, dass die Isotopen-Signatur der Schneedecke eine solide Aufzeichnung des über die Wintersaison herrschenden Wetters bietet, jedoch nur solange kein Schmelzwasser vorhanden ist. Insgesamt nimmt die Schneemetamorphose in allen drei Studien eine zentrale Rolle ein. Als wesentlicher Einflussfaktor erwies sich dabei das Temperatur-Regime, das eine grundsätzliche Unterteilung in kalte (trockene) und warme (feuchte) Schneeverhältnisse zulässt: Bei Kälte hatte die Rekristallisation der Schneedecke keinen messbaren Einfluss auf den Erhalt der untersuchten Ionen und Spurenelementen. Nur bei den stabilen Wasserisotopen war eine leichte Glättung des Signals durch die trockene Metamorphose erkennbar. Unter warmen Bedingungen wurde durch das vorhandene Schmelzwasser jedoch ein präferentieller Verlust von Ionen und Spurenelementen beobachtet und die stabilen Wasserisotope zeigten eine wesentliche Verschiebung des ursprünglichen Signals. Im Hinblick auf die globale Erwärmung und den Anstieg der Nullgradgrenze sind dies wichtige Erkenntnisse. In Zukunft wird auch höher gelegenes Gletschereis zunehmend durch Schmelzwasser beeinflusst sein und das Wissen über Umweltindikatoren, die nicht oder weniger auf Schmelzwasserkontakt reagieren, gewinnt an Bedeutung.

vii

Summary

During a snowfall, chemical compounds from the atmosphere are deposited on the ground. If the snow does not melt completely, as is the case on polar and alpine glaciers, these impurities remain inside the snow cover. Compacted by subsequent snowfalls, the snow gradually turns into firn and later into ice. Deposits thus trapped in the glacier ice form invaluable archives of past climates and original atmospheric composition. These archives can be brought back to the surface through ice cores for later analysis. However, in order to make the information contained therein comprehensible, the processes that led to the embedding must be understood. In particular, the recrystallization of the snow cover by snow metamorphism can lead to a redistribution or a loss of the embedded compounds. Because snow on the Earth's surface is physically relatively close to its melting point, there is a high vapor pressure. Similar to red-hot structural steel, which is highly ductile and continuously transforms its shape, the snow structure also changes. Inside the snow cover, the high vapor pressure causes water molecules of the snow crystals to be released constantly (sublimation) and to be deposited on adjacent ice surfaces (re-sublimation). This structural transformation can also have an influence on the contaminants present and can cause the movement of chemical compounds to different positions in the snow cover. Thus, a closer examination of the processes taking place during this rearrangement is needed, because the chemical composition found in the ice might no longer correspond to that originally deposited by the snow crystals, leading to errors when estimating past atmospheric conditions from ice cores. In this thesis we investigated the influence of snow metamorphism on the (re)distribution of ions, trace elements and stable water isotopes in the snow cover. They all represent important environmental indicators of past atmospheric conditions and play an important role in the interpretation of ice cores.

+ 2+ Our first study investigated the rearrangement of ammonium (NH4 ) and five other ions (Ca ,

- - + 2- Cl , F , Na und SO4 ) during metamorphism with a series of elution experiments. Ammonium in ice cores is an important proxy for forest fires and for biogenic and anthropogenic emissions. It also determines the acid-base balance and the buffer capacity in snow. For the elutions, snow samples from artificial and natural snow were exposed to controlled metamorphism in the laboratory for up to three months. As the storage period progressed, the effects of the structural rearrangement varied: Ions

+ - - with a high solubility in the ice (NH4 , F und Cl ) were incorporated deeper into the ice structure of the

Summary snow particles. They became less and less washed out during the elutions. Ions with lower solubility

2+ 2- + (Ca , SO4 und Na ), on the other hand, were excluded from the ice lattice and were therefore more

+ exposed to the elution water. This finding implies that NH4 is better protected from possible melt influences as the natural snow cover is ageing. At the same time, it is less available for chemical reactions at the surface. Melt effects of snow were also the subject of our second study, in which we investigated the preservation of trace elements in snow during metamorphism (winter) and snow melting (spring). Like ions, trace elements are important environmental proxies. For example, trace element analyses from ice cores have shown how the composition of the atmosphere changed with the beginning of industrialization and, thus, that there is a direct relationship between air pollution and human activities. In winter 2017, we dug five snow profiles on the high alpine snow field Weissfluhjoch in the Swiss Alps above Davos (2536 m a.s.ls, 46°49′47″ N 9°48′33″ E) at monthly intervals and analyzed the trace elements contained in the snow. The comparison of the profiles showed that deposited elements are stably enclosed in the snow cover during high winter. As the snow melts, certain elements are increasingly washed out, while others are only slightly reduced. The latter appear to be more resistant to melt water fluctuation. This preferential reduction is an important factor in the interpretation of ice cores, whose ice has been exposed to melting periods. In our third study, we investigated how the atmospheric isotope signal is deposited in the snow cover by four-month water vapor isotope measurements in combination with snow sampling. Water

18 vapor isotopes in ice cores can be used for temperature reconstruction. The ratio of heavier (H2O ) to

16 lighter (H2O ) water isotopes changes depending on atmospheric temperature. The signal deposited in snow is a complex function of atmospheric transport, cloud formation, precipitation and changes occurring after deposition, including metamorphism. Our isotope measurement study took place between January and June 2017 on the Weissfluhjoch. Similar to the previous study, we dug five snow profiles at monthly intervals. We assigned the deposited isotope signal to individual precipitation events and monitored the vertical isotope profiles inside in the snow cover. The classification of the events into precipitation from warm or cold air masses showed clear characteristics: The snow layers

18 formed by snowfall from warm air masses were more enriched (higher proportion H2O ) than the snow layers deposited by cold air masses. The comparison of the five snow profiles further showed that the deposited isotope signature remains very stable during the cold winter months (January to March). Only a slight smoothing of the signal was observed, which could be attributed to the snow metamorphism. The melting in early summer (June) caused a significant shift of the original signal. Our results show that the isotope signature of the snowpack provides a solid record of the weather throughout the winter season, but only as long as there is no meltwater percolation.

Summary

Overall, snow metamorphism was the primary process affecting the preservation of ions and trace elements in all three studies. This direct influence of snow metamorphism can be divided into either cold (dry) or warm (wet) snow conditions. Under cold conditions, the recrystallization of the snowpack did not affect the preservation of the ions or trace elements investigated but led to a slight smoothing of up to 2‰ in the stable water isotopes. Under warm conditions, flushing of meltwater caused a preferential loss of major ions and trace elements. Additionally, stable water isotopes showed a substantial shift from the original signal due to the presence of melt water. These findings have important implications within the context of interpreting glacial climate records under a changing climate. An increase in elevation of the 0°C isotherm will increase the influence of meltwater on glacial ice and firn thus altering the behaviors of major ions and trace elements. Knowledge of the behavior of these environmental proxies in meltwater will therefore become increasingly relevant for interpretation of glacial ice core records.

1 Introduction

This chapter gives an overview on the archival properties of snow in the cryosphere. Further, the various instruments used in experiments in this thesis are introduced. Finally, the objectives and the outline of this thesis are described.

1.1 Project Overview

1.1 Project Overview This thesis is part of the interdisciplinary project "Microscale Distribution of Impurities in Snow and Glacier Ice" (MiSo). The objective of the project is to investigate the fate of environmental proxies in alpine snow. We define environmental proxies as major ions, trace elements, isotopes and microplastics. Of special interest was the question of how the dynamics of snow impact chemical reactions that take place in snow and how this affects the preservation of environmental chemical compounds that are used as archives to reconstruct atmospheric conditions in the distant past. To bring the different research fields together, the project was conducted in a close collaboration between the WSL Institute for Snow and Avalanche Research (SLF) and the Paul Scherrer Institute (PSI). Three theses were written as part of the MiSo project. The first thesis focused on the behavior of rare environmental chemicals (trace elements) during melting and their potential as reconstruction proxies in Alpine snow and melt-aﬀected glacier ice (Avak, 2019). Further, an analytical method for the direct in situ analysis of trace elements in glacier ice was developed. The second thesis was dedicated to the reaction of ozone with bromide, particularly with respect to the interfacial process and the temperature dependence of this reaction (Edebeli, 2018). In addition, the partitioning of bromide between the surface and the bulk of the ice during snow metamorphism was investigated. This thesis now focuses on the influence of snow metamorphism on the distribution of major ions, trace elements, stable water isotopes and microplastics embedded in the snowpack. The rest of this chapter is structured as follows: Section 1.2 gives a wide overview of international research in snow and environmental chemistry. The experimental work and the instruments developed and used for this thesis are summarized in Section 1.3. In section 1.4 the study site is introduced, where the field experiments took place. Finally, Section 1.5 provides information on the research questions and the subject of the publications written over the course of this thesis.

1.2 Snow – an environmental proxy Snow and ice cover large parts of the Earth's surface and thus strongly influence the climate and life on Earth. This cover can be either seasonal if its melts completely during summer (as snow in the Alps), or permanent (Antarctic and Greenland inland ice). Together with other areas in which water is frozen (glaciers, sea ice and permafrost regions), snow forms the cryosphere. The cryosphere is an important water resource, modulates Earth’s energy balance through its high albedo, and influences global atmospheric circulation (Anderson and Neff, 2008; Dominé and Shepson, 2002; Grannas et al., 2013). The cryosphere also serves as an archive of the Earth's historical atmospheric composition and past climate (Alley, 2010; Langway, 2008). Chemical compounds in the atmosphere are deposited on the

1.2 Snow – an environmental proxy ground during snowfall (Heintzenberg and Rummukainen, 1993). These deposits may contain both locally produced components and components advected by synoptic scale air masses (Barrie and Hoff, 1985; Cooper and Alley, 2010; Grannas et al., 2013). In cold areas, where the snow does not melt, the components in the snow are compacted into firn and later into ice. Embedded in glacial ice or ice shields, the atmospheric composition can be preserved for millennia (Jouzel et al., 2007; Wolff et al., 2010). Various types of impurities found globally in snow and ice can be used as tracers. Major ions (defined as elements whose concentration is greater than 1 ppm in seawater) provide information about past changes in anthropogenic pollution, atmospheric transport, forest fires, and atmospheric temperature (Eichler et al., 2011; Kaspari et al., 2007; Kellerhals et al., 2010; Legrand and Mayewski, 1997). Concentrations of trace elements can give information about past volcanic activity (Gabrielli et al., 2014; Kaspari et al., 2009; Kellerhals et al., 2010), origin and variations of atmospheric mineral dust (Kaspari et al., 2009) and about the development of atmospheric anthropogenic pollution (Barbante et al., 2004; Beaudon et al., 2017; Correia et al., 2003; Eichler et al., 2017; Eyrikh et al., 2017; Schwikowski et al., 2004; Sierra-Hernández et al., 2018; Uglietti et al., 2015). Also, modern microplastics pollution can be detected globally in snow (Bergmann et al., 2019). Stable hydrogen isotopes represent another important parameter for the reconstruction of past climate changes (Craig, 1961; Cuffey et al., 1995; Dansgaard, 1964; Johnsen et al., 1995; Masson-Delmotte et al., 2008). The

16 16 18 stable water isotopic composition of precipitation (H2O , HDO , H2O ) is related to the air temperature (Dansgaard, 1953; Epstein and Mayeda, 1953) and has a long tradition of being used as a paleoclimate proxy. All climatological reconstructions derived from ice cores are based on the implicit assumption that the original composition in new snow is transferred into ice and remains there unchanged. However, this is not always the case. After deposition, snow is subject to dynamic ageing. The local rearrangements during dry metamorphism completely renew and thereby transform the structure of the original ice crystals. Pinzer et al. (2012) showed a characteristic lifetime of 2-3 days for an ice volume at a gradient of 50 K/m, which is typical for high alpine snowpacks. It seems likely that this highly dynamic reconstruction of the ice framework has an influence on the embedded impurities. Recent studies suggest that the vapor flux inside a snowpack causes isotopic fractionation and therefore alters the original signal (Christner et al., 2017; Ebner et al., 2017; Moser and Stichler, 1975; Sokratov and Golubev, 2009; Steen-Larsen et al., 2014). Also, ions and trace elements in snow (re)distribute by post-depository changes (Beine et al., 2002; Blunier et al., 2005; Domine et al., 2004; Legrand and Mayewski, 1997; Röthlisberger et al., 2002). On a microscale, the recrystallization can affect the distribution inside and outside of the ice lattice, i.e. substances that were on the crystal surface can move to the inside and vice versa (Cragin et al., 1996; Hewitt et al., 1991; Trachsel et al.,

1.3 Instrument development and setup

2019). If substances are on the crystal surface and if they are volatile enough, they might be exposed to vertical redistribution in the snow cover through the vapor phase or, if close to the snow surface, in exchange-processes with the atmosphere or photochemical reactions (Abbatt, 2013; Abbatt et al., 2012; Bartels-Rausch et al., 2008, 2013, 2014; Dominé and Shepson, 2002; Grannas et al., 2007; Steffen et al., 2007; Taillandier et al., 2007). In addition, if snow starts melting, meltwater percolation can relocate impurities (Eichler et al., 2001; Grannas et al., 2013; Li et al., 2006; Mann et al., 2011; Meyer et al., 2009; Meyer and Wania, 2008). This effect can be enhanced by current global warming and may occur in colder regions, where melt events so far have been rare. Despite a large number of field and laboratory investigations and computer simulations, central questions regarding the underlying molecular processes of inclusion and rearrangement remain unanswered (Abbatt et al., 2012; Bartels- Rausch et al., 2013, 2014; Grannas et al., 2007; Hullar and Anastasio, 2016; McNeill et al., 2012; Steffen et al., 2007; Wren and Donaldson, 2011). This thesis investigates the influence of snow metamorphism and melt-processes on the distribution of compounds as major ions, trace elements incorporated in the snowpack, surface chemistry reactions with ozone, and the migration of stable water isotopes.

1.3 Instrument development and setup Various instruments have been developed and used in experiments for this thesis - both in the laboratory and in the field. The most important of them and their actual application are briefly introduced in the following section.

1.3.1.1 Droplet Chopper (PSI) – Artificial snow

Figure 1.1: a) Droplet Chopper with 4 nozzles spraying the solution to produce artificial droplets. b) Artificial snow production setup including Droplet Chopper (left), Cryo container (middle) and stock solution (right).

In order to produce homogeneous snow samples with a known amount of impurities, an apparatus was needed that had a very high production capacity and could also be cleaned entirely to remove any contamination. The existing snowmaker at SLF (Schleef et al. 2013) could not be used, as it could not

1.3 Instrument development and setup be sufficiently cleaned to remove any contaminants. An existing droplet sprayer at PSI (Bartels-Rausch et al., 2013) had an insufficient production rate, but formed the basis for the development of a new device. The droplet chopper built for this thesis used Nitrogen (pressure 5 bar, flowrate 0.5l/min) in order to spray a solution of water spiked with defined amounts of ions out of 4 nozzles (Fig. 1.1a). The influx of the nozzles was coupled with a rotating disc to create an intermittent flow. The sprayed droplets had a diameter between 100-1000 μm depending on the settings (pressure and rotation velocity). Prior to the first production, the optimal settings were empirically determined and possible contamination was monitored by evaluating blank samples. The droplets were sprayed directly into liquid nitrogen, in order to minimize segregation during the freezing process (Fig. 1.1b). In the PSI cold room at -20 °C, the frozen ice droplets were separated from the liquid nitrogen and sieved to a size fraction of 300-600 μm in diameter. They were filled into sealed containers and prepared for transport to Davos.

1.3.2 Metamorphism chamber (SLF) – “Aging” snow

Figure 1.2: Installation of Metamorphism Box with (dummy) sample holders put in place, before the surrounding space was filled up with ice (at the bottom) and snow above. b) Removal of the last sample of a particular series to perform elution experiments. The gaps were refilled with dummy containers.

The metamorphism box in the SLF cold laboratory (outside dimensions 140 cm x 70 cm x 60 cm), exposes snow to an artificial temperature gradient. The temperature of the snow surface is controlled by the air temperature set inside the lab. A heating plate at the bottom increases the temperature and thereby maintains a gradient. The containers holding the snow samples were mounted onto a heating/cooling plate set to -4 °C at the base of a metamorphism box (Fig. 1.2a). The empty space around the sample holders was filled with sieved snow to create a homogeneous temperature field. The snow samples were stored in the box at a temperature gradient of 40 K/m for defined periods of up to 90 days. At the end of each storage period, the corresponding containers were removed and prepared for the elution experiments (Fig. 1.2b). In order to keep the conditions stable throughout the

1.3 Instrument development and setup experiment, empty spaces from removed containers were filled up with snow-filled dummy containers.

1.3.3 Micro-computer tomography scanner (SLF) – Snow structure analysis

Figure 1.3: a) CT-container filled with snow and ready to mount in the microCT. b) Reconstruction of the CT- image reveals the three-dimensional fully connected porous ice structure of snow. The scan corresponds to a new snow sample, lateral length of the section shown is 3 mm.

The X-ray computed tomography scanner microCT at the SLF (Scanco microCT 40, SCANCO Medical AG, Switzerland) allows for the three-dimensional analyses of the snow structure (Fig. 1.3b). Parameters such as density and specific surface area SSA can be determined from the scans. Thus, we were able to follow the evolution of the snow structures of the samples in the metamorphism box and characterize the collected field samples. The microCT was operated at a temperature of -15 °C and had a resolution of 10 µm. The snow samples were cut into a 15 x 15 x 60 mm column and mounted into a 20 mm microCT sample holder for scanning (Fig. 1.3a). The reconstructed microCT images were filtered with a Gaussian filter (support 2 voxels, standard deviation 1 voxel) and the threshold for segmentation was applied according to Hagenmuller (2014).

1.3 Instrument development and setup

1.3.4 Elution experiments (SLF) – Investigation of ion (re)distribution in the snow

Figure 1.4: a) Elution setup inside the cold laboratory for parallel elutions. b) Soaked snow sample inside the elution container being dripped from top with 0° C ultrapure water.

We used elution to investigate the microscale distribution of ions in snow because this method reveals the accessibility of different ions in complex multiphase settings. We could not use an organic solvent (freezing point lower than 0°) to exclude any melting because the analysis of organic liquids in the ion chromatograph was impossible. Thus, we decided to use ultra-pure water as eluent. In order to prevent melting during the elution, the eluent water, the snow samples, all instruments, and the laboratory were set to a temperature of -0.2 °C for at least 2 hours prior to the elution experiments. The elution setup was assembled from sterile medical technology components in order to prevent possible contamination through the equipment (Fig. 1.4a). Two different elution principles (water inflow from the top or from the bottom) were evaluated. To minimize the dilution, the final setup was designed to rinse the samples with a constant drip rate of 140 ml/h from the top (Fig. 1.4b). After a specific saturation of the snow samples (up to 1:45 hours) the eluate started to drip out. The eluate was collected at the outflow of the elution column in 15 ml PP vials (Fig 1.4b). Ten eluates per sample were collected to derive a concentration sequence. For the analysis, the collected eluates were transported to the PSI.

1.3.5 Ion chromatography (IC) and plasma mass spectrometry (ICP-SF-MS) (PSI) The concentrations of the major ions in the snow samples and from the different elution samples were determined using ion chromatography (IC) at the PSI lab with a detection limit in the low parts per billion (ppb). The device used was a Metrohm (Herisau, Switzerland) 850 Professional IC combined with an 872 Extension Module and an 858 Professional Sample Processor autosampler. All samples were kept at -20 °C until analysis. Trace elements (TE) in snow are present at even lower concentrations than ions, usually in a nano- to milligrams per liter range. Therefore, the TE snow samples had to be analyzed using discrete

1.3 Instrument development and setup inductively coupled plasma sector field mass spectrometry (ICP-SF-MS, Element 2, Thermo Fisher Scientific, Bremen, Germany) with a detection limit in the low parts per trillion (ppt). Before melting at room temperature, the samples were acidified with concentrated ultra-pure HNO3. The samples were injected by an auto sampler (CETAC ASX-260, Teledyne Cetac, Omaha NE, United States) in combination with a micro concentric nebulizer.

1.3.6 Picarro laser spectrometry (Weissfluhjoch) – Water vapor isotope measurement

Figure 1.5: a) Small hut on Weissfluhjoch test side with inlet for isotope measurement mounted close to the stair. b) Picarro cavity ring-down spectrometer with autosampler unit for daily calibration.

For measuring the stable water isotope composition of ambient vapor in the air, a Picarro cavity ringdown spectrometer (WS-CRDS, L2130-I, Picarro Inc., Santa Clara CA, United States) was installed on the Weissfluhjoch (Fig 1.5b). The system was placed inside a small hut with a regulated temperature around 16°C. Calibration measurements of known liquid water samples were performed using an autosampler (PAL HTC, CTC-Analytics, Zwingen, Switzerland) and a vaporizer once a day. A semi- automatic calibration was achieved by installing a remote-controlled relay. Ambient air was pumped in via a down-looking inlet mounted outside the small mountain hut at ~4 m above ground (between 2 and 4 m above the snow surface, Fig 1.5a). This installation has been in operation since its initial setup in January 2017 and has thus recorded a data set of more than 2.5 years. The measurement is planned to be continued. The collected snow samples were transported frozen to the PSI laboratory and were melted there directly before the analysis at room temperature. For the determination of δD and δ18O, 1 ml aliquots were analyzed in a further Picarro analyzer (WS-CRDS, L2130-i Analyzer, Picarro, Santa Clara, CA, United States) in the laboratory.

1.3 Instrument development and setup

1.3.7 Sampling snow pit – Collecting impurity records

Figure 1.6: a) Sampling team inside the snow pit protected with clean room overalls to prevent any contamination of the samples. b) Filled sampling container on top of the polycarbonate sampler.

Over five days a complete impurity profile of the snow cover on the Weissfluhjoch was recorded. The excavation of the snow profile (approx. 2 x 2 m) was carried out analogous to the traditional profile recording. After this step, special precautions had to be taken: All team members were outfitted with clean room overalls, respirator face masks, and ultra-clean plastic gloves to avoid contamination during the sampling (Fig. 1.6a). All sampling tools and containers had been rinsed with ultra-pure water in the lab and were sealed in bags until their use in the snow pit. The equipment was protected from direct sunlight during the entire sampling process. The samplings were carried out starting from the surface towards the ground with a resolution of 6 cm. Sampling procedure: - Prior to collecting samples, a few centimeters of the sampling snow wall were cut off with a clean spatula to eliminate possible contamination by the preceding excavation - A rectangular (15 × 24 cm) polycarbonate sampler was inserted horizontally into the profile wall. - Polypropylene sampling containers were pushed vertically towards the sampler (Fig 1.6b). - For each different species (ions, trace elements, and stable water isotopes), separate vials

were used. For trace element samples, the tubes were prepared with 0.2 M HNO3. In order to prevent possible cross-contamination, the area on the sampler was divided into different areas and the corresponding vials were always punctured in the same area. - With a spatula, the full vials were turned back upside-up, sealed and for temporarily cooling placed in the snowpack. - After having collected all samples of one layer, the sampler was pulled out of the profile wall and the remaining snow was removed. - The sampler was aligned 6 cm lower and inserted again into the snow wall.

1.4 Study site

This procedure was repeated until the ground was reached. For the lowest layer, the sample containers were filled directly down to the ground. Finally, the snow profile was backfilled with snow to re- establish a homogeneous snow surface within the field site. The samples were packed into an insulated box and transported directly across the ski slopes to the SLF cold laboratory.

1.4 Study site

Figure 1.7: Overview of the Weissﬂuhjoch test site of the WSL Institute for Snow and Avalanche Research SLF at an elevation of 2536 m a.s.l. in the Swiss Alps. The field is fully equipped with sensors recording different snow cover properties.

The field work for this thesis with respect to snow pit sampling and water vapor isotopes measurements was performed at the Weissfluhjoch study site above Davos, Switzerland (2536 m a.s.ls, 46°49′47″ N 9°48′33″ E). The site has a long history of research and was established in 1936 as the first Swiss snow laboratory for avalanche research. Since then, it has been used for various studies and the measurement installations have been continuously expanded (Marty and Meister, 2012). Today the study site is part of the CryoNET and is fully equipped with sensors recording different snow cover properties such as new snow and total snow height, air and snow surface temperature, and turbulent fluxes (Fig 1.7). Further, there are daily manual measurements including snow-pack density measurements with the Snow Micro Penetrometer SMP (Calonne et al., 2019) and a two-weekly snow pit snow classification. All this together offered an ideal basis for our research work. At the same time, the careful coordination with other scientific projects was demanding: The space on the predefined profile lines within the field is limited and each profile must be agreed upon. In terms of cleanliness, our requirements were very demanding, and a parallel recording with the traditional snow profile team was out of question.

1.5 Goals and Outline

1.5 Goals and Outline To overcome the gaps in process-understanding of distribution and redistribution of compounds in snow while accounting for the changing microstructure, the following goals have been identified:

To develop a method to study the distribution of the ions incorporated in snow Snow metamorphism has an impact on the distribution of embedded impurities; however, quantification of this distribution is missing. The first goal was therefore to develop a method to study the distribution of major ions in the lab including detailed monitoring of structural changes in specific surface area SSA and heat fluxes. To apply the method to artificial and natural snow exposed to temperature gradient metamorphism The method developed here is applied to artificial snow with concentration of ions at typical natural concentrations, as well as to natural snow. The conditioning of the samples took place in the lab, using a controlled temperature gradient. Previous similar comparisons have failed mainly because of an insufficient description of the process in the snowpack. Due to the existing infrastructure in the SLF cold lab and on the Weissfluhjoch measurement site, a clear identification of the different processes taking place is possible. To investigate the effect of snow metamorphism on stable oxygen isotopes in a natural snowpack With the knowledge of the redistribution of the ions, the campaign will be extended to study the behavior of δ18O in the snowpack during winter season. Similar to the first two objectives, a characterization of the conditions and processes in the snowpack be the focus of in this study.

All the objectives listed above were pursued and the results are described in the following chapters.

+ In Chapter 2 we study the microscale rearrangement of NH4 and five other major ions during dry snow metamorphism. Artificial and natural snow samples were stored several months at a controlled temperature gradient and in isothermal conditions in the lab. Using a series of elution experiments we were able to investigate the location and therefore the redistribution of the diﬀerent ions. Furthermore, the processes of temperature gradient and isothermal metamorphism are discussed in detail. Chapter 3 is devoted to study the impact of melting on the preservation of trace elements (TE) in a snowpack. To monitor the behavior of TEs during the snow season, we conducted a number of snow pit investigations at the WFJ: Three during the cold winter of 2017 (January, February, March); one at the beginning of the melt season (April); and one shortly before the snow was completely melted (June). The results allowed us to group different TE and describe their potential as proxies for natural and anthropogenic pollution in melt-aﬀected snow or firn cores.

1.5 Goals and Outline

In Chapter 4 we show how isotope signals deposited in an alpine snowpack can be assigned to individual precipitation events and how dry and wet snow metamorphism can alter the snow isotopic record. The study is based on the installed continuous water vapor measurement at WFJ in combination with a series of five snow pit isotope profiles and weekly surface samples. Chapter 5 summarizes the conclusions of the presented studies. Finally, we put forward ideas for future research in this field. Appendix A investigates how tiny plastic particles are transported through the atmosphere and washed out of the air by snowfall events. Snow samples from remote locations such as Tschuggen (Davos, Swiss Alps) and densely populated European sites (Bremen, Bavaria) were investigated. Microplastics could be detected in all investigated snow samples. Appendix B gives a best-practice overview for sampling different types of impurities inside an alpine snowpack.

References Introduction

References Abbatt, J. (2013). Arctic snowpack bromine release. Nat. Geosci. 6, 331–332. doi:10.1038/ngeo1805. Abbatt, J. P. D. D., Thomas, J. L., Abrahamsson, K., Boxe, C., Granfors, A., Jones, A. E., et al. (2012). Halogen activation via interactions with environmental ice and snow in the polar lower troposphere and other regions. Atmos. Chem. Phys. 12, 6237–6271. doi:10.5194/acp-12-6237- 2012. Alley, R. B. (2010). Reliability of ice-core science: historical insights. J. Glaciol. 56, 1095–1103. doi:10.3189/002214311796406130. Anderson, P. S., and Neff, W. D. (2008). Boundary layer physics over snow and ice. Atmos. Chem. Phys. 8, 3563–3582. doi:10.5194/acp-8-3563-2008. Avak, S. E. (2019). Impact and Implications of Melting on the Preservation of Trace Elements in High- Alpine Snow and Glacier Ice. Available at: https://boristheses.unibe.ch/1356/. Barbante, C., Schwikowski, M., Döring, T., Gäggeler, H. W., Schotterer, U., Tobler, L., et al. (2004). Historical record of European emissions of heavy metals to the atmosphere since the 1650s from alpine snow/ice cores drilled near Monte Rosa. Environ. Sci. Technol. 38, 4085–4090. doi:10.1021/es049759r. Barrie, L. A., and Hoff, R. M. (1985). Five years of air chemistry observations in the Canadian Arctic. Atmos. Environ. 19, 1995–2010. doi:10.1016/0004-6981(85)90108-8. Bartels-Rausch, T., Huthwelker, T., Jöri, M., Gäggeler, H., and Ammann, M. (2008). Interaction of gaseous elemental mercury with snow surfaces: Laboratoryinvestigation. Environ. Res. Lett. 3. doi:10.1088/1748-9326/3/4/045009. Bartels-Rausch, T., Jacobi, H. W., Kahan, T. F., Thomas, J. L., Thomson, E. S., Abbatt, J. P. D., et al. (2014). A review of air-ice chemical and physical interactions (AICI): Liquids, quasi-liquids, and solids in snow. Atmos. Chem. Phys. 14, 1587–1633. doi:10.5194/acp-14-1587-2014. Bartels-Rausch, T., Wren, S. N., Schreiber, S., Riche, F., Schneebeli, M., and Ammann, M. (2013). Diffusion of volatile organics through porous snow: Impact of surface adsorption and grain boundaries. Atmos. Chem. Phys. 13, 6727–6739. doi:10.5194/acp-13-6727-2013. Beaudon, E., Gabrielli, P., Sierra-Hernández, M. R., Wegner, A., and Thompson, L. G. (2017). Central Tibetan Plateau atmospheric trace metals contamination: A 500-year record from the Puruogangri ice core. Sci. Total Environ. 601–602, 1349–1363. doi:10.1016/J.SCITOTENV.2017.05.195. Beine, H. J., Dominé, F., Simpson, W., Honrath, R. E., Sparapani, R., Zhou, X., et al. (2002). Snow-pile and chamber experiments during the Polar Sunrise Experiment ‘Alert 2000’: exploration of nitrogen chemistry. Atmos. Environ. 36, 2707–2719. doi:10.1016/S1352-2310(02)00120-6. Bergmann, M., Mützel, S., Primpke, S., Tekman, M. B., Trachsel, J., and Gerdts, G. (2019). White and wonderful? Microplastics prevail in snow from the Alps to the Arctic. Sci. Adv. 5, eaax1157. doi:10.1126/sciadv.aax1157. Blunier, T., Floch, G. L., Jacobi, H., and Quansah, E. (2005). Isotopic view on nitrate loss in Antarctic surface snow. Geophys. Res. Lett. 32, L13501. doi:10.1029/2005GL023011. Christner, E., Kohler, M., and Schneider, M. (2017). The influence of snow sublimation and meltwater evaporation on δD of water vapor in the atmospheric boundary layer of central Europe. Atmos. Chem. Phys. 17, 1207–1225. doi:10.5194/acp-17-1207-2017. Cooper, C. D., and Alley, F. C. (2010). Air pollution control: a design approach. Long Grove: Waveland Press INC.

References Introduction

Correia, A., Freydier, R., Delmas, R. J., Simões, J. C., Taupin, J.-D., Dupré, B., et al. (2003). Trace elements in South America aerosol during 20th century inferred from a Nevado Illimani ice core, Eastern Bolivian Andes (6350 m asl). Atmos. Chem. Phys. 3, 1337–1352. doi:10.5194/acp-3-1337-2003. Cragin, J. H., Hewitt, A. D., and Colbeck, S. C. (1996). Grain-scale mechanisms influencing the elution of ions from snow. Atmos. Environ. 30, 119–127. doi:10.1016/1352-2310(95)00232-N. Craig, H. (1961). Isotopic Variations in Meteoric Waters. Science 133, 1702–3. doi:10.1126/science.133.3465.1702. Cuffey, K. M., Clow, G. D., Alley, R. B., Stuiver, M., Waddington, E. D., and Saltus, R. W. (1995). Large Arctic Temperature Change at the Wisconsin-Holocene Glacial Transition. Science 270, 455–458. doi:10.1126/science.270.5235.455. Dansgaard, W. (1953). The Abundance of O 18 in Atmospheric Water and Water Vapour. Tellus 5, 461– 469. doi:10.1111/j.2153-3490.1953.tb01076.x. Dansgaard, W. (1964). Stable isotopes in precipitation. Tellus 16, 436–468. doi:10.3402/tellusa.v16i4.8993. Dominé, F., and Shepson, P. B. (2002). Air-snow interactions and atmospheric chemistry. Science 297, 1506–1510. doi:10.1126/science.1074610. Domine, F., Sparapani, R., Ianniello, A., and Beine, H. J. (2004). The origin of sea salt in snow on Arctic sea ice and in coastal regions. Atmos. Chem. Phys. 4, 2259–2271. doi:10.5194/acp-4-2259-2004. Ebner, P. P., Steen-Larsen, H. C., Stenni, B., Schneebeli, M., Steinfeld, A., Ebner, P. P., et al. (2017). Experimental observation of transient δ18O interaction between snow and advective airflow under various temperature gradient conditions. Cryosphere 11, 1733–1743. doi:10.5194/tc-11- 1733-2017. Edebeli, J. (2018). Multiphase Chemical Reactivity in Cold Regions. doi:10.3929/ETHZ-B-000313948. Eichler, A., Schwikowski, M., and Gäggeler, H. W. (2001). Meltwater-induced relocation of chemical species in Alpine firn. Tellus, Ser. B Chem. Phys. Meteorol. 53, 192–203. doi:10.3402/tellusb.v53i2.16575. Eichler, A., Tinner, W., Brütsch, S., Olivier, S., Papina, T., and Schwikowski, M. (2011). An ice-core based history of Siberian forest fires since AD 1250. Quat. Sci. Rev. 30, 1027–1034. doi:10.1016/j.quascirev.2011.02.007. Eichler, J., Kleitz, I., Bayer-Giraldi, M., Jansen, D., Kipfstuhl, S., Shigeyama, W., et al. (2017). Location and distribution of micro-inclusions in the EDML and NEEM ice cores using optical microscopy and in situ Raman spectroscopy. Cryosphere 11, 1075–1090. doi:10.5194/tc-11-1075-2017. Epstein, S., and Mayeda, T. (1953). Variation of O18 content of waters from natural sources. Geochim. Cosmochim. Acta 4, 213–224. doi:10.1016/0016-7037(53)90051-9. Eyrikh, S., Eichler, A., Tobler, L., Malygina, N., Papina, T., and Schwikowski, M. (2017). A 320 Year Ice- Core Record of Atmospheric Hg Pollution in the Altai, Central Asia. Environ. Sci. Technol. 51, 11597–11606. doi:10.1021/acs.est.7b03140. Gabrielli, P., Hardy, D. R., Kehrwald, N., Davis, M., Cozzi, G., Turetta, C., et al. (2014). Deglaciated areas of Kilimanjaro as a source of volcanic trace elements deposited on the ice cap during the late Holocene. Quat. Sci. Rev. 93, 1–10. doi:10.1016/J.QUASCIREV.2014.03.007. Grannas, A. M., Bogdal, C., Hageman, K. J., Halsall, C., Harner, T., Hung, H., et al. (2013). The role of the global cryosphere in the fate of organic contaminants. 13, 3271–3305. doi:10.5194/acp-13- 3271-2013.

References Introduction

Grannas, A. M., Jones, A. E., Dibb, J., Ammann, M., Anastasio, C., Beine, H. J., et al. (2007). An overview of snow photochemistry: Evidence, mechanisms and impacts. Atmos. Chem. Phys. 7, 4329–4373. doi:10.5194/acp-7-4329-2007. Hagenmuller, P., Chambon, G., Flin, F., Morin, S., and Naaim, M. (2014). Snow as a granular material: assessment of a new grain segmentation algorithm. Granul. Matter 16, 421–432. doi:10.1007/s10035-014-0503-7. Heintzenberg, J., and Rummukainen, M. (1993). Airborne particles in snow. J. Glaciol. 39, 239–244. doi:10.1017/S0022143000015896. Hewitt, A. D., Cragin, J. H., and Colbeck, S. C. (1991). Effects of crystal metamorphosis on the elution of chemical species from snow. Proc. 48th East. Snow Conf. Guelph, Ontario. Hullar, T., and Anastasio, C. (2016). Direct visualization of solute locations in laboratory ice samples. Cryosph. 10, 2057–2068. doi:10.5194/tc-10-2057-2016. Johnsen, S. J., Dahl-Jensen, D., Dansgaard, W., and Gundestrup, N. (1995). Greenland palaeotemperatures derived from GRIP bore hole temperature and ice core isotope profiles. Tellus B Chem. Phys. Meteorol. 47, 624–629. doi:10.3402/tellusb.v47i5.16077. Jouzel, J., Masson-Delmotte, V., Cattani, O., Dreyfus, G., Falourd, S., Hoffmann, G., et al. (2007). Orbital and millennial Antarctic climate variability over the past 800,000 years. Science 317, 793–6. doi:10.1126/science.1141038. Kaspari, S., Mayewski, P. A., Handley, M., Osterberg, E., Kang, S., Sneed, S., et al. (2009). Recent increases in atmospheric concentrations of Bi, U, Cs, S and Ca from a 350-year Mount Everest ice core record. J. Geophys. Res. 114, D04302. doi:10.1029/2008JD011088. Kaspari, S., Mayewski, P., Kang, S., Sneed, S., Hou, S., Hooke, R., et al. (2007). Reduction in northward incursions of the South Asian monsoon since ∼1400 AD inferred from a Mt. Everest ice core. Geophys. Res. Lett. 34. doi:10.1029/2007GL030440. Kellerhals, T., Brütsch, S., Sigl, M., Knüsel, S., Gäggeler, H. W., and Schwikowski, M. (2010). Ammonium concentration in ice cores: A new proxy for regional temperature reconstruction? J. Geophys. Res. Atmos. 115, D16123. doi:10.1029/2009JD012603. Langway, C. C. (2008). The history of early polar ice cores. Cold Reg. Sci. Technol. 52, 101–117. doi:10.1016/J.COLDREGIONS.2008.01.001. Legrand, M., and Mayewski, P. (1997). Glaciochemistry of polar ice cores: A review. Rev. Geophys. 35, 219–243. doi:10.1029/96RG03527. Li, Z., Edwards, R., Mosley-Thompson, E., Wang, F., Dong, Z., You, X., et al. (2006). Seasonal variability of ionic concentrations in surface snow and elution processes in snow–firn packs at the PGPI site on Ürümqi glacier No. 1, eastern Tien Shan, China. Ann. Glaciol. 43, 250–256. doi:10.3189/172756406781812069. Mann, E., Meyer, T., Mitchell, C. P. J., and Wania, F. (2011). Mercury fate in ageing and melting snow: Development and testing of a controlled laboratory system. J. Environ. Monit. 13, 2695–2702. doi:10.1039/c1em10297d. Masson-Delmotte, V., Hou, S., Ekaykin, A., Jouzel, J., Aristarain, A., Bernardo, R. T., et al. (2008). A review of antarctic surface snow isotopic composition: Observations, atmospheric circulation, and isotopic modeling. J. Clim. 21, 3359–3387. doi:10.1175/2007JCLI2139.1. McNeill, V. F., Grannas, A. M., Abbatt, J. P. D., Ammann, M., Ariya, P., Bartels-Rausch, T., et al. (2012). Organics in environmental ices: sources, chemistry, and impacts. Atmos. Chem. Phys. 12, 9653– 9678. doi:10.5194/acp-12-9653-2012. Meyer, T., Lei, Y. D., Muradi, I., and Wania, F. (2009). Organic contaminant release from melting snow. Influence of chemical partitioning. Environ. Sci. Technol. 43, 657–662. doi:10.1021/es8020217.

References Introduction

Meyer, T., and Wania, F. (2008). Organic contaminant amplification during snowmelt. Water Res. 42, 1847–1865. doi:10.1016/j.watres.2007.12.016. Moser, H., and Stichler, W. (1975). Deuterium and oxygen-18 contents as an index of the properties of snow covers. in Snow Mechanics, Proceedings of the Grindelwald Symposium, April 1974 (Grindelwald), 442 pp. Available at: http://hydrologie.org/redbooks/a114/iahs_114_0122.pdf [Accessed August 29, 2019]. Pinzer, B. R., Schneebeli, M., and Kaempfer, T. U. (2012). Vapor flux and recrystallization during dry snow metamorphism under a steady temperature gradient as observed by time-lapse microtomography. Cryosph. 6, 1141–1155. doi:10.5194/tc-6-1141-2012. Röthlisberger, R., Hutterli, M. A., Wolff, E. W., Mulvaney, R., Fischer, H., Bigler, M., et al. (2002). Nitrate in Greenland and Antarctic ice cores: A detailed description of post-depositional processes. Ann. Glaciol. 35, 209–216. doi:10.3189/172756402781817220. Schleef, S., Jaggi, M., Löwe, H., and Schneebeli, M. (2014). An improved machine to produce nature- identical snow in the laboratory. J. Glaciol. 60, 94–102. doi:10.3189/2014JoG13J118. Schwikowski, M., Barbante, C., Doering, T., Gaeggeler, H. W., Boutron, C., Schotterer, U., et al. (2004). Post-17th-Century Changes of European Lead Emissions Recorded in High-Altitude Alpine Snow and Ice. Environ. Sci. Technol. 38, 957–964. doi:10.1021/es034715o. Sierra-Hernández, M. R., Gabrielli, P., Beaudon, E., Wegner, A., and Thompson, L. G. (2018). Atmospheric depositions of natural and anthropogenic trace elements on the Guliya ice cap (northwestern Tibetan Plateau) during the last 340 years. Atmos. Environ. 176, 91–102. doi:10.1016/J.ATMOSENV.2017.11.040. Sokratov, S. A., and Golubev, V. N. (2009). Snow isotopic content change by sublimation. J. Glaciol. 55, 823–828. doi:10.3189/002214309790152456. Steen-Larsen, H. C., Masson-Delmotte, V., Hirabayashi, M., Winkler, R., Satow, K., Prié, F., et al. (2014). Climate of the Past What controls the isotopic composition of Greenland surface snow? 10, 377– 392. doi:10.5194/cp-10-377-2014. Steffen, A., Douglas, T., Amyot, M., Ariya, P., Aspmo, K., Berg, T., et al. (2007). A synthesis of atmospheric mercury depletion event chemistry linking atmosphere, snow and water. Atmos. Chem. Phys. Discuss. 7, 10837–10931. doi:10.5194/acpd-7-10837-2007. Taillandier, A. S., Dominé, F., Simpson, W. R., Sturm, M., and Douglas, T. A. (2007). Rate of decrease of the specific surface area of dry snow: Isothermal and temperature gradient conditions. J. Geophys. Res. Earth Surf. doi:10.1029/2006JF000514. Trachsel, J. C., Avak, S. E., Edebeli, J., Schneebeli, M., Bartels-Rausch, T., Bruetsch, S., et al. (2019). Microscale Rearrangement of Ammonium Induced by Snow Metamorphism. Front. Earth Sci. 7, 194. doi:10.3389/feart.2019.00194. Uglietti, C., Gabrielli, P., Cooke, C. A., Vallelonga, P., and Thompson, L. G. (2015). Widespread pollution of the South American atmosphere predates the industrial revolution by 240 y. Proc. Natl. Acad. Sci. U. S. A. 112, 2349–54. doi:10.1073/pnas.1421119112. Wolff, E. W., Chappellaz, J., Blunier, T., Rasmussen, S. O., and Svensson, A. (2010). Millennial-scale variability during the last glacial: The ice core record. Quat. Sci. Rev. 29, 2828–2838. doi:10.1016/J.QUASCIREV.2009.10.013. Wren, S. N., and Donaldson, D. J. (2011). Exclusion of nitrate to the air-ice interface during freezing. J. Phys. Chem. Lett. 2, 1967–1971. doi:10.1021/jz2007484.

2 Microscale rearrangement of ammonium induced by snow

metamorphism

Published in Frontiers in Earth Science: Trachsel Jürg C., Avak Sven E., Edebeli Jacinta, Schneebeli Martin, Bartels-Rausch Thorsten, Brütsch Sabina, Eichler Anja. (2019). Microscale Rearrangement of Ammonium Induced by Snow Metamorphism. Frontiers in Earth Science. https://doi.org/10.3389/feart.2019.00194

DOI=10.3389/feart.2019.00194

ISSN=2296-6463 2.1 Introduction

Abstract Earth’s snowpack hosts chemical impurities that exchange with the overlying air, strongly impacting atmospheric chemistry. Being embedded in firn and glacier ice formed from surface snow, impurities provide the basis for reconstructing past atmospheric composition from ice core records. The location of these impurity compounds within the snow critically controls their reactivity and preservation.

+ Ammonium (NH4 ) determines acid-base equilibria and buffer capacity within the snow and is a fundamental ice-core proxy, especially for biogenic, forest fire, and anthropogenic emissions. However, the redistribution during snow metamorphism affecting snow chemical reactivity and potential post-depositional relocation is not understood so far. Here, we study the rearrangement of

+ 2+ – – + 2– NH4 and five other major ions (Ca , Cl , F , Na and SO4 ) during dry snow metamorphism using a series of elution experiments. Artificial and natural snow samples were stored for up to three months at a controlled temperature gradient of 40 K/m and isothermal conditions. The different types of snow increase complexity as natural and artificial snow have a different history that impacts their physical properties and the initial distribution of impurities. Further, we test our findings using natural snow samples taken from different depths of a natural snowpack at the Weissfluhjoch (Swiss Alps) to confirm the impact of our laboratory results to the cryosphere. With progressing temperature gradient metamorphism, snow structures in natural and artificial snow converged and ions with high solubility

+ - - in ice (NH4 , F and Cl ) were incorporated into the less accessible ice interior. In contrast, ions with

2+ 2- + lower solubility (Ca , SO4 and Na ) became better accessible for the eluent. Our results show that the redistribution during snow metamorphism is strongly dependent on the temperature gradient, exposure time and chemical composition. This study allowed for the first time to explicitly relate the

+ general low relocation proneness of NH4 during post-depositional processes such as meltwater percolation to the preferred incorporation of this ion into the less accessible ice interior during snow

+ metamorphism. Furthermore, our results imply that with ongoing aging of a snowpack, NH4 , independent of its primary location, is less available for chemical reactions at the air-ice interface.

2.1 Introduction Ice cores recovered from cold glaciers are invaluable archives of past climate and atmospheric composition, covering up to 800,000 years (Members EPICA Community, 2004). Among the various types of impurities trapped in the ice, major ions provide information about past changes in anthropogenic pollution, atmospheric transport, forest fires, and atmospheric temperature (Eichler et al., 2011; Kaspari et al., 2007; Kellerhals et al., 2010; Legrand and Mayewski, 1997; Thompson et al., 2013). During snowfall, atmospheric constituents such as aerosol particles and gases are scavenged and deposited by wet and/or dry deposition. After deposition in polar or high-altitude regions, snow

2.1 Introduction is successively transformed into firn and glacier ice, leading to an embedding of chemical impurities including major ions. However, the ice-core concentration records of major ions are not only determined by their initial atmospheric concentrations. Post-depositional processes can strongly alter the originally deposited signal. Such processes include wind erosion (Dansgaard et al., 1973; Lorius et al., 1969), re-emission of volatile compounds from the snow surface (Bartels-Rausch et al., 2008; Lalonde et al., 2002; Röthlisberger et al., 2002), sublimation (Stichler et al., 2001), migration within the snow and firn layer (Saltzman, 1995; Wolff, 1996), and relocation during melt-water percolation (Eichler et al., 2001; Grannas et al., 2013; Li et al., 2006). With increasing global temperatures, the fate of impurities during melt events is of increasing concern (Eichler et al., 2001; Grannas et al., 2013; Meyer and Wania, 2008). Species embedded in the interior of the snow matrix are less prone to being washed away by melt or rainwater than those species located at the air-ice interface. Eichler et al. (2001) have shown that in sections of ice core records that had experienced melt processes traces of impurities with high solubility in ice can still reliably be used as archives to reconstruct past atmospheric composition. Further interest in the location of impurities comes from their role in chemistry. Snow is a fully connected porous ice structure and readily exchanges trace gases and aerosols with the overlying air (Bartels-Rausch et al., 2014; Dominé and Shepson, 2002; Grannas et al., 2007). These fluxes are an interplay of physical exchange processes and chemistry. The chemical complexity of snow with its multiphase and multicomponent character has recently raised much interest (Abbatt et al., 2012; Bartels-Rausch et al., 2014; Kahan et al., 2014). Laboratory and field studies have shown that fluxes originate from impurities hosted in various compartments of snow. Volatile impurities forming a solid solution may portion to the gas phase, as shown for formaldehyde (Barret et al., 2011a, 2011b). Surface adsorption and related fluxes from and to the air-ice interface is of high relevance for halogens, whereas contributions from their solid solution are thought to be negligible (Abbatt et al., 2012). Similarly, the rate of higher-order chemical reactions critical depends on the location with significant differences for reactions occurring at the air-ice interface compared to those in liquid fractions of the snow, such as brine (Bartels-Rausch et al., 2014) or aerosol deposits (Bartels-Rausch et al., 2014; Koop et al., 2000). Chemistry in solid solutions is generally negligible. That the chemistry critically depends on the origin of impurities and the chemical properties of these compartments rather than simply the total impurity concentrations was recently impressively illustrated by a field study on halogen chemistry in arctic snow (Pratt et al., 2013). While it is clear that the fate of impurities and thus their preservation in ice cores and their role in atmospheric chemistry, critical depends on their location - in particular, the distribution between solid solution, liquid fraction, and at the air-ice interface- the role of snow dynamics on this distribution is essentially open. Dry metamorphism is the most common dynamic process of snow “aging”. Snow

2.1 Introduction on Earth’s ground is almost always exposed to a temperature gradient, for example between bedrock and the overlying air. Consequently, the relatively warmer ice surface sublimates and the water vapor is deposited on the cold side of the structure. Studies of snow exposed to a temperature gradient have shown that dry metamorphism leads to complete restructuring of the snowpack, i.e., to a mass turnover of the entire ice mass of up to 60 % per day at a gradient of ~50 K/m (Pinzer et al., 2012; Pinzer and Schneebeli, 2009). Although physical changes during snow metamorphism are well studied, very little is known about concurrent rearrangement of chemical species. Hewitt et al. (1991)

- - 2- investigated the effect of snow metamorphism on the redistribution of Cl , NO3 and SO4 . This study

- - 2- suggested incorporation of Cl and NO3 into the ice interior and exclusion of SO4 to the air-ice interface with progressing snow metamorphism. Motivation to study the incorporation of ions during metamorphism came from its influence on ionic concentrations in meltwater, where excluded ions lead to ionic pulses early in the melt with potentially adverse impacts on the environmental systems.

+ Of the major ions in ice cores, NH4 is an essential proxy for biogenic, forest fire, and anthropogenic agricultural emissions (Döscher et al., 1996; Eichler et al., 2009, 2011; Fuhrer et al.,

+ 1996; Kellerhals et al., 2010). NH4 is widely used to date ice cores by counting annual layers due to the strong seasonal variability (Fuhrer et al., 1993). The majority of previous studies point to good

+ preservation of NH4 concentration records even in areas strongly affected by melting (Alps (Eichler et al., 2001), South America (Ginot et al., 2010), Tian Shan (Li et al., 2006), Svalbard (Virkkunen et al., 2007). The good preservation has been suggested to be due to rearrangement processes during snow

+ metamorphism, including a burial of NH4 into the ice matrix, not accessible by meltwater. To the best

+ of our knowledge, no studies on the relocation of ammonium (NH4 ) during snow metamorphism are available, so far. Further interest is motivated by NH3 being the most abundant alkaline gas in the

+ atmosphere. Consequently, NH3/NH4 concentrations in the air and snow strongly impact the uptake

+ of acidic species into ice (Hoog et al., 2007). Therefore, relocation of NH4 during snow metamorphism might significantly change acid-base equilibria and buffer capacities within the snow microstructure, with direct consequences for air–ice chemical exchanges and chemical reactivity (Bartels-Rausch et al., 2014).

+ 2+ In this work, we monitor the redistribution of NH4 with respect to five other major ions (Ca ,

– – + 2– Cl , F , Na , and SO4 ) during temperature gradient snow metamorphism with a series of elution experiments. Elution was chosen as method because it reveals the accessibility of different ions in complex multiphase settings and thus gives an indication of their distribution in snow. Due to the complexity of the underlying mechanisms, this study focuses on the overall change in accessibility of

+ NH4 rather than disentangling the underlying processes in detail. To address the diversity of snow in the environment, we investigate changes to the accessibility of ions in natural and artificial snow. Great care was taken to work with snow samples with concentration levels of solutes less than 1000 µg/l to

2.2 Methods ensure direct relevance and comparability of the results to the Alpine snowpack. The concentrations are comparable to the study of Hewitt et al. (1991) but significantly reduced compared to most laboratory studies (Grannas et al., 2007; Hullar and Anastasio, 2016) and correspond to the range measured in Alpine snow (Baltensperger et al., 1993; Schwikowski et al., 1999; Wagenbach et al., 1988). One key aspect of the study was to expose the snow samples to well defined temperature settings. For comparison, snow samples were stored at isothermal and temperature gradient conditions to clarify the impact of the vastly different metamorphism conditions. Metamorphism was performed for artificial and natural snow samples in a well-controlled laboratory setting for up to three months. The structure of the samples was monitored using computer tomography. Additionally, we investigated structural differences and ion distribution in samples taken from different depths of a natural snowpack that had undergone natural snow metamorphism to discuss the relevance of our findings for environmental settings. This approach allowed, for the first time, investigation of the

+ microscale redistribution of NH4 with respect to other major ions during snow metamorphism.

2.2 Methods This study combines the preparation of doped artificial snow or sampling of natural snow,

2+ – – + + 2 metamorphism experiments, and recording elution series of the ions Ca , Cl , F , Na , NH4 , and SO4 . All tools and containers used during snow production, sampling, and elution were carefully pre- cleaned 5 times with ultrapure water (18 MΩcm quality, arium® pro, Sartorius, Göttingen, Germany).

2.2.1 Preparation of artificial and natural snow samples

2.2.1.1 Preparation of artificial snow Artificial snow was produced by shock freezing sprayed droplets of a sample solution in liquid nitrogen. This method of snow production has already been successfully applied in other studies (Bartels-Rausch et al., 2004, 2013; Kerbrat et al., 2010; Matykiewiczová et al., 2007). The sample solutions with defined

2+ – – + + 2– concentrations of the major ions Ca , Cl , F , Na , NH4 , and SO4 were prepared by dissolving specific amounts of salts (p.a. quality, CaCl2, NaF, and (NH4)2SO4) in ultrapure water (18 MΩcm quality, arium® pro, Sartorius, Göttingen, Germany). In the cold room of the Paul Scherrer Institute at -20 °C the ice droplets were sieved to a size fraction of 300-600 μm in diameter. About 60 g of the snow samples were filled in pre-cleaned 160 ml polypropylene (PP) containers (Faust, Schaffhausen, Switzerland). The bottom of each container was already pre-filled with 4 x 10 ml layers of frozen ultrapure water, to prevent the formation of an air gap at the bottom of the snow sample caused by the vapor flux during the aging. The containers were stored isothermally at -20 °C until the beginning of the metamorphism experiment. The bulk concentrations, cBulk, of the different major ions in the sieved snow samples were

2.2 Methods determined in triplicate analyses using ion chromatography (IC). Concentrations varied between ~380 and 1000 µg/l (Table 2.1).

Table 2.1: Comparison of the different bulk concentrations CBulk in µg/l. Stdev. was calculated out of 3 samples. Age Ammonium Calcium Chloride Fluoride Sodium Nitrate Sulfate days µg/l µg/l µg/l µg/l µg/l µg/l µg/l artificial snow - 381  10 548  42 1003  25 851  55 1013  57 - 1024  30 natural snow - 1029  5 99  7 52  4 8  0.2 18  2 1838  47 939  6 0 1029  5 99  7 52  4 8  0.2 18  2 1838  47 939  6 natural 30 138  8 27  4 25  4 < 0.5 10  1 218  12 83  9 snowpack 60 11  1 22  1 44  1 < 0.5 18  1 257  8 36  1 90 14  1 157  5 19  2 0.58  0.02 17  0.4 196  10 104  4

2.2.1.2 Sampling of natural snow at WFJ After a major snowfall event on April 4th, 2017, 20 kg of fresh snow were collected from the Weissfluhjoch site (WFJ) in the Swiss Alps above Davos (2536 m a.s.ls, 46°49′47″ N 9°48′33″ E). The site is a reference CryoNet station (Marty and Meister, 2012). The snow was sampled into a plastic box using a pre-cleaned Teflon shovel. The plastic box was sealed and placed in an insulated container during the transport. In a -20 °C laboratory at the WSL-Institute for Snow and Avalanche Research SLF, the snow was stirred using a Teflon rod to obtain a homogeneous snow sample. Bulk concentrations, cBulk, of the major ions in the WFJ snow samples were determined in triplicate analyses by IC and are presented in Table 2.1. The natural snow samples were subsampled in 160 ml PP containers (as described in Section 2.2.1.1.) and isothermally stored at -20 °C until the beginning of the metamorphism experiment.

2.2.2 Laboratory-controlled and natural snow metamorphism In total, we examined three different combinations of snow type and metamorphism: (a) artificial snow exposed to a laboratory-controlled temperature gradient (b) natural snow exposed to a laboratory-controlled temperature gradient (c) natural snow from diff. depths of a snow pit, that was exposed to natural metamorphism The combinations and the number of samples are listed in Table 2.2.

2.2.2.1 Laboratory-controlled metamorphism experiment The 160 ml PP containers holding the snow samples were mounted onto a heating plate set to -4 °C at the base of a metamorphism box (Figure 2.1A). The metamorphism box (outside dimensions: 140 cm x 70 cm x 60 cm, located at the SLF) offered space for up to 28 containers. The box was filled

2.2 Methods with a 4 cm ice layer, that is, to the same height as the ice layer in the sample containers. The remaining space around the sample holders was filled with sieved snow (about 6 cm depth to the cap of the sample holders). The metamorphism box was covered with a thin aluminum plate to inhibit sublimation of the filled-in snow. The laboratory temperature was held at -8 °C. Effective temperatures at the top and bottom of the 6 cm thick snow samples were -6.5 °C and -4.1 °C, respectively, corresponding to a temperature gradient of 40 K/m. The snow samples were stored in the metamorphism box for 0, 3, 6, 12, 30, 60, and 90 days, respectively. At the end of each storage period, the corresponding containers were removed and prepared for the elution experiments. In order to keep the conditions steady throughout the experiment, empty spaces from removed containers were filled up with snow-filled dummy containers.

Figure 2.1: (A) Scheme of the metamorphism box showing two sample containers (28 samples were stored); (B) Scheme of the elution setup. This setup was installed in the laboratory at 0 °C, a total of three elution stations could be run simultaneously.

In addition to the snow samples aged at the defined temperature gradient in the metamorphism box, we stored samples of artificial and natural snow at isothermal conditions (-20 °C) for a 90-day period. Samples were stored inside a multilayer box out of steel plates (1 cm thick) and styrofoam (2 and 5 cm thick) according to Löwe et al. (2011) to maintain a homogeneous temperature of -20 °C. In Table 2.2, an overview of all used containers is given. For the artificial snow samples, we used two containers for the 0-, 6-, 12- and 30-day batches respectively: One for the elution experiment and one for monitoring structural changes using X-ray micro-computed tomography (microCT). For the 3-day batch, we had

2.2 Methods only one container (elution), for the 60-day batch 4 containers (triplicate elution/microCT) and for the 90-day isothermal sample only 1 container (microCT) as well. For the natural snow, we used 4 containers for each 0-, 3-, 6-, 12-, 30-, 60-, 90-day batch (triplicate elution/microCT), and two containers (elution/microCT) for the 90-day isothermal batch. As initial state “0 day” we define all states of metamorphism up to and including a duration of 24 hours.

Table 2.2: Combinations of metamorphism type and snow examined in this study. The numbers of sample containers for the elution experiment and the microCT measurements are separated by a forward slash.

Metamorphism time (days) 0 3 6 12 30 60 90 90 isotherm

Metamorphism Samples number of sample containers: Elution / microCT

Controlled (Lab) Artificial snow 1/1 1/- 1/1 1/1 1/1 3/1 -/- -/1

Controlled (Lab) Natural snow 3/1 3/1 3/1 3/1 3/1 3/1 3/1 1/1

Nature Natural snowpack 3/1 -/- -/- -/- 3/1 3/1 3/1 -/-

2.2.2.2 Natural metamorphism: Samples of natural snow from a WFJ snow pit In addition to the collected surface snow (section 2.1.2), we sampled 1 kg of snow at 3 different depths of a snow pit at the field site, WFJ. Each sample corresponded to a snowfall in January, February, March, and April 2017 and an age of 90, 60, 30 and 0 days, respectively. This snow was filled into pre- cleaned plastic boxes and transported frozen to the SLF cold lab. In the laboratory at -20 °C, the snow was stirred with a Teflon rod and four 60 g subsamples of each sample were filled into 160 ml PP containers (Table 2.2, triplicate elution plus microCT sample). Bulk concentrations, cBulk, of the different major ions in the snow pit samples are presented in Table 2.1. The containers were isothermally stored at -20 °C until the beginning of the elution. These samples were not exposed to artificial temperature gradient metamorphism in the metamorphism box. The age of the snow in the snow profile was determined based on the density evolution. Daily characterization of the snowpack at the WFJ field is conducted by the SLF every winter season; this includes density measurements with the Snow Micro Penetrometer SMP (Calonne and Richter et al., 2019). These measurements allowed us to monitor the stratigraphic evolution within the snowpack and to determine the age and depth of particular snow layers.

2.2.3 Elution Experiments The elution experiments were performed in the cold room of the SLF at a temperature of 0 °C. All snow samples were tempered at -0.2 °C for 2 hours prior to the elution experiments. Then the snow samples were extracted by removing the ice bottom using a ceramic knife and transferred into the elution

2.2 Methods column. The elution was performed in a 10 cm long plastic column with a diameter of 6 cm (Figure 2.1B). The samples were rinsed with an eluent of ultrapure water (0 °C). The eluent flow was set to 140 ml/h with the help of a drop regulator. After a specific saturation time of between 30 minutes and 1:45 hours (Table 2.3) the eluate started to drip out at the outflow of the column and was collected in 15 ml PP vials (Sarstedt, Nümbrecht, Germany). We collected 10 eluates for each elution experiment: the first 4 with a volume of 3 ml, the second 4 with 6 ml, and finally 2 with 12 ml volume. During the elution experiment, the surface temperature of the elution station was monitored by an infrared thermometer to stay within -0.5 ± 0.5 °C. We assume that melting of the samples at this temperature and applying a 0 °C eluent is negligible but cannot be excluded entirely. After the collection of the 10 eluates, the snow sample in the elution column, including the retained ultrapure water, was allowed to melt at room temperature. Two aliquots were collected in 50 ml PP vials (Sarstedt, Nümbrecht, Germany). Procedural blanks taken during the elution experiments were always close to the detection limit for all investigated ions.

Table 2.3: Saturation and Elution Time in [hh:mm]. The saturation time is the time from the start of the elution until the first drop of eluate drained out below the sample. Elution time represents the total time until the 10th Eluate was collected, respectively 60 ml of eluate had drained out and the sample was transferred to melting.

artificial snow natural snow natural snowpack

Age (days) saturation time elution time saturation time elution time saturation time elution time 0 00:49 02:12 01:35 03:28 01:35 03:28 3 00:39 01:57 01:44 03:42 - - 6 01:17 03:27 01:40 03:20 - - 12 01:35 03:37 01:21 03:07 - - 30 01:00 03:13 01:15 03:00 01:26 03:30 60 00:56 02:56 01:18 02:56 01:19 03:24 90 - - 01:06 02:51 00:34 02:42 90 isothe. - - 01:35 03:08 - -

2.2.4 Vertical distribution of ions after temperature gradient storage To investigate the vertical distribution of the ion concentration within one sample, we split the microCT samples of the natural snow batch (0, 3, 6, 12, 60, 90, and 90 days isothermal, Table 2.2) into two parts: One part was used to determine the microstructure in the microCT (Section 2.2.6.). The second part was used to analyze the vertical distribution of the ions. The 6 cm high snow samples were vertically cut into five ~1.2 cm slices. The potentially contaminated surface of these slices was removed with a pre-cleaned ceramic knife and the decontaminated samples stored in 50 ml PP containers at -20 °C until IC analysis.

2.3 Results

2.2.5 Ion chromatography (IC) Samples were kept at -20 °C until analysis at the Paul Scherrer Institute. Concentrations of the major

2+ + + - - 2- cations Ca , Na , NH4 , and anions Cl , F , SO4 in the different elution and snow samples were determined using ion chromatography (IC). A Metrohm (Herisau, Switzerland) 850 Professional IC combined with an 872 Extension Module and an 858 Professional Sample Processor autosampler was used for the analyses. Cations were separated using a Metrosep C4 column (Metrohm) and 2.8 mM

-1 HNO3 as eluent at a flow rate of 1 ml min . Anions were separated using a Metrosep A Supp 10 column

(Metrohm) and were eluted stepwise using first, a 1.5 mM Na2CO3/0.3 mM NaHCO3 (1:1 mixture)

-1 eluent, then an 8 mM Na2CO3/1.7 mM NaHCO3 (1:1 mixture) eluent at a flow rate of 0.9 ml min . Possible instrumental drifts were monitored by measuring an in-house standard after every 20th

2+ - - + + 2- sample. Detection limits were 10, 1, 0.5, 0.5, 0.5, and 5 µg/l for Ca , Cl , F , Na , NH4 , and SO4 -, respectively.

2.2.6 X-ray microtomography (microCT) One sample of each batch (Table 2.2) was scanned for structural analysis using the microCT. The snow samples were extracted from the snow sample holder by separating the snow from the ice base layer. The snow samples were cut into a 1.5 x 1.5 x 6 cm column and mounted into a microCT sample holder (diameter 2 cm). The X-ray computed tomography scanner (Scanco microCT 40, SCANCO Medical AG, Switzerland) at the SLF was operated at a temperature of -15 °C and had a resolution of 10 µm. The reconstructed microCT images were filtered with a Gaussian filter (support 2 voxels, standard deviation 1 voxel) and the threshold for segmentation was applied according to Hagenmuller (2014). Structural parameters of the segmented ice structure were extracted with the software tools of the microCT device (Image Processing Language, Scanco Medical).

2.3 Results

2.3.1 Structural development during metamorphism

2.3.1.1 Temperature gradient metamorphism The microCT images in Figure 2.2 show the structural characteristics of the snow samples at different stages in metamorphism. These images display the drastic structural changes in the snow samples during metamorphism. The initial structure of the two types of snow, natural and artificial, differed strongly. In artificial snow, the individual spheres of homogeneous diameter are partially recognizable while the natural snow shows the typical characteristics of fresh snow with branching (Figures 2.2 and 2.3, first column). In both types, the bonds between the individual particles are still little developed, so that the shape of the original deposited crystals can be presumed. Already after 12 days

2.3 Results

Figure 2.2: Structural development of the snow samples during metamorphism imaged by microCT. Solid ice is black, interstitial air white. Cross-sections are shown for artificial snow (upper row) and natural snow (middle row) after 0, 12, 30, 60, and 90 days of storage under an artificial gradient of 40 K/m and 90 days isothermal storage at -20 °C. In the lower row cross-sections of 0, 30, 60 and 90 day old snow from different depths of the snow pit at WFJ are shown. These 4 samples originate from 4 different snowfall events at different stages in natural metamorphism. of storage time at a gradient of 40 K/m, the apparent differences between the two types of snow were much smaller. A significant coarsening of the structure was observed and a fully connected porous ice structure had grown in both the artificial and the natural snow. The reduction of the specific surface area (SSA) during this re-crystallization is qualitatively visible: With increasing exposure time to the temperature gradient, the coarsening of the snow particles is further enhanced (Figure 2.2, 3rd to 5th column). The natural snowpack samples as third snow type showed a similar structural development. In fact, its 90-day sample had undergone the largest transformation compared to the initial state at day 0 (Figure 2.2, last row). The evolution of the SSA of the individual snow samples during the metamorphism experiment is shown in Figure 2.4. SSA of natural snow decreased by a factor of 2.2 from 27.0 m2/kg to 12.2 m2/kg within 30 days. The initial decrease was less pronounced for artificial snow, showing a decrease from 15.9 m2/kg to 12.2 m2/kg during the first month. The development of the SSA in our experiments concur with results from experimental as well as modelling studies on dry metamorphism (Flanner and Zender, 2006; Pinzer et al., 2012; Taillandier et al., 2007). The snow samples taken from the WFJ depth profile showed similar SSA decrease during the first 30 days compared to the natural snow aged in the laboratory at 40 K/m (Figure 2.4). Afterward, the SSA of the WFJ snow pit samples decreased below that of the natural and artificial samples exposed to a controlled temperature gradient.

2.3 Results

Figure 2.3: Qualitative evolution (3D microstructure) of artificial and natural snow under an artificial temperature gradient of 40 K/m imaged by microCT after 0 days (left) and 12 days (right). The dimensions of the observed sub-volume of the measured field of view are 3 x 3 x 0.5 mm3.

The natural snow samples had a lower density (~400 kg/m3) than the artificial snow samples (~600 kg/m3). As the containers were sealed, bulk density remained almost constant during the course of the experiment.

Figure 2.4: SSA evolution of natural (yellow diamonds) and artificial snow (black dots) aged at the artificial temperature gradient of 40 K/m and after 90 days isothermal storage at -20 °C (grey bar, half-filled symbols; the blue squares represent SSA changes in 0, 30, 60, and 90-day old snow from different depths of the snow pit at WFJ. These 4 samples originate from 4 different snowfall events at different stages in natural metamorphism.

2.3 Results

2.3.1.2 Isothermal metamorphism The two samples isothermally stored at -20 °C for 90 days behaved differently than the ones exposed to the temperature gradient: Both artificial and natural snow samples experienced much less change without a temperature gradient, as shown in Figure 2.4. This is consistent with previous studies (Kaempfer et al., 2005; Taillandier et al., 2007). For artificial snow, the decrease in SSA from 15.9 m2/kg to 12.4 m2/kg after 90 days was ~22 %, corresponding to an aging time of ~6-30 days at a gradient of 40 K/m. The SSA of natural snow declined from 27 m2/kg to 19.7 m2/kg after 90 days of isothermal storage; this decrease of ~27 % corresponds to an aging time of only ~3-6 days at a gradient of 40 K/m.

+ 2.3.2 Redistribution of NH4 and other major ions during snow metamorphism

2.3.2.1 Elution experiments

2+ + Figure 5 presents the release profile of Ca and NH4 during the elution of WFJ snow samples exposed to controlled metamorphism for 0 and 30 days. Concentrations of all six investigated major ions

st th + 2+ decreased between the 1 and 10 eluate, as exemplarily shown for NH4 and Ca in Figure 2.5. The concentration in the first eluate, cA, reflects the concentration of impurities that are best accessible by the eluent (Figures 5A and B, left grey bars). Their concentration in the eluate is given by the amount of eluted ions and the volume of eluent. In this work, concentrations were normalized to the bulk concentrations, cBulk, of the original snow sample to allow for direct comparison between the different

+ 2+ ions. Since cBulk of NH4 is about 10 times larger than that of Ca in the natural snow sample (Table 2.1),

+ 2+ Figure 2.5A reveals that the amount of NH4 in the first eluate exceeds that of Ca by this same factor of 10 at the beginning of the metamorphism experiment (0 days). The concentration in the residual elution sample cR reflects the concentration of impurities that are least accessible by the eluent (Figures 2.5A and B, right grey bars).

+ 2+ Figure 2.5: Elution profiles of NH4 (red) and Ca (blue) of the natural snow sample after 0 days (A) and 30 days (B) temp. gradient metamorphism. Concentrations are normalized to the respective bulk concentration. cA th denotes the 1 eluate and cR the molten residual sample. Error bars represent 1  uncertainties of 3 snow samples.

2.3 Results

Generally, the ions with the highest concentration ratio cA/cBulk in the first eluate show the lowest concentration ratio cR/cBulk in the residual rest and vice versa (Figures 2.5A and B, right grey bars). A mass balance could be achieved within an uncertainty of < 40 %. The shape of the release profiles in Figure 2.5, where the majority of the ion load is found in the first few eluates is typical for contaminants that are readily eluted (Grannas et al., 2013). For example, laboratory elution experiments of water- soluble organics with melt water have shown that ¾ of each species is found in the first quarter of total eluates (Meyer et al., 2009a). This shape of the elution curve and in particular the absence of increasing concentration ratio towards the later part of the elution further indicates that release from insoluble aerosol deposits or other processes acting on longer time scales, such as chromatographic effects, can be neglected (Grannas et al., 2013; Mann et al., 2011; Meyer et al., 2009a, 2009b, 2011; Plassmann et al., 2010). Support comes from a study by Hewitt (1989) showing that ice surfaces do not lead to a

- - 2- chromatographic separation of Cl , NO3 , and SO4 in elution experiments. In this work, we use the first eluate and the residual elution sample for further analysis.

+ The relative concentration in the residual elution sample cR/cBulk are smaller than 1 for NH4 and

2+ Ca in both sampling sets of 0 and 30 days (Figure 2.5, right grey bars). Obviously, cR of each, which is given by the amount of ions not accessible to the eluent and the dilution of the residual snow with eluent, is smaller than the respective bulk concentration. To assess the dilution, the mass of residual water and of the snow sample was noted for each experiment. Assuming that the water filling the void space in the residual sample still contains ions (0.2 concentration ratio is a typical value in the 10th eluate, Figure 2.5), a dilution to a concentration ratio of ~0.7-0.8 can be estimated. Clearly, the cR/cBulk

+ 2+ ratio of 0.3 and 0.5 for NH4 and 0.4 and 0.3 for Ca at 0 and 30 days (Figure 2.5), respectively, are significantly lower. Despite the uncertainties in potential melting and refreezing of ice by the eluent (see below), this back-of-the-envelope estimate carefully suggests that the remaining snow after the elution is depleted in both ion concentration relative to their bulk concentrations. Since the accessible ion content was removed during the elution, these depleted ratios confirm qualitatively the ion’s and ice’s mass balance. As the dilution factors despite their large uncertainties are the same for all ions in a specific sample, we will focus on discussing their relative behavior in this work.

+ The relative increase of the NH4 cR/cBulk concentration ratio after 30 days compared to that of

2+ + Ca shows that NH4 is preferable found in the least accessible fraction after metamorphism. The comparison of the relative concentrations cA/cBulk in the first eluate for the samples aged for 0 and 30 days (Figures 2.5A vs 2.5B, left grey bars), show a strong trend as well: Ca2+ became even better

+ accessible after 30 days. On the contrary, the NH4 concentration ratio in the first eluate did not significantly change between day 0 and 30.

2.3 Results

Figure 2.6: Temporal evolution of normalized ion concentrations cA/cBulk for the accessible section (A-C) and cR/cBulk for residual section (D-F). (A, D) controlled metamorphism, artificial snow; (B, E) controlled metamorphism, natural snow; (C, F) natural metamorphism, natural snowpack. Error bars represent 1  uncertainties of three replicated samples. Data points within the grey bars (panel B and E) represent a sample isothermal stored for 90 days at -20 °C.

2.3.2.2 Temporal changes of ion concentrations in first eluate and residual snow

Figure 2.6 shows the temporal trend of the normalized ion concentrations in the first eluate, cA/cBulk

(Figures 2.6A-C), and cR/cBulk in the molten residual sample (Figures 2.6D-F). The 1 uncertainty for samples, where a 3-fold determination was performed, is also shown. The partially large uncertainties result from potential sampling inhomogeneities, alterations in snow metamorphism (between different sites in the metamorphism box or in the field), variations during the elution process and analytical errors.

+ Generally, the NH4 concentration ratio showed a pronounced increase in the residual sample

2- 2+ with time by about a factor of 2-3 (6D,F) and ~8 (6E), whereas SO4 and Ca ratios showed an increase in the first eluate. Depending on the different snow types, we observed varying trends with time for

- - + Cl , F , and Na , the reasons for which will be discussed later:

(a) Controlled metamorphism, artificial snow (Figures 2.6A and D)

+ - - + There was a continuous rise of the NH4 , Cl , F , and Na concentration ratios in the residual snow during

+ - the two months of aging; NH4 and Cl showed the strongest increase. In the first eluate, concentrations ratios of all ions decreased during the first 12 days and afterward there was a strong enhancement in

2- 2+ the SO4 and Ca ratios.

2.3 Results

(b) Controlled metamorphism, natural snow (Figures 2.6B and E)

- + - 2- There was a continuous increase of the F , NH4 , Cl , and SO4 concentration ratios in the residual snow

+ - during the first month; NH4 and F showed the strongest enrichment. Between one and three months, the concentration ratios of the four ions remained rather stable. The normalized concentrations of Na+ and Ca2+ stayed almost constant with time. In the first eluate, we observed declining concentration ratios of all major ions during the first 12 days, followed by increasing values, most pronounced for Na+ and Ca2+, during the first month. Concentration ratios of all ions remained relatively constant after the first 30 days.

+ - NH4 concentrations in the residual sample increased with time, whereas the concentrations of Cl ,

2- + 2+ - SO4 , Na , and Ca did not significantly change. The concentration of F was below the detection limit.

2- 2+ In the first eluate, concentrations of all ions peaked after 30 days. After 90 days, SO4 and Ca had the highest enrichment in the first eluate compared to other ions. Note: the 4 different snow samples (0, 30, 60, 90 days) resulted from 4 different precipitation events and the metamorphism proceeded under unknown conditions.

2.3.2.3 Temporal changes in ion concentrations under isothermal conditions Ion concentration ratios in the first eluate and the residual sample from the isothermally stored reference samples showed that there was little to no redistribution of the ions without temperature gradient compared to the samples of the same age exposed to temperature gradient (Figures 2.6B and E, grey inset). Chemical redistribution after 90 days of isothermal storage corresponds to the change after ~6-12 days at the temperature gradient.

+ 2.3.2.4 Enrichment of NH4 in the residual sample (gradient/isothermal)

+ NH4 shows an overall very high ratio in the residual samples, indicating its low tendency to be expelled from the ice bulk (Figure 2.6 D-F). After 30 days this ratio was significantly higher than that of any other ion except F- in the natural snow.

+ To qualitatively compare the enrichment of NH4 with that of the other ions in the residual sample and to correct for potential bias related to changes in the elution behavior and dilution caused by structural changes of the samples with time, we investigated the ratio of the ion concentration

+ + cR/cBulk (ion) to the NH4 concentration ratio cR/cBulk (NH4 ), as shown in Figure 2.7.

2.4 Discussion

+ Figure 2.7: Concentration ratio of the individual ions in relation to NH4 in the residual sample. Shown are results for (A) controlled metamorphism, artificial snow; (B) controlled metamorphism, natural snow; (C) natural metamorphism, natural snowpack. Data points within the grey inset (panel B) represent a sample isothermally stored for 90 days at -20 °C. Error bars represent 1  uncertainties from analyzing three different samples.

+ The comparison clearly shows that in artificial snow NH4 had the highest ratio in the residual sample

2+ 2- compared to all other ions during the whole experiment, whereas Ca and SO4 revealed always the

+ - 2+ 2- lowest (Figure 2.7A). Furthermore, records of the Na , F and Ca , SO4 ratios agreed well for the whole 60 days period (Figure 2.7A). This is different for natural snow and the natural snowpack, where this “grouping” was not observed. Here, at the beginning of the experiment, the Ca2+ ratio was more

+ enhanced in the residual snow compared to NH4 (Figures 2.7B and 2.7C). This changed after ~12 days in natural snow and ~30 days in the natural snowpack. In natural snow F- in the residual snow sample

+ was higher in relation to NH4 right from the begin - this was even more pronounced with increasing storage time. We discuss this finding in section 2.4.3.2. Changes of the ions’ distribution in the samples stored isothermally for 90 days was insignificant compared to the same age with gradient (Figure 2.7B).

2.3.2.5 Vertical distribution of ions after temperature gradient storage The comparison of the concentration ratio of the ions within the 5 parts (~1.2 cm) of the vertically split snow samples did not show any significant differences. No vertical gradient was formed during the different stages of snow metamorphism.

2.4 Discussion Snow metamorphism alters the ice matrix in which impurities are embedded, and in particular, continues creating new ice surfaces. We debate the processes involved in snow metamorphism in the following two sections. This forms the basis for discussing the driving processes for the relocation of

2.4 Discussion chemical species, especially why the structural and chemical changes were more pronounced during temperature gradient metamorphism compared to isothermal metamorphism.

2.4.1 Structural change under temperature gradient metamorphism A temperature gradient in snow creates a water vapor pressure gradient. This gradient induces a diffusive flux from high-pressure zones (warmer side of ice particles) to low pressure zones (colder side of ice particles) (Pinzer et al., 2012; Sokratov and Maeno, 1997). The net vapor flux is not a continuous flow-through from the bottom to the top of a snowpack. Rather it takes place between adjacent crystals. Yosida et al. (1955) describes this process as “hand to hand transport”. Further process details are described in Pinzer et al. (2012). According to Pinzer et al. (2012) numerical simulations and experiments show that the vapor flux is almost structure-independent. Therefore, a constant temperature gradient leads to a constant water vapor flux. Some earlier studies based on simplified snow geometries concluded that the vapor diffusion coefficient is enhanced (Colbeck, 1993; Sommerfeld et al., 1987), while others (Sokratov and Maeno, 2000; Voitkovskii et al., 1988) concluded that there is no enhancement. A detailed discussion of this issue is given in Pinzer et al. (2012, p. 1042). Over time, the local rearrangements completely renew and thereby transform the structure of the original ice crystals. The lifetime of a particular ice volume is defined by the complete mass turnover and allows calculating its re-crystallization rate. Pinzer et al. (2012) showed a characteristic lifetime of 2-3 days for an ice volume at a gradient of 50 K/m. Accordingly, during our storage period of 90 days at a gradient of 40 K/m, the entire ice structure was continuously sublimated and rebuilt up to 40 times. This continuous rebuild is accompanied by a decrease in SSA (Taillandier et al., 2007), as observed in all experiments (Figure 2.4). However, even though the SSA decrease decelerated during the artificially induced metamorphism experiments (Figure 2.4), the re-crystallization rate did not change due to the constant temperature gradient. Figure 2.4 shows that the initial SSA decrease in the artificial snow was much less pronounced compared to natural snow. This difference is due to the absence of fine crystal structures that would induce rapid SSA reduction during metamorphism. The absence of small structures in artificial snow resulted, on the one hand, from the round shape of the snow particles due to the production method and, on the other hand, from the subsequent sieving (section 2.2.1.1). After 12 days, the artificial and the natural snow showed a comparable pattern: An anisotropic structure was formed from the isotropic starting material as shown in Figure 2.3. Moreover, the SSA values converged to a very similar value within an uncertainty of ~10 %; in agreement to earlier work concluding that the SSA of artificial snow agrees well with that of metamorphosed natural snow (Bartels-Rausch et al., 2004) The samples taken directly from the different depths of the WFJ snow pit had been exposed to varying temperature gradients in the field. Within the temperature profile of a natural snowpack often

2.4 Discussion a diurnal cycle is visible: The lower temperatures at night cool the snow surface, leading to stronger gradients at night than during the day, while warming by the sun during the day leads to an increase in the snow surface temperature which may reverse the gradient in the snowpack. The SSA after 60 and 90 days of natural metamorphism was lower compared to that of the laboratory study. This is caused by the lower density and the higher average temperature of the WFJ site (Kaempfer et al., 2005).

2.4.2 Structural changes under isothermal conditions The development of the SSA under isothermal condition showed only a small change (Figure 2.4, dotted lines). This can be explained by a much lower re-crystallization rate. In absence of a temperature gradient, the re-crystallization is driven by the differences in surface energy (Löwe et al., 2011). There is a natural tendency to minimize those differences by reducing the local curvature of particular ice structures (Kerbrat et al., 2008). Dendritic snow crystals are thereby progressively rounded (Flin et al., 2004). The flux due to the curvature reduction is significantly smaller than the one due to the temperature gradient and therefore, this process is much slower (Kaempfer et al., 2005). Figure 2.2A (last column) clearly shows that the resulting structural changes are relatively small. The images confirm that the structure is still isotropic: After 90 days of isothermal storage, the original structure of the artificial snow was still visible. Only the bonding between the original particles increased. In natural snow, the single crystals were generally preserved and visible. However, these crystals were visibly rounded, as the edged contours had disappeared due to the reduction of the surface energy (Löwe et al., 2011). In contrast to the temperature gradient metamorphism, changes in SSA during isothermal storage can be used as direct measure of the re-crystallization rate: A reduction in SSA of 22 % for artificial snow represents roughly a mass turnover of around 80 % during 90 days. The estimation is based on the conversion of the SSA into equivalent sphere diameters (D=6/SSA) and the resulting change in volume of the spheres. This compares to a turnover of up to 4000 % during temperature gradient metamorphism, corresponding to the 40 re-crystallization cycles during the same time interval. This strong transformation, in which the entire ice structure is altered, forms the basis of our hypothesis that temperature gradient metamorphism has a strong effect on the (re-)distribution of the chemical species contained in snow, whereas less chemical changes are expected under isothermal conditions.

+ 2.4.3 Rearrangement of NH4 and other major ions during snow metamorphism Progressing snow metamorphism in the laboratory and the field led to an enhanced incorporation of

+ NH4 into less accessible ice interior regions. This occurred for all snow types. In the following we will

2.4 Discussion

+ relate structural changes of the snow during metamorphism to the observed locality of NH4 in the snow.

2.4.3.1 Initial chemical characterization of the snow a) Artificial snow:

2+ - - + + 2– The artificial snow was produced by shock freezing a stock solution of Ca , Cl , F , Na , NH4 , and SO4 and stored at -20 °C. The initial location of impurities in ice is a strong function of concentration, species, and freezing conditions (Bartels-Rausch et al., 2014). Generally, since the solubility of impurities is much lower in ice than in water, solutes are largely expelled from the forming ice when freezing from a solution. A second phase at air-ice interfaces, ice-ice interfaces of grain boundaries, or in internal micropockets forms (Section 2.4.3.2.). Interestingly, the analysis of the artificial snow

2+ 2 revealed prominent ion-pairs that persist during the entire metamorphism period. Here, Ca and SO4 ,

+ - + - Na and F , NH4 and Cl are grouped (Figures 2.6A, D and 2.7A). It appears that we have the following salts precipitating during the shock-freezing of the artificial snow: NaF at -3 °C, CaSO4 and NH4Cl at - 15 °C as predicted by salt-water binary phase diagrams (Lewis et al., 2010; Melnikov, 1997; Purdon and Slater, 1946; Stefan-Kharicha et al., 2018; Thomas and Dieckmann, 2010). These ions-pairs may stay close also during the metamorphism experiments. The location of the second phase and other reservoirs in the snow will be discussed below. b) Natural snow: The natural snow used for the 90-day metamorphism experiment in the laboratory contained mainly anthropogenically derived NH4NO3 and (NH4)2SO4, explaining the majority of the ion balance

2+ + (Table 2.1). The origin of the other ions is less certain: Ca (dust, CaCO3, CaSO4), Na (sea salt NaCl,

- - NaCO3), Cl (NaCl, NH4Cl), F (NaF, NH4F). In natural snow, the initial location of an ion in the snow structure is determined by scavenging processes in the atmosphere. Aerosol particles containing major ions can be incorporated in the snow crystal matrix, by acting as ice nuclei. On the other hand, direct impaction of aerosol particles on snow crystals or accumulation of supercooled water droplets onto ice crystal surfaces (riming) may lead to the enrichment of these species at the edges of the crystal

+ (Mosimann et al., 2002; Pruppacher and Klett, 2004). NH4 is less likely to be within the ice nuclei (contrary to e.g. Ca2+), but rather scavenged by the latter two processes. With progressing snow metamorphism, there is a strong relocation of the chemical compounds. Throughout our elution experiment, we monitored this change of ion location by studying their accessibility during elution with water.

2.4 Discussion

2.4.3.2 First eluate vs residual sample: What do we learn with respect to accessibility? Here, we differentiate between accessibility and preservation based on the elution behavior. The accessible fraction of the water-soluble ions are those that are eluted from the snow sample in the first 3 ml eluate after a saturation time in the order of 1 to 2 h (Table 2.3). Preserved are those ions that are not washed out even after 2-3 h of complete elution. Ions that are present at the ice surface are directly dissolved with the first eluent. Further, ions in grain boundaries are well known for their relatively fast diffusion and thus, effective transport of impurities from the interior of ice to the surface (Huthwelker et al., 2006). A back-of-the-envelope calculation based on the water self-diffusion in grain boundaries of 7  10-13 m2/s (Lu et al., 2009) suggests that within 1 h of saturation time during our elution experiments (Table 2.3), impurities could migrate 100 µm through grain boundaries. To the best of our knowledge, the diffusion of ions from grain boundaries into interstitial water has not been investigated or compared to the release from the ice interface. Concerning gas - ice exchange processes, Bartels-Rausch et al. (2004, 2013) has shown that grain boundaries have no significant impact on the adsorption of volatile trace gases. Extrapolation to the elution, i.e. the transfer from ice to water, is hampered by the very different time scales of the investigations and by the very distinct behavior of strong acids at the ice-air interface (Kong et al., 2017; Zimmermann et al., 2016). Given these uncertainties, we refrain from attributing the source of the accessible fraction of ions. Despite this uncertainty, the important point is that ions from either reservoir are accessible to infiltrating water in natural and artificial snow.

2+ 2- + The fact that Ca , SO4 , and Na showed higher relative concentrations in the first eluate compared to other ions reflects the ease of their accessibility. A large fraction of these ions is either located at the air-ice interface or the ice-ice interface (within 100 µm of the surface). This finding is in

+ 2+ 2- line with their low solubility in ice. In an analysis of ice core data, solubilities of Na , Ca , and SO4 in the order of 2-10 µg/l have been suggested (Eichler et al., 2001). Thus, the low solubility of the latter 3 ions in ice and taken that we worked with concentrations of 500-1000 µg/l in artificial snow and 20- 1000 µg/l in natural snow favors their presence at interfaces. In natural snow, an accumulation of Ca2+ in the residual part is visible on day 0 (Figures 2.6 and 2.7). Ca2+ seems to have been exceptionally well protected from the eluent at the beginning of the experiment. This may be an indication that Ca2+ acted as ice nuclei (Section 2.4.3.1.). This finding is in good agreement with observations that mineral dust particles preferably act as ice nuclei (Szyrmer and Zawadzki, 1997).

+ - - In general, NH4 , F , and Cl were found at very low concentration ratios in the first eluate, but with the highest concentration ratios in the residual sample. This general finding is consistent with their high solubility in ice. Phase diagrams of solid solutions are sparse, and we currently lack predictive capabilities of solute concentration in ice. Based on the atomic radii of N and F being comparable to

+ - that of an O atom, NH4 and F are generally thought to substituting water molecules, in the ice lattice

2.4 Discussion

(Hobbs, 1974; Pruppacher and Klett, 2010). Besides substitution, solutes may also fill interstitial space

+ - in crystals. Hobbs (1974) gives a solubility of about 2 g/l for NH4 and F . As detailed by Hobbs and others (Dominé et al., 2008; Huthwelker et al., 2006) such solubility measurements are delicate and the apparent solubility can be significantly enhanced in presence of grain boundaries or if the thermodynamic equilibrium is not established, or if micropockets forms. We therefore caution when referring to this value. For HCl a solid solution solubility of 3 mg/l was derived by Thibert and Dominé (1997) in thermodynamically carefully controlled experiments on single crystalline ice. Whether this solubility reflects that of Cl- or of HCl is open to debate, as is the mechanism. However, concentrations

+ - - of NH4 , F , and Cl in our snow samples are well below the solubility limits reported in the literature and suggests that they generally could be quantitatively incorporated into the ice structure. Furthermore, micropockets can host impurities in ice. They are potentially present in shock- frozen ice (Hullar and Anastasio, 2016), such as our artificial snow, as well as in ice cores (Eichler et al., 2017, 2019). In principle, micropockets may be composed of salt deposits or liquid brine. In ice cores, for example, sulfate deposits appear to be common (Eichler et al., 2017, 2019). This might explain why we find sulfate at similar concentration ratio in the residual samples of natural snow as chloride, for which a very high solubility in ice crystals was found. Detailed laboratory studies have revealed that the number of micropockets in ice frozen from the liquid critically depends on the freezing rate and conditions as well as on the concentration of salt (Hullar and Anastasio, 2016; McFall et al., 2018). In shock-frozen ice, evidence was found for the presence of solutes in micropockets with a radius below 1 µm. In our shock frozen samples, we used significantly lower concentrations of ions by about a factor of 1000, also to minimize the occurrence of micropockets. However, with a total ion load of 4 10-6 mol, one can estimate a total liquid volume of 5 µl in the artificial snow. This would correspond to thousands of 1 µm (radius) micropockets in each shock-frozen ice sphere of the artificial snow. For this back-of- the-envelope estimate, a typical concentration of brine that is in equilibrium with ice at -5 °C of 1.4 M was used. In summary, we have discussed solid solution, interfaces, micropockets and condensation nuclei as potential location of impurities in our snow samples. Despite the inability to clearly identify the location in our samples, we argued that impurities at the air-ice interface and to some extent at the ice-ice interface would represent easily accessible compartments. Whether the impurities form patches, micropockets or are present as induvial molecules there, is beyond the scope of this work. Solid solutions and internal grain micropockets represent reservoirs where the solutes are rather inaccessible.

2.4 Discussion

2.4.3.3 Impact of structural changes during metamorphism on the distribution of ions During the first ~12 days of the snow metamorphism, the concentration ratio of all accessible ions in

+ - - the first eluate decreased. After ~12 days, NH4 , Cl , and F concentrations were gradually enhanced in

2- 2+ the less accessible interior of the ice matrix, whereas SO4 and Ca accumulated at the ice surface. Here we investigate how those findings relate to structural changes. The general decrease in the concentration ratios of all ions during the first 12 days was concurrent with a reduction in the SSA (Figure 2.4). However, the SSA decline during the first 12 days was much more pronounced in the natural snow (drop from 27.0 m2/kg to 15.5 m2/kg) than in the artificial snow (drop from 15.9 m2/kg to 13.5 m2/kg) (Section 2.3.1.1). Changes in the accessible fraction cA/cBulk of the ions in artificial and natural snow were rather similar in the meantime (decrease by a factor of ~1.2-2 between day 0 and 12 (Figures 2.6A and B). Therefore, we conclude that changes in the SSA did not directly influence the accessibility of ions during the elution. This is not surprising. The surface area does indeed play a crucial role for exchange processes such as adsorption from the gas phase (Bartels-Rausch et al., 2002; Kerbrat et al., 2008). In this study, however, the amount of accessible ions is not the result of a dynamic partitioning equilibrium. Rather, the absence of the SSA impact again confirms that the elution is not limited by the interface-water flux during elution. In the first few days at best, the grain boundaries could play a role. Riche et al. (2012) demonstrated a loss of internal grain boundaries in artificial snow of up to 40 % at isothermal storage at a temperature of -5 °C. In our experiments, we expect even faster changes due to the high water vapor turnaround in temperature gradient metamorphism. With a gradient, grain boundaries move with the growing ice, eventually forming ice-ice interfaces when two growing areas meet (Figure 2.8). How impurities respond to this is essentially unknown. However, the loss of internal grain boundaries during the initial stage of metamorphism would lead to a less effective transport of impurities from the interior of the ice to accessible surfaces and thus explain the decreasing concentrations of the ions.

Another hypothesis, that the initial decrease of cA could be the result of changes in the accessibility due to changes in residence time of the eluent, was not confirmed. In artificial snow, the shortest saturation time is observed for day 0 and day 3 (Table 2.3). This can be explained by the poorly connected structure at the beginning of the experiment (Section 2.3.1.1) that led to a higher permeability. Thus, we expected a lower uptake of ions from the ice surface, as well as less exchange with potential pores and consequently a lower cA. For day 6 and 12 the saturation time increased significantly without increasing cA. Similarly, in the case of natural snow there was no correlation between residence time and cA; during the first 12 days, the longest saturation time was contrasted with the smallest cA. We conclude that the elution times have no influence on the results of the elution experiments. This is an important finding with regard to a further process that could influence the elution; wet snow metamorphism. As soon as there is liquid water present in the snowpack, new

2.4 Discussion processes of transformation and particle growth occur (Brun, 1989). From the lack of correlation, we conclude that possible wet snow metamorphism processes play no significant role during the elution procedure. This finding is confirmed by the elution results of the isothermal samples (Figures 2.6B and E). Although these samples occasionally have the longest elution times (Table 2.3), there is barely any change in the ion distribution. To conclude the most likely explanation for the observed decrease in the concentration ratios of all accessible ions in the first eluate during the initial ~12 days of the snow metamorphism is its relation to changes in the grain boundary content - and the re-location of impurities initially hosted there - driven by the continual complete re-structuring of the ice matrix during the temperature gradient metamorphism. In the second phase of the performed elution experiment, between day 12 and day 30, despite

2- 2+ negligible changes in the SSA, there was a significant enhancement of SO4 , Ca concentration ratios compared to the other ions at the accessible ice surface in artificial and natural snow samples. At the

+ - - same time, NH4 , Cl , and F ratios increased in the ice interior. Diffusion might act on these timescales to redistribute impurities particularly along grain- boundaries. Yet, diffusion is driven by concentration gradients (for example, between the interfaces and the interior). Due to the lack of knowledge about diffusion constants, we could only speculate why

+ 2+ this should lead to a separation of NH4 and Ca . More importantly, diffusion as the main process can be excluded by comparing these results to those of the isothermal experiments, showing no such separation. The clear difference to the isothermal data strongly indicates that the observed distribution manifests themselves during the temperature gradient metamorphism. Furthermore Hörhold et al. (2012) suggest that the densification of firn is enhanced by impurities. The preferred occurrence of Ca2+ at the accessible ice surface is in support of this hypothesis, as Ca2+ could enhance grain-boundary mobility during the densification process. Volatile solutes can migrate with the water vapor from warm to cold areas of snow. Wagnon et al. (1999) have

- - - detected a displacement of Cl , F and NO3 in firn layers at high sulfuric acid concentrations. Indeed, the sorption to growing ice has also been found to be enhanced significantly for strong, volatile acids (De Angelis and Legrand, 1994; Dominé et al., 1995; Domine and Rauzy, 2004; Kärcher et al., 2009; Kärcher and Basko, 2004; Ullerstam and Abbatt, 2005). To test the relevance of this process, the concentration of ions was determined vertically along the sample. These vertical concentration profiles (Section 2.3.2.5) give no indication of an enrichment along the vapor flux. This is in line with the initial presence of non-volatile compounds in our work (such as NH4NO3 and (NH4)2SO4 in the natural snow (Section 2.4.3.1.), and a low overall acidity, preventing the formation of volatile species. If therefore only pure water vapor sublimates during the metamorphism, the ions must remain at the interfaces of the remaining ice. Cragin et al. (1996) sketches a similar mechanism. However, this was based on

2.4 Discussion the false assumption that large snow crystals continue to grow at the expense of small ones; although older and younger particles and ice surfaces exist in parallel, the maximum lifetime of an ice volume is limited to about 3 days (Section 2.4.1). The noteworthy result is thus that the rearrangement continues over much longer timescales and is not related to changes in the SSA. Indeed, Pinzer et al. (2012) have shown that SSA is only a poor indicator of the dynamics in the snow during metamorphism and therefore not well suited to characterize the redistribution of embedded impurities. The change in SSA is only in the first phase of metamorphism related to vapor flux (and therefore, re-crystallization) because a quasi-steady state is reached with respect to the SSA, but not for the re-crystallization during a constant temperature gradient driven water vapor flux.

Figure 2.8: Visualization of redistribution of impurities (red dots, green crosses) during temperature gradient metamorphism. The reconstruction of the ice matrix is driven by the vapor transport between neighboring ice interfaces (white arrows). Ice mass sublimates (yellow area) and deposits at the nearest cold interface leading to the growth of ice there (dark blue area). Impurities from the sublimating ice remain at the newly formed ice interface. The specific tendency of impurities to be incorporated into the growing ice (red dots show higher tendency than green crosses) ultimately drives the observed changes in accessibility during temperature gradient metamorphism.

At first view, the tendency of the ions to be hosted in the interior of ice - either as solid solution or in micropockets - goes along with their solubility in ice. For species where the solubility is very likely

2+ 2- exceeded such as Ca and SO4 , we find a larger part at the interfaces (Figure 2.8, green crosses). The explanation by solubility, however, does not go far enough to explain the observed relocation of ions at lower concentration than their solubility. If solubility would be the only argument, there should be

+ - - no difference and no change with time between the ions NH4 , F , Cl , since solubility limit in ice was not reached for any of them. We therefore propose a more complex picture describing the fate of chemicals during temperature gradient metamorphism. It is based on the distribution equilibria between solid solution, partitioning to interfaces, and into micropockets. First it is important to note that ion redistribution in the isothermally stored samples was insignificant compared to the temperature gradient samples (Figure 2.7B), giving strong experimental evidence for the pivotal role of the re-crystallization of the ice matrix. Therefore, exchange between the solute’s reservoirs is

2.4 Discussion limited at isothermal conditions. Secondly, change in the distribution seems to continue for more than 60 days. This timescale is much longer than the lifetime of the ice matrix and the question arises why the preferential distribution of solutes is not reached earlier, when the ice matrix is completely reformed for the first time: Clearly, the evaporating ice leaves the impurities in whatever compartment behind irrespective of the specific compartment (Section 2.4.3.2.), remaining at the newly formed interfaces (Figure 2.8). Ion-free ice grows at colder neighboring areas on interfaces that might hold ions with an ice growth rate of roughly 2 nm/s. For this estimation we assumed an optically equivalent diameter D = 0.4 mm (D = 6/SSA) and a lifetime of an imaginary ice particle of about 60 h (section 2.4.1). The ice growth rate has thus the same order of magnitude as the migration distance based on typical

16 14 2 diffusivities in ice that are in the range of 7  10 to 2  10 m /s for HCl and HNO3 (Thibert and

+ Dominé, 1997, 1998). Thus NH4 and any impurity with both a diffusivity in that range and beneficial energetics of stabilization in the ice, can - driven by the strong concentration gradient - migrate into the newly formed ice forming a solid solution there. Taken the uncertainties in diffusivity, this estimate servers as rough upper limit. For ions that have a lower tendency to enter and be stabilized in the ice a smaller fraction enters the ice and a larger part remains at their initial location. With proceeding metamorphism, this location will end as air-ice interface, i.e. the location of these ions will more and more shift towards the accessible interfaces. Our data indicate that ions with a high solubility in ice also have a high tendency to be efficiently incorporated into growing ice (Figure 2.8, red dots), presumably because the same physical properties are at play in both processes.

+ For the qualitative comparison of the NH4 incorporation into the ice interior regions in relation to that of the other 5 ions during snow metamorphism, we formed enrichment sequences based on

+ the ratio of the normalized ion concentration to the normalized NH4 concentration in the residual snow after 90 days (Figure 2.7). We obtained the following enrichment sequences for the 3 different experiments:

+ - - + 2- 2+ (a) lab-controlled metamorphism, artificial snow: NH4 > Cl > F ~ Na > SO4 ~ Ca

- + - 2- 2+ + (b) lab-controlled metamorphism, WFJ snow: F > NH4 > Cl ~ SO4 ~ Ca > Na

+ - + 2- 2+ - (c) natural metamorphism, WFJ snow: NH4 > Cl ~ Na > SO4 ~ Ca (F was below detection limit)

+ After 90 days of laboratory-controlled and natural snow metamorphism, NH4 was always one of the two ions most enriched in the less accessible ice interior. The differences in the obtained enrichment sequences and thus the incorporation of the ions into the growing ice can be explained by different factors. The absolute concentration of ions plays a major role in the distribution of impurities in ice, as has been observed for trace elements in natural snow (Avak et al., 2019) and in earlier laboratory studies (Bartels-Rausch et al., 2014; Workman and Reynolds, 1950). For example, F- concentrations in the artificial snow are two orders of magnitude

2.4 Discussion higher than that of the WFJ snow samples (Table 2.1). At higher concentrations, solubility might have been exceeded, explaining the less favorable F- incorporation in the less accessible ice interior in artificial compared to natural snow. Furthermore, chemical composition has a strong influence on the

+ - - solubility of certain ions in ice. As an example, NH4 enhances the solubility of F and Cl in ice (Gross and Svec, 1997).

2.4.3.4 Environmental implications Our findings can directly be applied to interpret results of elution studies in the field. Climate warming induces melt processes significantly influencing the preservation of atmospheric pollutants in high- altitude snowpacks and glacier ice. Understanding the post-depositional behavior of different compounds is crucial for future reconstruction of past environmental conditions from these natural

+ archives. A variety of studies has investigated the elution behavior of NH4 and other major ions from melting snowpacks and glacier ice. Table 2.4 presents a compilation of published elution sequences.

+ This compilation contains only sites where NH4 was studied together with other major ions. Furthermore, for better clarity, the listed sequences contain only the 6 major ions investigated in our study, even though further ions may be available in some of the works.

+ - The compilation shows that in all elution studies (except one) NH4 is together with Cl the least mobile

+ ion. Our study allows us, for the first time, to explicitly relate the preservation of NH4 at various sites

+ affected by melting around the world to the burial of NH4 into the less accessible ice interior during

2- 2+ + snow metamorphism. Furthermore, the observed higher mobility of SO4 and Ca compared to NH4 in the elution studies (Table 2.4) can be clearly related to a stronger gradual enrichment of these two ions at the accessible ice surface during snow metamorphism, favoring a meltwater relocation. The obvious differences between the published elution sequences might be explained by the different metamorphism stages and temperature gradients of the investigated snowpack, as well as varying chemical composition, different sources and concentration levels, and acid input (Brimblecombe et al., 1987, 1988; Cragin et al., 1996; Gross and Svec, 1997; WU et al., 2018). Another factor is the partial occurrence of wet snow metamorphism in some study areas compared to the assumed pure dry metamorphism investigated in our work (Section 2.4.3.3). However, despite the dissimilarities

+ between the different study sites, the general low mobility of NH4 due to the gradual in the less

+ accessible ice interior indicates that NH4 is one of the best-suited environmental proxies even in snowpacks and ice cores affected by melting. The establishment of such meltwater-resistant proxies is particularly relevant as many high-mountain glaciers worldwide are retreating or are in danger of melting.

2.5 Conclusion

Table 2.4: Compilation of published elution sequences.

# Authors Location Sequence

2- + 2+ + − 1 Brimblecombe et al. (1985) Cairn Gorm Mt., Scotland SO4 > NH4 > Ca > Na > Cl

2– 2+ + + − − 2 Eichler et al. (2001) Upper Grenzgletscher, Swiss Alps SO4 > Ca > Na > NH4 ~ F > Cl

2– 2+ + – + 3 Li et al. (2006) Urumqui glacier No. 1, Tian Shan SO4 > Ca > Na > Cl > NH4

2– 2+ + – + 4 Virkkunen et al. (2007) Lomonosovfonna, Svalbard SO4 ~ Ca > Na ~ Cl > NH4

2- 2+ − + − + 5 Ginot et al. (2010) Chimborazo, Ecuador SO4 > Ca > F > Na > Cl > NH4

2– 2+ + – + 6 You et al. (2015) Urumqi Glacier No. 1, Tian Shan SO4 > Ca ~ Na > Cl > NH4

2+ 2− + + − 7 Wang et al. (2018) Baishui Glacier No. 1, Tibetan Plateau Ca > SO4 > NH4 > Na > Cl

2- 2+ + - + 8 Avak et al. (2019) Weissfluhjoch, Swiss Alps SO4 ~ Ca > Na ~ Cl > NH4

2.5 Conclusion

+ 2+ + – – In this study, we investigated the redistribution of NH4 and five other major ions (Ca , Na , Cl , F , and

2– SO4 ) during dry snow metamorphism of different snow types, i.e. shock-frozen artificial snow and natural snow. Our results show that temperature gradient snow metamorphism in the laboratory and

+ in the field leads to the incorporation of NH4 into less accessible ice interior regions. Exposed to a gradient of 40 K/m, the complete ice matrix of every sample was continuously rebuilt within two to three days, induced by the water vapor flux. We propose that the continuous reconstruction drives

+ - - the selective incorporation of ions with a high solubility in ice (NH4 , F , and Cl ) into interior regions

2+ 2- and the rejection of others (including Ca , SO4 ) to the exterior (air-ice or ice-ice interface). This distribution of the ions results from the specific balance of ice growth rate, diffusivity, chemical composition, and the tendency to enter the ice matrix. The absence of a relation to changes in the specific surface area confirms that this parameter is not a valid indicator of water vapor fluxes in temperature gradient metamorphism. During isothermal snow metamorphism, the chemical rearrangement was not significant. This can be explained by the low re-crystallization rate, which was two orders of magnitude lower than in the experiments with a temperature gradient of 40 K/m. With

+ progressing snow metamorphism, NH4 was always one of the two ions most enriched in the interior ice regions relative to the total amount. The comparability of our results for all studied snow

+ types suggests that the redistribution of NH4 in metamorphosing snow is independent on the snow type and thus, the findings can be widely applied to different snowpacks worldwide. This work

+ explicitly links the observed preservation of NH4 at a number of melt-affected study sites to the

+ incorporation into the ice interior during snow metamorphism. Accordingly, NH4 is one of the best- suited environmental proxies in snowpacks, firn and ice cores affected by melting.

Acknowledgements We are grateful for the technical assistance of Matthias Jaggi (SLF) and Mario Birrer (PSI). We would also like to thank Margret Matzl for her help in evaluating the microCT data. And we thank Bettina Richter for the help with Python. We greatly appreciate the comments and suggestions of the two anonymous referees, which helped to improve the clarity of the manuscript.

References Chapter 2

References Abbatt, J. P. D. D., Thomas, J. L., Abrahamsson, K., Boxe, C., Granfors, A., Jones, A. E., et al. (2012). Halogen activation via interactions with environmental ice and snow in the polar lower troposphere and other regions. Atmos. Chem. Phys. 12, 6237–6271. doi:10.5194/acp-12-6237- 2012. Avak, S. E., Trachsel, J. C., Edebeli, J., Brütsch, S., Bartels‐Rausch, T., Schneebeli, M., et al. (2019). Melt‐ Induced Fractionation of Major Ions and Trace Elements in an Alpine Snowpack. J. Geophys. Res. Earth Surf., 2019JF005026. doi:10.1029/2019JF005026. Baltensperger, U., Schwikowski, M., Gäggeler, H. W., Jost, D. T., Beer, J., Siegenthaler, U., et al. (1993). Transfer of atmospheric constituents into an alpine snow field. Atmos. Environ. Part A, Gen. Top. 27, 1881–1890. doi:10.1016/0960-1686(93)90293-8. Barret, M., Domine, F., Houdier, S., Gallet, J. C., Weibring, P., Walega, J., et al. (2011a). Formaldehyde in the Alaskan Arctic snowpack: Partitioning and physical processes involved in air-snow exchanges. J. Geophys. Res. Atmos. 116, D00R03. doi:10.1029/2011JD016038. Barret, M., Houdier, S., and Domine, F. (2011b). Thermodynamics of the formaldehyde-water and formaldehyde-ice systems for atmospheric applications. J. Phys. Chem. A 115, 307–317. doi:10.1021/jp108907u. Bartels-Rausch, T., Eichler, B., Zimmermann, P., Gäggeler, H. W., and Ammann, M. (2002). The adsorption of nitrogen oxides on crystalline ice. Atmos. Chem. Phys. 2, 431–468. doi:10.5194/acpd-2-431-2002. Bartels-Rausch, T., Guimbaud, C., Gäggeler, H. W., and Ammann, M. (2004). The partitioning of acetone to different types of ice and snow between 198 and 223 K. Geophys. Res. Lett. 31, L16110. doi:10.1029/2004GL020070. Bartels-Rausch, T., Huthwelker, T., Jöri, M., Gäggeler, H., and Ammann, M. (2008). Interaction of gaseous elemental mercury with snow surfaces: Laboratoryinvestigation. Environ. Res. Lett. 3. doi:10.1088/1748-9326/3/4/045009. Bartels-Rausch, T., Jacobi, H. W., Kahan, T. F., Thomas, J. L., Thomson, E. S., Abbatt, J. P. D., et al. (2014). A review of air-ice chemical and physical interactions (AICI): Liquids, quasi-liquids, and solids in snow. Atmos. Chem. Phys. 14, 1587–1633. doi:10.5194/acp-14-1587-2014. Bartels-Rausch, T., Wren, S. N., Schreiber, S., Riche, F., Schneebeli, M., and Ammann, M. (2013). Diffusion of volatile organics through porous snow: Impact of surface adsorption and grain boundaries. Atmos. Chem. Phys. 13, 6727–6739. doi:10.5194/acp-13-6727-2013. Brimblecombe, P., Clegg, S. L., Davies, T. D., Shooter, D., and Tranter, M. (1987). Observations of the preferential loss of major ions from melting snow and laboratory ice. Water Res. 21, 1279–1286. doi:10.1016/0043-1354(87)90181-3. Brimblecombe, P., Clegg, S. L., Davies, T. D., Shooter, D., and Tranter, M. (1988). The loss of halide and sulphate ions from melting ice. Water Res. 22, 693–700. doi:10.1016/0043-1354(88)90180-7. Brimblecombe, P., Tranter, M., Abrahams, P. W., Blackwood, I., Davies, T. D., and Vincent, C. E. (1985). Relocation and Preferential Elution of Acidic Solute through the Snowpack of a Small, Remote, High-Altitude Scottish Catchment. Ann. Glaciol. 7, 141–147. doi:10.3189/S0260305500006066. Brun, E. (1989). Investigation on Wet-Snow Metamorphism in Respect of Liquid-Water Content. Ann. Glaciol. 13, 22–26. doi:10.3189/S0260305500007576.

References Chapter 2

Calonne, N., Richter, B., Löwe, H., Cetti, C., Judith, Herwijnen, A. Van, et al. (2019). The RHOSSA campaign: Monitoring the seasonal evolution of an alpine snowpack up to daily resolution. in prep. Colbeck, S. C. (1993). The vapor diffusion coefficient for snow. Water Resour. Res. 29, 109–115. doi:10.1029/92WR02301. Cragin, J. H., Hewitt, A. D., and Colbeck, S. C. (1996). Grain-scale mechanisms influencing the elution of ions from snow. Atmos. Environ. 30, 119–127. doi:10.1016/1352-2310(95)00232-N. Dansgaard, W., Johnsen, S. J., and Clausen, H. B. (1973). Stable isotope glaciology. Meddelelser Om Gronl. 197, 1–54. De Angelis, M., and Legrand, M. (1994). Origins and variations of fluoride in Greenland precipitation. J. Geophys. Res. 99, 1157. doi:10.1029/93JD02660. Dominé, F., Albert, M., Huthwelker, T., Jacobi, H.-W. W., Kokhanovsky, A. A., Lehning, M., et al. (2008). Snow physics as relevant to snow photochemistry. Atmos. Chem. Phys. 8, 171–208. doi:10.5194/acp-8-171-2008. Domine, F., and Rauzy, C. (2004). Influence of the ice growth rate on the incorporation of gaseous HCl. Atmos. Chem. Phys. 4, 2513–2519. doi:10.5194/acp-4-2513-2004. Dominé, F., and Shepson, P. B. (2002). Air-snow interactions and atmospheric chemistry. Science 297, 1506–1510. doi:10.1126/science.1074610. Dominé, F., Thibert, E., Silvente, E., Legrand, M., and Jaffrezo, J.-L. (1995). Determining past atmospheric HCl mixing ratios from ice core analyses. J. Atmos. Chem. 21, 165–186. doi:10.1007/BF00696579. Döscher, A., Gäggeler, H. W., Schotterer, U., and Schwikowski, M. (1996). A historical record of ammonium concentrations from a glacier in the Alps. Geophys. Res. Lett. 23, 2741–2744. doi:10.1029/96GL02615. Eichler, A., Brütsch, S., Olivier, S., Papina, T., and Schwikowski, M. (2009). A 750 year ice core record of past biogenic emissions from Siberian boreal forests. Geophys. Res. Lett. 36, L18813. doi:10.1029/2009GL038807. Eichler, A., Schwikowski, M., and Gäggeler, H. W. (2001). Meltwater-induced relocation of chemical species in Alpine firn. Tellus, Ser. B Chem. Phys. Meteorol. 53, 192–203. doi:10.3402/tellusb.v53i2.16575. Eichler, A., Tinner, W., Brütsch, S., Olivier, S., Papina, T., and Schwikowski, M. (2011). An ice-core based history of Siberian forest fires since AD 1250. Quat. Sci. Rev. 30, 1027–1034. doi:10.1016/j.quascirev.2011.02.007. Eichler, J., Kleitz, I., Bayer-Giraldi, M., Jansen, D., Kipfstuhl, S., Shigeyama, W., et al. (2017). Location and distribution of micro-inclusions in the EDML and NEEM ice cores using optical microscopy and in situ Raman spectroscopy. Cryosphere 11, 1075–1090. doi:10.5194/tc-11-1075-2017. Eichler, J., Weikusat, C., Wegner, A., Twarloh, B., Behrens, M., Fischer, H., et al. (2019). Impurity Analysis and Microstructure Along the Climatic Transition From MIS 6 Into 5e in the EDML Ice Core Using Cryo-Raman Microscopy. Front. Earth Sci. 7, 20. doi:10.3389/feart.2019.00020. Flanner, M. G., and Zender, C. S. (2006). Linking snowpack microphysics and albedo evolution. J. Geophys. Res. 111, D12208. doi:10.1029/2005JD006834. Flin, F., Brzoska, J. B., Lesaffre, B., Coléou, C., and Pieritz, R. A. (2004). Three-dimensional geometric measurements of snow microstructural evolution under isothermal conditions. Ann. Glaciol. 38, 39–44. doi:10.3189/172756404781814942.

References Chapter 2

Fuhrer, K., Neftel, A., Anklin, M., and Maggi, V. (1993). Continuous measurements of hydrogen peroxide, formaldehyde, calcium and ammonium concentrations along the new grip ice core from summit, Central Greenland. Atmos. Environ. Part A, Gen. Top. 27, 1873–1880. doi:10.1016/0960-1686(93)90292-7. Fuhrer, K., Neftel, A., Anklin, M., Staffelbach, T., and Legrand, M. (1996). High-resolution ammonium ice core record covering a complete glacial-interglacial cycle. J. Geophys. Res. Atmos. 101, 4147– 4164. doi:10.1029/95JD02903. Ginot, P., Schotterer, U., Stichler, W., Godoi, M. A., Francou, B., and Schwikowski, M. (2010). Influence of the Tungurahua eruption on the ice core records of Chimborazo, Ecuador. Cryosphere 4, 561– 568. doi:10.5194/tc-4-561-2010. Grannas, A. M., Bogdal, C., Hageman, K. J., Halsall, C., Harner, T., Hung, H., et al. (2013). The role of the global cryosphere in the fate of organic contaminants. 13, 3271–3305. doi:10.5194/acp-13- 3271-2013. Grannas, A. M., Jones, A. E., Dibb, J., Ammann, M., Anastasio, C., Beine, H. J., et al. (2007). An overview of snow photochemistry: Evidence, mechanisms and impacts. Atmos. Chem. Phys. 7, 4329– 4373. doi:10.5194/acp-7-4329-2007. Gross, G. W., and Svec, R. K. (1997). Effect of ammonium on anion uptake and dielectric relaxation in laboratory-grown ice columns. J. Phys. Chem. B 101, 6282–6284. doi:10.1021/jp963213c. Hagenmuller, P., Chambon, G., Flin, F., Morin, S., and Naaim, M. (2014). Snow as a granular material: assessment of a new grain segmentation algorithm. Granul. Matter 16, 421–432. doi:10.1007/s10035-014-0503-7. Hewitt, A. D., Cragin, J. H., and Colbeck, S. C. (1989). Does snow have ion chromatographic properties? in Proceedings of the 46th Annual Eastern Snow Conference (Quebec City), 165–171. Hewitt, A. D., Cragin, J. H., and Colbeck, S. C. (1991). Effects of crystal metamorphosis on the elution of chemical species from snow. Proc. 48th East. Snow Conf. Guelph, Ontario. Hobbs, P. (1974). Ice physics. Oxford, Clarendon Press, 837 pp. Hoog, I., Mitra, S. K., Diehl, K., and Borrmann, S. (2007). Laboratory studies about the interaction of ammonia with ice crystals at temperatures between 0 and −20°C. J. Atmos. Chem. 57, 73–84. doi:10.1007/s10874-007-9063-0. Hörhold, M. W., Laepple, T., Freitag, J., Bigler, M., Fischer, H., and Kipfstuhl, S. (2012). On the impact of impurities on the densification of polar firn. Earth Planet. Sci. Lett. 325–326, 93–99. doi:10.1016/j.epsl.2011.12.022. Hullar, T., and Anastasio, C. (2016). Direct visualization of solute locations in laboratory ice samples. Cryosph. 10, 2057–2068. doi:10.5194/tc-10-2057-2016. Huthwelker, T., Ammann, M., and Peter, T. (2006). The uptake of acidic gases on ice. Chem. Rev. 106, 1375–1444. doi:10.1021/cr020506v. Kaempfer, T. U., Schneebeli, M., and Sokratov, S. A. (2005). A microstructural approach to model heat transfer in snow. Geophys. Res. Lett. 32, 1–5. doi:10.1029/2005GL023873. Kahan, T. F., Wren, S. N., and Donaldson, D. J. (2014). A pinch of salt is all it takes: Chemistry at the frozen water surface. Acc. Chem. Res. doi:10.1021/ar5000715. Kärcher, B., Abbatt, J. P. D., Cox, R. A., Popp, P. J., and Voigt, C. (2009). Trapping of trace gases by growing ice surfaces including surface-saturated adsorption. J. Geophys. Res. 114, D13306. doi:10.1029/2009JD011857. Kärcher, B., and Basko, M. M. (2004). Trapping of trace gases in growing ice crystals. J. Geophys. Res. Atmos. 109, n/a-n/a. doi:10.1029/2004JD005254.

References Chapter 2

Kaspari, S., Mayewski, P., Kang, S., Sneed, S., Hou, S., Hooke, R., et al. (2007). Reduction in northward incursions of the South Asian monsoon since ∼1400 AD inferred from a Mt. Everest ice core. Geophys. Res. Lett. 34. doi:10.1029/2007GL030440. Kellerhals, T., Brütsch, S., Sigl, M., Knüsel, S., Gäggeler, H. W., and Schwikowski, M. (2010). Ammonium concentration in ice cores: A new proxy for regional temperature reconstruction? J. Geophys. Res. Atmos. 115, D16123. doi:10.1029/2009JD012603. Kerbrat, M., Huthwelker, T., Gäggeler, H. W., and Ammann, M. (2010). Interaction of nitrous acid with polycrystalline ice: Adsorption on the surface and diffusion into the bulk. J. Phys. Chem. C 114, 2208–2219. doi:10.1021/jp909535c. Kerbrat, M., Pinzer, B., Huthwelker, T., Gäggeler, H. W., Ammann, M., and Schneebeli, M. (2008). Measuring the specific surface area of snow with X-ray tomography and gas adsorption: Comparison and implications for surface smoothness. Atmos. Chem. Phys. 8, 1261–1275. doi:10.5194/acp-8-1261-2008. Kong, X., Waldner, A., Orlando, F., Artiglia, L., Huthwelker, T., Ammann, M., et al. (2017). Coexistence of physisorbed and solvated HCl at warm ice surfaces. J. Phys. Chem. Lett. 8, 4757–4762. doi:10.1021/acs.jpclett.7b01573. Koop, T., Kapilashrami, A., Molina, L. T., and Molina, M. J. (2000). Phase transitions of sea-salt/water mixtures at low temperatures: Implications for ozone chemistry in the polar marine boundary layer. J. Geophys. Res. Atmos. 105, 26393–26402. doi:10.1029/2000JD900413. Lalonde, J. D., Poulain, A. J., and Amyot, M. (2002). The role of mercury redox reactions in snow on snow-to-air mercury transfer. Environ. Sci. Technol. 36, 174–178. doi:10.1021/es010786g. Legrand, M., and Mayewski, P. (1997). Glaciochemistry of polar ice cores: A review. Rev. Geophys. 35, 219–243. doi:10.1029/96RG03527. Lewis, A., Nathoo, J., Reddy, T., Randall, D., Zibi, L., and Jivanji, R. (2010). Novel technology for recovery of water and solid salts from hypersaline brines: Eutectic Freeze Crystallization. Report to WRC. Available at: https://www.coolseparations.nl/wp-content/uploads/2017/12/A.-Lewis-Novel- Technology-for-Recovery-of-Water-and-Solid-Salts-from-Hypersaline-Brines-1.pdf [Accessed June 11, 2019]. Li, Z., Edwards, R., Mosley-Thompson, E., Wang, F., Dong, Z., You, X., et al. (2006). Seasonal variability of ionic concentrations in surface snow and elution processes in snow–firn packs at the PGPI site on Ürümqi glacier No. 1, eastern Tien Shan, China. Ann. Glaciol. 43, 250–256. doi:10.3189/172756406781812069. Lorius, C., Merlivat, L., and Hagemann, R. (1969). Variation in the mean deuterium content of precipitations in Antarctica. J. Geophys. Res. 74, 7027–7031. doi:10.1029/JC074i028p07027. Löwe, H., Spiegel, J. K., and Schneebeli, M. (2011). Interfacial and structural relaxations of snow under isothermal conditions. J. Glaciol. 57, 499–510. doi:10.3189/002214311796905569. Lu, H., McCartney, S. A., and Sadtchenko, V. (2009). H/D exchange kinetics in pure and HCl doped polycrystalline ice at temperatures near its melting point: Structure, chemical transport, and phase transitions at grain boundaries. J. Chem. Phys. 130, 054501. doi:10.1063/1.3039077. Mann, E., Meyer, T., Mitchell, C. P. J., and Wania, F. (2011). Mercury fate in ageing and melting snow: Development and testing of a controlled laboratory system. J. Environ. Monit. 13, 2695–2702. doi:10.1039/c1em10297d. Marty, C., and Meister, R. (2012). Long-term snow and weather observations at Weissfluhjoch and its relation to other high-altitude observatories in the Alps. Theor. Appl. Climatol. 110, 573–583. doi:10.1007/s00704-012-0584-3.

References Chapter 2

Matykiewiczová, N., Klánová, J., and Klán, P. (2007). Photochemical degradation of PCBs in snow. Environ. Sci. Technol. 41, 8308–8314. doi:10.1021/es0714686. McFall, A. S., Edwards, K. C., and Anastasio, C. (2018). Nitrate photochemistry at the air–ice interface and in other ice reservoirs. Environ. Sci. Technol. 52, 5710–5717. doi:10.1021/acs.est.8b00095. Melnikov, I. A. (1997). The Arctic sea ice ecosystem. Gordon and Breach. Members EPICA Community (2004). Eight glacial cycles from an Antarctic ice core. Nature 429, 623– 628. doi:10.1038/nature02599. Meyer, T., Lei, Y. D., Muradi, I., and Wania, F. (2009a). Organic contaminant release from melting snow. 1. Influence of chemical partitioning. Environ. Sci. Technol. 43, 657–662. doi:10.1021/es8020217. Meyer, T., Lei, Y. D., Muradi, I., and Wania, F. (2009b). Organic contaminant release from melting snow. 2. Influence of snow pack and melt characteristics. Environ. Sci. Technol. 43, 663–668. doi:10.1021/es8020233. Meyer, T., Lei, Y. D., and Wania, F. (2011). Transport of polycyclic aromatic hydrocarbons and pesticides during snowmelt within an urban watershed. Water Res. 45, 1147–1156. doi:10.1016/j.watres.2010.11.004. Meyer, T., and Wania, F. (2008). Organic contaminant amplification during snowmelt. Water Res. 42, 1847–1865. doi:10.1016/j.watres.2007.12.016. Mosimann, L., Weingartner, E., and Waldvogel, A. (2002). An analysis of accreted drop sizes and mass on rimed snow crystals. J. Atmos. Sci. 51, 1548–1558. doi:10.1175/1520- 0469(1994)051<1548:aaoads>2.0.co;2. Pinzer, B. R., and Schneebeli, M. (2009). Snow metamorphism under alternating temperature gradients: Morphology and recrystallization in surface snow. Geophys. Res. Lett. 36, L23503. doi:10.1029/2009GL039618. Pinzer, B. R., Schneebeli, M., and Kaempfer, T. U. (2012). Vapor flux and recrystallization during dry snow metamorphism under a steady temperature gradient as observed by time-lapse microtomography. Cryosph. 6, 1141–1155. doi:10.5194/tc-6-1141-2012. Plassmann, M. M., Meyer, T., Lei, Y. D., Wania, F., McLachlan, M. S., and Berger, U. (2010). Theoretical and experimental simulation of the fate of semifluorinated n-alkanes during snowmelt. Environ. Sci. Technol. 44, 6692–6697. doi:10.1021/es101562w. Pratt, K. A., Custard, K. D., Shepson, P. B., Douglas, T. A., Pöhler, D., General, S., et al. (2013). Photochemical production of molecular bromine in Arctic surface snowpacks. Nat. Geosci. 6, 351–356. doi:10.1038/ngeo1779. Pruppacher, H. R., and Klett, J. D. (2004). Microphysics of Clouds and Precipitation. Dordrecht: Kluwer Academic Publishers doi:10.1007/0-306-48100-6_12. Pruppacher, H. R., and Klett, J. D. (2010). “Cooling of Moist Air,” in Atmospheric and Oceanographic Sciences Library, 485–501. doi:10.1007/978-0-306-48100-0_12. Purdon, F. F., and Slater, V. W. (1946). Aqueous Solution and the Phase Diagram. Edward Arnold And Company. Riche, F., Bartels-Rausch, T., Schreiber, S., Ammann, M., and Schneebeli, M. (2012). Temporal evolution of surface and grain boundary area in artificial ice beads and implications for snow chemistry. J. Glaciol. 58, 815–817. doi:10.3189/2012JoG12J058. Röthlisberger, R., Hutterli, M. A., Wolff, E. W., Mulvaney, R., Fischer, H., Bigler, M., et al. (2002). Nitrate in Greenland and Antarctic ice cores: A detailed description of post-depositional processes. Ann. Glaciol. 35, 209–216. doi:10.3189/172756402781817220.

References Chapter 2

Saltzman, E. S. (1995). “Ocean/Atmosphere Cycling of Dimethylsulfide,” in Ice Core Studies of Global Biogeochemical Cycles (Berlin, Heidelberg: Springer Berlin Heidelberg), 65–89. doi:10.1007/978- 3-642-51172-1_4. Schwikowski, M., Brütsch, S., Gäggeler, H. W., and Schotterer, U. (1999). A high-resolution air chemistry record from an Alpine ice core: Fiescherhorn glacier, Swiss Alps. J. Geophys. Res. Atmos. 104, 13709–13719. doi:10.1029/1998JD100112. Sokratov, S. A., and Maeno, N. (1997). Heat and mass transport in snow under a temperature gradient. in Snow Engineering: Recent Advances, eds. Izumi, Nakamura, and Sack (Rotterdam: Balkema), 49–54. Sokratov, S. A., and Maeno, N. (2000). Effective water vapor diffusion coefficient of snow under a temperature gradient. Water Resour. Res. 36, 1269–1276. doi:10.1029/2000WR900014. Sommerfeld, R. A., Friedman, I., and Nilles, M. (1987). The Fractionation of Natural Isotopes During Temperature Gradient Metamorphism of Snow. Seas. Snowcovers Physics, Chem. Hydrol., 95– 105. doi:10.1007/978-94-009-3947-9_5. Stefan-Kharicha, M., Kharicha, A., Mogeritsch, J., Wu, M., and Ludwig, A. (2018). Review of Ammonium Chloride-Water Solution Properties. J. Chem. Eng. Data 63, 3170–3183. doi:10.1021/acs.jced.7b01062. Stichler, W., Schotterer, U., Fröhlich, K., Ginot, P., Kull, C., Gäggeler, H., et al. (2001). Influence of sublimation on stable isotope records recovered from high-altitude glaciers in the tropical Andes. J. Geophys. Res. Atmos. 106, 22613–22620. doi:10.1029/2001JD900179. Szyrmer, W., and Zawadzki, I. (1997). Biogenic and Anthropogenic Sources of Ice-Forming Nuclei: A Review. Bull. Am. Meteorol. Soc. 78, 209–228. doi:10.1175/1520- 0477(1997)078<0209:BAASOI>2.0.CO;2. Taillandier, A. S., Dominé, F., Simpson, W. R., Sturm, M., and Douglas, T. A. (2007). Rate of decrease of the specific surface area of dry snow: Isothermal and temperature gradient conditions. J. Geophys. Res. Earth Surf. doi:10.1029/2006JF000514. Thibert, E., and Dominé, F. (1997). Thermodynamics and Kinetics of the Solid Solution of HCl in Ice. J. Phys. Chem. B 101, 3554–3565. doi:10.1021/jp962115o. Thibert, E., and Dominé, F. (1998). Thermodynamics and Kinetics of the Solid Solution of HNO 3 in Ice. J. Phys. Chem. B 102, 4432–4439. doi:10.1021/jp980569a. Thomas, D. N., and Dieckmann, G. S. (2010). Sea Ice: Second Edition. , eds. D. N. Thomas and G. S. Dieckmann Wiley doi:10.1002/9781444317145. Thompson, L. G., Mosley-Thompson, E., Davis, M. E., Zagorodnov, V. S., Howat, I. M., Mikhalenko, V. N., et al. (2013). Annually resolved ice core records of tropical climate variability over the past ∼1800 years. Science 340, 945–950. doi:10.1126/science.1234210. Ullerstam, M., and Abbatt, J. P. D. (2005). Burial of gas-phase HNO3 by growing ice surfaces under tropospheric conditions. Phys. Chem. Chem. Phys. 7, 3596–3600. doi:10.1039/b507797d. Virkkunen, K., Moore, J. C., Isaksson, E., Pohjola, V., Perämäki, P., Grinsted, A., et al. (2007). Warm summers and ion concentrations in snow: Comparison of present day with Medieval Warm Epoch from snow pits and an ice core from Lomonosovfonna, Svalbard. J. Glaciol. 53, 623–634. doi:10.3189/002214307784409388. Voitkovskii, K. F., Golubev, V. N., Sazonov, A. V., and Sokratov, S. A. (1988). Novye dannye o koeffitsiente diffuzii vodyanogo para v snege [New data on diffusion coefficient of water vapor in snow]. Mater. Glyatsiologicheskikh Issled. Data Glaciol. Stud. 63, 76–81.

References Chapter 2

Wagenbach, D., Münnich, K. O., Schotterer, U., and Oeschger, H. (1988). The anthropogenic impact on snow chemistry at Colle Gnifetti, Swiss Alps. Ann. Glaciol. 10, 183–187. doi:10.1017/s0260305500004407. Wagnon, P., Delmas, R. J., and Legrand, M. (1999). Loss of volatile acid species from upper firn layers at Vostok, Antarctica. J. Geophys. Res. Atmos. 104, 3423–3431. doi:10.1029/98JD02855. Wang, S., Shi, X., Cao, W., and Pu, T. (2018). Seasonal variability and evolution of glaciochemistry at an alpine temperate glacier on the southeastern Tibetan Plateau. Water (Switzerland) 10, 114. doi:10.3390/w10020114. Wolff, E. W. (1996). “Location, Movement and Reactions of Impurities in Solid Ice,” in Chemical Exchange Between the Atmosphere and Polar Snow (Berlin, Heidelberg: Springer Berlin Heidelberg), 541–560. doi:10.1007/978-3-642-61171-1_23. Workman, E. J., and Reynolds, S. E. (1950). Electrical phenomena occurring during the freezing of dilute aqueous solutions and their possible relationship to thunderstorm electricity. Phys. Rev. 78, 254–259. doi:10.1103/PhysRev.78.254. WU, G., LI, P., ZHANG, X., and ZHANG, C. (2018). Using a geochemical method of dissolved and insoluble fractions to characterize surface snow melting and major element elution. J. Glaciol., 1–11. doi:10.1017/jog.2018.87. Yosida, Z., and Colleagues (1955). Physical studies on deposited snow. 1.: Thermal properties. Contrib. from Inst. Low Temp. Sci. 7, 19–74. Available at: http://hdl.handle.net/2115/20216. You, X., Li, Z., Edwards, R., and Wang, L. (2015). The transport of chemical components in homogeneous snowpacks on urumqi glacier no. 1, eastern tianshan mountains, central Asia. J. Arid Land 7, 612–622. doi:10.1007/s40333-015-0131-z. Zimmermann, S., Kippenberger, M., Schuster, G., and Crowley, J. N. (2016). Adsorption isotherms for hydrogen chloride (HCl) on ice surfaces between 190 and 220 K. Phys. Chem. Chem. Phys. 18, 13799–13810. doi:10.1039/c6cp01962e.

3 Melt-induced fractionation of major ions and trace elements in an Alpine snowpack

Published in Journal Geophysical Research Atmospheres: Avak, S. E., Trachsel, J. C., Edebeli, J., Brütsch, S., Bartels‐Rausch, T., Schneebeli, M., et al. (2019). Melt‐Induced Fractionation of Major Ions and Trace Elements in an Alpine Snowpack. J. Geophys. Res. Earth Surf. 124, 2019JF005026. doi:10.1029/2019JF005026.

3.1 Introduction

Abstract Understanding the impact of melting on the preservation of atmospheric compounds in high-Alpine snow and glacier ice is crucial for future reconstruction of past atmospheric conditions. However, detailed studies investigating melt-related changes of such proxy information are rare. Here, we present a series of five snow pit profiles of 6 major ions and 34 trace elements at Weissfluhjoch, Switzerland, collected between January and June 2017. Atmospheric compounds were preserved during the cold season while melting towards the summer resulted in preferential loss of certain species from the snowpack or enrichment at the base of the snowpack. Increasing mobilization of

+ - + - 2+ 2- major ions with meltwater (NH4 < Cl ~ Na < NO3 ~ Ca ~ SO4 ) can be related to their stronger enrichment at ice crystal surfaces during snow metamorphism. Results for trace elements show that less abundant element such as Ce, Eu, La, Mo, Nd, Pb, Pr, Sb, Sc, Sm, U, and W were best preserved and may still serve as tracers to reconstruct past natural and anthropogenic atmospheric emissions from melt-affected snow pit and ice core records. The obtained elution behavior matches the findings from another high-Alpine site (upper Grenzgletscher) for major ions and the large majority of investigated trace elements. Both studies indicate that water solubility and location at the microscopic scale are likely to determine the relocation behavior with meltwater and also suggest that the observed species dependent preservation from melting snow and ice are representative for the Alpine region, reflecting Central European atmospheric aerosol composition.

3.1 Introduction Major ions (MIs) and trace elements (TEs) serve as important proxies for reconstructing past environmental conditions from high-Alpine snow pits (e.g., Gabrieli et al., 2011; Greilinger et al., 2016; Hiltbrunner et al., 2005; Kuhn et al., 1998; Kutuzov et al., 2013; Nickus et al., 1997) and ice cores (e.g., Döscher et al., 1996; Eichler et al., 2000; Preunkert et al., 2000; Schwikowski et al., 1999, 2004). For instance, concentration records of ammonium, mainly released from livestock breeding and agriculture (Döscher et al., 1996; Schwikowski et al., 1999), nitrate, primarily emitted by traffic (Döscher et al., 1995; Preunkert et al., 2003; Wagenbach et al., 1988), sulfate, typical for fossil fuel burning (Döscher et al., 1995; Preunkert et al., 2001; Schwikowski et al., 1999), and lead, a heavy metal mainly emitted by mining activities, metal production, coal combustion, or the use of leaded gasoline (Schwikowski et al., 2004), revealed the strong impact of Western European industry and society on the atmosphere over the last decades. Concentration records of Ca, Mg, or the rare-earth elements can be used to reconstruct historic variations of mineral dust emissions to the atmosphere (Gabrieli et al., 2011; Gabrielli et al., 2008). However, post-depositional melting induced by climate warming can significantly alter concentration records of atmospheric trace species from high-altitude glaciers and

3.1 Introduction snowpack as shown for MIs, organic pollutants, or water stable isotopes (Eichler et al., 2001; Herreros et al., 2009; Kang et al., 2008; Müller-Tautges et al., 2016; Pavlova et al., 2015; Sinclair & MacDonell, 2016; You et al., 2015). As glaciers, which have served as environmental archives to assess the natural and anthropogenic impact on the atmosphere, are progressively in danger of being affected by melting (Zhang et al., 2015), there is an increasing need to understand the impact of melting on the preservation of various environmental proxies in these archives. Investigation of the fate of MIs during melting of snowpack and glacier ice has been the subject of several studies (Eichler et al., 2001; Ginot et al., 2010; Grannas et al., 2013; Kang et al., 2008; Lee et al., 2008; Z. Li et al., 2006; Virkkunen et al., 2007; Wang et al., 2018; Zong-Xing et al., 2015). Preferential elution of certain ions relative to others has been observed, being strongly dependent on the respective geographical location of the snowpack.

2- + As a common feature, SO4 is generally significantly depleted with melting, while NH4 appears to be preserved despite meltwater percolation. In the majority of these studies available in the literature, either melting had strongly affected the MI records making the initial conditions before melting unknown, or a direct comparison of concentration records before and after melting to quantify the degree of depletion has not been performed. In contrast to MIs, the effect of melting on the fate of TEs in snow has still not been well characterized. To the best of our knowledge, only a few studies have addressed the influence of melting on TE concentration records in snow pits. Wong et al. (2013) artificially infiltrated meltwater into Greenland snow pits to investigate melt effects on 12 different TE records. They observed that mineral dust particle-bound TEs remained immobile during meltwater percolation resulting in the preservation of their seasonal chemical signal. On the contrary, Zhongqin et al. (2007) reported from the analyses of snow-firn pits taken in the Central Asian Tien Shan Mountains that meltwater percolation during the summer may have eluted the five investigated TEs (Al, Cd, Fe, Pb, Zn) from the snow-firn pack. The TE records from the analysis of a meltwater-affected firn part of an Alpine ice core indicated that meltwater percolation led to preferential loss of certain TEs (Avak et al., 2018). Water- insoluble TEs and low-abundant water-soluble TEs remained largely immobile with meltwater. Since this study proposed a geographical site-specificity for the preferential elution of TEs (Avak et al., 2018), it remains unclear how representative these findings are for the Alpine region. The high-Alpine snow field site at Weissfluhjoch (WFJ), Switzerland, is well-suited for snowpack studies. Numerous studies have been conducted at this site focusing on snow characterization, snow mechanics, snow metamorphism, and the development of measurement methods since 1936 (Marty & Meister, 2012). So far, studies investigating the chemical impurities in the snowpack at WFJ focused

2- - - + + + 2+ 2+ on SO4 , NO3 , Cl , K , Na , NH4 , Ca , and Mg . Baltensperger et al. (1993) compared consecutive measurements of surface snow sampled from January to March 1988 to that from a snow pit taken at the end of March at WFJ. The snow pit samples were found to be representative of precipitation

3.2 Materials and Methods deposition during the winter. Snow pit sampling, sample handling procedures, and chemical analysis of MIs were further refined by Schwikowski et al. (1997). In addition, the seasonal variability in deposition and the different emission sources of impurities were reflected in the vertical distribution of MIs in the snowpack (Schwikowski et al., 1997). Here, we present the first study monitoring the post-depositional fate of an extensive set of environmentally relevant atmospheric impurities in the high-Alpine snowpack at WFJ. We monitored six MIs and 34 TEs during an entire winter/spring season and investigated their behavior during melting of the snowpack in early summer. Five snow pits sampled from January to June 2017 allowed capturing both the initial, undisturbed records of atmospheric compounds in the snowpack and the records after melting had occurred during the warmer season. Impurity profiles reflecting dry conditions (without significant melting) were compared with profiles reflecting wet conditions to systematically investigate the impact of melting on the preservation of atmospheric tracers in high-Alpine snow conditions.

3.2 Materials and Methods

3.2.1 Study site and meteorological setting Five snow pit samplings were conducted at the high-Alpine snow field site Weissfluhjoch (WFJ) of the WSL-Institute for Snow and Avalanche Research (SLF), Eastern Switzerland (2536 m a.s.l., 46°49′47″ N 9°48′33″ E) at regular time intervals in the winter, spring, and early summer seasons in 2017. Sampling dates were January (Jan) 25th, February (Feb) 22nd, March (Mar) 21st, April (Apr) 17th, and June (Jun) 1st. Earlier snow sampling studies suggest uniform spatial snow deposition and insignificant perturbations of the snow stratigraphy due to strong winds at this site (Baltensperger et al., 1993; Schwikowski et al., 1997). Samples were obtained from the snow pits either one day before or after the weekly density measurements with 3 cm resolution, which were part of an extensive snowpack monitoring program of the SLF during the winter season 2016/17 (Calonne et al., 2016). The snow heights during the samplings were 87 cm (Jan 25th), 126 cm (Feb 22nd), 185 cm (Mar 21st), 166 cm (Apr 17th), and 83 cm (Jun 1st, Figure 3.1). Snow surface (via an infrared radiometer) and 2-m air temperatures (by an automated weather station) indicated that dry conditions without significant melting prevailed for the first three sampling dates on Jan 25th, Feb 22nd, and Mar 21st (Figure 3.1). Only 6 days with a mean 2- m air temperature above 0 °C occurred during the period January, 1st until Mar 21st (1.3 ± 0.8 °C on average). Partial melting was identifiable in the snow pit on the fourth sampling date on Apr 17th due to 13 days between Mar 21st and Apr 17th with mean 2-m air temperatures above 0 °C (1.8 ± 1.3 °C on average). The snowpack was entirely wet on the fifth sampling date on Jun 1st due to another 25 days between Apr 17th and Jun 1st with mean 2-m air temperatures above 0 °C (4.8 ± 3.4 °C on average)

3.2 Materials and Methods

Figure 3.1: 2-m air and snow surface temperatures, and daily precipitation rate and snowpack heights at the Weissfluhjoch field site, Swiss Alps, during the winter season of 2016/17. Snow pit samplings were conducted both in the cold season, where dry conditions without significant melting prevailed, and in the warm season, where severe melting of the snowpack occurred.

3.2.2 Snow pit sampling As snow is particularly sensitive to contamination on a trace amount level due to its low impurity content, precautions were taken during sampling. Sterile clean room overalls (Tyvek® IsoClean®, DuPont, Wilmington DE, United States), particulate respirator face masks (3M, Maplewood MN, United States), and ultra-clean plastic gloves (Semadeni, Ostermundigen, Switzerland) were worn during the sampling. All tools were carefully rinsed with ultra-pure water (18 MΩ cm quality, arium® pro, Sartorius, Göttingen, Germany) prior to use. Snow pits were sampled with a vertical resolution of 6 cm down to the bottom of the snowpack by pushing a custom-built rectangular (15 x 24 cm) sampler made from polycarbonate into the pit wall. To allow for sufficient sample volume, the snow was filled into 50 mL polypropylene vials (Sarstedt, Nümbrecht, Germany) by pushing them twice with the opening facing downwards into the snow. Separate vials were used for MI, water stable isotope (WSI), and TE analysis. Polypropylene vials were pre-cleaned five times with ultra-pure water for MI and WSI samples. For TE samples, the tubes were pre-cleaned five times with ultra-pure water (18 MΩ cm quality, Milli-Q®

Element, Merck Millipore, Burlington MA, United States) plus once with 0.2 M HNO3 prepared from

TM ultra-pure HNO3 (Optima , Fisher Chemical, Loughborough, United Kingdom). Samples for TE analysis were taken from the snow at the front part of the sampler to prevent possible cross contamination of the vials used for MI sampling with HNO3.

3.2 Materials and Methods

3.2.3 Major ion, water stable isotope and trace element analysis A total of 324 samples for MI, WSI, and TE analysis were kept frozen at -20 °C until analysis at the Paul Scherrer Institute.

+ + 2+ - - 2- MIs (Na , NH4 , Ca , Cl , NO3 , and SO4 ) present in the snow pit samples were analyzed after melting at room temperature using ion chromatography (IC, 850 Professional IC equipped with a 872 Extension Module Liquid Handling and a 858 Professional Sample Processor auto sampler, Metrohm, Herisau, Switzerland). Cations were separated using a Metrosep C4 column (Metrohm) and 2.8 mM

-1 HNO3 as eluent at a flow rate of 1 mL min . Anions were separated using a Metrosep A Supp 10 column

(Metrohm) and were eluted stepwise using first, a 1.5 mM Na2CO3/0.3 mM NaHCO3 (1:1 mixture)

-1 eluent, then an 8 mM Na2CO3/1.7 mM NaHCO3 (1:1 mixture) eluent at a flow rate of 0.9 mL min . Possible instrumental drifts were monitored by measuring an in-house standard after every 20th sample. The precision of the method was ~5%. WSI samples were melted at room temperature and 1 mL aliquots were analyzed for δD and δ18O using a wavelength-scanned cavity ring down spectrometer (WS-CRDS, L2130-i Analyzer, Picarro, Santa Clara CA, United States). Samples were injected into the vaporizer (A0211, Picarro, Santa Clara CA, United States) using a PAL HTC-xt autos ampler (LEAP Technologies, Carrboro NC, United States). Three in-house standards were measured after every tenth sample for calibration and to monitor instrumental drifts. The measurement uncertainty was <0.1‰ for δ18O and <0.5‰ for δD. Snow pit samples were melted at room temperature, acidified with concentrated ultra-pure

HNO3 to 0.2M (1% v/v), and analyzed 3-4 h after acidification following the same procedure as described in (Avak et al., 2018) using discrete inductively coupled plasma sector field mass spectrometry (ICP-SF-MS, Element 2, Thermo Fisher Scientific, Bremen, Germany). Low (LR, R = 300) or medium resolution (MR, R = 4000) data was acquired for Ag (LR), Al (MR), Ba (LR), Bi (LR), Ca (MR), Cd (LR), Ce (LR), Co (MR), Cs (LR), Cu (LR & MR), Eu (LR), Fe (MR), La (LR), Li (LR & MR), Mg (MR), Mn (MR), Mo (LR), Na (MR), Nd (LR), Ni (LR & MR), Pb (LR), Pr (LR), Rb (LR), Sb (LR), Sc (MR), Sm (LR), Sr (LR), Tl (LR), Th (LR), U (LR), V (MR), W (LR), Yb (LR), Zn (LR & MR) and Zr (LR). If both LR and MR resolution data was available, MR data was used for further data evaluation. Quantification of intensity values was performed by internal and external calibration using a Rh-standard and seven liquid standards covering the typical concentration range of TEs in natural snow and ice samples, respectively. Linear regressions of the calibration curves consistently revealed correlation coefficients >0.999.

3.2.4 Data Evaluation IC raw data was processed using the MagIC Net 3.2 software (Metrohm, Herisau, Switzerland). Sample concentrations were not blank-corrected as the concentrations of the blank, determined by analyzing

3.3 Results and Discussion ultrapure water, were below the detection limit (DL). Sample concentrations below the DL were substituted with half the value of the DL. On average, 10% of the measurement values were below the DL. ICP-MS raw data was evaluated following the method described by Knüsel et al. (2003). Concentrations were blank-corrected by subtracting a measurement blank consisting of four measurements of ultra-pure 0.2 M HNO3. The instrumental DL was defined as 3σ of the measurement blank and concentrations below the DL were substituted with half of the value of the DL. 109Ag was excluded from the data set for further evaluation and discussion as concentrations of all samples were below the DL. Total depths of the snow profiles were converted to water equivalents (w.eq.) by multiplying with the respective density. As Apr 17th had the largest snow depth, the profiles on Jan 25th, Feb 22nd, Mar 21st, and Jun 1st were aligned relative to the profile on Apr 17th, with 0 cm w.eq. being the surface on this day and assuming the same bottom depth for all profiles. All five profiles cover the depth interval 45-65 cm w.eq. The profiles on Feb 22nd, Mar 21st, Apr 17th, and Jun 1st cover 36-65 cm w.eq., the profiles on Feb 22nd, Mar 21st, and Apr 17th cover 30-65 cm w.eq., and the profiles on Mar 21st and Apr 17th cover 5-65 cm w.eq. The sample at the base of each snow pit was omitted from all chemical profiles to exclude a possible influence of the soil at WFJ.

3.3 Results and Discussion

3.3.1 Major ions (MIs)

3.3.1.1 Comparison of the five MI concentration profiles

+ - 2+ 18 The chemical profiles of selected MIs (NH4 , Cl , Ca ) and the δ O records in the five snow pits are shown in Figure 3.2. The general MI patterns of the four profiles, reflecting the winter and spring periods (January-April), show a strong correspondence in the respective overlapping parts. Apart from possible changes caused by melting (Apr 17th), slight concentration differences in the profiles can be attributed to the spatial variability of impurities within the snowpack as the locations of the individual snow pit samplings were several meters (up to 20 m) apart. The resemblance between the four winter and spring snow pits is also visible for the corresponding δ18O profiles which support the depth assignment of the different snow pits.

+ - 2+ The chemical profiles of NH4 , Cl , and Ca are exemplary for the different emission sources and transport characteristics of MIs in Alpine snow and ice. The depths 0-30 cm w.eq. (Mar 21st and Apr 17th) indicate precipitation occurring in spring after Feb 22nd, while 30-65 cm w.eq. represent winter

+ precipitation (Figure 3.1). The NH4 springtime concentrations are roughly three times higher than

+ wintertime concentrations. NH4 detected in snow pits and ice cores from European high-Alpine sites

3.3 Results and Discussion

+ is nowadays generally of anthropogenic origin (Döscher et al., 1996; Gabrieli et al., 2011). NH4 also shows a pronounced maximum towards summer due to enhanced emissions from agricultural

+ activities, and stronger convection (Baltensperger et al., 1997). Higher concentrations of NH4 in the springtime snowfall compared to wintertime snowfall at WFJ were also observed by Schwikowski et

2+ - + al. (1997). Ca and Cl , unlike NH4 , have higher concentrations in the wintertime snow than in spring. At Alpine sites, Ca2+ and Cl- are indicative of the input from mineral dust (Bohleber et al., 2018; Schwikowski et al., 1995; Wagenbach et al., 1996) and marine sea salt aerosols (Eichler et al., 2004), respectively. A very pronounced peak in the Ca2+ and Na+ profiles between 50 and 60 cm w.eq. (Figure 3.2) can be most likely attributed to a Saharan dust event. This is corroborated by back trajectory analysis for the beginning of January 2017 and measurements by the Swiss Federal Office of Meteorology and Climatology (MeteoSwiss) at the Jungfraujoch high-Alpine research station on January 5th, 2017. After several weeks with temperatures above 0 °C, the snowpack was very wet from meltwater (and rainwater) at the beginning of the summer (Jun 1st; Figure 3.1). The uppermost (surface) sample of the snow pit from Jun 1st (red dotted line in Figure 3.2) is most likely influenced by residual snow impurities after surface melting and wet or dry deposition between Apr 17th and Jun 1st (Figure 3.1) and was thus, not taken into consideration. The δ18O profile of Jun 1st reveals a strong smoothing compared to the previous profiles, indicative of strong melting (Thompson et al., 1993). However, a signature between 45-65 cm w.eq. resembling that from the previous profiles is still visible. This indicates that at this depth meltwater percolation was concentrated to the porous space in the snowpack and primarily percolated along ice surfaces, keeping the ice matrix in principle preserved. The MI profiles of Jun 1st compared to those of the first three sampling dates are either almost

+ + - 2+ - 2- unaltered (NH4 ), slightly depleted (Na , Cl ), or strongly depleted (Ca , NO3 , SO4 ), most likely due to different elution behavior of the investigated ions with meltwater (Eichler et al., 2001).

3.3.1.2 Preferential elution of MIs: Elution sequence and discussion

To quantify and compare the apparent preferential elution of MIs, a concentration ratio cwet/cdry for the overlapping depth (45-65 cm w.eq.) was calculated for each MI (Table 3.1). cwet corresponds to the

st integrated MI concentration for the wet profile of Jun 1 , whereas cdry represents the mean of the integrated concentration profiles during dry periods (Jan 25th, Feb 22nd, Mar 21st). The profiles on Apr 17th were not included as the snowpack was neither completely dry nor very wet. Corresponding to

+ - + - 2+ 2- the concentration ratio, an elution sequence was established: NH4 < Cl ~ Na < NO3 ~ Ca ~ SO4 ,

+ 2- where NH4 is the least mobile ion and SO4 , the most mobile ion. Similar elution sequences were reported by Eichler et al. (2001), Ginot et al. (2010), and Zong-xing et al. (2015), who observed a

2- 2+ + - particularly leaching of SO4 and Ca , while NH4 and Cl were retained in melting firn and snowpack.

3.3 Results and Discussion

+ - 2+ Figure 3.2: Concentration profiles of NH4 , Cl , and Ca of the five snow pits taken at the Weissfluhjoch field site during winter/spring (black/blue curves) and early summer (red curve). For comparison, the corresponding δ18O records are shown. The grey dashed line indicates the bottom of the snowpack. Profile depths were aligned to the one of April 17th and 0 cm w.eq. reflects the surface of the snowpack on this day. The surface sample of the snow pit from Jun 1st is shown with a dashed line (see text for details). The shaded area indicates the depth interval (45- 65 cm w.eq.) used for calculation of the concentration ratio and determining an elution sequence.

Elution sequences of MIs were also determined through laboratory studies (Cragin et al., 1996; Tranter et al., 1992; Tsiouris et al., 1985). We recently performed an elution experiment where, for the first time, homogenous impurity-doped artificial snow was exposed to well-defined snow metamorphism conditions prior to leaching with 0 °C ultra-pure water. This experiment was conducted to determine enrichment differences of MIs between ice interiors and surfaces after metamorphism (Trachsel et al., 2017). Snow undergoes drastic structural transformation cycles during metamorphism (Pinzer et al., 2012), which may result in significant impurity redistribution. Our elution sequence at the WFJ agrees well with the findings of the laboratory-based elution experiment (Figure 3.3). MIs having a higher

+ - solubility in ice such as NH4 and Cl (Feibelman, 2007; Hobbs, 1974), showed less mobility most likely

3.3 Results and Discussion due to incorporation into the crystal interior during snow metamorphism. They can be incorporated either by substituting water molecules located on lattice sites of the ice crystal (Zaromb & Brill, 1956) or by occupying interstitial spaces of the crystal structure (Petrenko & Whitworth, 2002). On the other

2+ 2- hand, Ca and SO4 were enriched on ice surfaces with progressing snow metamorphism (Figure 3.3); explaining their availability for mobilization with meltwater. This is supported by previous location studies of salts in ice. Mulvaney et al. (1988) used a combination of scanning electron microscopy

(SEM) and energy-dispersive X-ray microanalysis (EDXMA) to show that H2SO4 concentrations at triple junctions are several orders of magnitude higher compared to grain interiors in polar ice from

Antarctica. Similar observations for H2SO4 and HNO3 using Raman spectroscopy were reported by

Fukazawa et al. (1998). Accumulation of MgSO4 at grain boundaries in Greenland GISP2 ice was revealed by low-vacuum SEM-EDX (Baker et al., 2003). The varying elution behavior of MIs observed

st - 2+ 2- in the snow pit of Jun 1 with NO3 , Ca and SO4 being most heavily depleted can therefore be explained by their microscopic locations on ice surfaces, exposing them to relocation during melting.

+ + 2+ - 2- Figure 3.3: Time evolution of relative enrichment of Na , NH4 , Ca , Cl , and SO4 concentrations at the ice surface determined by leaching artificial snow, homogeneously doped with known concentrations of MIs and exposed for different time periods to a temperature gradient mimicking snow metamorphism, with 0 °C ultra-pure water (adapted from Trachsel et al., 2017).

Our findings show that the atmospheric composition of MIs is well preserved in the snowpack at WFJ during the cold season. Melting during the warm season leads to preferential leaching of MIs depending on their microscopic location either on the ice surface or in the ice interior. The loss of MIs

2+ 2- + is particularly significant for Ca and SO4 . In contrast, the strong persistence of NH4 emphasizes that

3.3 Results and Discussion

+ NH4 can still serve as environmental tracer for the interpretation of snow pit and ice core records affected by melting.

3.3.2 Trace elements (TEs)

3.3.2.1 TE concentration profiles Concentrations of Co, Fe, Ce, Sb, Ca, and Sr in the snow pits on Jan 25th, Feb 22nd, Mar 21st, Apr 17th, and Jun 1st are exemplarily shown in Figure 3.4. The general pattern of the profiles of the first four snow pits (Jan to Apr) agree well in the overlapping depths. As for the MI profiles, misalignment in peaks can be attributed to small-scale spatial variability of impurities in the snowpack. The reproducibility of chemical profiles with ultra-trace level concentrations (e.g. Ce and Sb) from different sampling campaigns demonstrates that significant contamination during the snow pit samplings did not occur. ICP-MS measurements of Ca and Na concentration profiles are in good agreement with the records of Ca2+ and Na+ concentrations obtained by IC (Figure 3.2 and 3.4). The uppermost (surface) sample of the snow pit from Jun 1st revealed highly elevated concentrations for many TEs (red dotted line in Figure 3.4). This is most likely due to residual snow impurities after surface melting and TEs that reached the snowpack by wet or dry deposition between Apr 17th and Jun 1st (Figure 3.1). For this reason, the uppermost sample of the snow pit from Jun 1st was not taken into consideration for the following discussion. The 34 TEs can be either of geogenic origin or emitted by anthropogenic sources. At high-Alpine sites, Al, Ba, Bi, Ca, Cs, Fe, Li, Mg, Mn, Na, Rb, Sr, Th, Tl, U, W, Zr, and the rare-earth elements (Ce, Eu, La, Nd, Pr, Sc, Sm, Yb) are mainly deposited with mineral dust (Gabrieli et al., 2011; Gabrielli et al., 2008) whereas Ag, Cd, Co, Cu, Mo, Ni, Pb, Sb, V, and Zn in Alpine snow and ice are characteristic of anthropogenic atmospheric pollution (Barbante et al., 2004; Gabrieli et al., 2011; Gabrielli et al., 2008; Schwikowski et al., 2004; Van de Velde et al., 1999, 2000). The chemical profiles of Ca, Ce, Fe, and Sr (Figure 3.4) are representative of TEs originating from mineral dust particles and reveal an enrichment in the depths reflecting winter precipitation (30-65 cm w.eq.) where dry conditions prevailed (Jan 25th, Feb 22nd, Mar 21st). These TEs show a major peak between 50 and 60 cm w.eq. due to a Saharan dust

+ event beginning of January 2017 (Section 3.3.1.1). As explained above for NH4 , TEs indicative of anthropogenic influence such as Sb (Figure 3.4), are enriched in depths reflecting spring precipitation deposited after Feb 22nd (0-30 cm w.eq.). This is due to the typical seasonality of convective air mass transport from the planetary boundary layer to high-Alpine sites occurring in spring and summer (Baltensperger et al., 1997). Based on the concentration profiles on Jun 1st, the 34 investigated TEs were divided into three groups (Table 3.1). Group 1 (Ce, Eu, La, Mo, Nd, Pb, Pr, Sb, Sc, Sm, U, W) is characterized by an almost

3.3 Results and Discussion

Figure 3.4: Concentration profiles of Co, Fe, Ce, Sb, Ca, and Sr of the five snow pits taken at the Weissfluhjoch field site between winter/spring (black/blue curves) and early summer (red curve). The grey dashed line indicates the bottom of the snowpack. Profile depths were aligned to the one of April 17th and 0 cm w.eq. reflects the surface of the snowpack on this day. The surface sample of the snow pit from Jun 1st is shown with a dashed line (see text for details). The shaded area indicates the depth interval (45-65 cm w.eq.) used for calculation of the concentration ratio. TEs were classified as “influenced by soil”, or allocated to group 1 or 2 (retained or depleted concentration profile) according to the pattern of Jun 1st).

3.3 Results and Discussion unbiased profile pattern between 45 and 65 cm w.eq. compared to the dry condition profiles (e.g. Ce, Sb, Figure 3.4). This indicates that group 1 TEs were not affected by melting. Group 2 TEs (Ba, Bi, Ca, Cd, Na, Sr, Th, Zn, Zr) were strongly depleted on Jun 1st compared to winter and springtime records (as shown for Ca and Sr in Figure 4). Group 3 TEs (Al, Co, Cs, Cu, Fe, Li, Mg, Mn, Ni, Rb, Tl, V, Yb) exhibit a strong enrichment below 55 cm w.eq. close to the soil (see e.g. Co and Fe, Figure 3.4). The bottom of the snowpack on Jun 1st was very wet, probably enriched with impurities not of atmospheric origin, but present in the soil at WFJ. We therefore attribute the strong increase in concentration below 55 cm w.eq. to the contact with soil and denote this group as “influenced by soil” (i.b.s). Group 3 TEs were therefore not included in further evaluation.

3.3.2.2 Different preservation of TEs under melt conditions A quantitative classification of TE fractionation during melting of the snowpack (group 1 and 2) was performed by calculating a concentration ratio cwet/cdry for each TE in the overlapping depths (see above, Table 3.1). The highest concentration ratio was obtained for La indicative of its well preserved concentration record on Jun 1st, whereas the concentration profile of Zr is most severely affected by melting, having the lowest concentration ratio. Concentration ratios of TEs classified as group 1 are close to 1 and range between 1.5 ± 1.1 (La) and 0.84 ± 0.36 (Pb). Concentration ratios of group 2 TEs are significantly below 1 and range between 0.55 ± 0.23 (Cd) and 0.13 ± 0.08 (Zr). Arranging the concentration ratios according to size and plotting this rank against the ratio indicates that each TE belonging to group 1 and 2 was differently preserved in the snowpack during melting (Figure 3.5). The preservation of group 1 TEs with observed concentration ratios >0.7 (Ce, Eu, La, Mo, Nd, Pb, Pr, Sb, Sc, Sm, U, W), is in excellent agreement with findings from another Alpine site (Avak et al., 2018, Figure 3.5). At this ~180 km distant site, meltwater percolation led to post-depositional disturbance of a 16 m firn section of a high-Alpine ice core from the upper Grenzgletscher (Monte Rosa massif, southern Swiss Alps, 4200 m a.s.l., 45°55′28″ N 7°52′3″ E). Preferential elution of TEs led to significant concentration depletion in the records of Ba, Ca, Cd, Co, Mg, Mn, Na, Ni, Sr, and Zn, whereas the seasonality in the Al, Bi, Ce, Cs, Cu, Eu, Fe, La, Li, Mo, Nd, Pb, Pr, Rb, Sb, Sc, Sm, Tl, Th, U, V, W, Yb, and Zr records was well preserved (Avak et al., 2018). The different behavior of TEs during meltwater percolation was related to their varying water solubility and location at a microscopic scale. TEs mainly present in water-insoluble mineral dust particles at the upper Grenzgletscher site (Al, Fe, Zr, and the rare-earth elements) were enriched on firn grain surfaces and mostly preserved, since their insolubility in water resulted in immobility with meltwater. TEs likely present in water-soluble compounds (Ba, Bi, Ca, Cd, Co, Cs, Cu, Li, Mg, Mn, Mo, Na, Ni, Pb, Rb, Sb, Sr, Tl, Th, U, V, W, Zn; Avak et al., 2018; Birmili et

3.3 Results and Discussion al., 2006; Greaves et al., 1994; Hsu et al., 2010; T. Li et al., 2015) revealed variable mobility with meltwater presumably due to the different microscopic location of the ions in the ice structure.

Figure 3.5: Rank of preservation plotted against the concentration ratio cwet/cdry in the overlapping part (45-65 cm w.eq. depth) for each TE classified into group 1 (retained concentration profile) or 2 (depleted concentration st profile). cwet corresponds to the integrated TE concentration of the wet profile (Jun 1 ), whereas cdry represents the mean of the integrated concentration profiles during dry conditions (Jan 25th, Feb 22nd, Mar 21st). Symbols in bold indicate a similar behavior as observed for high-Alpine firn affected by meltwater percolation (Avak et al., 2018). Circle sizes represent the mean concentrations in the dry snow pits where no melting occurred.

For those rather water-soluble TEs, the concentration at the upper Grenzgletscher site was found to be primarily decisive determining either incorporation into the ice interior during snow metamorphism or segregation to ice surfaces because of exceeded solubility limits in ice. In correspondence with the upper Grenzgletscher study and the work reported by Wong et al. (2013), also at WFJ, records of investigated TEs present as water-insoluble mineral dust particles (rare- earth elements: Ce, Eu, La, Nd, Pr, Sc, Sm) are well preserved most probably due to their immobility with meltwater. Contrariwise, Zhongqin et al. (2007) observed a depletion of mineral dust-bond elements, such as Fe and Al at Urumqi glacier No. 1 (eastern Tien Shan) in summer and relate this to meltwater relocation. However, it cannot be excluded that the low concentrations in summer observed for all investigated elements are due to dilution effects during the wet season. There is also strong agreement in the behavior of rather water-soluble TEs between upper Grenzgletscher and WFJ, showing a dependency on the concentration level. While TEs present in ultra- trace amounts tend to be preserved, more abundant TEs were preferentially eluted (Figure 3.5). The only three TEs revealing a different behavior between the two sites are Bi, Th and Zr, which still show a preservation of the seasonality at upper Grenzgletscher, but significant depletion at WFJ. One

3.3 Results and Discussion explanation could be a higher proportion of the water-soluble fraction of Bi, Th and Zr deposited during winter at WFJ, favoring higher meltwater mobility. Another possible reason is differing concentrations. However, Th and Zr concentrations are not significantly different between the two sites. Only the higher Bi concentrations at WFJ could explain stronger melt-induced relocation. Table 3.1: Concentration ratio, classificationa, mean concentration of the dry snow pits (Jan 25th, Feb 22nd, Mar 21st), and DLs of MIs and TEs investigated in this study.

Concentration Mean concentration Detection limit (ng L-1)

ratio / Classification (ng L-1) Na+ 0.62 ± 0.45 17000 ± 2840 500

+ NH4 0.77 ± 0.27 26300 ± 4580 500 Ca2+ 0.37 ± 0.18 71500 ± 11400 10000 Cl- 0.64 ± 0.43 34000 ± 4980 1000

- NO3 0.37 ± 0.11 249000 ± 21200 1000

2- SO4 0.36 ± 0.10 57300 ± 5330 5000 Al i.b.s. 4870 ± 792 115 Ba 0.39 ± 0.26 / Group 2 262 ± 67 3.5 Bi 0.50 ± 0.28 / Group 2 4.5 ± 0.91 0.01 Ca 0.46 ± 0.24 / Group 2 110000 ± 21400 498 Cd 0.55 ± 0.23 / Group 2 1.9 ± 0.26 0.27 Ce 1.5 ± 1.1 / Group 1 14 ± 2.9 0.03 Co i.b.s. 24 ± 4.5 0.22 Cs i.b.s. 2.1 ± 0.33 0.07 Cu i.b.s. 48 ± 5.6 2.7 Eu 1.1 ± 0.79 / Group 1 0.51 ± 0.10 0.13 Fe i.b.s. 5990 ± 1180 245 La 1.5 ± 1.1 / Group 1 6.1 ± 1.2 0.03 Li i.b.s. 6.1 ± 0.90 3.8 Mg i.b.s. 37000 ± 6600 30 Mn i.b.s. 612 ± 110 0.69 Mo 1.1 ± 0.59 / Group 1 3.1 ± 0.46 1.8 Na 0.50 ± 0.32 / Group 2 15900 ± 2620 1280 Nd 1.2 ± 0.92 / Group 1 7.7 ± 1.7 0.01 Ni i.b.s. 101 ± 22 9 Pb 0.84 ± 0.36 / Group 1 180 ± 15 2.7 Pr 1.1 ± 0.84 / Group 1 1.8 ± 0.38 0.01 Rb i.b.s. 15 ± 2.1 0.54 Sb 0.88 ± 0.36 / Group 1 9.1 ± 1.2 0.13 Sc 1.5 ± 1.4 / Group 1 1.8 ± 0.41 0.17 Sm 1.2 ± 0.88 / Group 1 2.0 ± 0.46 0.02 Sr 0.37 ± 0.21 / Group 2 270 ± 55 3.4 Th 0.45 ± 0.30 / Group 2 1.4 ± 0.25 0.03 Tl i.b.s. 0.39 ± 0.04 0.01 U 1.5 ± 0.99 / Group 1 2.3 ± 0.64 0.01 V i.b.s. 19 ± 2.5 0.37 W 1.2 ± 0.69 / Group 1 0.54 ± 0.07 0.18 Yb i.b.s. 0.63 ± 0.13 0.02 Zn 0.47 ± 0.36 / Group 2 2780 ± 463 28 Zr 0.13 ± 0.08 / Group 2 5.4 ± 0.75 0.26 aTEs were classified depending on whether the profile of the snow pit of Jun 1st showed a strong enrichment from the soil at Weissfluhjoch below 55 cm w.eq. (i.b.s./influenced by soil), an unbiased profile pattern compared to the dry condition profiles (group 1), or a heavily depleted concentration curve (group 2).

3.4 Conclusion

These findings in the preservation of TEs corroborate our previous hypothesis (Avak et al., 2018) that for ice cores and snow pits from Alpine areas, representing Central European atmospheric aerosol composition, and affected by partial melting, Ce, Eu, La, Mo, Nd, Pb, Pr, Sb, Sc, Sm, U, and W may still be used as robust environmental proxies. However, chemical profiles of Ba, Ca, Cd, Na, Zn, and Sr are prone to be depleted by melting. Records of Bi, Th and Zr behaved differently at the two Alpine sites. The potential of TEs classified as “i.b.s.” here and previously found to be preserved (Al, Cs, Cu, Fe, Li, Rb, Tl, V, Yb) or depleted (Co, Mg, Mn, Ni) in firn (Avak et al., 2018) could not be further investigated because of the influence of the soil in the presence of meltwater.

3.4 Conclusion The analysis of five snow pits sampled during the winter/spring season of 2017 at WFJ is presented here. Comparison of impurity profiles representing dry and wet snow conditions allowed observing the fate of atmospheric trace species during melting of the snowpack. While the atmospheric composition is well preserved during the winter, melting in the spring and early summer caused a preferential loss of certain MIs and TEs. We propose that the elution behavior of MIs is depending on their microscopic location, which is most likely defined by redistribution processes occurring during

2+ - + + - 2- + snow metamorphism. Among the six investigated MIs (Ca , Cl , Na , NH4 , NO3 , SO4 ), NH4 was

2+ 2- retained strongest, probably due to its preferred location in the ice interior. Ca and SO4 concentrations were significantly depleted; this can be explained by their predominant occurrence on ice crystal surfaces. Variable mobility was also observed for TEs, with Ce, Eu, La, Mo, Nd, Pb, Pr, Sb, Sc, Sm, U, and W being the best-preserved elements. In tendency, high concentrations of TEs were found to favor relocation with meltwater and thus, low abundant TEs were preferably retained. The major ions and 18 out of 21 investigated TEs revealed a consistent behavior with meltwater percolation at two Alpine sites, WFJ and upper Grenzgletscher, which are 180 km apart. This indicates that this proposition is particularly valid for the Alpine region, reflecting Central European atmospheric

+ aerosol composition. Based on the observations at the two Alpine sites, we propose that NH4 , TEs deposited as water-insoluble particles, and TEs of water-soluble particles occurring in low concentrations may still be applicable as environmental proxies in snow pits and ice cores affected by melting. Corroborating the existence of meltwater-resistant proxies is particularly relevant as many high-mountain glaciers worldwide, which provide avenues for assessing natural and anthropogenic impact on the atmosphere, are increasingly affected by melting.

Acknowledgements

Acknowledgements The authors would like to thank Philipp Baumann and Matthias Jaggi (both from SLF) for support during the sampling campaigns of February 22nd and June 1st, respectively. The authors also acknowledge Franziska Scholder-Aemisegger (Institute for Atmospheric and Climate Science, ETH Zurich) for performing back trajectory analyses and Martine Collaud (Swiss Federal Office of Meteorology and Climatology) for providing a list of Saharan dust event detected at the Jungfraujoch high-Alpine research station between November 2016 and February 2017. Funding was provided by the Swiss National Science Foundation (SNSF) under Grant No. 155999. The corresponding data will be available at the NOAA (National Oceanic and Atmospheric Administration) data center for paleoclimate (ice core sites) after acceptance of the paper: http://www.ncdc.noaa.gov/data-access/paleoclimatology-data. Finally, we greatly appreciate the comments and suggestions of the three anonymous referees which helped to improve the clarity of the manuscript.

References Chapter 3

References Avak, S. E., Schwikowski, M., & Eichler, A. (2018). Impact and implications of meltwater percolation on trace element records observed in a high-Alpine ice core. Journal of Glaciology, 64(248), 877– 886. https://doi.org/10.1017/jog.2018.74 Baker, I., Cullen, D., & Iliescu, D. (2003). The microstructural location of impurities in ice. Canadian Journal of Physics, 81(1–2), 1–9. https://doi.org/10.1139/p03-030 Baltensperger, U., Schwikowski, M., Gäggeler, H. ., Jost, D. ., Beer, J., Siegenthaler, U., et al. (1993). Transfer of atmospheric constituents into an alpine snow field. Atmospheric Environment. Part A. General Topics, 27(12), 1881–1890. https://doi.org/10.1016/0960-1686(93)90293-8 Baltensperger, U., Gäggeler, H. W., Jost, D. T., Lugauer, M., Schwikowski, M., Weingartner, E., & Seibert, P. (1997). Aerosol climatology at the high-alpine site Jungfraujoch, Switzerland. Journal of Geophysical Research: Atmospheres, 102(D16), 19707–19715. https://doi.org/10.1029/97JD00928 Barbante, C., Schwikowski, M., Döring, T., Gäggeler, H. W., Schotterer, U., Tobler, L., et al. (2004). Historical Record of European Emissions of Heavy Metals to the Atmosphere Since the 1650s from Alpine Snow/Ice Cores Drilled near Monte Rosa. Environmental Science & Technology, 38(15), 4085–4090. https://doi.org/10.1021/es049759r Birmili, W., Allen, A. G., Bary, F., & Harrison, R. M. (2006). Trace Metal Concentrations and Water Solubility in Size-Fractionated Atmospheric Particles and Influence of Road Traffic. Environmental Science & Technology, 40(4), 1144–1153. https://doi.org/10.1021/es0486925 Bohleber, P., Erhardt, T., Spaulding, N., Hoffmann, H., Fischer, H., & Mayewski, P. (2018). Temperature and mineral dust variability recorded in two low-accumulation Alpine ice cores over the last millennium. Climate of the Past, 14(1), 21–37. https://doi.org/10.5194/cp-14-21-2018 Calonne, N., Cetti, C., Fierz, C., Van Herwijnen, A., Jaggi, M., Löwe, H., et al. (2016). A unique time series of daily and weekly snowpack measurements at Weissfluhjoch, Davos, Switzerland. In International snow science workshop proceedings 2016. International snow science workshop, ISSW 2016, Breckenridge, CO, USA, October 2-7 (pp. 684–689). Cragin, J. H., Hewitt, A. D., & Colbeck, S. C. (1996). Grain-scale mechanisms influencing the elution of ions from snow. Atmospheric Environment, 30(1), 119–127. https://doi.org/10.1016/1352- 2310(95)00232-N Döscher, A., Gäggeler, H. W., Schotterer, U., & Schwikowski, M. (1995). A130 years deposition record of sulfate, nitrate and chloride from a high-alpine glacier. Water, Air, & Soil Pollution, 85(2), 603– 609. https://doi.org/10.1007/BF00476895 Döscher, A., Gäggeler, H. W., Schotterer, U., & Schwikowski, M. (1996). A historical record of ammonium concentrations from a glacier in the Alps. Geophysical Research Letters, 23(20), 2741–2744. https://doi.org/10.1029/96GL02615 Eichler, A., Schwikowski, M., & Gäggeler, H. W. (2000). An Alpine ice-core record of anthropogenic HF and HCl emissions. Geophysical Research Letters, 27(19), 3225–3228. https://doi.org/10.1029/2000GL012006 Eichler, A., Schwikowski, M., & Gäggeler, H. W. (2001). Meltwater-induced relocation of chemical species in Alpine firn. Tellus B: Chemical and Physical Meteorology, 53(2), 192–203. https://doi.org/10.3402/tellusb.v53i2.16575 Eichler, A., Schwikowski, M., Furger, M., Schotterer, U., & Gäggeler, H. W. (2004). Sources and distribution of trace species in Alpine precipitation inferred from two 60-year ice core

References Chapter 3

paleorecords. Atmospheric Chemistry and Physics Discussions, 4(1), 71–108. https://doi.org/10.5194/acpd-4-71-2004 Feibelman, P. J. (2007). Substitutional NaCl hydration in ice. Physical Review B, 75(21), 214113. https://doi.org/10.1103/PhysRevB.75.214113 Fukazawa, H., Sugiyama, K., Mae, S., Narita, H., & Hondoh, T. (1998). Acid ions at triple junction of Antarctic ice observed by Raman scattering. Geophysical Research Letters, 25(15), 2845–2848. https://doi.org/10.1029/98GL02178 Gabrieli, J., Carturan, L., Gabrielli, P., Kehrwald, N., Turetta, C., Cozzi, G., et al. (2011). Impact of Po Valley emissions on the highest glacier of the Eastern European Alps. Atmospheric Chemistry and Physics, 11(15), 8087–8102. https://doi.org/10.5194/acp-11-8087-2011 Gabrielli, P., Cozzi, G., Torcini, S., Cescon, P., & Barbante, C. (2008). Trace elements in winter snow of the Dolomites (Italy): A statistical study of natural and anthropogenic contributions. Chemosphere, 72(10), 1504–1509. https://doi.org/10.1016/j.chemosphere.2008.04.076 Ginot, P., Schotterer, U., Stichler, W., Godoi, M. A., Francou, B., & Schwikowski, M. (2010). Influence of the Tungurahua eruption on the ice core records of Chimborazo, Ecuador. The Cryosphere, 4(4), 561–568. https://doi.org/10.5194/tc-4-561-2010 Grannas, A. M., Bogdal, C., Hageman, K. J., Halsall, C., Harner, T., Hung, H., et al. (2013). The role of the global cryosphere in the fate of organic contaminants. Atmospheric Chemistry and Physics, 13(6), 3271–3305. https://doi.org/10.5194/acp-13-3271-2013 Greaves, M. J., Statham, P. J., & Elderfield, H. (1994). Rare earth element mobilization from marine atmospheric dust into seawater. Marine Chemistry, 46(3), 255–260. https://doi.org/10.1016/0304-4203(94)90081-7 Greilinger, M., Schöner, W., Winiwarter, W., & Kasper-Giebl, A. (2016). Temporal changes of inorganic ion deposition in the seasonal snow cover for the Austrian Alps (1983–2014). Atmospheric Environment, 132, 141–152. https://doi.org/10.1016/j.atmosenv.2016.02.040 Herreros, J., Moreno, I., Taupin, J.-D., Ginot, P., Patris, N., De Angelis, M., et al. (2009). Environmental records from temperate glacier ice on Nevado Coropuna saddle, southern Peru. Advances in Geosciences, 22, 27–34. https://doi.org/10.5194/adgeo-22-27-2009 Hiltbrunner, E., Schwikowski, M., & Körner, C. (2005). Inorganic nitrogen storage in alpine snow pack in the Central Alps (Switzerland). Atmospheric Environment, 39(12), 2249–2259. https://doi.org/10.1016/j.atmosenv.2004.12.037 Hobbs, P. V. (1974). Ice Physics. Oxford: Oxford University Press. Hsu, S.-C., Wong, G. T. F., Gong, G.-C., Shiah, F.-K., Huang, Y.-T., Kao, S.-J., et al. (2010). Sources, solubility, and dry deposition of aerosol trace elements over the East China Sea. Marine Chemistry, 120, 116–127. https://doi.org/10.1016/j.marchem.2008.10.003 Kang, S., Huang, J., & Xu, Y. (2008). Changes in ionic concentrations and δ18O in the snowpack of Zhadang glacier, Nyainqentanglha mountain, southern Tibetan Plateau. Annals of Glaciology, 49(October 2006), 127–134. https://doi.org/10.3189/172756408787814708 Knüsel, S., Piguet, D. E., Schwikowski, M., & Gäggeler, H. W. (2003). Accuracy of Continuous Ice-Core Trace-Element Analysis by Inductively Coupled Plasma Sector Field Mass Spectrometry. Environmental Science & Technology, 37(10), 2267–2273. https://doi.org/10.1021/es026452o Kuhn, M., Haslhofer, J., Nickus, U., & Schellander, H. (1998). Seasonal development of ion concentration in a high alpine snow pack. Atmospheric Environment, 32(23), 4041–4051. https://doi.org/10.1016/S1352-2310(97)00216-1 Kutuzov, S., Shahgedanova, M., Mikhalenko, V., Ginot, P., Lavrentiev, I., & Kemp, S. (2013). High- resolution provenance of desert dust deposited on Mt. Elbrus, Caucasus in 2009-2012 using

References Chapter 3

snow pit and firn core records. The Cryosphere, 7(5), 1481–1498. https://doi.org/10.5194/tc-7- 1481-2013 Lee, J., Nez, V. E., Feng, X., Kirchner, J. W., Osterhuber, R., & Renshaw, C. E. (2008). A study of solute redistribution and transport in seasonal snowpack using natural and artificial tracers, 243–254. https://doi.org/10.1016/j.jhydrol.2008.05.004 Li, T., Wang, Y., Li, W. J., Chen, J. M., Wang, T., & Wang, W. X. (2015). Concentrations and solubility of trace elements in fine particles at a mountain site, southern China: regional sources and cloud processing. Atmospheric Chemistry and Physics, 15(15), 8987–9002. https://doi.org/10.5194/acp-15-8987-2015 Li, Z., Edwards, R., Mosley-Thompson, E., Wang, F., Dong, Z., You, X., et al. (2006). Seasonal variability of ionic concentrations in surface snow and elution processes in snow–firn packs at the PGPI site on Ürümqi glacier No. 1, eastern Tien Shan, China. Annals of Glaciology, 43, 250–256. https://doi.org/10.3189/172756406781812069 Marty, C., & Meister, R. (2012). Long-term snow and weather observations at Weissfluhjoch and its relation to other high-altitude observatories in the Alps. Theoretical and Applied Climatology, 110(4), 573–583. https://doi.org/10.1007/s00704-012-0584-3 Müller-Tautges, C., Eichler, A., Schwikowski, M., Pezzatti, G. B., Conedera, M., & Hoffmann, T. (2016). Historic records of organic compounds from a high Alpine glacier: influences of biomass burning, anthropogenic emissions, and dust transport. Atmospheric Chemistry and Physics, 16(2), 1029– 1043. https://doi.org/10.5194/acp-16-1029-2016 Mulvaney, R., Wolff, E. W., & Oates, K. (1988). Sulphuric acid at grain boundaries in Antarctic ice. Nature, 331(6153), 247–249. https://doi.org/10.1038/331247a0 Nickus, U., Kuhn, M., Baltensperger, U., Delmas, R., Gäggeler, H. W., Kasper, A., et al. (1997). SNOSP: Ion deposition and concentration in high alpine snow packs. Tellus B, 49(1), 56–71. https://doi.org/10.1034/j.1600-0889.49.issue1.4.x Pavlova, P. A., Jenk, T. M., Schmid, P., Bogdal, C., Steinlin, C., & Schwikowski, M. (2015). Polychlorinated Biphenyls in a Temperate Alpine Glacier: 1. Effect of Percolating Meltwater on their Distribution in Glacier Ice. Environmental Science & Technology, 49(24), 14085–14091. https://doi.org/10.1021/acs.est.5b03303 Petrenko, V. F., & Whitworth, R. W. (2002). Physics of Ice. Oxford: Oxford University Press. https://doi.org/10.1093/acprof:oso/9780198518945.001.0001 Pinzer, B. R., Schneebeli, M., & Kaempfer, T. U. (2012). Vapor flux and recrystallization during dry snow metamorphism under a steady temperature gradient as observed by time-lapse microtomography. The Cryosphere, 6(5), 1141–1155. https://doi.org/10.5194/tc-6-1141-2012 Preunkert, S., Wagenbach, D., Legrand, M., & Vincent, C. (2000). Col du Dôme (Mt Blanc Massif, French Alps) suitability for ice-core studies in relation with past atmospheric chemistry over Europe. Tellus B: Chemical and Physical Meteorology, 52(3), 993–1012. https://doi.org/10.3402/tellusb.v52i3.17081 Preunkert, S., Legrand, M., & Wagenbach, D. (2001). Sulfate trends in a Col du Dôme (French Alps) ice core: A record of anthropogenic sulfate levels in the European midtroposphere over the twentieth century. Journal of Geophysical Research: Atmospheres, 106(D23), 31991–32004. https://doi.org/10.1029/2001JD000792 Preunkert, S., Wagenbach, D., & Legrand, M. (2003). A seasonally resolved alpine ice core record of nitrate: Comparison with anthropogenic inventories and estimation of preindustrial emissions of NO in Europe. Journal of Geophysical Research: Atmospheres, 108(D21). https://doi.org/10.1029/2003JD003475

References Chapter 3

Schwikowski, M., Seibert, P., Baltensperger, U., & Gäggeler, H. W. (1995). A study of an outstanding Saharan dust event at the high-alpine site Jungfraujoch, Switzerland. Atmospheric Environment, 29(15), 1829–1842. https://doi.org/10.1016/1352-2310(95)00060-C Schwikowski, M., Novo, A., Baltensperger, U., Delmas, R., Gäggeler, H. W., Kasper, A., et al. (1997). Intercomparison of snow sampling and analysis within the alpine-wide snowpack investigation (SNOSP). Water, Air, & Soil Pollution, 93(1–4), 67–91. https://doi.org/10.1007/BF02404748 Schwikowski, M., Brütsch, S., Gäggeler, H. W., & Schotterer, U. (1999). A high-resolution air chemistry record from an Alpine ice core: Fiescherhorn glacier, Swiss Alps. Journal of Geophysical Research: Atmospheres, 104(D11), 13709–13719. https://doi.org/10.1029/1998JD100112 Schwikowski, M., Brütsch, S., Gäggeler, H. W., & Schotterer, U. (1999). A high-resolution air chemistry record from an Alpine ice core: Fiescherhorn glacier, Swiss Alps. Journal of Geophysical Research: Atmospheres, 104(D11), 13709–13719. https://doi.org/10.1029/1998JD100112 Schwikowski, M., Barbante, C., Doering, T., Gaeggeler, H. W., Boutron, C., Schotterer, U., et al. (2004). Post-17th-Century Changes of European Lead Emissions Recorded in High-Altitude Alpine Snow and Ice. Environmental Science & Technology, 38(4), 957–964. https://doi.org/10.1021/es034715o Sinclair, K. E., & MacDonell, S. (2016). Seasonal evolution of penitente glaciochemistry at Tapado Glacier, Northern Chile. Hydrological Processes, 30(2), 176–186. https://doi.org/10.1002/hyp.10531 Thompson, L. G., Mosley-Thompson, E., Davis, M., Lin, P. N., Yao, T., Dyurgerov, M., & Dai, J. (1993). “Recent warming”: ice core evidence from tropical ice cores with emphasis on Central Asia. Global and Planetary Change, 7(1–3), 145–156. https://doi.org/10.1016/0921-8181(93)90046- Q Trachsel, J. C., Schneebeli, M., Avak, S. E., Eichler, A., Edebeli, J., & Bartels-Rausch, T. (2017). MISO: More than a Japanese Seasoning. In Annual Report 2016 of the Laboratory of Environmental Chemistry (p. 15). Villigen PSI. Retrieved from https://www.psi.ch/luc/AnnualReportsEN/LUC_Annual_Report_2016.pdf Trachsel, J. C., Avak, S. E., Edebeli, J., Brütsch, S., Bartels-Rausch, T., Schneebeli, M., & Eichler, A. (2019). Microscale re-arrangement of ammonium induced by snow metamorphism. In Preparation. Tranter, M., Tsiouris, S., Davies, T. D., & Jones, H. G. (1992). A laboratory investigation of the leaching of solute from snowpack by rainfall. Hydrological Processes, 6(2), 169–178. https://doi.org/10.1002/hyp.3360060205 Tsiouris, S., Vincent, C. E., Davies, T. D., & Brimblecombe, P. (1985). The Elution of Ions Through Field and Laboratory Snowpacks. Annals of Glaciology, 7, 196–201. https://doi.org/10.3189/S0260305500006169 Van de Velde, K., Ferrari, C., Barbante, C., Moret, I., Bellomi, T., Hong, S., & Boutron, C. (1999). A 200 Year Record of Atmospheric Cobalt, Chromium, Molybdenum, and Antimony in High Altitude Alpine Firn and Ice. Environmental Science & Technology, 33(20), 3495–3501. https://doi.org/10.1021/es990066y Van de Velde, K., Boutron, C. F., Ferrari, C. P., Moreau, A.-L., Delmas, R. J., Barbante, C., et al. (2000). A two hundred years record of atmospheric cadmium, copper and zinc concentrations in high altitude snow and ice from the French-Italian Alps. Geophysical Research Letters, 27(2), 249– 252. https://doi.org/10.1029/1999GL010786 Virkkunen, K., Moore, J. C., Isaksson, E., Pohjola, V., Perämäki, P., Grinsted, A., & Kekonen, T. (2007). Warm summers and ion concentrations in snow: comparison of present day with Medieval Warm Epoch from snow pits and an ice core from Lomonosovfonna, Svalbard. Journal of Glaciology, 53(183), 623–634. https://doi.org/10.3189/002214307784409388

References Chapter 3

Wagenbach, D., Münnich, K. O., Schotterer, U., & Oeschger, H. (1988). The Anthropogenic Impact on Snow Chemistry at Colle Gnifetti, Swiss Alps. Annals of Glaciology, 10, 183–187. https://doi.org/10.3189/S0260305500004407 Wagenbach, D., Preunkert, S., Schäfer, J., Jung, W., & Tomadin, L. (1996). Northward Transport of Saharan Dust Recorded in a Deep Alpine Ice Core. In S. Chester & R. Guerzoni (Eds.), The Impact of Desert Dust Across the Mediterranean (pp. 291–300). https://doi.org/10.1007/978-94-017- 3354-0_29 Wang, S., Shi, X., Cao, W., & Pu, T. (2018). Seasonal Variability and Evolution of Glaciochemistry at An Alpine Temperate Glacier on the Southeastern Tibetan Plateau. Water, 10(2), 114. https://doi.org/10.3390/w10020114 Wong, G. J., Hawley, R. L., Lutz, E. R., & Osterberg, E. C. (2013). Trace-element and physical response to melt percolation in Summit (Greenland) snow. Annals of Glaciology, 54(63), 52–62. https://doi.org/10.3189/2013AoG63A602 You, X., Li, Z., Edwards, R., & Wang, L. (2015). The transport of chemical components in homogeneous snowpacks on Urumqi Glacier No. 1, eastern Tianshan Mountains, Central Asia. Journal of Arid Land, 7(5), 612–622. https://doi.org/10.1007/s40333-015-0131-z Zaromb, S., & Brill, R. (1956). Solid Solutions of Ice and NH4F and Their Dielectric Properties. The Journal of Chemical Physics, 24(4), 895–902. https://doi.org/10.1063/1.1742629 Zhang, Q., Kang, S., Gabrielli, P., Loewen, M., & Schwikowski, M. (2015). Vanishing High Mountain Glacial Archives: Challenges and Perspectives. Environmental Science & Technology, 49(16), 9499–9500. https://doi.org/10.1021/acs.est.5b03066 Zhongqin, L., Chuanjin, L., Yuefang, L., Feiteng, W., & Huilin, L. (2007). Preliminary results from measurements of selected trace metals in the snow–firn pack on Ürümqi glacier No. 1, eastern Tien Shan, China. Journal of Glaciology, 53(182), 368–373. https://doi.org/10.3189/002214307783258486 Zong-Xing, L., Qi, F., Xiao-yan, G., Yan, G., Yan-hui, P., Ting-ting, W., et al. (2015). The evolution and environmental significance of glaciochemistry during the ablation period in the north of Tibetan Plateau, China. Quaternary International, 374, 93–109. https://doi.org/10.1016/j.quaint.2014.06.071

4 Recording frontal passages with the ẟ18O of Alpine snow: How post-depositional processes alter the precipitation signal in the snowpack

Manuscript in preparation: Jürg Trachsel, Franziska Scholder-Aemisegger, Sven Avak, Anja Eichler, Martin Schneebeli. Recording frontal passages with the ẟ18O of Alpine snow: How post-depositional processes alter the precipitation signal in the snowpack.

Abstract The vertical profile of the stable water isotope composition of Alpine snow is a complex function of atmospheric transport, cloud formation, snow deposition and alterations that occurs during snow metamorphism. So far, this chain of processes including both pre- and post-depositional effects has never been investigated in detail. In this study, we present a continuous four-month time series of hourly water vapor isotope measurements in the near-surface atmosphere in combination with weekly surface snow samples and a series of five snow pit ẟ18O profiles sampled every four weeks at Weissfluhjoch, Switzerland. The campaign took place between January and June 2017. During these 4 months, new snow was measured on more than 40 days with a maximum snow height of 205 cm, which is below the long-term climatological average. The maximum amplitude of the ẟ18O variations measured in the vertical profiles is 13‰. The variability in the ẟ18O of the snow cover was assigned to individual frontal precipitation events and associated with precipitation from cold or warm airmasses. The temporal evolution of simulated snow density from the snow cover model SNOWPACK was used to relate individual snow layers in the snow cover to snowfall events. The events were classified into precipitation from different airmasses based on weather maps and air parcel back-trajectory analyses using the ERA5 reanalysis dataset from the European Centre for Medium Range Weather Forecasts. Layers on the snow surface originating from snowfall formed in warm airmasses are generally more enriched (associated with higher ẟ18O in the range -16‰ to -10‰) compared to layers, which were deposited during snowfall events occurring in cold airmasses with ẟ18O in the range -24‰ to -20‰. The air parcels with subtropical or southern midlatitude moisture sources thus lead to higher values of ẟ18O than air parcels with subpolar and northern midlatitude moisture sources. In exceptional cases, fresh snow with low ẟ18O occurred during warm front passages with deep clouds reaching into very cold environments in the upper troposphere. Inside the snow cover, the isotopic signature was preserved in the snow layers during the winter season from January to March and smoothed by only 1-2 ‰ due to dry metamorphism. Wet metamorphism during spring resulted in an enrichment of ẟ18O in the snow cover. Hence, a relatively uniform shift of up to 9‰ in the absolute values of the vertical ẟ18O profile was observed during the melting phase, while the vertical variability patterns were conserved. Our results thus show a good preservation of the vertical ẟ18O variability in the snowpack. The ẟ18O of the seasonal Alpine snow cover therefore provides an interesting record of the thermodynamic characteristics of frontal precipitation events during the winter season.

4.1 Introduction

4.1 Introduction The ẟ18O records in ice cores from the polar ice caps have a long tradition for their use as paleoclimate proxies reflecting past temperature fluctuations (Craig, 1961; Cuffey et al., 1995; Dansgaard, 1964; Johnsen et al., 1995; Masson-Delmotte, 2005). The method is based on the observation that the stable

16 16 18 water isotopic composition of precipitation (H2O , HDO , H2O ) is related to the air temperature (Dansgaard, 1953; Epstein and Mayeda, 1953). Cold airmasses are generally associated with lower ẟ18O than warm air masses, a well-known property described by Dansgaard (1964) at the climatological timescale by his “isotope effects”. Several reasons for the decrease in the observed ẟ18O in vapor and precipitation behind cold fronts have been discussed in the literature such as the advection of colder air masses (Gedzelman and Lawrence, 1990), the formation of clouds at different heights (Rindsberger et al., 1990), the progressive depletion due to rain out (Celle-Jeanton et al., 2004) and effects of below cloud interaction (Aemisegger et al., 2015; Graf et al., 2019). Araguás-Araguás et al. (1998) demonstrated that temperature correlations do not apply in all regions of the world: In high precipitation areas (e.g. South Pacific Monsoon) the seasonal variability can be controlled by the amount effect as described by Rozanski et al., (1993). The isotopic composition of a given water sample is usually expressed in the so‐called δ notation

18 in units of permille (‰) as ẟ O = (Rsample/RVSMOW2-1) x 1000 with the isotope ratio R defined as the ratio of the concentration of heavier to lighter isotopes (18O/16O) and standardized to the Vienna Standard Mean Ocean Water (VSMOW2). When following the water vapor in an air parcel from its oceanic evaporative moisture source to the precipitation sink, several processes including mixing and phase change processes can affect the ẟ18O of an air parcel. Phase change processes are associated with so- called isotopic fractionation effects. Ocean evaporation is a process during which heavy water molecules preferentially stay in the liquid phase due to their stronger bonds (equilibrium fractionation effect) and smaller diffusivities through an unsaturated laminar layer (non-equilibrium fractionation effect). These isotope fractionation processes are dependent on the thermodynamic conditions (temperature and relative humidity) of the environment during the phase change. The isotope ratio of the freshly formed water vapor Rsample is thus lower than the isotope ratio of ocean water (usually very close to RVSMOW2). The δ‐values of atmospheric waters are therefore mostly negative. Samples with high (less negative) δ18O values compared to the δ18O of ocean water (0‰) have a greater proportion of heavy isotopes and are referred to as more enriched in heavy isotopes. Similarly, samples with low (more negative) δ18O values compared to seawater have a smaller proportion of heavy isotopes and are referred to as depleted in heavy isotopes. Large-scale atmospheric transport is a non-fractionating process, whereas cloud formation and precipitation lead to a preferential removal of heavy water molecules in the air parcel.

4.1 Introduction

At the millennial timescale, rapid climate fluctuations during so-called Dansgaard-Oeschger events (Bond et al., 1993; Dansgaard et al., 1993) have been recorded with changes of up to 4-6‰ within half a century in ẟ18O from Greenland ice cores. At the weather system timescale, the passage of atmospheric fronts leads to similar changes (4-8‰) in ẟ18O of atmospheric vapor and precipitation within a few hours (Aemisegger et al., 2015; Gedzelman and Lawrence, 1990; Pfahl et al., 2012). Continuous records of the isotope composition of near-surface water vapor are available (Aemisegger et al., 2012; Benetti et al., 2017; Bonne et al., 2013), but due to the lack of high-resolution isotope profiles in the snowpack, the interactions between the atmosphere and the snow remain largely unknown (Christner et al., 2017). Baltensberger et al. (1993) found a good agreement between the measured ẟ18O in new snow and samples derived from a snow pit at the beginning of spring at the Weissfluhjoch site (WFJ) in the Swiss Alps. However, it remains unclear how abrupt changes in the isotope signal of the precipitation during frontal events are transferred into the snow cover. Several studies reported a rather modest correlation between temperature and ẟ18O in ice cores from Alpine regions, especially for long-term observations (Bohleber et al., 2013, 2018; Mariani et al., 2014). The reason for this could be post-depositional phenomena, which modify the ice core data (Mariani et al., 2014). Snow deposited on the ground is exposed to different processes, such as recrystallization, sublimation and wind-pumping, that possibly alter the isotope signal in the snowpack (Beria et al., 2018). Stichler et al. (2001) have isolated sublimation at the surface of the snow cover and diffusive mixing of water vapor within the firn as main processes for altering the isotopic record in high-altitude glaciers. Furthermore, changes in the surface snow isotope composition could be linked to water vapor exchange and wind pumping (Steen-Larsen et al., 2014). Within the snow cover, snow metamorphism (Ebner et al., 2017) and water flow during melt or rain events (Lee et al., 2009) can affect the isotopic signal. In the following, we review the most important findings on post-depositional processes published in the literature so far taking place a) at the snow surface, b) in the snow interior) and c) during snowmelt. a) Snow surface processes It is important to note that sublimation and other atmospheric exchange processes can only take place in the uppermost centimeters of a snowpack because deeper layers are decoupled from the atmosphere by the isolating snow above. The depth to which the exchange can take place depends on the permeability and tortuosity of the snow and wind effects at the surface (Johnsen et al., 2000; Waddington et al., 1996). Stichler et al. (2001) have observed an influence of wind effects in the range of 5-10 cm depth with respect to the ẟ18O. Sublimation is the most discussed process and the scientific discourse on whether sublimation influences the isotope signal of the snow is still ongoing. Two contradictory theories exist: On the one hand, there is the theory of a layer-by-layer total sublimation

4.1 Introduction

(onion model), which does not lead to a modification of the isotope ratios in the underlaying snow (Ambach et al., 1968; Friedman et al., 1991; Johnsen, 1973). On the other hand, there are studies which have found an enrichment of heavy isotopes in the remaining snow due to fractionation at the air-ice interface (Gurney and Lawrence, 2004; Moser and Stichler, 1975; Neumann et al., 2008; Sokratov and Golubev, 2009; Stichler et al., 2001). In addition to sublimation (net mass loss), water vapor exchange between the near-surface atmosphere and the snow surface is possible. Experimental data in the laboratory have shown that the isotope ratio in the snow changes when air flows through it with a different ratio (Ebner et al., 2017). This finding is supported by current field studies for Greenland and Antarctica (Casado et al., 2018; Madsen et al., 2019; Ritter et al., 2016; Steen-Larsen et al., 2014; Touzeau et al., 2016). The investigated areas receive relatively low precipitation amounts, which results in a long exposure of the snow surface to the atmosphere. b) Processes in the snow interior Traditionally isotopic changes within the snow or firn are explained by diffusion (Johnsen, 1973; Stichler et al., 2001b; Whillans and Grootes, 1985). Even though the observed smoothing of the vertical profiles in firn cores could be linked with temperature and density differences in the firn (Johnsen, 1973), the diffusive process as such remains a black box. Sommerfeld et al. (1987) stated that vapor flux within the snowpack alters the ratio of stable isotopes after deposition. Sokratov and Golubev (2009) attribute the observed fractionation to recrystallization and the associated diffusion of water molecules through air. Recently, more detailed studies have shown that the long-known hand to hand process of vapor flux (Yosida, 1955) also influences the isotopic composition of the snow (Ebner et al., 2017; Steen-Larsen et al., 2014). Snow metamorphism describes the recrystallization that snow undergoes after deposition: Snow on Earth is almost always exposed to a temperature gradient. In alpine regions, the gradient usually occurs between the ground (0°C at places without permafrost) and the air above the snow cover (surface air temperature depends on weather conditions). The gradient can, therefore, be positive or negative. This gradient causes the relatively warmer ice surface (higher water vapor pressure) of every snow particle to sublimate. The generated water vapor accumulates on neighboring relatively colder surface (lower water vapor pressure) of the ice structure (Pinzer et al., 2012; Sokratov and Maeno, 1997). Over time, the local rearrangements completely renew and thereby transform the structure of the original ice crystals. Pinzer et al. (2012) showed a characteristic lifetime of 2-3 days for an ice volume when exposed to a gradient of 50 K/m, which is typical for high alpine snowpacks at the beginning of the winter season. How this complete rearrangement of the snowpack affects the isotope signal in detail is not yet clear (Sturm and Benson, 1997).

4.1 Introduction c) Melt processes A homogeneous snow cover always melts at the surface. Depending on the dominating components of the radiation balance the melt zone is at the very surface (dominating sensible and latent heat fluxes), or a few centimeters at depth in the case when shortwave radiation penetrates into the snow. The snowpack reaches the melting point by heat diffusion and when the liquid water flows through it. Which of these two processes dominates, depends on the infiltration rate? Denoth (1982) introduced two idealized states: - Pendular System: the capillary tension is high enough to retain most of the meltwater penetrating from the surface into the snow cover. The flow rate is low. - Funicular System: at higher water content, the capillary flow is overcome by gravitational flow and higher flow rates occur, corresponding to the infiltration rate. Recent simulations by Wever et al. (2014) have shown that there are many transitional states between funicular and pendular system. Furthermore, there is a feedback of wet snow metamorphism to the flow pattern inside the snowpack (Avanzi et al., 2016; Schneebeli, 1995). The influence of meltwater on the snow isotopic profiles has already been the subject of various studies. Snowmelt preferentially discharges isotopically light water, thereby enriching the residual snowpack in heavier isotopes (Albert and Hardy, 1995; Feng et al., 2002; Soulsby et al., 2000; Taylor et al., 2001). Feng et al. (2002) obtain a total isotopic variation between 1‰ and 4‰ in ẟ18O depending on the melt conditions. Lee et al. (2010) have experimentally simulated snow melting by artificial watering of the snow cover. Based on their result, they developed a model that simulates the hydraulic exchange between mobile and immobile water and isotopic exchange between liquid water and ice within the snowpack. As soon as liquid water is present in the snow cover, the process of wet metamorphism is initiated. In contrast to other porous media, e.g. as in soils, the liquid water leads to a coarsening of the snow structure (Colbeck, 1986; Raymond and Tusima, 1979). The “system snow” reduces the surface free energy by a growth of the large particles at the expense of the smaller particles, during this process, the melted mass is deposited on the larger structures (Albert and Krajeski, 1998). The growth rate is strongly dependent on the liquid water content (Brun, 1988) and is many times faster than during dry metamorphism. The contribution of wet snow metamorphism to the change in the isotope signal during the melting period is unclear but is likely key in the widely observed “melt-out effect” of the snow cover (Ala-Aho et al., 2017), i.e. the progressive enrichment of both the snow cover and the meltwater as the melt season progresses.

The short review of important atmospheric and snow physical processes that shape the isotope signals in the snow cover emphasizes, that the formation of climate archives from the snow deposition to the

4.2 Experimental methods final embedding of the isotope signals into ice cores is a long chain of complex pre- and post- depositional processes that only starts to be explored (Casado et al., 2018; Madsen et al. 2019). The aim of this paper is to show how the isotope signals associated with abrupt changes of airmass with distinct thermodynamic properties are archived in a seasonal alpine snow cover that is exposed at the same time to post-depositional modification. A separation and quantitative description of the post-depositional processes affecting the isotopic composition of the Alpine snow is complex. With our collection of five snow isotope profiles and simultaneous water vapor isotope measurements in the air before and during the melt season (Section 4.2), we would like to contribute to a better understanding of the role of these processes and how they alter the initially deposited atmospheric signal. To follow the isotope signals associated with frontal passages from the snow deposition into the snow cover, we first present the winter evolution and the different precipitation events in Section 4.3. The results and discussion in Section 4.4 are organized in two parts. First, the archival of frontal passages in the seasonal snowcover is discussed by analyzing six major precipitation events that explain a large proportion of the evolution of the seasonal snow cover in winter 2016-2017 at the measurement site. Second, the impact of the post-depositional processes affecting the embedded isotope signal in the snowpack is analysed. Finally, conclusions from this study on a seasonal Alpine snow cover are drawn and an outlook with ideas for future studies on interannual ẟ18O signals recorded in shallow ice cores and their link to frontal passages is given (Section 4.5).

4.2 Experimental methods

4.2.1 Overview of measurements The Weissfluhjoch WFJ measurement site of the WSL Institute for Snow and Avalanche Research is located above Davos at 2,536 m above sea level (46°49′47″N 9°48′33″E). The site is a reference CryoNet station (Marty and Meister, 2012). In winter 2017 our campaign to monitor the isotope signal in the air at ~2 m above the snow surface as well as in the snowpack took place from January to June (Fig. 4.1). The water vapor isotope composition was continuously measured from 3 February onwards with a Picarro cavity ringdown spectrometer (L2130-i), with a short interruption from 3 to 12 March due to a broken inlet tubing. On five sampling days, a snow pit was dug and the entire snowpack was sampled with a resolution of 6 cm as further described in Section 4.2.2.1. Besides samples for the isotope analysis, samples for trace elements, ions and black carbon were collected. The chemical results are described in Avak et al. (2019), and in this work we focus on the water isotope data. In parallel to our campaign, an extensive measurement program to characterize the snow cover on the WFJ took place (Calonne et al., 2019). Among other parameters, a weekly density profile (3 cm) and a layer by layer classification

4.2 Experimental methods of the entire snow cover were compiled. For optimal comparability, the different measurements were taken directly before or after each other, the maximum separation was two days.

Figure 4.1: Overview performed measurements on WFJ with temporal and spatial resolution of the different sampling parameters. a) Field site Weissfluhjoch in the Swiss Alps, 2536 m a.s.l. (left); heated hut with Picarro analyzer inside and inlet on the left side of the roof. b) Sampling Team in clean overalls inside snow pit (left); sampling snow surface (middle); filled tubes inside snow pit before being packed for isolated transport to the lab (right).

4.2.2 Stable water isotope measurements

4.2.2.1 Snow sampling The isotope samples of the entire snow pit were collected on 25 January, 22 February, 21 March, 17 April and 1 June with a vertical resolution of 6 cm. Snow depths during sampling varied between 185 cm (21 March) and 83 cm (1 June). Earlier studies conducted at the WFJ by Baltensperger et al. (1993) and Schwikowski et al. (1997) point to a uniform snow deposition. However small spatial differences caused by topography and wind drift cannot be completely excluded, since the positions of the individual snow profiles were up to 20 m apart. In addition to the five snow pits, a weekly surface profile of the top 20 cm with a resolution of 3 cm was sampled (3 samples directly at the surface, 7 samples downwards). For these surface profiles, tubes with a diameter of 3 cm were horizontally pushed into the snow cover using a template.

4.2 Experimental methods

In order to prevent any contamination of the samples, all instruments and containers were rinsed with milliQ water 5 times before use. Furthermore, clean overalls and respirator face masks were worn during sampling. The sampling was carried out starting from the surface to the ground. For this purpose, a rectangular (15 × 24 cm) polycarbonate sampler was inserted horizontally into the profile wall. The corresponding 50 ml polypropylene containers (Sarstedt, Nümbrecht, Germany) were then pushed vertically towards the sampler. A detailed description of the sampling is given in Avak et al. (2019). The samples were hermetically sealed and transported frozen to the laboratory of the Paul Scherer Institute PSI.

4.2.2.2 Analysis of δD and δ18O from snow samples in the lab The frozen samples were melted in the PSI laboratory at room temperature. For the determination of δD and δ18O 1 ml aliquots were filled. A wavelength-scanned cavity ring down spectrometer (WS-CRDS, L2130-i Analyzer, Picarro, Santa Clara, CA, USA) at PSI was used for the analysis. The samples were injected into the evaporator (A0211, Picarro, Santa Clara, CA, USA) using PAL HTC-xt autosampler (LEAP Technologies, Carrboro, NC, USA). Three internal standards were measured after every tenth sample to calibrate and monitor instrumental drifts. The uncertainty of a measurement is given as <0.1‰ for δ18O and <0.5‰ for δD (Avak et al., 2019)

4.2.2.3 Measuring the stable water isotope composition of the vapor phase Measurements of the stable water isotope composition of ambient vapor were performed with a cavity ring-down spectrometer (CRDS) from CRYOS-EPFL, version L2130-i from Picarro. The inlet was situated outside a mountain hut at ~4 m above ground (between 2 and 4 m above the snow surface). The CRDS system was located inside a temperature regulated room with temperatures around 16°C. A ~8 m long PTFE tube (1/4” I.D.), inner volume 8 l, was heated to a temperature of 40°C and flushed with a KNF Pump at a rate of ~10 l/min to minimized memory effects. Calibration measurements of known liquid water samples were performed using an autosampler and a vaporizer once a day. Three standards were used with isotopic compositions that spread the CRDS measured range of isotope -values in the vapor phase at the measurement site (Table 4.1). For each standard six injections were done, of which the first three were discarded to minimize memory effects, due to previous injections with different standards. The autosampler injections were done at a water vapor mixing ratio of 20’000 ppmv into a vaporizer heated at 140°C. To correct for the known instrument-specific water vapor mixing ratio dependent bias in laser spectrometric measurements (e.g. Aemisegger et al., 2012) a 3rd order polynomial fit to a series of calibration runs performed at different water vapor mixing ratios with standard 1 was used. This series of calibration runs were performed in the lab using a dew point generator (Li-610, Fig. 4.2b) similar to the water vapor mixing ratio tests described in Aemisegger et al. 2012 and Sodemann et al. 2017. The water vapor mixing

4.3 Winter evolution and association of precipitation events ratios were calibrated using the same dew point experiment (Fig. 4.2a). We consider these two correction functions as instrument characteristics that remained constant during the campaign. This proved to be a good assumption during previous campaigns (Aemisegger et al., 2014; Sodemann et al., 2017). The daily calibration runs were used to normalize the ambient vapor data to the VSMOW- VSLAP2 scale and to correct for instrument drifts. Table 4.1: Isotope composition of the standards (determined by isotope-ratio mass spectrometry) used for calibration of the vapor isotope measurements.

Isotope variable Standard 1 Standard 2 Standard 3 δ 18O [‰] -11.42 ± 0.2‰ -24.89 ± 0.73‰ -40.56 ± 0.30‰ δ 2H [‰] -82.05 ± 0.6‰ -153.9 ± 1.06‰ -325.83 ± 0.90‰

a) b)

Figure 4.2: a) Linear calibration curve for the water vapor mixing ratio, based on dew point generator measurements at different dew point temperatures between 12°C and -4°C (left). b) Correction functions for the water vapor mixing ratio dependent isotope bias (right) derived from the dew point experiment.

4.3 Winter evolution and association of precipitation events

4.3.1 Evolution of the seasonal snow cover The winter season 2017 was comparatively dry for the Swiss Alps with several cold spells in January. The snow depth on the WFJ was almost consistently below average (Zweifel et al., 2017). The measurement site was snowed in on 4 November with the passage of a warm front and shortly after, on 6 November a cold front (Tab. 4.2 and Fig 4.3). At the beginning of January there was only 40 cm of snow. In the first half of January, several snow falls caused the snow height to accumulate to about 1 m. The largest 24 h snow fall event was measured on 1 February with 47cm fresh snow. The remaining February was again too warm and too dry compared to the long-term average and the snow depth decreased slightly mainly due to compaction. March was unusually warm too with relatively large amounts of precipitation in the first half of the month that led to a significant increase in snow depth.

4.3 Winter evolution and association of precipitation events

The maximum snow height for the whole winter was reached on 10 March at 205 cm. Thereafter the snow depth decreased again significantly until mid-April, mainly due to compaction. Several snowfall events from mid-April to early May led to a relatively steady snow depth. From the second week of May onwards the snow began to melt, which quickly and continuously reduced the snow depth with an average decrease of 5 to 6 cm per day. On the 1st of June there was only 83 cm of snow left. Two weeks later (June 14th) the measuring field was completely free of snow (Fig 4.3).

Figure 4.3: Development of the snow cover at the WFJ field site in the Swiss Alps during the winter 2016/2017. The 2‐m air (dashed black line) and snow surface temperatures (solid black line) and daily precipitation rate (dark blue bars) and snowpack heights (light blue bars) are shown. The snow pit samplings, indicated with red vertical lines, were conducted both in the cold season, where dry conditions without signiﬁcant melting prevailed, and in the warm season, where severe melting of the snowpack occurred. The letters refer to particular snow layers in the snow cover with a specific ẟ18O signature (see Tab. 4.2 and Fig. 4.10). Figure adapted from Avak et al., 2019.

4.3.2 Weather system-based classification of precipitation events The reanalysis dataset ERA5 from the European Centre for Medium-Range Weather Forecasts (ECMWF, Copernicus Climate Change Service (C3S), (2017)) served as a databases to analyze the weather situation during the major 26 precipitation events at the WFJ measurement site listed in Table 4.2. All precipitation events of at least 5 cm of fresh snow in 24 h were analyzed. Maps of the equivalent potential temperature e interpolated to a pressure level of 700 hPa were visually inspected to classify events into cold or warm precipitation events. The horizontal gradient of e in the lower troposphere is often used in objective front identification methods as a thermal variable to detect large-scale mobile fronts (Schemm et al., 2015). The fact that e is conserved during moist adiabatic motion and thus reflects both the temperature and humidity of an airmass, makes it a powerful variable for

4.3 Winter evolution and association of precipitation events identifying airmasses of different origin and with contrasting thermodynamic characteristics (i.e. to separate cold and dry polar air masses from subtropical warm and moist air masses). Over high orography such as the Alps a pressure level of 700 hPa for identifying near surface fronts using the e field has proven adequate (Jenkner et al., 2010).

Additionally, to the manual analysis of e maps over Europe, the airmass transport history of air parcels arriving in a vertical column above the WFJ site was analyzed. The trajectories were calculated every 6 hours, for 10 days back in time with an output of the air parcels’ positions every hour along the back trajectories. The trajectories were started from the surface up to 500 hPa in vertical steps of 50 hPa above the site. These trajectories were computed based on the three-dimensional wind fields from the ERA5 dataset using LAGRANTO version 2.0 (Sprenger and Wernli, 2015; Wernli and Davies, 1997). For each precipitation event, the moisture source regions of the air parcels precipitating above the WFJ measurement site (i.e. showing a decrease in specific humidity at the arrival point at a relative humidity of at least 80%) were determined based on the algorithm presented in Sodemann et al. (2008) and applied as in Aemisegger (2018). To obtain event-based precipitation estimates, the individual six- hourly time steps were weighted using the sum of the large-scale and convective precipitation rates from ERA5 interpolated to the WFJ location.

4.3.3 Snowpack model for associating snow layers to precipitation events The allocation of the different layers in the snow profile to the corresponding precipitation events was performed based on the seasonal density development simulated by the snow cover model SNOWPACK version v1473 (Bartelt and Lehning, 2002). Based on meteorological input data, SNOWPACK simulates the formation and development of snow layers including metamorphism and settlement effects (Lehning et al., 2002). The simulation for the WFJ are based on air temperature, relative humidity, snow surface temperature, wind speed, short- and longwave radiation from the local weather station at the WFJ site. Recent studies showed a good agreement between the simulated and measured density and snow height development for WFJ (Neige et al., 2019, Richter et al., 2019). Furthermore, SNOWPACK is able to simulate the liquid water content and liquid water flow in snow using the Richards equation (Wever et al., 2014). We use this information in combination with local weather data for the interpretation of the processes taking place during the melt season (Section 4.4.5.2).

4.4 Results and Discussion

Table 4.2: List of major snowfall events with total amounts of at least 5 mm classified into cold and warm events based on the synoptic flow between 1.11.2016 and 1.6.2017. Dates refer to manual snow observation days from 8 am to 8 am local time. The weather charts refer to Fig. 4.7, the ẟ18O profile column refers to the profiles in Fig. 4.5. The highlighted cold (blue) and warm (red) events are discussed in more detail in the text. WF denotes a warm front, CF a cold front, OF an occluded front and L a low-pressure system with a pressure minimum located over the Alps. The type of frontal system was identified by eye using maps of e at 700 hPa from ERA5.

Nr. Start End Precipitating New snow Front T Weather ẟ18O airmass [cm] [°C] chart profile 1 04.11.16 06.11.16 warm 5 WF -1 A 2 06.11.16 08.11.16 cold 39 CF -13 Fig. 4a B 3 09.11.16 11.11.16 warm 21 WF -8 C 4 18.11.16 20.11.16 warm 12 CF -5 C 5 25.11.16 28.11.16 warm 7 OF -5 C 6 18.12.16 20.12.16 cold 5 - -11 D 7 24.12.16 27.12.16 warm 8 OF -5 - 8 02.01.17 06.01.17 cold 38 CF -15 D 9 07.01.17 09.01.17 cold 13 WF/OF -14 D 10 11.01.17 14.01.17 warm 49 WF -8 Fig. 4b E 11 14.01.17 17.01.17 cold 19 CF -18 Fig. 4c F 12 31.01.17 02.02.17 warm 56 WF -7 G 13 03.02.17 09.02.17 Alternating 22 CF/WF -10 H, I cold/warm /OF 14 10.02.17 11.02.17 cold 5 L -10 J 15 21.02.17 22.02.17 warm 7 WF -3 Fig. 4d K 16 23.02.17 25.02.17 cold 5 CF -10 L 17 27.02.17 03.03.17 Alternating 31 -8 L CF cold/warm 18 04.03.17 08.03.17 cold 60 CF -9 Fig. 4e M 19 08.03.17 10.03.17 warm 59 WF -4 Fig. 4f N 20 18.03.17 19.03.17 warm 15 WF -2 O 21 21.03.17 23.03.17 warm 8 L -3 P 22 03.04.17 05.04.17 warm 11 L -4 P 23 15.04.17 17.04.17 warm 40 CF/OF -6 P 24 18.04.17 20.04.17 cold 23 CF/OF -11 25 25.04.17 27.04.17 warm 16 WF -2 26 27.04.17 29.04.17 cold 21 CF -9 27 01.05.17 02.05.17 cold 22 CF -7 28 02.05.17 05.05.17 warm 11 L -4 29 06.05.17 08.05.17 warm 24 OF -2 30 08.05.17 10.05.17 cold 5 CF -5

4.4 Results and Discussion

4.4.1 Physical properties of the snow cover The temperature, snow density and manual snow type profiles provide important information on the thermodynamic properties and the type of metamorphism (dry or wet) that affects the vertical

4.4 Results and Discussion structure of the snow cover and might alter the isotopic signature. In preparation for the evaluation of the isotope signal in the snow, this section discusses the evolution of the physical properties of the snow cover. The temperature record of the air at 2 m above the ground shows that during the entire measurement campaign there were repeatedly short periods in which the zero-degree limit was exceeded (Fig. 4.3). However, this does not necessarily lead to a melting of the snow at the surface, due to longwave radiation loss. More reliable is the snow surface temperature where the zero-degree temperature indicates a melting surface snow. An overview of the thermal characteristics of the snowpack on the five profile sampling days is given in Table 4.3. Table 4.3: Thermal characteristics of winter 2016/17 on WFJ. Air temperature (TA) was measured automatically 2m above ground (ENET, WFJ2). Snow surface temp (TSS) was measure automatically (ENET, WFJ2). For number of days that are on or above 0° C the 24h mean temperature was calculated. Date Snow Air Water Sampling date snow Aver. TSS Nr. of days Aver. TA since Nr. of days TA Nr. of days with depth since last TSS = 0°C since last sampling above 0° C measured (cor. traditional sampling last sampling since last discharge snow profile) sampling (Lysimeter) [cm] [°C] [-] [°C] [-] [-] 25 Jan 2017 87 - - - - - (24 Jan) 22 Feb 2017 126 -11.0 ± 3.6 0 -4.9 ± 2.1 0 0 (22 Feb) 21 Mar 2017 185 -7.6 ± 2.6 0 -3.4 ± 3.4 5 0 (21 Mar) 17 Apr 2017 166 -4.7 ± 1.3 0 -0.2 ± 2.5 14 0 (19 Apr) 1 Jun 2017 83 -3.3 ± 4.2 12 0.3 ± 5.9 24 21 (29 Mai)

Using the traditional snow profiles sampled on the five profile sampling days, it becomes possible to estimate the impact of surface melting on the snow cover properties. In Figure 4.4, the melt affected areas classified as melt forms MF, respectively melt-freeze crust (Fierz et al., 2009), are marked in red colour. In the snow profile of 24 January, melt forms were found at the height of 18.5-17.5, 22.0-20.0 and 30.5-29 cm. These can also be found in the profile of 22 February (26.0-24.0 cm). Two additional melt events were detected in the top area 112.0-111.8 and 118.0-117.7 cm. All melt affected areas were limited to a thickness of 0.2 to 2 centimetres. In the profile from 21 March the former melt events were still visible in the lowest part of the snow cover (below 23 cm) and two more were found at the height of 174.5-173.5 and 185-182.5 cm. The snow profile on 17 April showed a completely different pattern: the top 56 cm (194-138 cm) were still classified as "decomposing and fragmented precipitation particles PP” and “rounded grains RG", but below there were exclusively melt forms MF present. Furthermore, the snow cover was completely isothermal underneath a height of 120 cm –

4.4 Results and Discussion a clear indicator for liquid water inside the snowpack. On 1 June, the snow cover was melted down to 83 cm and consisted completely of melt forms and melt crusts MF.

Figure 4.4: Snow profiles including snow classification after IACS, δ18O profile (black), density (orange), hand hardness (as column, from 1 Finger F (soft) to Knife K (hard)). Hand hardness is a qualitative parameter and largely correlates with the density profile. It is useful in identifying layer boundaries within the profile. The hand hardness abbreviations are PP for Precipitation Particles; DF for Decomposing and Fragmented precipitation particles, RG for Rounded Grains; FC for Faceted Crystals; DH for Depth Hoar; SF for Surface Hoar; MF for Melt Forms; ICE for Ice Formations.

The combination of temperature data and snow profiles thus indicates that dry conditions were dominating up to and including the third sampling day on 21 March. Individual short-term melting events can be clearly delimited to individual layers of a few centimeters in the snow profile. These melt events did not have any obvious impact on the isotope profile, since they are associated with both layers that are rather depleted in heavy isotopes (e.g. 23 cm on 25 January) and layers that are rather enriched (e.g. 20 cm on 25 January). In the fourth profile sampled on 17 April the majority of the snow cover was already isothermal and thus wet. On the last sampling day on 1 June, about 35% of the snow cover had melted off (calculated based on the water equivalent, Table 4.1) and the snow cover was completely saturated (maximum water content 12%). This enables a separation of the five sampled profiles into ones affected by dry (25 Jan, 22 Feb and 22 Mar) and ones affected by wet metamorphism (17 Apr and 1 Jun).

4.4.2 Five profiles of ẟ18O in snow Figure 4.4 illustrates the isotope profiles in combination with the traditional snow profiles. The high variability within the snow cover both in the isotope signature and in the density profile is evident. The snow profile of 21 March (Fig 4.4c) with the highest snow depth of winter shows a maximum amplitude of the 18O signal of 12‰. The snow density in March varies by a factor of two, between 230 kg/m3 and 440 kg/m3. The origin of this variability must be considered separately: The variable isotope signal is primarily determined by different frontal passages, large-scale advection of air masses with different

4.4 Results and Discussion origins and histories, and cloud processes. Although the same processes also determine the physical properties of the snow (such as density and grain shape), they only define them in new snow (Kikuchi et al., 2013). The structural parameters recorded in the snow pit are primarily defined by post- depositional processes such as snow metamorphism and settlement, while our hypothesis is that the ẟ18O signal is still mainly the one originally deposited. This contrasting impact of post-depositional processes on the isotope signal compared to the snow density also explains, why we find no correlation between structural parameters at the day of sampling such as density or the snow type classification and snow ẟ18O. In order to compare the different δ18O profiles among each another, the snow depth was converted into water equivalent (w.eq.) by multiplication with the corresponding density of the traditional snow profile (Fig 4.4). This eliminates settling effects taking place during the season. Furthermore, for Figure 4.5 all profiles were aligned to the one with the maximal snow depth (17 April). Thus 67 cm w.eq represents the common ground horizon, while the surface of the snow cover varies as a function of the snow depth. The range 45 to 67 cm w.eq is represented in all five snow profiles. The examination of the first three profiles (Jan, Feb and Mar) reveals similar characteristics: Two wide local δ18O maxima are visible: The local maximum C is at a height of 50 - 60 cm w.eq (-16.5‰) and the local maximum K at a height of 27 - 30 cm w.eq (-12‰). Layer C was deposited in January and can be assigned to an extratropical cyclone (Section 4.4.4.2). Layer K has been deposited by snowfalls on 21 February, one day before the second profile was sampled. The absolute minimum value is -22‰ at a height of 44 cm w. eq. (local minimum F) and was formed by a warm precipitation event (Section 4.4.4.1). The profile of 17 April generally shows a similar pattern as the previous one in March However, in the upper half, a distinct smoothing is visible: both local minima (18 and 33 cm w. eq., local minima M and J) and local maxima (10 and 27 cm w. eq., local maxima N and K) were attenuated by 2-3 ‰. From the snow physics data (Chapter 4.4.1) we know that at this time the snow cover was already isothermal and therefore liquid water was present. Therefore, wet snow metamorphism is likely a key player in these small shifts as will be discussed below (Section 4.4.4.5). The δ18O profile of 1 June reveals the largest differences compared to the previous profiles: On the one hand, the snow

4.4 Results and Discussion

Figure 4.5: Records δ18O in five snow pits at WFJ above Davos in winter 2016/2017. The ground surface is at 67 cm w. eq. and the surface of the snow cover at different snow water depths depending on the snow cover extent. The letters indicate individual local maxima and minima in δ 18O, which were associated with individual warm or cold phases of the precipitation events listed in Table 4.2. The first three snow pits, 25 Jan 2017 (black), 22 Feb 2017 (red) and 21 Mar 2017 (blue), were taken in cold conditions, 17 Apr 2017 (green) was isothermal and partial melted, during the last one on 01 June 2017 (violet) the snowpack was wet and percolating meltwater was observed. depth was reduced by about 50% and on the other hand, the entire profile was shifted towards more enriched δ18O values than in the other four profiles. The general shape including local minima and maxima (between 50 and 60 cm w.eq.) however remains the same. In the following sections, we first describe the atmospheric processes affecting the variability in 18O of water vapor and fresh snow during the measurement period (Section 4.4.3), then we discuss the archival of the atmospheric isotope signals in the snow cover (Section 4.4.4) and, finally, we analyze the snow physical processes that have altered the isotopic signal between the different profile days (Section 4.4.5).

4.4.3 Variability in air and snow surface δ18O signal The δ18O of water vapor measured between 1 February and 10 May 2017 on WFJ shows very large synoptic timescale variability induced by different airmasses with contrasting isotope signatures (Fig. 4.6). Hourly mean values between -40‰ and -18‰ were measured at the site, with a slight increasing trend from winter to spring. On a multiday average the δ18O increases from -28‰ in the beginning of

4.4 Results and Discussion

February to -20‰ in the beginning of May (not shown). Frontal passages and changes of airmasses have been shown to drive the synoptic timescale variability in water vapor isotope signals measured in the pre-Alps and the northern Swiss plateau region earlier on (Aemisegger et al., 2014, 2015). The measurements from WFJ presented here, are the first multi-month hourly dataset of stable water isotopes in water vapor from a high alpine snow-covered site. Several frontal passages show a marked decrease in the δ18O signal of water vapor such as on 18 February, on 25 February, in mid-March, or at the end of April. Mostly these sharp decreases in δ18O occur during the arrival of a cold air mass and are associated with a cold front passage. One exception with a short very strong δ18O drop over a few hours during a warm front passage will be discussed in detail in Section 4.4.4.3. The latter sharp decrease in vapor δ18O was probably due to the arrival of air parcels that have experienced important rain out of heavy isotopes due to precipitation formation underway and/or possibly due to strong convective downdrafts. During periods of warm air advection from the North Atlantic such as on 21-22 February high δ18O in water vapor were measured (see also Section 4.4.4). The δ18O timeseries measured in water vapor thus reflects changes in airmasses and shows a positive correlation with local surface temperature measurements at the synoptic timescale (r=0.46 for 36h averages).

Figure 4.6: Combination of air and snow signal of δ18O. Upper Panel: Dashed line: δ18O signal in snow surface from weekly sampling (top 3 cm, triplicate analysis), solid line: δ18O signal in air from continuous vapor measurement (calculated to 1 hour mean). The missing signal between 04. and 13. March is due to an instrument failure. In the lower panel, air temperature (TA) measured 2m above ground and snow surface temperature (TSS) and the daily fresh snow amount (measured at 08:00 local time) is shown.

The surface snow samples show a similar bi-weekly behavior as the vapor data and a trend towards more enriched values in spring compared to winter. The surface snow is more enriched than the vapor

4.4 Results and Discussion at 2-4 m above the snow surface by 5‰ on average. The equilibrium ice hypothetically formed from the measured water vapor would be slightly more enriched in heavy isotopes than the observed fresh snow. Using the equilibrium fractionation factors from Merlivat and Nief (1967) at temperatures in the range 0 to -20°C we obtain snow δ18O values of between -3‰ and -12‰. The fact that the hypothetical equilibrium ice formed from ambient vapor at the measurement site is mostly more enriched than the sampled surface snow indicates that the surface snow was formed at higher altitudes from water vapor that was more depleted in heavy isotopes than the near-surface vapor. The differences between surface snow and vapor δ18O tend to be smaller during days with precipitation than during days without. Large differences of 25‰ can occur on days such as the 26th of February, when strong cold air advection brought very depleted water vapor without fresh snow. These large differences are due to the fact that the surface snow mainly reflects the isotope signature of the previous snowfall, in this case a warm precipitation event from 21 February with a comparatively high snow δ18O and the vapor is strongly depleted due to the cold advection. During the dry days without precipitation following the cold event of the 25 February, the surface snow shows a slight enrichment signal of 2‰ in δ18O, possibly due to preferential sublimation of the light water molecules. Consequently, an enrichment in heavy molecules in the remaining surface snow can be observed (Stichler et al., 2001). A similar trend can be identified in the period 22 to 27 March. Events with marked positive and negative anomalies in the δ18O of surface snow can be found again in the five sampled snow profiles. For example, the maximum δ18O in surface snow of -9‰ on the 14th of March stems from the warm precipitation event Nr. 19 (Table 4.2) and gets archived as peak “N” in the snow profiles of March and April (Fig. 4.5). Similarly, the surface snow δ18O minimum on 5th of March originally comes from the fresh snow on that day associated with a cold front passage and precipitation event Nr.18. This minimum can be found in the March and April profiles again in the form of the local minimum “M”.

4.4.4 Archival of frontal passages in the seasonal snow cover Based on the findings of the last sections we formulate the hypothesis that the observed δ18O variability in the snow profiles is mainly induced by the advection of airmasses with different thermodynamic properties. Snow formed from cold airmasses generally corresponds to depleted snow layers, snow from warm airmasses corresponds to enriched snow layers. In the following, we discuss 6 snowfall events (listed in Table 4.2) and their archival in the snow cover. We first discuss the large- scale meteorological conditions associated with these events and then their corresponding δ18O signature in the snow cover.

4.4 Results and Discussion

4.4.4.1 Six examples of cold and warm precipitation events For further analysis we have selected the six precipitation events with the largest influence on the characteristics of the winter 2016/2017 among the 30 major snowfall events listed in Table 4.2. Here, we discuss the large-scale weather context of the events highlighted in Table 4.2 in blue for cold events and in red for warm events. Each of these events contributed to the observed δ18O variability in the five snow profiles (Figs. 4.4 and 4.5). Precipitation during cold events resulted from forced orographic uplift of airmasses from the northern part of the North Atlantic (Fig. 4.7) with a large fraction of moisture sources located North of 45°N (Fig. 4.A1). The large-scale context of these events is characterized by an elongated trough reaching far South and favoring the advection of polar and subpolar airmasses over the North Atlantic towards central Europe and Switzerland (Fig. 4.7a,c,e). In all three cases, a ridge is building up over the eastern North Atlantic. The northerly flow towards the Alps in the lower troposphere is established by a relatively weak cyclone located over northeastern Europe and a meridionally elongated anticyclone in the eastern North Atlantic. Precipitation during these cold events forms in an environment of anomalously low e (Fig. 4.8a,c,e). Warm precipitation events at the WFJ measurement site are generally associated with a strong zonal flow (Fig. 4.7a,c,e) with warm and moist air transport from the subtropical and midlatitude North Atlantic (South of 45°N, Fig. 4.8d,f and Fig. 4.A1) and/or from the Mediterranean (Fig. 4.8e). In all three cases, a deep surface cyclone is moving zonally over northwestern Europe. In two cases (21 February and 8 March, Fig 4.7d,f) a zonally elongated anticyclone is present over southwestern Europe. Note that often several consecutive days of precipitation are associated with two distinct events, one warm and one cold event depending on the dominating airmass in which the clouds are formed (e.g. events 10 & 11 are associated with a cold front passage or 18 & 19 associated with a warm front passage in Table 4.2). The three exemplary cold events are associated with a low δ18O anomaly in the sampled snow profiles (local minima B, F, M in Fig. 4.5). As opposed to the low δ18O of cold airmasses, warm airmasses led to several δ18O local maxima in the snow profiles (E, K, N in Fig. 4.5). These six events shortly presented here thus confirm the hypothesis above with warm airmasses leading to more enriched snow layers and cold airmasses to more depleted snow layers. In the next Section, we will take a closer look at the 18O archival of a multi-day precipitation period from 11 to 17 January with a warm event from 11 to 13 January and a cold event from 13 to 17 January during the passage of winter storm Egon in northern Germany.

100

4.4 Results and Discussion

Figure 4.7: Six selected precipitation events, their large-scale atmospheric flow context and air parcel history. The trajectories are shown in lines colored after the air parcel’s pressure. In grey contours the sea level pressure, in blue contours the geopotential height at 700 hPa. The black cross indicates the location of the WFJ measurement site.

101

4.4 Results and Discussion

Figure 4.8: Six selected precipitation events and their large-scale atmospheric flow context. In colored shading the equivalent potential temperature in K at 700 hPa and in black contours the sea level pressure. The black cross indicates the location of the WFJ measurement site.

102

4.4 Results and Discussion

4.4.4.2 Event analysis winter storm Egon: 11-17 January 2017 The analysed precipitation period consists of event 10 and 11 (Table 4.3, Fig. 4.7b,c and 4.8b,c) and led to the deposition of the local maximum E and local minimum F within the snowpack (Fig. 4.5). A sharp decrease in the δ18O is visible in the January profile in Figure 4.5 on a height between 44 and 51 cm w.eq.: The δ18O signal rises first from -18‰ to -15‰ before dropping to -24‰ (E to F in Fig. 4.5). The snow in this part of the profile can be attributed to a 7-day precipitation event from 11 to 17 January (E and F in Fig. 4.3, Fig. 4.7b,c and 4.8b,c): The extratropical cyclone Egon West of Brittany first transported warm air towards Switzerland, with a temperature of -4.3°C on the WFJ in the evening of the 12 January. At 00 UTC on 13 January, the cyclone was located over northern Germany (Fig. 4.8b). A warm front on 12 January brought 10 cm of fresh snow, with a snowfall line at 1400 m.a.s.l. (Avalanche Bulletin SLF, 13.01.2017). The warm front can be seen in the vertical cross section in Fig. 4.9a with the warm air gliding upwards along the moist isentropes and producing an elongated mixed phase cloud with large amounts of supercooled cloud liquid water content and snowfall over the alpine ridge. The local maximum value of -15‰ in δ18O of the snow profile of 25 January can be attributed to this snowfall. On 13 January, a cold front followed (Fig. 4.7c, Fig. 4.9b). The temperatures decreased significantly on the WFJ. The deep cold airmass extending vertically up to 500 hPa reached WFJ at around 3 UTC on 13 January (Fig. 4.9b). At the front, a mixed phase cloud reaching up to 400 hPa can be observed in Figure 4.9b. The arrival of the relatively moist and unstable stratified polar air led to intensive snowfall on the Northern side of the Alps which amounted to 36 cm of fresh snow within 24 hours. The lowest values of -22‰ in the profile can be assigned to this snow. The local maximum in the snow δ18O profile in E and the local minimum in F, respectively the associated temperature decrease can be attributed directly to the passage of this cold front. The e contrast of ~10 K between the two airmasses can be clearly seen in the cross sections shown in Figure 4.9.

103

4.4 Results and Discussion

Figure 4.9: Longitudinal cross section of the warm front at 00 UTC on 12 January in a) and the cold front at 3 UTC on 13 January 2017 in b), showing the equivalent potential temperature in filled contours and hydrometeors in colored contour lines from ERA5. Cloud liquid water content is shown in blue, rainwater content in red, cloud ice water content in white, and snow water content in grey. All hydrometeor contour lines are shown for 10 to 110 mg/kg in steps of 20 mg. The magenta contour line shows the 0°C isotherm. The green cross indicates the location of the WFJ measurement site. The geographical location of the cross section in shown in Figure 4.4b.

Although, in most cases, precipitation from warm air masses was associated with a high δ18O in snow and precipitation from cold air masses with a low δ18O in snow, exceptions to this pattern exist. One event of particularly low δ18O in snow deposited during a warm front passage and precipitation produced from a warm air mass was observed on 18-19 March 2017 and will be discussed in the next Section.

4.4.4.3 Event analysis of a warm front on 18-19 March 2017 In the association of the different precipitation events with a warm or cold airmass in Section 4.3.2 we isolated an event with exceptionally low δ18O during a warm precipitation event: On 18-19 March surface snow and air temperatures were relatively high, i.e. close to the freezing point and very wet

104

4.4 Results and Discussion snow fell on these two days during the passage of a warm front. The flow towards the Alps was zonal as for the other three warm precipitation events discussed above, with orographically forced uplift of the warm and moist airmass from the North Atlantic (Fig. 4.10a). The longitudinal cross section across the Alps in Figure 4.10c shows very deep clouds with large specific hydrometeor water contents forming at the front with a vertical extent reaching up to the tropopause level. It is likely that precipitation prior to the arrival of the airmass at the WFJ and the unusually deep clouds resulted in snowfall deposited on the WFJ that was very depleted in heavy isotopes compared to other precipitation events from warm air masses. The analysis showed that frontal passages, as the most typical weather phenomenon in the Alp, can lead to rapid change in water vapor and new snow water isotopes. The majority of the warm and cold precipitation events analyzed here show a strong impact of the large-scale transport of airmasses with cold, polar airmasses leading to precipitation with a low δ18O and warm airmasses with subtropical and midlatitude North Atlantic sources leading to precipitation with a high δ18O. Some exceptions nevertheless exist, with the exceptional warm front case from 18-19 March that led to snowfall with low δ18O, even if the precipitation was formed a warm airmass. This finding is in agreement with a study using idealized numerical model simulations with the isotope-enabled model COSMOiso from Dütsch et al. (2016), who showed that the effect of horizontal transport was the main driver of the contrasting δ18O signals in the cold and the warm sector of extratropical cyclones, and that effects of vertical transport, the vertical extent of the clouds and fractionation due to phase changes in and below the clouds can be important at smaller scales close to the fronts. This finding could be helpful to explain the observed poor correlation of the isotopic signal and temperature in alpine firn cores by Bohleber et al. (2013) and Mariani et al. (2014).

105

4.4 Results and Discussion

Figure 4.10: Overview of the large-scale conditions from ERA5 at 12 UTC on 18 March for the precipitation event from a warm airmass on 18-19 March (event 20) during a warm front passage over the Alps. Back-trajectories arriving in the layer between the surface and 500 hPa colored with pressure in a) including sea level pressure in black contours and the geopotential height at 700 hPa in blue contours. Equivalent potential temperature (e) at 700 hPa distribution (colored contours) over Europe with sea level pressure in black contours in b). Longitudinal cross section through the Alps at 12 UTC on 18 March 2017 in c) with filled contours and hydrometeor contour lines as in Figure 4.9. The cross-section location is shown in b).

106

4.4 Results and Discussion

4.4.5 Post-depositional Processes After the discussion on the δ18O variability introduced by atmospheric processes into the snow cover, we now focus on the impact of post-depositional processes. By comparing the five isotope profiles as shown in Figure 4.5, we analyze the development of the isotopic signal deposited in the snow cover, throughout the snow season and discuss the impact of dry and wet metamorphism.

4.4.5.1 Preservation of the snow δ18O during cold winter despite dry metamorphism The good preservation of the isotope signal at low temperatures, as revealed in the first 3 sampled profiles (Jan, Feb, Mar), is consistent with many studies (Mosley-Thompson et al., 1985; Grootes and Stuiver, 1997; Jouzel et al., 1983; Masson-Delmotte et al., 2008). Several other studies also suggest that sublimation causes isotopic fractionation thus altering the isotope composition of the snow after its deposition (Christner et al., 2017; Ebner et al., 2017; Moser and Stichler, 1975; Sokratov and Golubev, 2009; Steen-Larsen et al., 2014). A closer comparison of the first three profiles in Figure 4.5 reveals a slight smoothing between the January and February profiles, and between the February and March profiles, in the range of around 1 to 2‰. For instance, the local maxima and minima in the January profile (48 and 51 cm w.eq) are missing in the February profile. A plausible explanation for the smoothing are cycles of sublimation, vapor transport and resublimation during dry metamorphism. The directed vapor flux generated by the temperature gradient migrates from crystal to crystal without creating a continuous flow-through from the bottom to the top (Yosida, 1955). This leads to a local mixing of the water molecules resulting in a slightly smoothed δ18O profile. However, in the first three profiles, no continuous enrichment can be seen, as it has been observed in other studies at the snow surface (Gurney and Lawrence, 2004; Moser and Stichler, 1975; Neumann et al., 2008; Sokratov and Golubev, 2009; Stichler et al., 2001). This can be the result of the different vapor flux paths at the surface and within the snow cover: At the surface sublimated water molecules can escape into the atmosphere. Assuming preferential volatilization of the lighter 16O molecules, the δ18O in the remaining snow is thereby increased. Within the snow cover, however, the transport path of water vapor is limited and the vapor molecules re- sublimate at the next ice crystal (Pinzer et al., 2012; Sokratov and Maeno, 1997). Therefore, a potential fractionation on a scale larger than the grain size has hardly any influence (Taylor et al., 2001). This explains the smoothing leading to smaller changes in δ18O in the profile without changing the bulk δ18O of the snowpack. The temporal change of the δ18O minimum at a height of 44.5 cm w.eq (local minimum F in Fig 4.5), demonstrates the challenges, which the interpretation of the isotope profiles entails. The minimum at this height reveals a continuous enrichment over the first 3 months: The δ18O value increased from -22.8‰ in January to -22.0‰ in February and finally to -21.1‰ in March. Until April,

107

4.4 Results and Discussion the δ18O value remained constant before changing to -16.7 ‰ due to melting processes (Section 4.4.1). Over the 4 months, this particular snow was exposed to various processes: The layer can be assigned to a cold precipitation event in mid-January (Table 4.2, event 11) following the passage of the extratropical cyclone Egon over Northern Germany. Until the following snowfall on 1 February, direct exchange with the atmosphere as well as sublimation processes were possible. The first sampling took place on 25 January during the period of atmospheric exposure. The snowfall from 31 January to 1 February covered the layer with 47 cm of new snow (event 12, Table 4.2) and thus stopped any surface exchange processes. Which part of the change of 0.8‰ visible in comparison with the profile from 22 February occurred before and after the snowfall event cannot be determined retrospectively. Until 21 March, the layer was enclosed even deeper in the snow cover due to additional precipitation events (Table 4.2). The δ18O signal increased by a further 1.1‰ during this period. The overlaying also led to continuous compaction accompanied by the settlement of the snow cover. The density of the corresponding layer increased from 170 kg/m3 in January to 350 kg/m3 in February to 450 kg/m3 in March (Fig 4.4). Although the traditional snow profiles (Fig. 4.4) show that the classification of faceted crystals FC for the layer has not changed, settlement can have affected the sampled δ18O profile: Despite a constant absolute resolution (6 cm), the water equivalent (or information density) per sample increases. If the vertical δ18O signal variation in the snow cover is large, this can have a significant influence. The δ18O change in the F minimum from January to February can be explained solely by settlement effects around the height of 44.5 cm w.eq. However, the change between the February and March profiles are too large to be explained by settlement only. To summarize, the δ18O signal in the snow profiles sampled during the cold phase of the winter 2016/2017 shows variations of ~9‰ between local maxima and minima in δ18O. These variations were associated with the δ18O signal of the fresh snow and only minor smoothing effects of 1-2‰ could be attributed to post-depositional processes related to dry metamorphism and/or compaction of the snow cover.

4.4.5.2 Smoothing and shifting of the δ18O record during spring During the last two profile sampling dates in April and June, the snowpack was wet (Section 4.4.1). As soon as rain- or melt-water is present in the snow cover, a complex system of recrystallization, mixing and fractionation processes is established (Taylor et al., 2001). In order to discuss the observed smoothing of up to 4‰ in Mai and the shift in June, we discuss in the following 3 sub-processes, that affect the isotopic composition of the snowpack during snowmelt. The separation into a), b), and c) is based on the funicular and pendular system in the snowpack (Denoth, 1980).

108

4.4 Results and Discussion a) Meltwater content Water from the surface percolates into the snow cover and is retained in there ("pendular system"): If this meltwater or rainwater has a different isotope concentration than the snow cover, the bulk concentration is changed. However, the maximum water content is limited to below 6%, so no major shift in concentration is to be expected. Slow wet snow metamorphism begins (see Brun 1986). b) Wet metamorphism Transformation of ice particles by wet snow metamorphism and mixing: The liquid water leads locally to the melting of the small crystals and to the growth of the large ones (Denoth, 1982). This leads to an isotopic rem ixing. The dissolved small particles mix with the pore water. Subsequently, the flow system has a decisive influence: In the pendular system, the water of the dissolved crystals is available for crystal growth (process b). In the funicular System, however, the percolation water washes out the water of the dissolved crystals. The latter system has a much greater potential for concentration changes. c) Fractionation Although fractionation processes take place both during melting and freezing of water, they are relevant in this case only during (re)-freezing: The melting particles dissolve completely (melting at the surface and metamorphic dissolution of the small particles) and a bulk δ18O is formed from the snow (Taylor et al., 2001). Thus, short-term fractionation processes have no effect. With crystal growth, however, an equilibrium fractionation takes place (during the transition liquid-solid, 18O prefers to go into the solid phase, which leads to an enrichment of the ice particles. The liquid water, in contrast, becomes isotopically lighter. A coupling to process b) can then occur. As soon as a liquid water flux is present, the fractionation process is intensified, as the concentration differences become larger due to less depleted fresh water.

Figure 4.11: Change in isotopic profile during melt in pendular and funicular system in an idealized snowpack (homogeneous layered, 0°C isothermal). Original δ18O profile (black), smoothed δ18O profile due to non- percolating meltwater (orange), shifted δ18O profile due to meltwater percolation (green). At the transition from a pendular to a funicular system the percolating water from the upper (idealized) layer begins to mix with the dissolving particles from the layer with different δ18O.

109

4.4 Results and Discussion

Figure 4.12: Snowpack simulation of liquid water content within the snow matrix in winter 2016/17. Sampling Days are marked with red bars.

The pendular and funicular systems can explain the differences in the profiles from 17 Apr and 1 June. On both sampling days, the snow cover was already wet and therefore wet snow metamorphism was in progress. The snowpack simulations in Figure 7 show that from 21 March, liquid water content LWC inside the snow cover is increasing. From 4 April on the snowpack is completely wet, but the lowest layers are not completely saturated to 4% until May. Therefore, it can be assumed that the pendular system dominated in April: the subprocesses a) and b) took place relatively locally, due to the limited water input from the surface (begin of melt season). This led to substantial smoothing in the April profile compared to the March profile (Fig. 4.5). Until June the water input increased. As described above, the processes b) and c) are thereby becoming dominant. The wet snow metamorphism runs much faster (the wetter the system the higher the growth rate of the grains). Moreover, the high-water flow drives the fractionation by a further effect: The liquid water in the snow cover, already depleted by the re-deposition process (wet snow metamorphism), is replaced much faster with “fresh” water. This fresh not yet depleted water now shows a stronger fractionation effect during the identical re-deposition processes. If this water from higher layers originates from snow with a higher δ18O, as it is often the case due to its deposition later in the winter season when temperatures are increased, the effect alteration of the existing signal is even more pronounced. In summary, the enhanced infiltration of water can lead to an enrichment of the entire snow profile as can be observed in the profile of 1 June Beside the clear shift, the δ18O variability is still

110

4.5 Conclusion and Outlook preserved to some extent with a preservation of the local maxima E, C, K and the local minima F and B (Fig. 4.5). If this profile is preserved due to refreezing, the isotopic signal does not represent the absolute δ18O of the originally deposited snow anymore. However, thanks to the conservation of the main vertical variability pattern it still reflects the variability of the original atmospheric signal associated with frontal precipitation events.

4.5 Conclusion and Outlook Between February and June 2017, we studied the variability in the δ18O of an Alpine snow cover over using a combination of five snow profiles sampled every month, weekly snow surface data and hourly water vapour isotope data in near-surface ambient air. By choosing six exemplary precipitation events associated with frontal passages, we showed how the variability induced by atmospheric transport processes is recorded in the snow cover. The measured water vapor signal helped to understand how the signal gets deposited into the snow cover. The water vapour δ18O signal shows very large synoptic timescale variability induced by frontal passages. The δ18O associated with frontal passages and different airmasses with contrasting isotope signatures dominates over the seasonal trends. The surface snow samples show a similar bi-weekly behavior as the vapor data and a trend towards more enriched values in spring compared to winter. The surface snow was found to be more enriched than the vapor measured 2-4 m above the snow surface by around 5‰ on average. The δ18O profile in the snow cover shows several local maxima an minima over the winter. The δ18O signal from different layers in the snow cover could be assigned to individual frontal precipitation events by using the density information of the profiles and by taking the settlement into account as simulated by the snow model SNOWPACK. A detailed study of the weather situation and airmass origin were done using the ERA5 reanalysis dataset and back-trajectories computed with LAGRANTO based on the three- dimensional wind fields from ERA5. Generally, local maxima in δ18O in the snow profiles could be associated with precipitation from warm air masses and local minima in δ18O with precipitation from cold air masses. This association revealed furthermore that precipitation, which lasted longer than one day, is often associated with frontal passages and was formed by different consecutive events (warm and cold). This results in a large variability of the 18O signal, which is deposited within a relatively limited vertical extent within the snowpack. In one exceptional case, precipitation from a warm air mass has led to a local minimum of the δ18O in snow, due to a particularly deep cloud and heavy rain out underway. This finding might explain the observed poor correlation of the isotopic signal and temperature in alpine firn cores found in previous studies by Bohleber et al. (2013) and Mariani et al. (2014). The comparison of the monthly snow profiles showed, that the δ18O signal in the snow cover was conserved during the cold winter months (January, February, March). Short melt events at the

111

4.5 Conclusion and Outlook surface had no influence on the signal in the profile above the selected sampling resolution of 6 cm. There was a slight smoothing of the individual peaks in the range of 1-2‰, which we attributed to dry snow metamorphism. As soon as daily temperatures increase and the snow cover starts to melt, the availability of water and/or flow through the snowpack becomes crucial. In the wet profile (April) without water flow (pendular system) a smoothing of about 4‰ was visible. Until June multiple days of continuous water flow could be measured and the profile was shifted by up to 9‰ towards less negative values. We attribute this to a faster exchange processes and the intensive wet snow metamorphism in the snowpack. However, the general shape of the profile is maintained, even though the absolute values have changed. The use of a melt-affected profile for reconstructing weather events is therefore possible, however, it is associated with additional challenges due to the overall enrichment of the snow cover. In times of global warming and melt-events up to the poles this can become more and more relevant. This paper makes a first step towards the study of how the variability in thermodynamic properties of the atmosphere associated with atmospheric transport can be recorded in the seasonal Alpine snow. We could show that frontal passages can be reconstructed from the isotope signal. This finding may be useful to reconstruct meteorological events from the past using high resolution shallow Alpine ice cores.

Acknowledgements We thank the CRYOS-EPFL group (M. Lehning and H. Huwald) for lending us the Picarro analyzer for this campaign on the WFJ. We are grateful for the analysis of the snow samples with another Picarro analyzer at PSI lab and for the help in collecting samples of Sabina Brütsch. We thank Bettina Richter for the help with Snowpack simulations and Python evaluation. Further we like to thank Jacinta Edebeli, Matthias Jaggi and and Neige Calonne for their help of collecting snow pit samples. We thank Pirmin Ebner and Heini Wernli for useful feedback on an earlier version of this manuscript. Funding was provided by the Swiss National Science Foundation (SNSF) under Grant No. 155999.

112

Appendix Chapter 4

Cold events Warm events a) b)

x x

c) d)

x x

e) f)

x x

x Weissfluhjoch

Figure 4.A1: Moisture sources of six selected precipitation events from cold and warm airmasses.

References Chapter 4

References Aemisegger, F. (2018). On the link between the North Atlantic storm track and precipitation deuterium excess in Reykjavik. Atmos. Sci. Lett. 19, e865. doi:10.1002/asl.865. Aemisegger, F., Pfahl, S., Sodemann, H., Lehner, I., Seneviratne, S. I., and Wernli, H. (2014). Deuterium excess as a proxy for continental moisture recycling and plant transpiration. Atmos. Chem. Phys. 14, 4029–4054. doi:10.5194/acp-14-4029-2014. Aemisegger, F., Spiegel, J. K., Pfahl, S., Sodemann, H., Eugster, W., and Wernli, H. (2015). Isotope meteorology of cold front passages: A case study combining observations and modeling. Geophys. Res. Lett. 42, 5652–5660. doi:10.1002/2015GL063988. Aemisegger, F., Sturm, P., Graf, P., Sodemann, H., Pfahl, S., Knohl, A., et al. (2012). Measuring variations of δ 18O and δ 2H in atmospheric water vapour using two commercial laser-based spectrometers: An instrument characterisation study. Atmos. Meas. Tech. 5, 1491–1511. doi:10.5194/amt-5-1491-2012. Ala-Aho, P., Tetzlaff, D., McNamara, J. P., Laudon, H., and Soulsby, C. (2017). Using isotopes to constrain water flux and age estimates in snow-influenced catchments using the STARR (Spatially distributed Tracer-Aided Rainfall-Runoff) model. Hydrol. Earth Syst. Sci. 21, 5089–5110. doi:10.5194/hess-21-5089-2017. Albert, M., and Krajeski, G. (1998). A fast, physically based point snowmelt model for use in distributed applications. Hydrol. Process. 12, 1809–1824. doi:10.1002/(SICI)1099- 1085(199808/09)12:10/11<1809::AID-HYP696>3.0.CO;2-5. Albert, M. R., and Hardy, J. P. (1995). Ventilation experiments in a seasonal snow cover. Biogeochem. Seas. snow-covered catchments. Proc. Symp. Boulder, 1995, 41–49. Available at: http://hydrologie.org/redbooks/a228/iahs_228_0041.pdf [Accessed August 30, 2019]. Ambach, W., Dansgaard, W., Eisner, H., and Møller, J. (1968). The altitude effect on the isotopic composition of precipitation and glacier ice in the Alps. Tellus 20, 595–600. doi:10.3402/tellusa.v20i4.10040. Araguás-Araguás, L., Froehlich, K., and Rozanski, K. (1998). Stable isotope composition of precipitation over southeast Asia. J. Geophys. Res. Atmos. 103, 28721–28742. doi:10.1029/98JD02582. Avak, S. E., Trachsel, J. C., Edebeli, J., Brütsch, S., Bartels‐Rausch, T., Schneebeli, M., et al. (2019). Melt‐ Induced Fractionation of Major Ions and Trace Elements in an Alpine Snowpack. J. Geophys. Res. Earth Surf., 2019JF005026. doi:10.1029/2019JF005026. Avanzi, F., De Michele, C., Morin, S., Carmagnola, C. M., Ghezzi, A., and Lejeune, Y. (2016). Model complexity and data requirements in snow hydrology: seeking a balance in practical applications. Hydrol. Process. 30, 2106–2118. doi:10.1002/hyp.10782. Baltensperger, U., Schwikowski, M., Gäggeler, H. W., Jost, D. T., Beer, J., Siegenthaler, U., et al. (1993). Transfer of atmospheric constituents into an alpine snow field. Atmos. Environ. Part A, Gen. Top. 27, 1881–1890. doi:10.1016/0960-1686(93)90293-8. Bartelt, P., and Lehning, M. (2002). A physical SNOWPACK model for the Swiss avalanche warning: Part I: numerical model. Cold Reg. Sci. Technol. 35, 123–145. doi:10.1016/S0165-232X(02)00074-5. Benetti, M., Steen-Larsen, H. C., Reverdin, G., Sveinbjörnsdóttir, Á. E., Aloisi, G., Berkelhammer, M. B., et al. (2017). Stable isotopes in the atmospheric marine boundary layer water vapour over the Atlantic Ocean, 2012–2015. Sci. Data 4, 160128. doi:10.1038/sdata.2016.128.

References Chapter 4

Beria, H., Larsen, J. R., Ceperley, N. C., Michelon, A., Vennemann, T., and Schaefli, B. (2018). Understanding snow hydrological processes through the lens of stable water isotopes. Wiley Interdiscip. Rev. Water 5, e1311. doi:10.1002/wat2.1311. Bohleber, P., Erhardt, T., Spaulding, N., Hoffmann, H., Fischer, H., and Mayewski, P. (2018). Temperature and mineral dust variability recorded in two low-accumulation Alpine ice cores over the last millennium. Clim. Past 14, 21–37. doi:10.5194/cp-14-21-2018. Bohleber, P., Wagenbach, D., Schöner, W., and Böhm, R. (2013). To what extent do water isotope records from low accumulation Alpine ice cores reproduce instrumental temperature series? Tellus, Ser. B Chem. Phys. Meteorol. 65, 1–17. doi:10.3402/tellusb.v65i0.20148. Bond, G., Broecker, W., Johnsen, S., McManus, J., Labeyrie, L., Jouzel, J., et al. (1993). Correlations between climate records from North Atlantic sediments and Greenland ice. Nature 365, 143– 147. doi:10.1038/365143a0. Bonne, J.-L., Masson-Delmotte, V., Cattani, O., Delmotte, M., Risi, C., Sodemann, H., et al. (2013). The isotopic composition of water vapour and precipitation in Ivittuut, Southern Greenland. Atmos. Chem. Phys. Discuss. 13, 30521–30574. doi:10.5194/acpd-13-30521-2013. Brun, E. (1988). Investigation on wet-snow metamorphism in respect of liquid-water content. Ann. Glaciol. 13, 22–26. Available at: http://adsabs.harvard.edu/abs/1988AnGla..13...22B [Accessed September 12, 2011]. Calonne, N., Richter, B., Löwe, H., Cetti, C., Judith, Herwijnen, A. Van, et al. (2019). The RHOSSA campaign: Monitoring the seasonal evolution of an alpine snowpack up to daily resolution. in prep. Casado, M., Landais, A., Picard, G., Münch, T., Laepple, T., Stenni, B., et al. (2018). Archival processes of the water stable isotope signal in East Antarctic ice cores. Cryosphere 12, 1745–1766. doi:10.5194/tc-12-1745-2018. Celle-Jeanton, H., Gonfiantini, R., Travi, Y., and Sol, B. (2004). Oxygen-18 variations of rainwater during precipitation: application of the Rayleigh model to selected rainfalls in Southern France. J. Hydrol. 289, 165–177. doi:10.1016/J.JHYDROL.2003.11.017. Christner, E., Kohler, M., and Schneider, M. (2017). The influence of snow sublimation and meltwater evaporation on δD of water vapor in the atmospheric boundary layer of central Europe. Atmos. Chem. Phys. 17, 1207–1225. doi:10.5194/acp-17-1207-2017. Colbeck, S. (1986). Statistics of coarsening in water-saturated snow. Acta Metall. 34, 347–352. doi:10.1016/0001-6160(86)90070-2. Copernicus Climate Change Service (C3S) (2017). ERA5: Fifth generation of ECMWF atmospheric reanalyses of the global climate. Copernicus Clim. Chang. Serv. Clim. Data Store. Available at: https://cds.climate.copernicus.eu/cdsapp#!/home. Craig, H. (1961). Isotopic Variations in Meteoric Waters. Science 133, 1702–3. doi:10.1126/science.133.3465.1702. Cuffey, K. M., Clow, G. D., Alley, R. B., Stuiver, M., Waddington, E. D., and Saltus, R. W. (1995). Large Arctic Temperature Change at the Wisconsin-Holocene Glacial Transition. Science (80-. ). 270, 455–458. doi:10.1126/science.270.5235.455. Dansgaard, W. (1953). The Abundance of O 18 in Atmospheric Water and Water Vapour. Tellus 5, 461– 469. doi:10.3402/tellusa.v5i4.8697. Dansgaard, W. (1964). Stable isotopes in precipitation. Tellus 16, 436–468. doi:10.3402/tellusa.v16i4.8993.

115

References Chapter 4

Dansgaard, W., Johnsen, S. J., Clausen, H. B., Dahl-Jensen, D., Gundestrup, N. S., Hammer, C. U., et al. (1993). Evidence for general instability of past climate from a 250-kyr ice-core record. Nature 364, 218–220. doi:10.1038/364218a0. Denoth, A. (1980). The Pendular-Funicular Liquid Transition in Snow. J. Glaciol. 25, 93–98. doi:10.3189/s0022143000010315. Denoth, A. (1982). The pendular-fenicular liquid transition and snow metamorphism. J. Glaciol. 28, 357–364. doi:10.1017/S0022143000011692. Dütsch, M., Pfahl, S., and Wernli, H. (2016). Drivers of δ2H variations in an idealized extratropical cyclone. Geophys. Res. Lett. 43, 5401–5408. doi:10.1002/2016GL068600. Ebner, P. P., Steen-Larsen, H. C., Stenni, B., Schneebeli, M., Steinfeld, A., Ebner, P. P., et al. (2017). Experimental observation of transient δ18O interaction between snow and advective airflow under various temperature gradient conditions. Cryosphere 11, 1733–1743. doi:10.5194/tc-11- 1733-2017. Epstein, S., and Mayeda, T. (1953). Variation of O18 content of waters from natural sources. Geochim. Cosmochim. Acta 4, 213–224. doi:10.1016/0016-7037(53)90051-9. Feng, X., Taylor, S., Renshaw, C. E., and Kirchner, J. W. (2002). Isotopic evolution of snowmelt 1. A physically based one-dimensional model. Water Resour. Res. 38, 35-1-35–8. doi:10.1029/2001WR000814. Fierz, C., Armstrong, R. L., Durand, Y., Etchevers, P., Greene, E., McClung, D. M., et al. (2009). The international classification for seasonal snow on the ground ( UNESCO , IHP ( International Hydrological Programme )– VII , Technical Documents in Hydrology , No 83 ; IACS ( Internation ... UNESCO-IHP. Paris. Friedman, I., Benson, C., and Gleason, J. (1991). Isotopic changes during snow metaporphism. Stable Isot. Geochemistry A Tribut. to Samuel Epstein 3, 211–221. Available at: https://www.geochemsoc.org/files/8114/1269/7654/SP-3_211-222_Friedman.pdf [Accessed August 29, 2019]. Gedzelman, S. D., and Lawrence, J. R. (1990). The Isotopic Composition of Precipitation from Two Extratropical Cyclones. Mon. Weather Rev. 118, 495–509. doi:10.1175/1520- 0493(1990)118<0495:TICOPF>2.0.CO;2. Graf, P., Wernli, H., Pfahl, S., and Sodemann, H. (2019). A new interpretative framework for below- cloud effects on stable water isotopes in vapour and rain. Atmos. Chem. Phys 19, 747–765. doi:10.5194/acp-19-747-2019. Grootes, P. M., and Stuiver, M. (1997). Oxygen 18/16 variability in Greenland snow and ice with 10-3- to 105-year time resolution. J. Geophys. Res. Ocean. 102, 26455–26470. doi:10.1029/97JC00880. Gurney, S. D., and Lawrence, D. S. L. (2004). Seasonal trends in the stable isotopic composition of snow and meltwater runoff in a subarctic catchment at Okstindan, Norway. Nord. Hydrol. 35, 119– 137. Available at: https://iwaponline.com/hr/article-pdf/35/2/119/363523/119.pdf [Accessed August 24, 2019]. Jenkner, J., Sprenger, M., Schwenk, I., Schwierz, C., Dierer, S., and Leuenberger, D. (2010). Detection and climatology of fronts in a high-resolution model reanalysis over the Alps. Meteorol. Appl. 17, 1–18. doi:10.1002/met.142. Johnsen, S., Clausen, H., Cuffey, K. M., Hoffmann, G., Schwander, J., and Creyts, T. (2000). Diffusion of stable isotopes in polar firn and ice: the isotope effect in firn diffusion. in Physics of Ice Core Records, 121-140 (Hokkaido: Physics of Ice Core Records, 121-140), 121–140. Available at:

116

References Chapter 4

https://eprints.lib.hokudai.ac.jp/dspace/bitstream/2115/32465/1/P121-140.pdf [Accessed August 29, 2019]. Johnsen, S. J. (1973). Stable isotope homogenization of polar firn and ice. 210–219. Johnsen, S. J., Dahl-Jensen, D., Dansgaard, W., and Gundestrup, N. (1995). Greenland palaeotemperatures derived from GRIP bore hole temperature and ice core isotope profiles. Tellus B Chem. Phys. Meteorol. 47, 624–629. doi:10.3402/tellusb.v47i5.16077. Jouzel, J., Merlivat, L., Petit, J. R., and Lorius, C. (1983). Climatic information over the last century deduced from a detailed isotopic record in the south pole snow. J. Geophys. Res. 88, 2693–2703. doi:10.1029/JC088iC04p02693. Kikuchi, K., Kameda, T., Higuchi, K., and Yamashita, A. (2013). A global classification of snow crystals, ice crystals, and solid precipitation based on observations from middle latitudes to polar regions. Atmos. Res. 132–133, 460–472. doi:10.1016/J.ATMOSRES.2013.06.006. Lee, J., Feng, X., Posmentier, E. S., Faiia, A. M., and Taylor, S. (2009). Stable isotopic exchange rate constant between snow and liquid water. Chem. Geol. 260, 57–62. doi:10.1016/j.chemgeo.2008.11.023. Lehning, M., Bartelt, P., Brown, B., and Fierz, C. (2002). A physical SNOWPACK model for the Swiss avalanche warning: Part III: meteorological forcing, thin layer formation and evaluation. Cold Reg. Sci. Technol. 35, 169–184. doi:10.1016/S0165-232X(02)00072-1. Madsen, M. V., Steen-Larsen, H. C., Hörhold, M., Box, J., Berben, S. M. P., Capron, E., et al. (2019). Evidence of Isotopic Fractionation During Vapor Exchange Between the Atmosphere and the Snow Surface in Greenland. J. Geophys. Res. Atmos. 124, 2932–2945. doi:10.1029/2018JD029619. Mariani, I., Eichler, A., Jenk, T. M., Brönnimann, S., Auchmann, R., Leuenberger, M. C., et al. (2014). Temperature and precipitation signal in two Alpine ice cores over the period 1961-2001. Clim. Past 10, 1093–1108. doi:10.5194/cp-10-1093-2014. Marty, C., and Meister, R. (2012). Long-term snow and weather observations at Weissfluhjoch and its relation to other high-altitude observatories in the Alps. Theor. Appl. Climatol. 110, 573–583. doi:10.1007/s00704-012-0584-3. Masson-Delmotte, V. (2005). GRIP Deuterium Excess Reveals Rapid and Orbital-Scale Changes in Greenland Moisture Origin. Science (80-. ). 309, 118–121. doi:10.1126/science.1108575. Masson-Delmotte, V., Hou, S., Ekaykin, A., Jouzel, J., Aristarain, A., Bernardo, R. T., et al. (2008). A review of antarctic surface snow isotopic composition: Observations, atmospheric circulation, and isotopic modeling. J. Clim. 21, 3359–3387. doi:10.1175/2007JCLI2139.1. Merlivat, L., and Nief, G. (1967). Fractionnement isotopique lors des changements d‘état solide-vapeur et liquide-vapeur de l’eau à des températures inférieures à 0°C. Tellus 19, 122–127. doi:10.1111/j.2153-3490.1967.tb01465.x. Moser, H., and Stichler, W. (1975). Deuterium and oxygen-18 contents as an index of the properties of snow covers. in Snow Mechanics, Proceedings of the Grindelwald Symposium, April 1974 (Grindelwald), 442 pp. Available at: http://hydrologie.org/redbooks/a114/iahs_114_0122.pdf [Accessed August 29, 2019]. Mosley-Thompson, E., Kruss, P. D., Thompson, L. G., Pourchet, M., and Grootes, P. (1985). Snow Stratigraphic Record at South Pole: Potential for Paleoclimatic Reconstruction. Ann. Glaciol. 7, 26–33. doi:10.3189/s0260305500005863. Neumann, T. A. A., Albert, M. R. R., Lomonaco, R., Engel, C., Courville, Z., and Perron, F. (2008). Experimental determination of snow sublimation rate and stable-isotopic exchange. Cambridge University Press doi:10.3189/172756408787814825.

117

References Chapter 4

Pfahl, S., Wernli, H., and Yoshimura, K. (2012). The isotopic composition of precipitation from a winter storm-a case study with the limited-area model COSMO iso. Atmos. Chem. Phys 12, 1629–1648. doi:10.5194/acp-12-1629-2012. Pinzer, B. R., Schneebeli, M., and Kaempfer, T. U. (2012). Vapor flux and recrystallization during dry snow metamorphism under a steady temperature gradient as observed by time-lapse microtomography. Cryosph. 6, 1141–1155. doi:10.5194/tc-6-1141-2012. Raymond, C. F., and Tusima, K. (1979). Grain Coarsening of Water-Saturated Snow. J. Glaciol. 22, 83– 105. doi:10.3189/S0022143000014076. Rindsberger, M., Jaffe, S., Rahamin, S., and Gat, J. R. (1990). Patterns of the isotopic composition of precipitation in time and space: data from the Israeli storm water collection program. Tellus B 42, 263–271. doi:10.1034/j.1600-0889.1990.t01-2-00005.x. Ritter, F., Steen-Larsen, H. C., Werner, M., Masson-Delmotte, V., Orsi, A., Behrens, M., et al. (2016). Isotopic exchange on the diurnal scale between near-surface snow and lower atmospheric water vapor at Kohnen station, East Antarctica. Cryosph. 10, 1647–1663. doi:10.5194/tc-10-1647- 2016. Rozanski, K., Araguás-Araguás, L., and Gonfiantini, R. (1993). “Isotopic Patterns in Modern Global Precipitation,” in (American Geophysical Union (AGU)), 1–36. doi:10.1029/GM078p0001. Schemm, S., Rudeva, I., and Simmonds, I. (2015). Extratropical fronts in the lower troposphere-global perspectives obtained from two automated methods. Q. J. R. Meteorol. Soc. 141, 1686–1698. doi:10.1002/qj.2471. Schneebeli, M. (1995). Development and stability of preferential flow paths in a layered snowpack. Biogeochem. Seas. snow-covered catchments. Proc. Symp. Boulder, 1995, 89–95. Available at: https://iahs.info/uploads/dms/iahs_228_0089.pdf [Accessed August 30, 2019]. Sodemann, H., Aemisegger, F., Pfahl, S., Bitter, M., Corsmeier, U., Feuerle, T., et al. (2017). The stable isotopic composition of water vapour above Corsica during the HyMeX SOP1 campaign: Insight into vertical mixing processes from lower-tropospheric survey flights. Atmos. Chem. Phys. 17, 6125–6151. doi:10.5194/acp-17-6125-2017. Sodemann, H., Schwierz, C., and Wernli, H. (2008). Interannual variability of Greenland winter precipitation sources: Lagrangian moisture diagnostic and North Atlantic Oscillation influence. J. Geophys. Res. 113, D03107. doi:10.1029/2007JD008503. Sokratov, S. A., and Golubev, V. N. (2009). Snow isotopic content change by sublimation. J. Glaciol. 55, 823–828. doi:10.3189/002214309790152456. Sokratov, S. A., and Maeno, N. (1997). Heat and mass transport in snow under a temperature gradient. in Snow Engineering: Recent Advances, eds. Izumi, Nakamura, and Sack (Rotterdam: Balkema), 49–54. Sommerfeld, R. A., Friedman, I., and Nilles, M. (1987). The Fractionation of Natural Isotopes During Temperature Gradient Metamorphism of Snow. Seas. Snowcovers Physics, Chem. Hydrol., 95– 105. doi:10.1007/978-94-009-3947-9_5. Soulsby, C., Malcolm, R., Helliwell, R., Ferrier, R. C., and Jenkins, A. (2000). Isotope hydrology of the Allt a’ Mharcaidh catchment, Cairngorms, Scotland: Implications for hydrological pathways and residence times. Hydrol. Process. 14, 747–762. doi:10.1002/(SICI)1099- 1085(200003)14:4<747::AID-HYP970>3.0.CO;2-0. Sprenger, M., and Wernli, H. (2015). The LAGRANTO Lagrangian analysis tool – version 2.0. Geosci. Model Dev. 8, 2569–2586. doi:10.3929/ETHZ-B-000103598.

118

References Chapter 4

Steen-Larsen, H. C., Masson-Delmotte, V., Hirabayashi, M., Winkler, R., Satow, K., Prié, F., et al. (2014). Climate of the Past What controls the isotopic composition of Greenland surface snow? 10, 377– 392. doi:10.5194/cp-10-377-2014. Stichler, W., Schotterer, U., Fröhlich, K., Ginot, P., Kull, C., Gäggeler, H., et al. (2001a). Influence of sublimation on stable isotope records recovered from high-altitude glaciers in the tropical Andes. J. Geophys. Res. Atmos. 106, 22613–22620. doi:10.1029/2001JD900179. Stichler, W., Schotterer, U., Fröhlich, K., Ginot, P., Kull, C., Gäggeler, H., et al. (2001b). Influence of sublimation on stable isotope records recovered from high-altitude glaciers in the tropical Andes. J. Geophys. Res. Atmos. 106, 22613–22620. doi:10.1029/2001JD900179. Sturm, M., and Benson, C. S. (1997). Vapor transport, grain growth and depth-hoar development in the subarctic snow. J. Glaciol. 43, 42–59. Available at: http://cat.inist.fr/?aModele=afficheN&cpsidt=2705176 [Accessed September 12, 2011]. Taylor, S., Feng, X., Kirchner, J. W., Osterhuber, R., Klaue, B., and Renshaw, C. E. (2001). Isotopic evolution of a seasonal snowpack and its melt. Water Resour. Res. 37, 759–769. doi:10.1029/2000WR900341. Touzeau, A., Landais, A., Stenni, B., Uemura, R., Fukui, K., Fujita, S., et al. (2016). Acquisition of isotopic composition for surface snow in East Antarctica and the links to climatic parameters. Cryosphere 10, 837–852. doi:10.5194/tc-10-837-2016. Waddington, E. D., Cunningham, J., and Harder, S. L. (1996). “The Effects Of Snow Ventilation on Chemical Concentrations,” in Chemical Exchange Between the Atmosphere and Polar Snow (Berlin, Heidelberg: Springer Berlin Heidelberg), 403–451. doi:10.1007/978-3-642-61171-1_18. Wernli, H., and Davies, H. C. (1997). A Lagrangian-based analysis of extratropical cyclones. I: The method and some applications. Q. J. R. Meteorol. Soc. 123, 467–489. doi:10.1256/smsqj.53810. Wever, N., Jonas, T., Fierz, C., and Lehning, M. (2014). Model simulations of the modulating effect of the snow cover in a rain-on-snow event. Hydrol. Earth Syst. Sci. 18, 4657–4669. doi:10.5194/hess-18-4657-2014. Whillans, I. M., and Grootes, P. M. (1985). Isotopic diffusion in cold snow and firn. J. Geophys. Res. 90, 3910–3918. doi:10.1029/JD090iD02p03910. Yosida, Z. (1955). Physical Studies on Deposited Snow. System 7, 19–74. Available at: http://hdl.handle.net/2115/20216. Zweifel, B., Pielmeier, C., Marty, C., and Techel, F. (2017). Schnee und Lawinen in den Schweizer Alpen. Hydrologisches Jahr 2016/17. Davos doi:https://doi.org/10.3929/ethz-a-000008971.

119

5 Conclusions and Outlook

Conclusion and Outlook

This thesis demonstrated the direct effect of snow metamorphism on the microscale re-distribution of ammonium and other ions and trace elements embedded in the snow as well as on the concentration of stable water isotopes. Stable water isotope evolution was monitored in the field, while ion and trace element behavior was investigated through both laboratory experiments and field observations. In addition, the global circulation of microplastic particles was investigated by analyzing snow samples from various regions. Snow metamorphism is a highly dynamic process that takes place continuously in nature and thus constantly changes all snow cover worldwide. Recrystallization rate in the snow cover is variable and strongly depends on local conditions such as absolute temperature and temperature gradient in the snowpack, and meteorological effects like precipitation or turbulent fluxes. These physical processes during snow metamorphism affect incorporated environmental proxies throughout the transformation from snow to ice. Understanding of these processes are therefore of fundamental importance for the interpretation of ice core records. Crucially, in light of global warming and the increase in elevation of the 0°C isotherm, knowledge of different proxies that are less affected by meltwater is gaining in relevance. In the following section, the findings from my three studies are summarized within this context.

+ - - 2+ 2- The first study investigated the micro-scale distribution of NH4 , F , Cl Ca and SO4 in snow of different stages of metamorphism. These ions were selected as they all are important environmental proxies and we expected differences in the way they are incorporated into ice. In a first step, controlled laboratory experiments were chosen for the study, allowing assessment of the influence of snow metamorphism on distribution of the selected ions. Importantly, the effects of wind pumping could be excluded in these controlled experiments. For artificial snow, the concentration of the contained ions could be defined and the intensity of recrystallization during metamorphism could be calculated for all snow types. In a second step, the experiments were compared to natural snow in the field.

The results of elution experiments showed that temperature gradient metamorphism in both

+ the laboratory and the field lead to the uptake of NH4 in less accessible interior ice regions. These findings demonstrate our laboratory results are transferable to natural studies, which has important implications for future processes studies on snow metamorphism. The chosen elution analysis method was successfully implemented in snow but proved to be demanding and only partially suitable for a

122

Conclusion and Outlook quantitative assessment of the embedding processes of compounds into snow. A direct determination method, which was not applicable within the framework of this project, could be more effective for quantitative statements (see Outlook: Analytics). However, while elution is only partially suitable for a quantitative assessment of the embedding processes of compounds into snow, no alternative experimental methodology is currently known or could be developed within the framework of this project.

The experiments demonstrated that recrystallization rate during metamorphism has a substantial influence on chemical rearrangement. While preferential incorporation of ammonium under a temperature gradient became clearly visible, rearrangement observed under isothermal conditions was insignificant. These observations demonstrate diffusive processes in the ice are too slow to give rise to a substantial redistribution of ions in snow. Our results suggest that recrystallization

+ - - drives the selective incorporation of ions with high solubility in ice (group A1: NH4 , F and Cl ) into inner

2+ 2- regions and the exclusion of others (group B1: Ca , SO4 ) to the outside (air-ice or ice-ice interface).

Group A1 thus becomes more resistant to meltwater.

Transferred to nature, these findings mean that group A1 is better protected from possible melt influences with ongoing recrystallization of the snow cover. At the same time, the ions are less available for chemical reactions on the surface. In contrast, group B1 becomes more exposed to water fluxes. This can lead to a high ion concentration in the meltwater (ionic pulses or acid flushes) at the beginning of the melting season, which negatively affects vegetation and watercourses at higher altitudes. With highly contaminated snow in combination with a high snow depth, this amplification could endanger fauna at lower elevations.

Snow metamorphism also influences the distribution of trace elements (TE). We found the following groups of TE: A2 with immobility to meltwater percolation (Ag, Al, Bi, Cu, Cs, Fe, Li, Mo, Pb, Rb, Sb, Th,

Tl, U, V, W, Zr, Ce, Eu, La, Nd, Pr, Sc, Sm, Yb) and B2 with depleted TEs (Ba, Ca, Cd Co, Mg, Mn, Na, Ni,

Sr, Zn). Mobility was determined by the comparison of five snow profiles, recorded during Winter 2017.

Results demonstrated water-soluble TEs are incorporated into the ice matrix and thus protected from meltwater flushing depending on their concentration and solubility in ice. In contrast to the ions, there are also non-water-soluble TEs. These were also not washed out because they are generally not transportable by (melt)water. The verification of the preferential loss of TEs by meltwater from group

B2 (water-soluble TEs with low solubility in ice) versus from group A2 (water-soluble TEs with high

123

Conclusion and Outlook solubility in ice and water-insoluble TEs) is an important finding with regard to the interpretation of ice cores containing ice previously exposed to periods of melt.

In addition to the influence of snow metamorphism, comparison of five snow profiles also revealed interesting atmospheric transport patterns: Different origins could be identified for the distinct peaks of trace element concentrations. Several Sahara dust exports were registered, represented by high mineral concentrations in particular snow layers. The increase in the

+ concentration of NH4 in the upper snow layers (spring snow) can be attributed to the convective mixing of the boundary layer. Due to higher temperatures in spring, air stratification is less homogeneous, and more exchange occurs. As a result, substances from biogenic and anthropogenic sources from the valleys (mostly agricultural emissions) reach higher atmospheric layers and lead to higher concentrations in the deposited snow during snowfalls.

Determination of the redistribution of trace elements was not successful using the same elution experimental method as in the ion study. Compared to ions, the lower solubility of TEs in snow towards water meant that only minimal amounts of substances could be mobilized in the elution experiments.

As a result, the concentrations measured in the eluate were often at or below the detection limit, making a reliable assessment impossible. Future studies quantifying redistribution processes of TEs during snow metamorphism would therefore require direct methods such as laser ablation ICP-MS

(see outlook).

The isotope measurement campaign, which took place between January and June 2017, showed the isotope signal deposited in the alpine snow cover during frontal precipitation strongly correlated with the advection of air masses with different characteristics (warm and cold fronts). The isotope signature of the snowpack thus provides a solid record of the weather throughout the winter season. Although the isotope signal corresponds well to the air masses, this does not necessarily imply that it correlates with local near-surface air temperature. This was illustrated by a two-day precipitation event where a very low (depleted) ẟ18O signal was deposited at relatively high near-surface air temperature due to cold air entrapment in a warm front. This finding could be helpful to explain the observed poor correlation of the isotopic signal and the temperature in alpine firn cores in other studies. It is important to note that the measured surface air temperature does not need to be equal to cloud formation temperature. The combination of the measured isotope signal in snow on the ground, in the air above the ground, and the use of a reanalysis dataset facilitated additional deductions on

124

Conclusion and Outlook microphysical processes taking place in the atmosphere. Discussed processes include preferential rainout effects, supersaturation in clouds that increases dexcess and decreases ẟ18O in the precipitation, and the growth of ice (lower saturation vapor pressure) at the cost of evaporating supercooled liquid droplets (high saturation vapor pressure).

The comparison of five isotopic snow profiles showed that the deposited isotope signal remains stable during the cold winter months (January to March) and redistribution effects due to dry snow metamorphism are less relevant in an alpine snowpack. Melting in early spring then caused a significant shift of the original signal – although without completely distorting its original shape. The isotopes therefore have to be somehow classified between the groups A1, A2 and B1, B2, as the signature remains despite meltwater flows (as in group A). However, in order to use isotope profiles influenced by melt as proxies, additional corrections would be necessary. In particular, the occurrence of fractionation processes during snowmelt (in combination with wet snow metamorphism) makes the interpretation of a melt-influenced isotope signature even more complicated.

The impact of metamorphism on snow is usually described by the evolution of the specific surface area (SSA). In contrast, the laboratory experiments on ion redistribution (Chapter 2) showed

SSA is a poor indicator of the dynamics in the snow during metamorphism. Therefore SSA-development is not well suited to characterize the redistribution of embedded impurities. The change in SSA is only related to vapor flux and recrystallisation during the first phase of metamorphism because a quasi- steady state is reached with respect to the SSA, even though the recrystallization rate remains unchanged during a constant temperature gradient driven water vapor flux.

In conclusion, it can be said that the snow metamorphism played a central role in all three studies. The direct influence can be divided into cold (dry) and warm (wet) snow conditions: In cold conditions

(temperature gradient metamorphism), the recrystallization of the snowpack (monitored above a scale of two centimeters) did not affect the preservation of the ions or trace elements investigated as both temporarily remained well preserved in the snow cover. In the stable water isotopes, the dry metamorphism led to a slight smoothing of up to 2‰. On a microscale, however, a fractionation of ions and trace elements into the previously described groups A and B could also be observed. This is essential for the reaction under warm (wet) conditions: The microscopic redistribution of proxies strongly influences their robustness against meltwater fluxes.

125

Conclusion and Outlook

As soon as water is in the system, it becomes more complex. For the major ions and trace elements, the preferential loss of group B was observed, as they were flushed out by the water fluxes.

Stable water isotopes showed a substantial shift of the original signal. In addition to these observations and with the presence of water, wet snow metamorphism sets in. This is a highly dynamic process and runs much faster than dry metamorphism. The coupling of the influence of wet and dry metamorphism makes an interpretation of the results much more problematic. Controlled laboratory experiments are the only way to disentangle such complexities.

Complexity can be reduced by separating individual processes, which is difficult to infeasible in the field. A good instrumentation of the measurement field, as is the case on the Weissfluhjoch, can facilitate quantification of environmental influences and understanding of their impact (see Outlook:

Water vapor Isotope measurement series at the Weissfluhjoch). However, in nonlinear coupled systems such as water isotope fractionation, overlaps inevitably remain, which can have a significant influence on the sensitive system. For interpretation of the isotope signature in the snow cover, a reduction of the degrees of freedom is therefore a prerequisite for verifying the hypothesis set out in

Chapter 5 (see outlook: Isotope fractionation).

Laboratory experiments are suitable for improving process understanding, even if the results are not fully transferable to nature. Moreover, laboratory experiments also create decisive advantages with regard to the development of models for individual environmental processes: It is challenging to parameterize models with results from nature because there is always a certain number of degrees of freedom. For the development of robust physical models, however, the reduction of these degrees of freedom is a prerequisite (see outlook: Snow cover models).

Finally, our fourth study (Appendix A, microplastic) showed that microplastic could be detected even in the snow in remote regions of the Alps and the Arctic. It is necessary to further monitor the introduction of these microplastics as well as investigate the influence of these particles on the health of animals and humans (see outlook: Environmental monitoring).

126

Conclusion and Outlook

In light of the findings in this research, we suggest the following topics should be addressed in subsequent studies:

Environmental monitoring: The influence of metamorphism and melt processes on other important

environmental proxies such as microplastics and black carbon should be investigated. In the near

future, more and more glaciers worldwide will be affected by melt events and the identification

of meltwater-resistant proxies will become more relevant due to the continuing climate

warming.

CryoNet Extension: On the Weissfluhjoch, the SLF has been regularly recording snow cover every

winter since 1936. It should be considered whether the measurement of traditional parameters

such as grain size and grain shape can be extended to chemical substances or whether

environmental monitoring can become part of CryoNet.

Microplastic: The measurement and quantification of microplastics concentration in snow should be

further expanded. So far, little is known about the processes involved in the embedding of

microplastic. Via snowmelt, these particles could enter soils, rivers and drinking water. Given

that the highest concentrations found in this thesis occurred along roads, the current practice

of depositing road snow in creeks and lakes and related potential impacts should also be

investigated.

Analytics: Development of a direct determination method on the microscale must be advanced for a

quantitative investigation of the (re-)distribution of substances in snow and ice during

metamorphism. Laser ablation mass spectrometry (LA-ICP-MS) is a promising method: A laser

that can release tiny ice volumes and particles (ablation) is coupled by a carrier gas with a plasma

mass spectrometer (ICP-MS) that analyzes ions and trace elements. However, this method is still

under development but should become operational in the future.

Isotope fractionation: The fractionation processes during metamorphism (dry and wet) are

complicated and are superimposed in the field by additional factors (wind drift, fresh snow,

settlement). A quantitative separation of the effects influencing the isotope signal should be

advanced by a detailed laboratory study. In our opinion, this is preferable to further field studies

under the current state of research methodologies.

Water vapor isotope measurement series at the Weissfluhjoch: The measurement series of stable

oxygen isotopes in the air on the WFJ, which was established in January 2017, provides a unique

127

Conclusion and Outlook

data set of over 2.5 years (as of today, the measurements are still running). The data set is

extended by the automatic recording of meteorological data (CryoNet) and thus provides an

ideal basis for modeling.

Future water vapor isotope measurement: For a continuation of the measurement at the

Weissfluhjoch, the collection of additional parameters should be considered: a) Fresh snow

isotope composition sampled on event-based regularity b) Radar information on hydrometeors

and vertical structure of the cloud, c) Phase of the precipitation, liquid vs wet snow (April-June)

including size distribution (disdrometer). With respect to snowpit sampling, at least one full

profile directly before the melt-season starts should be recorded, since conditions are rather

well preserved during cold season.

Snow cover models: The rearrangement of substances contained in snow (especially isotopes but also

volatile substances) should be considered in development of future snow cover models. These

proxies provide a relevant link to the conditions in which snow precipitation has formed. It may

also be possible to derive an indirect link to the compaction behavior and thus improve

settlement modeling.

Solubility of substances in ice: Solubility in ice plays a central role for incorporation of substances into

ice (e.g., during dry metamorphism). However, only solubilities of a few substances are known

- - + (experimentally determined solubility in ice are available for Cl , F , and NH4 ). In addition,

solubility in ice must be determined for other atmospheric substances to confirm observed

incorporation of ions and trace elements in snow and ice and to allow further modeling.

In summary, this project gave a more detailed insight into the dynamic behavior and occurrence of environmental proxies embedded in snow. The constant transformation of the snow structure affects the substances contained in it at various scales. For the interpretation of any proxy signature in snow, the knowledge of the processes taking place in the snow cover is essential.

128

Appendix A White and wonderful? Microplastics prevail in snow from the Alps to the Arctic

Published in Science Advances: Bergmann, M., Mützel, S., Primpke, S., Tekman, M. B., Trachsel, J., and Gerdts, G. (2019). White and wonderful? Microplastics prevail in snow from the Alps to the Arctic. Sci. Adv. 5, eaax1157. doi:10.1126/sciadv.aax1157.

A.1 Introduction

Abstract Microplastics (MP) are ubiquitous and considerable quantities prevail even in the Arctic; however, there are large knowledge gaps regarding pathways to the North. To assess if atmospheric transport plays a role, we analyzed for the first-time snow samples from ice floes in Fram Strait. For comparison, we investigated snow samples from remote (Swiss Alps) and populated (Bremen, Bavaria) European sites. MPs were identified by Fourier-Transform Infrared Imaging in 20 of 21 samples. The MP concentration of Arctic snow was significantly lower (0–14.4 × 103 N L-1) than European snow (0.19– 154 × 103 N L-1) but still substantial. Polymer composition varied strongly, but varnish, rubber, polyethylene and polyamide dominated overall. Most particles were in the smallest size range with no saturation, implying that there are yet smaller particles beyond the current detection limit of 11 µm. Our data highlight that atmospheric transport and deposition can be significant pathways for MPs meriting more research.

A.1 Introduction Plastic pollution is a problem of growing environmental concern, because production rates have increased to 380 million t y-1 in 2015 (Geyer et al., 2017), and annual waste production is projected to rise to 3.4 billion t in the next 30 years (Silpa et al., 2018). Many countries still have inefficient waste management and water treatment systems allowing leakage to the environment (Jambeck et al., 2015), which is exacerbated by littering behavior. Mismanaged plastic waste could triple from 60 - 99 million metric tons (Mt) in 2015 to 155 - 265 Mt by 2060, assuming a business-as-usual scenario (Lebreton and Andrady, 2019). In addition, plastic is designed to be durable. Therefore, it persists in the environment for long periods of time. Thus, it is hardly surprising that plastic pollutants are ubiquitous and have been reported from environments close to urban centers, terrestrial areas, freshwater environments as well as from the shores of remote uninhabited islands, the sea surface, water column and deep seafloor (Bergmann et al., 2017b). Plastic pollutants have also reached polar regions including Arctic beaches (Bergmann et al., 2017a), sea ice (Obbard et al., 2014), water column (Grøsvik et al., 2018), sea surface (Lusher et al., 2015; Bergmann et al., 2016; Cózar et al., 2017; Kanhai et al., 2018) and the seafloor (Tekman et al., 2017). Under the influence of light, mechanic abrasion, waves and temperature fluctuations, plastic items fragment into smaller sizes and are termed microplastic (MP) when attaining sizes below 5 mm. Surprisingly, Arctic surface waters turned out to harbor the highest MP concentrations in a global comparison of MP quantities, despite their remoteness (Barrows et al., 2018). What is more, litter quantities have increased significantly on the deep Arctic seafloor over the past 15 years as highlighted in a time-series study (Tekman et al., 2017). This stimulated further research unveiling very high concentrations of MPs in the sediment of the deep

132

A.2 Results

Fram Strait (Bergmann et al., 2017c) and Arctic sea ice (Peeken et al., 2018) prompting the question: ‘How do all these MPs make it so far to the north?’ One possibility is aerial transport since a yet still limited number of studies found MPs in atmospheric fallout of the cities Dongguan (China) and Tehran (Cai et al., 2017; Dehghani et al., 2017). In France, MP concentrations in atmospheric fallout increased 5-fold after a rain event indicating that wet deposition could be a pathway of MPs to Earth’s surfaces (Dris et al., 2015) including the oceans. Despite the limited research on airborne plastics, its importance is obvious given the precautions that have to be taken in studies on MPs to reduce the risk of contamination through airborne MPs (Woodall et al., 2015). Further, airborne MPs represent a hitherto largely neglected route of exposure to humans and wildlife as such particles could be taken up through inhalation (Gasperi et al., 2018). Snow is a scavenger for diverse impurities and acts as a filter on the ground by dry deposition (Heintzenberg and Rummukainen, 1993; Zhao et al., 2015). Here, we sampled snow deposited on ice floes drifting in Fram Strait and on Svalbard to assess if atmospheric fallout is a pathway of MPs to the Arctic environment. To enable comparisons with previous studies in the Arctic the standardized automated FTIR imaging analysis was applied (Bergmann et al., 2017c; Primpke et al., 2017; Peeken et al., 2018; Primpke et al., 2018). The Arctic is still widely conceived as one of the last pristine environments on the Globe. To verify this view, we also assessed MP concentrations in snow close to more urban sites in northern Europe (Bremen City, Isle of Heligoland) and the Alps (Davos, Tschuggen, Bavaria) for comparison (Fig. A.1).

A.2 Results MPs and microfibers were found in all but one snow samples, ranging from 0.02 – 154 × 103 N L-1 and 0.043 – 10.2 × 103 N L-1, respectively (Fig. A.2, table S1). MPs accounted for 0 – 88 % (mean 12 %) of the filtered particles. The material type of fibers, however, could not be determined due to the applied analysis pipeline. Therefore, fiber numbers comprise both synthetic and natural fibers. Nevertheless, we present fiber concentrations to enable comparison with previously published studies, which reported primarily fibers.

A.2.1 Fiber and microplastic quantities at different locations MPs occurred at a mean concentration of 9.8 × 103 N L-1 (± standard error 6.9 × 103 N L-1) with highest quantities detected in the sample Bavaria 2 (154 × 103 N L-1), followed by Heligoland 2 (17.6 × 103 N L-1) and Ice Floe 9 (14.4 × 103 N L-1) (Fig. A.1a-b, table S1). Snow from Europe contained significantly more MPs (24.6 ± 18.6 × 103 N L-1) than samples from ice floes (Mann-Whitney U test: W = 54, p = 0.011). Still, a mean of 1.76 ± 1.58 × 103 N L-1 is substantial for a secluded location such as the Arctic. Figure A.3 illustrates images of MPs and fibers from different locations. Ice Floe 9 had a much higher MP

133

A.2 Results concentration than the other Arctic snow samples raising concern regarding the potential of contamination during sampling or from the helicopter used. Exclusion of this value decreases the mean to 0.18 ± 0.07 × 103 N L-1. Still, we think that this high level is unlikely due to contamination as varnish, the main polymer type detected, was not part of the equipment used and also present in high quantities in samples from Ice Floe 3, Bremen, the Alps and Heligoland, where no helicopter was used. In addition, the sample Ice Floe 9 was taken outside the area likely to be affected by helicopter turbulence. Since the second highest MP abundance was detected on Heligoland at a time of strong winds (18 m s-1, table S2) we tested if MP abundance was correlated with wind speed, but found no significant correlation (Spearman’s rank correlation test: ρ = -0.06, p = 0.80). However, wind direction may have played a role as the strong easterly winds may have transported particles from the neighboring dunes or from the mainland.

Figure A.1: Map of sampling locations for snow. (a) Sampling sites in the Arctic (ICE: ice floes, SV: Svalbard) (b) in Europe (HL: Heligoland, BR: Bremen, BV: Bavaria, Tsch: Tschuggen, DV: Davos) (c) Overview of all locations. Size of circles reflects microplastic particle quantities at log scale.

The maximum quantities of fibers were detected in snow from Ice Floe 4 (10.2 × 103 N L-1), followed by Heligoland 2 (2.75 × 103 N L-1) and Bavaria 3 (2.57 × 103 N L-1) (Fig. A.1a-b, table S1). Still, overall, snow from Europe contained significantly more fibers (mean: 1.431 ± 0.325 × 103 N L-1) than samples from Arctic ice floes (M-W: W = 96, p = 0.024), which had still a substantial mean (1.38 ± 1.10 × 103 N L-1) given their remoteness. Again, it could be argued that the exceptionally high fiber numbers from Ice 134

A.2 Results

Floe 4 may be due to contamination. Its exclusion would lead to a lower mean (0.28 ± 0.095 × 103 N L-1). However, we deem this unlikely as Ice Floe samples 5-8 were taken on the same large ice floe by the same personnel, so contamination of these samples should be similarly high. There was no significant correlation between fiber abundance and wind speed (Spearman: ρ = -0.17, p = 0.45). The abundance of fibers was positively correlated with MP abundance (Spearman: ρ = 0.62, p = 0.002).

A.2.2 Size of microplastics and fibers The size of MPs detected ranged between 11 – 500,475 µm. Eighty percent of all detected MPs were ≤25 μm and 98% of all particles were <100 μm. Overall, the amount of particles decreased with increasing size (Figure A.2c) with no asymptote reached in the smallest detectable size class. Table S2 comprises the size distribution of individual samples.

Figure A.2: Particles detected in snow samples collected at different locations from Europe to the Arctic. (a) Microplastic particle quantities recorded by FTIR at different locations; (b) Concentrations of microfibers detected at different locations (note: no polymers identified); (c) Box and whiskers plot of proportions of microplastic numbers in different size classes from all snow samples. The upper and lower boundary of the box indicates the 75th and 25th percentile, respectively. The line within the box marks the median, error bars indicate the 90th and 10th percentiles, black diamonds represent outliers; (d) Relative composition of polymers identified by FTIR at different locations.

135

A.2 Results

The length of fibers ranged between 65 – 14,314 µm. While 97% had a maximum length of 5 mm, 31% were shorter than 500 μm. In general, the fibers show an increasing trend towards shorter lengths but are not saturated in the lowest size class (fig. S1). European fibers were significantly longer compared with those from Arctic snow (M-W: W = 13,723, p = 0.0001) even if the two largest European fibers (14,314;13,704 µm) were excluded. Table S3 includes the size information of individual fibers marked with their region of origin.

A.2.3 Material composition The highest proportion of MPs in the total natural and synthetic particle load was found in snow from Ice Floe 1 (88%), followed by Bavaria 2 (67%) and Ice Floe 9 (37%) (table S2). There was no significant difference in the proportion of MP particles from European and Arctic snow (M-W: W= 170, p = 0.59). The composition varied considerably with 19 different polymer types found in total ranging between 2 (Ice Floe 4) and 12 types (Bavaria 2) per sample (Fig. A.2d, table S1). The number of polymers per sample was significantly higher in European (mean: 8.63 ± 0.80) compared with Arctic (mean: 5.14 ± 0.79) samples (M-W: W = 123, p = 0.013). Acrylates/polyurethanes/varnish/lacquer (hereafter: varnish) occurred most frequently (17 samples), followed by nitrile rubber (16 samples), polyethylene, polyamide and rubber type 3 (13; ethylene-propylene-diene rubber). The polymer composition of samples from Europe and the Arctic was significantly different (PERMANOVA: Pseudo- F = 2.43, p = 0.006). The dissimilarity in the polymer composition from European and Arctic samples was 67% and caused primarily by much higher abundances of polyamide, varnish, rubber type 3, nitrile rubber, ethylene-vinyl-acetate and polyethylene in European samples. By contrast, polystyrene, polyvinylchloride, polycarbonate, polylactic acid and polyimide occurred exclusively in Arctic snow.

A.2.4 Other particles Other particles detected in snow accounted for 22 – 100 % of the total particles (table S2) and comprised chitin, charcoal, coal, animal fur, plant fibers and sand. Except for coal, all of these were significantly more abundant in European snow (M-W: W = 114, p = 0.0015), which also explains the significant differences found in their composition (PERMANOVA: Pseudo-F = 7.75, p = 0.001). Particles assigned to “plant fibers” and “animal fur” contributed most to the 46% dissimilarity (SIMPER).

136

A.2 Results

a b c

200 µm 100 µm 100 µm d e f

2 mm 2 mm 2 mm

g h i

Figure A.3: Photographs of microplastics detected in snow. (a) polystyrene fiber from Svalbard 4 (Mohnbukta, 1,101 µm length); (b) polypropylene particle from Heligoland (256 µm diameter); (c) polyvinylchloride fiber from Ice Floe 8 (956 µm length). Note, these two fibers could be analyzed by FTIR as they happened to lay plane on the filter. (d-f) Aluminum oxide filter with enriched snow sample; (g-h) corresponding polymer-dependent false-color image after FTIR measurement and automated analysis. (d/g) Bavaria 2, the sample with most microplastics; (e/h) Ice Floe 9: the sample with the third most microplastics; (f/i) Bremen: intermediate microplastic numbers but many fibers

137

A.3 Discussion

A.3 Discussion This study provides the first data on contamination of snow by MPs. MP concentrations in snow were very high indicating significant contamination of the atmosphere. During its passage through the atmosphere, snow binds airborne particles and pollutants, which are eventually deposited on Earth’s surfaces, a phenomenon termed ‘scavenging’ (Zhao et al., 2015). Our data show that ‘scavenging’ represents an important pathway of MPs to land and ocean environments in Europe and the Arctic. Based on annual snow fall data we estimate an annual MP deposition of 8.8 ± 7.9 N m-2 (0 – 72 N m-2) in the Fram Strait, of 1.4 ± 0.4 N m-2 in Svalbard (0 – 2 N m-2) and 66 ± 60.1 N m-2 (0 – 308 N m-2) in the Alps. It should be noted, however, that these estimates come with very large uncertainties given the variability of the data.

A.3.1 Abundance of microplastic particles and microfibers Although MP levels were significantly higher in European (0.191–154 × 103 N L-1) compared with Arctic snow (0–14.4 × 103 N L-1) the conclusions drawn have to be treated with caution given the high variability and sparse spread of samples over large areas. Some variability may be due to the fact that snow from the Arctic was not freshly deposited unlike all European snow samples (except for Swiss snow) and could have been exposed to secondary dry deposition of airborne particles (Heintzenberg and Rummukainen, 1993) for an unknown period of time. Still, the lower concentrations found in Arctic snow is not surprising given the distance of the Arctic to densely populated source areas. Indeed, MP concentrations were also consistently higher in atmospheric fallout from densely populated urban compared with suburban or less densely populated sites in France and Dongguan (Dris et al., 2016; Cai et al., 2017). (Dehghani et al., 2017) recorded MPs in the range of 3–20 N g dry dust-1 from Tehran. However, quantities are not directly comparable as wet deposition is a more efficient mode of transport than dry deposition (Uematsu et al., 1985), and different methods were used. The same applies to MPs recorded in dry fallout from Dongguan (31–43 N m-2 day-1) (Cai et al., 2017). The highest MP concentration was found in Bavarian snow. Since we took this sample next to a country road, traffic could play a role in terms of automotive emissions, dispersion of settled MPs by cars and subsequent scavenging by falling snow. Experiments have shown that significant numbers of similar sized zinc sulfide particles placed on a road were swirled by cars (Sehmel, 1973). While the most abundant polymer types in this sample, rubber types 1 and 3, point to car tires as a potential source, no firm conclusions can be drawn due to current methodological constraints in the detection of specific rubber types. Still, the number of MPs in snow from a street in the city of Bremen, our most urban site (~568,000 inhabitants), was only intermediate (2 × 103 N L-1). This could be due to lower particle dispersion as freshly deposited snow was collected on a Sunday when traffic is low. By contrast, the Bavarian sample was taken on a Monday with higher car traffic. (Lonati et al., 2006) reported a 20%

138

A.3 Discussion lower emission of fine particles in Milano during week-ends due to lower particle dispersion by decreased traffic. Snow from the North Sea island Heligoland ranked second and fourth highest (17.6 × 103 N L-1), which is unexpected considering that it is inhabited by only ~1,200 people, and cars are banned. Although there was no significant correlation between MP levels and wind speed overall, strong easterly winds (18 m s-1) prevailing before and during the sampling time may have dispersed MPs from the island’s dune environment and possibly also transported particles from seawater or the mainland. The lowest European MP concentration was detected in a sample from the Swiss Alps (0.19 × 103 N L-1). This could be considered background contamination as the site was neither urban nor close to traffic. The MP levels in Arctic snow were still considerable with the third highest concentration (14.4 × 103 N L-1) originating from an ice floe. The magnitude difference between Ice Floe 9 and other ice floe samples is striking. Until we know more about atmospheric MP pollution, we can only speculate that atmospheric MP pollution may be variable in both time and space resulting in large differences of MPs falling out via snow. Local wind conditions may play a role. In addition, Arctic ice floes are often carried from the Central Arctic to the south (i.e. Fram Strait) by the Transpolar Drift once the sea ice breaks up in spring (Tekman et al., 2017; Peeken et al., 2018). During their drift, they may encounter different air masses carrying varying amounts of MP, which could cause the differences observed. Still, whatever the exact cause, it is surprising given the remoteness of the location and poses the question ‘Where does it all come from?’ It is conceivable that airborne MPs are emitted or dispersed locally by ships and by wind. Indeed, snow from the Vladivostok district contained particles of marine origin including sea urchins, algae and mollusk shells (Golokhvast, 2014). Still, the majority of particles likely originates from more distant regions. To shine a light on the issue we draw on knowledge of pathways of mercury, which is found in high concentrations in Arctic wildlife. A mass-balance approach indicates that wet deposition via snow is the main pathway of mercury to the Arctic ocean (Outridge et al., 2008). Three main routes of transport lead to an accumulation of mercury in the Arctic: (1) the Icelandic Low and North American High transport air masses from north America and western Europe (40%), (2) the Aleutian Low brings air masses from the north Pacific and east Asia (25%) and (3) the Siberian High delivers eastern European and Siberian air masses (15%) to the Arctic (Outridge et al., 2008). As with mercury, MPs may be blown over long distances to the Arctic from urban areas in all directions leading to unexpectedly high levels of MP in the Arctic atmosphere. This is likely exacerbated during high phases of the North Atlantic Oscillation (Eckhardt et al., 2003). Even large mineral particles (≤ 450 µm) can be transported from the Sahara to the north Atlantic over distances of 3,500 km by mechanisms such as rapid horizontal transport, turbulence, uplift in convective systems as well as electrical levitation of particles (van der Does et al., 2018). It has been estimated that pollen of willow and pine

139

A.3 Discussion

(10 - 200 μ) were transported from western Europe to the Arctic at 3,000 m altitude in only five days (Rousseau et al., 2004).

A.3.2 Comparison with microplastics levels in Arctic sea ice and deep-sea sediments Unfortunately, a direct comparison of our results on MP particles with previous data on airborne MPs or those from Arctic seawater is not possible as the latter two deal primarily with fibers (but see section on fibers below) or use different methods resulting in much lower numbers (Lusher et al., 2015; Kanhai et al., 2018; Morgana et al., 2018). We can, however, compare our results of Arctic ice floes with previous data on MPs in Arctic sea ice and deep-sea sediments (Bergmann et al., 2017c; Peeken et al., 2018), also to investigate if airborne MPs are a source to these compartments. These studies used the same standardized analytical methods and are thus comparable with our approach. MP concentrations ranged between 1.1 – 12 × 103 N L-1 in Arctic sea ice, which tends to concentrate particles by a factor of 100 relative to adjacent seawater (Peeken et al., 2018) and between 0.04 – 3.46 × 103 N L-1 in deep- sea sediments (Bergmann et al., 2017c). Given similar or higher levels in snow from ice floes (0 – 14.4 × 103 N L-1), it seems likely that they contribute to MPs in sea ice and also sink to the seafloor over time. MPs deposited on ice floes via snow may be embedded directly into the sea ice matrix through snow metamorphism, i.e. compaction, or through release into melt ponds and subsequent refreezing. Another route to these matrices could simply be through wet deposition onto the ocean surface before or during sea ice formation. Once in seawater MPs may sink to the seafloor. The composition of polymers in snow from ice floes seemed more similar to that of sea ice compared with deep-sea sediments. For example, varnish was one of the most frequent and abundant polymer types reported in both of these spheres but less important in sediments. Still, nitrile rubber and polyamide were important in both ice floe snow as well as deep-sea sediments and sea ice, such that these polymers may have been transported from the atmosphere to sea ice and the seafloor.

A.3.3 Microplastics size The size distribution of MPs in snow was surprisingly similar to MP sizes found in Arctic sea ice and deep-sea sediments (Bergmann et al., 2017c; Peeken et al., 2018). The majority of MPs detected in all of these spheres was in the smallest size range (11 µm). Scrutinized by scan-electron microscopy MP particles from atmospheric fallout of Dongguan revealed signs of weathering such that (Cai et al., 2017) hypothesized atmospheric degradation processes through collision and friction dynamics as well as chemical weathering due to higher irradiation and oxygen levels in the atmosphere. This could enhance fragmentation into smaller sizes. The fact that there is no size saturation in the lowest size range implies that there may be yet more particles in size categories below our current detection limit. In addition, smaller particles are probably more likely to be picked up and transported by air masses.

140

A.3 Discussion

No matter what the cause, our results likely underestimate MP quantities as was also concluded in previous studies (Bergmann et al., 2017c; Dris et al., 2017; Peeken et al., 2018). This highlights the need to quantify small particles for realistic assessments of MP pollution. Most studies currently focus on particles >200–300 µm.

A.3.4 Abundance of fibers All snow samples contained fibers ranging between 0.043–10.2 × 103 N L-1 with highest concentrations in snow from Ice Floe 4 followed by snow from Bremen. While it is unknown what proportion of these were synthetic polymers our fiber concentrations were at least 4–7 orders of magnitude higher than previous reports from outdoor and indoor environments near Paris (median = 0.0009 and 0.0054 N L-1, respectively) (Dris et al., 2017). Fiber and MP abundance in Dongguan were in the same order of magnitude as samples from Paris (Cai et al., 2017). This large discrepancy is likely due to methodological differences and differences in the underlying mechanisms, i.e. dry versus wet deposition. Indeed, (Dris et al., 2015) reported five times higher MP levels after a rain event and snow can scavenge aerosol particles up to 50 times more efficiently than rain (Zhao et al., 2015). Whatever the reasons, the high abundance of fibers in remote Arctic snow is striking whereas high levels in urban Bremen do not come as a surprise. The positive correlation between fiber and MP concentrations may indicate that they both come from similar sources. While 33% of the fibers detected near Paris were polymers, 23% of the fibers from Dongguan were made of plastic (Cai et al., 2017; Dris et al., 2017). If 28% (the mean of these two values) of the fibers from our study were also polymers this would add 0.012–2.86 × 103 N L-1 and result in a total MP load of 0.029–156.86 × 103 N L-1. However, this assumption has to be treated with caution since the proportion of natural fibers in our samples may differ from those of more urban areas.

A.3.5 Polymer composition The polymer composition of samples was very variable, even at sites, which were located in close proximity, e.g. Ice Floe 4–8, that came from the same ice floe. Scavenging of particles depends on the size of snow crystals, particle size, wind speed, air humidity and snow intensity (Zhao et al., 2015; Yuyan L. et al., 2018) to name a few of many factors, which may have influenced also the polymer composition. Variable polymer compositions were also reported from Arctic sea ice and deep-sea sediments (Bergmann et al., 2017c; Peeken et al., 2018). The number of polymer types found per sample was highest in Bavaria (12 types), which is not surprising given that this was also the sample with the highest MP abundance and sampled next to a country road where settled MPs may be subject to dispersion by traffic. The number of detected polymer types both per sample and for all samples was much higher in this study (19) compared with atmospheric fallout from China (4) and

141

A.3 Discussion

France (3)(Dris et al., 2016; Cai et al., 2017). This is likely due to wet deposition scavenging effects and methodological differences, because previous results were based on the analysis of visually pre- selected particles and considered only particles larger than 50 µm, which were also mostly fibers. Varnish was the most frequent and among the top five most abundant polymer types. It was also detected in ice cores and deep-sea sediments albeit less frequently (Bergmann et al., 2017c; Peeken et al., 2018). This reflects the widespread application of polymer-based varnish for protective coatings of surfaces including vehicles, ships, wind turbines, aquaculture and buildings. The building sector accounts for ~20% of the European plastic converter demand (PlasticsEurope, 2018). MPs may be emitted during construction on building sites, through abrasion of coated surfaces by wind and rain or when handling construction waste (Bertling et al., 2018). Future research is needed to quantify the importance and pathway of this source of MPs. Three different types of rubber were the next most frequent and the most abundant polymers. They may enter the environment as abrasion product from tires or as extremely durable synthetic rubber roofing membranes commonly used in roof construction (rubber type 3, ethylene propylene diene rubber). Abrasion from sealing gaskets (rubber type 1, sealing rubber), cable and shoe soles could be further sources of rubber, some of which belong to the most common sources of MPs in Germany (Bertling et al., 2018). While Arctic sea ice contained scant rubber particles, it was also found in deep-sea sediments (Bergmann et al., 2017c; Peeken et al., 2018), where nitrile rubber was one of the most frequent and abundant polymers. As it is very resistant to a wide range of temperatures, as well as to oil, gasoline and other chemicals it is widely used for hoses, seals, O-rings, synthetic leather, grommets, cable jacketing and transmission belts in offshore oil platforms and the automotive and aeronautical industry. Polyamide (including nylon) was amongst the top five most frequent and abundant MPs detected. It was also reported from a household and atmospheric fallout in France (Dris et al., 2017). Polyamide is widely used in synthetic fabrics, automotive applications, fisheries, sails, toothbrushes, packaging and carpets. It was more prevalent in European than in Arctic snow. Polyamide was, however, abundant in Atlantic and Arctic surface waters and deep-sea sediment (Enders et al., 2015; Lusher et al., 2015; Bergmann et al., 2017c; Kanhai et al., 2018; Morgana et al., 2018; Peeken et al., 2018). Thus, pathways other than atmospheric transport may play a more prominent role in the transport of polyamide to the Arctic. Likewise, polyethylene occurred only in low numbers in two ice floe samples but was the most abundant polymer type detected in Arctic sea ice, Atlantic surface waters (Enders et al., 2015; Lusher et al., 2015; Morgana et al., 2018; Peeken et al., 2018) and prevailed in most European samples as well as in atmospheric fallout of Dongguan (Cai et al., 2017) and a French household (Dris et al., 2017).

142

A.3 Discussion

Again, mechanisms other than atmospheric transport may be more important in polyethylene transport to the Arctic.

A.3.6 Health implications The large concentrations of MPs and microfibers in snow highlight the importance of the atmosphere as a source of airborne MPs and microfibers. Through this pathway, MPs likely find their way into soil and aquatic environments and therefore also into food chains. In populated areas, it is common practice to remove snow from streets and transport it ‘away’. Our results show that such locations should be chosen wisely so as to avoid contamination of sensitive areas. But what is more, MPs in snow has fallen out of the atmosphere and could therefore be considered an indicator of aerial MP pollution. This is relevant in the context of human and animal health, especially in terms of inhabited European sites, where residents may breathe in airborne MPs and fibers. However, although we know that airborne contamination of seafood during indoor food preparation and meals exceeds original MP concentrations (Catarino et al., 2018), there has been surprisingly little research about the inhalation risk of airborne MPs. It has been postulated that only the smallest sized MP fraction is respired into the deep lung while particles exceeding a length of 5 µm, a diameter < 3 µm and with a length-to- diameter ratio of 3:1 are subject to coughing or mucociliary clearance (Gasperi et al., 2018) such that they end up in the gastrointestinal tract. Still, the detection of MPs and other fibers of up to 135 µm length in lung tissues, including carcinoma, challenges this notion (Pimentel et al., 1975; Pauly et al., 1998). MPs in pulmonary tissues may persist for a long time as they are durable in body fluids (Gasperi et al., 2018). Chronic inhalation of MPs, especially in combination with adsorbed or added chemicals (Dehghani et al., 2017), may lead to health risks including respiratory irritation, allergic alveolitis, inflammation, fibrosis and genotoxicity (Gasperi et al., 2018). (Pauly et al., 1998) suggest that MPs may be considered candidate agents contributing to the risk of lung cancer, especially of non-smokers. The high MP concentrations detected in snow samples from continental Europe to the Arctic indicate significant air pollution and stress the urgent need for research on human and animal health effects focusing on airborne MPs. The high amount of MPs present in the atmosphere as indicated by significant concentrations in snow from continental Europe to the Arctic is puzzling at first. It is known, however, that snow either takes up aerosols or forms around aerosol nuclei containing pollutants, e.g. from vehicle exhaust particles (Nazarenko et al., 2017). In addition, dust emitted from Earth surfaces, deposition and dispersion between atmosphere, land surface and the aquatic realm could facilitate the transportation of MPs (Cai et al., 2017), but research is needed to verify this. One open question is also if aquatic environments act as a source of MPs to the atmosphere, e.g. during evaporation and storm events. This could also explain the high concentrations of MPs and fibers in secluded uninhabited regions such

143

A.4 Material and Methods as the Arctic. Another explanation is long-range transport by wind systems bringing anthropogenic particles from urban sites. Large dust particles are transported over distances of 3,500 km from the Sahara to the North Atlantic (van der Does et al., 2018). This is similar to the distance between our Arctic sites and Europe, which happens to be the most important pathway in terms of wind-driven transport of mercury to the Arctic. Our results highlight a hitherto neglected pathway of MPs to the Arctic Ocean, an ecosystem, which is already stressed by the effects of the climate crisis.

A.4 Material and Methods

A.4.1 Study sites Between 2015 and 2017, five ice floes drifting in the Arctic Fram Strait were visited by ship-based helicopters or dingy during three expeditions of the research icebreaker RV Polarstern (Fig. A.1 a, Table A.1). Surface snow was sampled with a pre-rinsed mug, steel spoon or soup ladle and transferred into containers made of polyvinylchloride, polyethylene or glass (see table S2 for more details). In March 2018, five samples were taken at different locations on Svalbard (Fig. 1a, Table A.1) by citizen scientists embarking on a land expedition by ski doo (Aemalire project). The citizens were instructed on contamination prevention and equipped with protocol forms, pre-rinsed 2-L stainless-steel containers (Ecotanca), porcelain mug, steel spoon and soup ladle for sampling. In February 2018, the surface of freshly deposited snow was transferred with a spoon from parking cars in the city of Bremen into glass jars (Fig. A.1b, Table A.1). Only surface snow was taken, leaving behind a thin layer of snow on top of car surfaces. One month later, freshly deposited surface snow was collected with a soup ladle on the Isle of Heligoland from the backyard and next to a pedestrian path in front of the Alfred Wegener Institute campus (Fig. A.1b, Table A.1) avoiding soil and bottom surfaces. In the Swiss Alps, surface snow was collected with a spoon at a snowfield in the surroundings of Tschuggen and in the village of Davos (Fig. A.1b). This snow had fallen two days before sampling (07/03/2018). In the Bavarian Alps (Germany), citizens transferred freshly fallen surface snow with a steel spoon from three different locations into glass jars that had been pre-rinsed with tap water (Fig. A.1b, Table A.1): sample 1 was taken next to a country lane, sample 2 was collected next to a country lane and a lake and sample 3 was taken in a green area with nearby anthropogenic activity. All teams worked with bare hands and the equipment was rinsed with Milli-Q water, tap water or snow. Except for the Bavarian samples, potential sample contamination from the sealing of glass jars was minimized by a sheet of tin foil. Samples were kept frozen but defrosted during transport to Heligoland or in the laboratory.

144

A.4 Material and Methods

Table A.1: Details of snow-sampling campaigns.

Wind Sample Area Longitude Latitude Cruise Date Sampler speed

°N °E (m s-1)

Ice Floe 1 Arctic 79.2576 2.3121 PS99.2 07 July 2016 AWI 3.1

Ice Floe 2 Arctic 78.5926 5.2376 PS99.2 02 July 2016 AWI 6

Ice Floe 3 Arctic 79.2576 2.3121 PS99.2 30 June 2016 AWI 9 - 10

Ice Floe 4a Arctic 80.0934 0.2253 PS108 30 August 2017 AWI 2.8 - 3.6

Ice Floe 5 Arctic 80.0922 0.2212 PS108 30 August 2017 AWI 2.8 - 3.6

Ice Floe 6 Arctic 80.0915 0.2203 PS108 30 August 2017 AWI 2.8 - 3.6

Ice Floe 7 Arctic 80.0910 0.2163 PS108 30 August 2017 AWI 2.8 - 3.6

Ice Floe 8 Arctic 80.0904 0.2156 PS108 30 August 2017 AWI 2.8 - 3.6

Ice Floe 9 Arctic 79.0315 3.5447 PS107 04 August 2017 AWI 2.8 - 3.6

Svalbard 1 Arctic 78.2118 16.5511 Aemelire 10 March 2018 Citizens 1 - 3

Svalbard 2 Arctic 77.5107 16.2567 Aemelire 14 March 2018 Citizens 2

Svalbard 3 Arctic 78.1433 15.4215 Aemelire 14 March 2018 Citizens 6

Svalbard 4 Arctic 78.2091 18.4517 Aemelire 16 March 2018 Citizens 2

Svalbard 5 Arctic 77.4586 17.0988 Aemelire 28 March 2018 Citizens 6

N Heligoland 1 Germany 54.1832 7.8884 - 17 March 2018 AWI 18

N Heligoland 2 Germany 54.1835 7.8883 - 17 March 2018 AWI 18

N 25 February Bremen Germany 53.0675 8.7931 - 2018 AWI 2.5

Bavaria 1 Alps 47.6505 11.4335 - 19 March 2018 Citizens 2.5

Bavaria 2 Alps 47.5837 11.3921 - 19 March 2018 Citizens 2.5

Bavaria 3 Alps 47.4367 11.2587 - 19 March 2018 Citizens 2.5

SLF Tschuggen Alps 46.7840 9.9210 - 07 March 2018 Davos 2.5

SLF Davos Alps 46.7986 9.8448 - 07 March 2018 Davos 2.5

145

A.4 Material and Methods

A.4.2 Contamination prevention and procedural blanks All personnel taking samples was instructed to position themselves against the wind and sample undisturbed snow in front of them with bare hands to avoid contamination. If not stated otherwise all laboratory ware was made of glass or stainless steel and thoroughly rinsed with Milli-Q water before use. All polymer-based items, which could not be replaced by alternatives (e.g. bottle caps, filter holders) were made of polytetrafluoroethylene (PTFE), a polymer that cannot be detected within the current FTIR imaging settings (Primpke et al., 2018). Airborne particles were filtered by dustboxes (DB1000, G4 pre-filtration, HEPA-H14 final filtration, Q = 950 m3 h-1, Möcklinghoff Lufttechnik) in laboratories for particle sorting and FTIR analyzes. All filtration steps were performed in a laminar flow cabinet (Scanlaf Fortuna, Labogene). Cotton laboratory coats and clothes were worn to reduce contamination from synthetic textiles. To account for possible contamination from plastic sample containers two procedural blanks were created in the laboratory by filling the PVC and PE containers used with Milli-Q water and storage at -20°C for four days to simulate the freezing and thawing process. To assess possible contamination of the Bavarian snow samples from used tap water and the resin sealing in lids, a similar glass jar was filled with tap water and frozen by the citizen scientists. All blanks were processed in the same way as the snow samples. For the plastic containers up to three FTIR imaging runs were performed due to low numbers found to obtain a reliable result, which represents the extent of MP contamination. No blanks were taken during field sampling. The amounts of MPs determined in the samples with available blanks were corrected by converting the quantity of MPs detected in the blank to the sample volumes (table S2). The number of particles L−1 was calculated for each sample based on the volume of melted snow. The polymer types, abundance and sizes of MPs in procedural blanks are available in table S4. Blanks from the PE container harbored 256 N L-1 and consisted primarily of polyethylene. The PVC flask contained 308 N L-1, most of which was polypropylene and polyvinylchloride. The glass jar that had been pre-rinsed with tap water contained the highest number of MP (27,243 N L-1) with a diverse mix of polymer types, mostly polyethylene and rubber types 1 and 3. To address the uncertainty of all results regarding possible contamination the propagation of uncertainty was calculated (Haave et al., 2019) as described in note S1.

A.4.3 Analytical procedure for the detection of (microplastic) plastics Following three overhead-twists, three times an aliquot (10 ml: 1 ml sample (melted snow) diluted with 9 ml Milli-Q) of each sample was analyzed by FlowCam (Fluid Imaging Technologies, Scarborough, USA) to visualize and quantify particle concentrations and sizes and thereby a potential area coverage. Based on this assessment a specific volume of the sample, ranging between 0.2–100% of the calculated volume for full area coverage was filtered onto an aluminum oxide filter (Anodisc, Ø = 25 mm, pore

146

A.4 Material and Methods size: 0.2 µm, Whatman, Germany). This step is essential to avoid a particle overload of the filter, which would lead to a total absorption of subsequent measurement by FTIR radiation and error in the resulting particle counts. Despite this precaution, the membrane of eight snow samples became clogged during filtration such that it was stopped and the filtrated volume recorded. After drying in a in the desiccator for 2 days, the filter was placed onto the calcium fluoride window of the FTIR-microscope and an overview image taken (40 x magnification). The particles on the filter were analyzed by a Hyperion 3000 μFTIR microscope equipped with a focal plane array (FPA) detector with 64 × 64 detector elements and connected to a TENSOR 27 spectrometer (Bruker Optics GmbH, Germany). A VIS objective (4 × magnification) and an IR objective (3.5 × magnification) along with an infrared range of 3,600−1,250 cm−1 was used for measurements performed with the software OPUS 7.5 (Bruker Optics). The measurement area of 20 × 20 FPA (14.1 × 14.1 mm) fields comprised the entire sample-filter surface and produced measurements of 400 single fields. Measurements were performed by a 3.5 × IR-objective in transmission mode with 32 scans per FPA field without binning at a resolution of 8 cm−1 allowing the detection of particles down to 11 µm in ~4.5 h.

A.4.4 Detection of fibers Because of the design of the particle analysis pipeline during FTIR image analysis fibers can currently not be identified reliably with this automated method. Still, to enable comparison with previous studies, the fibers present on each filter were counted and photographed using a stereomicroscope (Olympus SZX16, 8 × magnification). The total number of fibers per filter was extrapolated to 1 liter of melted snow. The total number of fibers found in the blanks (N = 83) was subtracted from the sample results. In addition, the length of fibers of suspected anthropogenic origin was measured using CellSens Micro Imaging tools (8 - 32 × magnification, Olympus, Germany).

A.4.5 Data analysis The FTIR imaging data were automatically processed (Primpke et al., 2017). Briefly, each spectrum in the measurement file was analyzed via two library searches to confirm polymer identity using an adaptable database design (Primpke et al., 2018). The library can be downloaded (https://link.springer.com/article/10.1007/s00216-018-1156-x). Each pixel identified was stored with its position, analysis quality and finally assigned polymer type into a file, which was subject to image analysis based on Python 3.4 scripts and Simple ITK functions (Primpke et al., 2017). This approach enabled the identification, quantification and size determination of all polymer particles whilst excluding human bias (Primpke et al., 2017). MP particles were assigned to size classes to reduce the complexity of the size distribution and for comparison with previous studies.

147

A.5 Supplementary materials

All statistical comparisons were made based on non-parametric statistics (Mann-Whitney U test, MINITAB 18, Statistica 13). We tested for Spearman’s rank correlations between wind speed on the sampling day and MP quantity as well as correlations between MP and fiber concentrations (Minitab 18; p > 0.05). The polymer composition of samples from Europe and the Arctic was compared by multivariate analyses (PERMANOVA, PRIMER-e version 6.1.16 with PERMANOVA 1.0.6) based on Bray−Curtis similarities of 4th-root transformed data of polymer types (Anderson and Walsh, 2013). The SIMPER routine of PRIMER-e was used to assess what polymer type contributed to the dissimilarity. The annual MP and fiber fallout was calculated for the areas sampled using mean annual snowfall values for Fram Strait (200 kg m-2 (Gallet et al., 2017)), Svalbard (450 kg m-2 (Winther et al., 2003)) and Davos (500 kg m-2 (Christoph, 2017)) as tentative estimates for MP and fiber deposition rates via snow. Bremen and Heligoland were not included because snow fall in these regions is ephemeral. The density of melted snow samples was assumed as water density (1 kg L-1)

A.5 Supplementary materials This supplementary chapter contains the following material: • Fig. S1. Length frequency (%) of microfibers found in Arctic and European snow. • Note S1. Calculation of propagation of uncertainty of results reg. possible contamination. • Table S1. Concentration of microplastics (MPs), fibers polymer types and other particles.

Further Supplementary material was not included to this chapter due to its volume. It is available at: https://advances.sciencemag.org/content/suppl/2019/08/12/5.8.eaax1157.DC1 • Table S2. Concentration of microplastics as well as size frequency distribution and metadata. • Table S3. Number, size and color of fibers measured in Arctic and European snow samples. • Table S4. Concentration, composition and size of microplastics detected in blank samples

148

A.5 Supplementary materials

Figure S1: Length frequency (%) of microfibers found in Arctic and European snow. Each bar represents a size bin of 100 µm. Note: these fibers were not identified as polymers

Note S1. Calculation of propagation of uncertainty of results regarding possible contamination. To address uncertainties due to the missing field blanks we calculated the uncertainty of our values with regards to the error of particle numbers of FTIR (5%, N(FTIR)), the filtration process and the minimum number (N(Low. N.) = 5 of Ice Floe 4) found in samples from glass containers following an approach similar to (1) using the following equation:

1 2 푁 2 ∆푁(푆푛표푤) = √( × (푁(퐹푇퐼푅) + 푁(퐿표푤. 푁. ))) + ( × ∆푉) 푉2 −푉2

This calculation lead to an average relative error of 37% of the presented values, while sample with low numbers of particles detected were generally having a higher value. Only two samples showed a relative error value above 100% (Ice Floe 3 and 4). Ice Floe 3 was already blank-corrected while Ice Floe 4 showed the lowest number of particles.

149

A.5 Supplementary materials

3486

16866

4225

669

16373

458

176

246

458

739

352

931

2676

1258

0.028

Davos

calculations

unless unless stated

and

5927

8245

690

1.3

191

1591

0.11

Tschuggen

400

182114

1143

286

124114

4400

1486

2573

1.9

6114

107

5498

0.018

roundedvalues

2715

67048

399

5144

2229

81606

31097

415

3389

1172

430

21571

1817

1348

11193

1537

66.5

154137

5903

8873

0.038

All values are All are givenas values N L

265

49880

3614

106747

651

867

193

651

210

1.6

2627

109

6770

0.042

Bavaria

based on non on based

209

67339

261

6226

104

1548

139

1927

2.6

2000

115

4372

0.058

Bremen

data

102200

400400

13400

4400

21000

400

200

12200

200

400

3800

400

2745

3.1

17600

2795

0.005

for for

44821

455357

4107

3036

25179

3393

536

2321

357

3750

357

536

838

2.2

11786

3048

0.006

Heligoland

1975

2469

123

1667

185

741

130

1111

120

0.016

1233

256

14469

103

0.8

122

3953

0.246

713

2923

179

1944

231

144

113

181

12.7

846

165

1303

0.195

7573

9029

194

1359

485

1498

3.6

680

194

0.010

984

5440

123

284

218

579

5.1

369

763

0.106

Svalbard Svalbard

200

23467

533

2067

867

11200

133

145

37.3

14400

216

579

0.015

31771

3786

15579

123

0.4

186

7187

0.140

of rounding calculations (see Supplementary Table (seeSupplementary calculations rounding of

528

3192

152

121

2.2

495

0.125

279

3246

1492

584

4.3

230

324

0.061

895

1561

218

153

774

7.5

218

266

0.092

28313

64096

482

361

241

10186

0.6

602

776

0.008

218

7.5

0.204

611

383

0.0

0.014

87.9

124

0.078

Ice Ice Floe

]

acetate

]

1. 1. Concentration ofmicroplastics (MPs), types polymer fibers and particles other detected samples. snow in

vinyl

particle [N]

Sand

Plant Plant fibers

Chitin Chitin

Coal Coal

Charcoal Charcoal

Animal fur Animalfur

Rubber type 3 type Rubber

Rubber type 1 type Rubber

Polyimide Polyimide

Ethylene

Polycaprolactone Polycaprolactone

Polylactic Polylactic acid

Polychloroprene Polychloroprene

s/varnish/lacquer

Acrylates/polyurethane

Polyester Polyester

Nitrile rubber Nitrile rubber

modified

Cellulose Cellulose chemical

Polyvinylchloride Polyvinylchloride

Polyamide Polyamide

Polycarbonate Polycarbonate

Polystyrene Polystyrene

Polypropylene Polypropylene

chlorinated chlorinated

Polyethylene Polyethylene

Polyethylene oxidized Polyethylene

Polyethylene Polyethylene

Fibers L [N

MPs MPs [%]

Total L MPs [N

Polymer Polymer

Total[N] particles

Filtrated [L] volume

otherwise. Some values diverge slightly as a divergeas slightly result otherwise.values Some Table Table 150

Acknowledgements We gratefully acknowledge the principal scientists and crews and of RV Polarstern cruises (PS93.2, PS99.2, PS107, PS108) as well as helicopter crews. We thank S. Flögel (GEOMAR), J. Hagemann, M. Huchler, M. Kaess, J. Lemburg, N. Lochthofen, J. Ludszuweit, J. Rapp, D. Scholz, M. Spill (AWI), M. Schneebeli and M. Jaggi (SLF Davos); K. Mützel and K. Wocher (citizens) for sampling snow. Snow from Svalbard was collected within the citizen-science project AEMELIRE by K. Müller (RIS-ID 10992). We thank M. Egger and three anonymous reviewers whose comments improved an earlier version of the manuscript.

References Appendix A

References Anderson, M.J., and Walsh, D.C.I. (2013). PERMANOVA, ANOSIM, and the Mantel test in the face of heterogeneous dispersions: What null hypothesis are you testing? Ecological Monographs 83(4), 557-574. doi: doi:10.1890/12-2010.1. Barrows, A.P.W., Cathey, S.E., and Petersen, C.W. (2018). Marine environment microfiber contamination: Global patterns and the diversity of microparticle origins. Environmental Pollution 237, 275-284. doi: https://doi.org/10.1016/j.envpol.2018.02.062. Bergmann, M., Lutz, B., Tekman, M.B., and Gutow, L. (2017a). Citizen scientists reveal: Marine litter pollutes Arctic beaches and affects wild life. Marine Pollution Bulletin 125(1-2), 535–540. doi: https://doi.org/10.1016/j.marpolbul.2017.09.055. Bergmann, M., Sandhop, N., Schewe, I., and D’Hert, D. (2016). Observations of floating anthropogenic litter in the Barents Sea and Fram Strait, Arctic. Polar Biology 39(3), 553-560. doi: 10.1007/s00300-015-1795-8. Bergmann, M., Tekman, M.B., and Gutow, L. (2017b). Marine litter: Sea change for plastic pollution. Nature 544(7650), 297-297. doi: 10.1038/544297a. Bergmann, M., Wirzberger, V., Krumpen, T., Lorenz, C., Primpke, S., Tekman, M.B., et al. (2017c). High Quantities of Microplastic in Arctic Deep-Sea Sediments from the HAUSGARTEN Observatory. Environmental Science & Technology 51(19), 11000–11010. doi: 10.1021/acs.est.7b03331. Bertling, J., Bertling, R., and Hamann, L. (2018). "Kunststoffe in der Umwelt: Mikro- und Makroplastik. Ursachen, Mengen, Umweltschicksale, Wirkungen, Lösungsansätze, Empfehlungen", (ed.) S.- u.E.U. Fraunhofer-Institut für Umwelt-. (Oberhausen). Cai, L., Wang, J., Peng, J., Tan, Z., Zhan, Z., Tan, X., et al. (2017). Characteristic of microplastics in the atmospheric fallout from Dongguan city, China: preliminary research and first evidence. Environmental Science and Pollution Research 24(32), 24928-24935. doi: 10.1007/s11356-017- 0116-x. Catarino, A.I., Macchia, V., Sanderson, W.G., Thompson, R.C., and Henry, T.B. (2018). Low levels of microplastics (MP) in wild mussels indicate that MP ingestion by humans is minimal compared to exposure via household fibres fallout during a meal. Environmental Pollution 237, 675-684. doi: https://doi.org/10.1016/j.envpol.2018.02.069. Christoph, M. (2017). GCOS SWE data from 11 stations in Switzerland. WSL Institute for Snow and Avalanche Research SLF. doi: 10.16904/15. Available: https://www.envidat.ch/dataset/gcos- swe-data. Cózar, A., Martí, E., Duarte, C.M., García-de-Lomas, J., van Sebille, E., Ballatore, T.J., et al. (2017). The Arctic Ocean as a dead end for floating plastics in the North Atlantic branch of the Thermohaline Circulation. Science Advances 3(e1600582). doi: 10.1126/sciadv.1600582. Dehghani, S., Moore, F., and Akhbarizadeh, R. (2017). Microplastic pollution in deposited urban dust, Tehran metropolis, Iran. Environmental Science and Pollution Research 24(25), 20360-20371. doi: 10.1007/s11356-017-9674-1. Dris, R., Gasperi, J., Mirande, C., Mandin, C., Guerrouache, M., Langlois, V., et al. (2017). A first overview of textile fibers, including microplastics, in indoor and outdoor environments. Environmental Pollution 221, 453-458. doi: http://dx.doi.org/10.1016/j.envpol.2016.12.013. Dris, R., Gasperi, J., Rocher, V., Saad, M., Renault, N., and Tassin, B. (2015). Microplastic contamination in an urban area: a case study in Greater Paris. Environmental Chemistry 12(5), 592-599. doi: http://dx.doi.org/10.1071/EN14167.

152

References Appendix A

Dris, R., Gasperi, J., Saad, M., Mirande, C., and Tassin, B. (2016). Synthetic fibers in atmospheric fallout: A source of microplastics in the environment? Marine Pollution Bulletin 104, 290-293. doi: http://dx.doi.org/10.1016/j.marpolbul.2016.01.006. Eckhardt, S., Stohl, A., Beirle, S., Spichtinger, N., James, P., Forster, C., et al. (2003). The North Atlantic Oscillation controls air pollution transport to the Arctic. Atmospheric Chemistry and Physics 3(5), 1769-1778. doi: 10.5194/acp-3-1769-2003. Enders, K., Lenz, R., Stedmon, C.A., and Nielsen, T.G. (2015). Abundance, size and polymer composition of marine microplastics ≥ 10 μm in the Atlantic Ocean and their modelled vertical distribution. Marine Pollution Bulletin 100(1), 70-81. doi: http://dx.doi.org/10.1016/j.marpolbul.2015.09.027. Gallet, J.-C., Merkouriadi, I., Liston, G.E., Polashenski, C., Hudson, S., Rösel, A., et al. (2017). Spring snow conditions on Arctic sea ice north of Svalbard, during the Norwegian Young Sea ICE (N-ICE2015) expedition. Journal of Geophysical Research: Atmospheres 122(20), 10,820-810,836. doi: 10.1002/2016jd026035. Gasperi, J., Wright, S.L., Dris, R., Collard, F., Mandin, C., Guerrouache, M., et al. (2018). Microplastics in air: Are we breathing it in? Current Opinion in Environmental Science & Health 1, 1-5. doi: https://doi.org/10.1016/j.coesh.2017.10.002. Geyer, R., Jambeck, J.R., and Law, K.L. (2017). Production, use, and fate of all plastics ever made. Science Advances 3(7), e1700782. doi: 10.1126/sciadv.1700782. Golokhvast, K.S. (2014). Airborne Biogenic Particles in the Snow of the Cities of the Russian Far East as Potential Allergic Compounds. Journal of Immunology Research 2014, 7. doi: 10.1155/2014/141378. Grøsvik, B.E., Prokhorova, T., Eriksen, E., Krivosheya, P., Horneland, P.A., and Prozorkevich, D. (2018). Assessment of Marine Litter in the Barents Sea, a Part of the Joint Norwegian–Russian Ecosystem Survey. Frontiers in Marine Science 5(72). doi: 10.3389/fmars.2018.00072. Haave, M., Lorenz, C., Primpke, S., and Gerdts, G. (2019). Different stories told by small and large microplastics in sediment - first report of microplastic concentrations in an urban recipient in Norway. Marine Pollution Bulletin 141, 501-513. doi: https://doi.org/10.1016/j.marpolbul.2019.02.015. Heintzenberg, J., and Rummukainen, M. (1993). Airborne particles in snow. Journal of Glaciology 39(132), 239-244. doi: 10.3189/S0022143000015896. Jambeck, J.R., Geyer, R., Wilcox, C., Siegler, T.R., Perryman, M., Andrady, A., et al. (2015). Plastic waste inputs from land into the ocean. Science 347(6223), 768-771. doi: 10.1126/science.1260352. Kanhai, L.D.K., Gårdfeldt, K., Lyashevska, O., Hassellöv, M., Thompson, R.C., and O'Connor, I. (2018). Microplastics in sub-surface waters of the Arctic Central Basin. Marine Pollution Bulletin 130, 8- 18. doi: https://doi.org/10.1016/j.marpolbul.2018.03.011. Lebreton, L., and Andrady, A. (2019). Future scenarios of global plastic waste generation and disposal. Palgrave Communications 5(1), 6. doi: 10.1057/s41599-018-0212-7. Lonati, G., Giugliano, M., and Cernuschi, S. (2006). The role of traffic emissions from weekends’ and weekdays’ fine PM data in Milan. Atmospheric Environment 40(31), 5998-6011. doi: https://doi.org/10.1016/j.atmosenv.2005.12.033. Lusher, A.L., Tirelli, V., O’Connor, I., and Officer, R. (2015). Microplastics in Arctic polar waters: the first reported values of particles in surface and sub-surface samples. Scientific Reports 5, 14947. doi: 10.1038/srep14947. Morgana, S., Ghigliotti, L., Estévez-Calvar, N., Stifanese, R., Wieckzorek, A., Doyle, T., et al. (2018). Microplastics in the Arctic: A case study with sub-surface water and fish samples off Northeast

153

References Appendix A

Greenland. Environmental Pollution 242, 1078-1086. doi: https://doi.org/10.1016/j.envpol.2018.08.001. Nazarenko, Y., Fournier, S., Kurien, U., Rangel-Alvarado, R.B., Nepotchatykh, O., Seers, P., et al. (2017). Role of snow in the fate of gaseous and particulate exhaust pollutants from gasoline-powered vehicles. Environmental Pollution 223, 665-675. doi: https://doi.org/10.1016/j.envpol.2017.01.082. Obbard, R.W., Sadri, S., Wong, Y.Q., Khitun, A.A., Baker, I., and Thompson, R.C. (2014). Global warming releases microplastic legacy frozen in Arctic Sea ice. Earth's Future 2(6), 2014EF000240. doi: 10.1002/2014EF000240. Outridge, P.M., Macdonald, R.W., Wang, F., Stern, G.A., and Dastoor, A.P. (2008). A mass balance inventory of mercury in the Arctic Ocean. Environmental Chemistry 5(2), 89-111. doi: https://doi.org/10.1071/EN08002. Pauly, J.L., Stegmeier, S.J., Allaart, H.A., Cheney, R.T., Zhang, P.J., Mayer, A.G., et al. (1998). Inhaled cellulosic and plastic fibers found in human lung tissue. Cancer Epidemiology Biomarkers & Prevention 7(5), 419-428. Peeken, I., Primpke, S., Beyer, B., Gütermann, J., Katlein, C., Krumpen, T., et al. (2018). Arctic sea ice is an important temporal sink and means of transport for microplastic. Nature Communications 9(1), 1505. doi: 10.1038/s41467-018-03825-5. Pimentel, J.C., Avila, R., and Lourenço, A.G. (1975). Respiratory disease caused by synthetic fibres: a new occupational disease. Thorax 30(2), 204-219. Primpke, S., Lorenz, C., Rascher-Friesenhausen, R., and Gerdts, G. (2017). An automated approach for microplastics analysis using focal plane array (FPA) FTIR microscopy and image analysis. Analytical Methods 9(9), 1499-1511. doi: 10.1039/C6AY02476A. Primpke, S., Wirth, M., Lorenz, C., and Gerdts, G. (2018). Reference database design for the automated analysis of microplastic samples based on Fourier transform infrared (FTIR) spectroscopy. Analytical and Bioanalytical Chemistry 410(21), 5131-5141. doi: 10.1007/s00216-018-1156-x. Rousseau, D.-D., Duzer, D., Etienne, J.-L., Cambon, G., Jolly, D., Ferrier, J., et al. (2004). Pollen record of rapidly changing air trajectories to the North Pole. Journal of Geophysical Research: Atmospheres 109(D6). doi: doi:10.1029/2003JD003985. Sehmel, G.A. (1973). Particle resuspension from an asphalt road caused by car and truck traffic. Atmospheric Environment (1967) 7(3), 291-309. doi: https://doi.org/10.1016/0004- 6981(73)90078-4. Silpa, K., Yao, L., Bhada-Tata, P., and Van Woerden, F. (2018). "What a Waste 2.0: A Global Snapshot of Solid Waste Management to 2050", in: Urban Development. (Washington, DC). Tekman, M.B., Krumpen, T., and Bergmann, M. (2017). Marine litter on deep Arctic seafloor continues to increase and spreads to the North at the HAUSGARTEN observatory. Deep-Sea Res. I 120, 88- 99. doi: http://dx.doi.org/10.1016/j.dsr.2016.12.011. Uematsu, M., Duce, R.A., and Prospero, J.M. (1985). Deposition of atmospheric mineral particles in the North Pacific Ocean. Journal of Atmospheric Chemistry 3(1), 123-138. doi: 10.1007/bf00049372. van der Does, M., Knippertz, P., Zschenderlein, P., Giles Harrison, R., and Stuut, J.-B.W. (2018). The mysterious long-range transport of giant mineral dust particles. Science Advances 4(12), eaau2768. doi: 10.1126/sciadv.aau2768. Winther, J.-G., Bruland, O., Sand, K., Gerland, S., Marechal, D., Ivanov, B., et al. (2003). Snow research in Svalbard—an overview. Polar Research 22(2), 125-144. doi: 10.3402/polar.v22i2.6451. Woodall, L.C., Gwinnett, C., Packer, M., Thompson, R.C., Robinson, L.F., and Paterson, G.L.J. (2015). Using a forensic science approach to minimize environmental contamination and to identify

154

References Appendix A

microfibres in marine sediments. Marine Pollution Bulletin 95, 40-46 doi: http://dx.doi.org/10.1016/j.marpolbul.2015.04.044. Yuyan L., Bo F., Yaxing S., Ting W., Zhizhong Z., and L., H. (2018). Scavenging of atmospheric particulates by snow in Changji, China. Global NEST Journal 20(3), 471 - 476. Zhao, S., Yu, Y., He, J., Yin, D., and Wang, B. (2015). Below-cloud scavenging of aerosol particles by precipitation in a typical valley city, northwestern China. Atmospheric Environment 102, 70-78. doi: https://doi.org/10.1016/j.atmosenv.2014.11.051

155

Appendix B Collecting snow samples of environmental proxies – Best practice

Best-practice guideline based on the procedures applied in the MiSo project for collecting samples.

B.1 Introduction

B.1 Introduction Snow cover analysis has a long history worldwide. In the Alps early snow research concentrated mostly on avalanches, which is why the recording of structural parameters such as density and grain shape was of great interest (Colbeck, 1987; Fierz et al., 2009; Marty and Meister, 2012; Pielmeier and Schneebeli, 2003a). In polar regions similar parameters were recorded to understand physical processes taking place (Colbeck, 1987). Around 1940, the usefulness of snow measurements for monitoring environmentally and hydrologically relevant processes was recognized. Since then, advances in analytics have led to various studies that have shown snow to be a dynamic material in which a variety of physical, chemical and biological processes take place (Abbatt, 2013; Bartels-Rausch et al., 2013, 2014; Beine et al., 2002; Dominé and Shepson, 2002; Grannas et al., 2007; McNeill et al., 2012). Despite the large number of field studies, there are hardly any uniform sampling procedures. On the one hand, the variety of substances require different prerequisites for sampling. On the other hand, methods for sampling are permanently evolving. In the following, we give a short overview of protocols which are useful for planning and performing a sampling study. While a harmonized standard procedure is missing, recording structural parameters are described in several protocols worldwide (AAA, 2016; CAA, 2007; Dürr et al., 2016; NZMSC, 2017). However, the mandatory parameters for snow sampling were clearly defined in the International Snow Classification (Fierz et al., 2009). We recommend the observer's manual of the WSL Institute for Snow and Avalanche Research SLF (Dürr et al., 2016) as it provides detailed information on the preparation of manually observed snow profiles and a corresponding field book is available (Darms et al., 2014). The detailed design, preparation and completion of field studies is described in the protocol of the USGS (Ingersoll et al., 2009). The USGS protocol includes standard operating procedures (SOP) for collecting samples, but does not describe the specific needs for sampling different substances (Ingersoll et al., 2009, pp. 21 ff.). Gallet et. al (2018) presented a procedure for sampling black carbon, water isotopes, major ions and microorganisms. In the following we present the sampling procedures implemented within the project "Microscale Distribution of impurities in Snow and Glacier Ice" (MiSo). This guideline focuses on taking snow samples to investigate major ions, trace element, water isotopes, black carbon and microplastic. This guideline extends the procedures mentioned above (Gallet et al., 2018; Ingersoll et al., 2009) and is a step forward into the direction of a standardized sampling protocol.

158

B.2 Sampling

B.2.1 Basics The aim of any sampling is to sample the impurities present in the snow cover without vertical discontinuity and without contamination. In addition to the actual sampling in the snow pit, further work steps such as approaching the measuring field with the entire equipment, reporting in the field (if necessary including geodata), protective equipment against contamination and the transport of the frozen samples back to the laboratory must also be planned. Therefore, a measurement campaign has to be carefully designed. If a parameter is not or only inadequately recorded while sampling, this can usually not be corrected subsequently. For the administrative planning we recommend the monitoring protocol of the USGS (Ingersoll et al., 2009, pp. 6-10).

B.2.2 Materials The necessary sampling equipment depends on the parameters which should be recorded. It should be available and organized at an early stage, so that it can be pre-prepared for sampling. Isotopes, ions, and black carbon can be collected by the identical sampling tools, but trace elements and microplastic require different tools. All tools and containers used for snow sampling in the MiSo project (analysis of ions and trace elements in ppb concentrations) were carefully pre-cleaned 5 times with ultrapure water (e.g. with 18 MΩcm quality, arium® pro, Sartorius, Göttingen, Germany). After cleaning, the tools were packed in rinsed plastic bags and sealed until use.

B.2.2.1 Basic equipment In Table 1 the tools necessary for a snow pit sampling are listed. The quantities and in particular the product types listed are to be understood as examples to facilitate the planning of a comparable campaign. Of course, other products can be applied as well.

B.2.2.2 Sampling containers The sampling of different species demands variable requirements. When collecting samples for ion analysis, the usage of metallic tools should be avoided (risk of contamination with ions from the metal), but metallic tools (besides glass) are exactly the right choice for taking microplastic samples, as these samples could be contaminated by any synthetic tools. The different requirements are listed in Table B.2.

159

B.2 Sampling

Table B.1: Listing of the sampling equipment. Cat. Tool # Type snow shovels 2 - stadia rod 2 - Traditional snow saw 1 - Equipment instruments for traditional profile recording - see Dürr et al. (2016) (density, temperature, etc.) avalanche Safety Equipment if necessary - - white spatula 2 Semadeni scraper, PP, Nr. 3563/3564 white small shovel 1 Semadeni shovel PP Nr. 2645/2650 General tools custom‐built rectangular (15 × 24 cm) sampler 1 material: PP for sampling sterile towels 100 Rohr AG, cleaning towel Viper P160, PE plastic bags 20 clean plastic gloves 50 Clean gloves, Coop, Switzerland Prevention of clean room overalls 3 Tyvek IsoClean, DuPont, Wilmington, USA contamination respirator face masks 3 (3M, Maplewood, USA) sampling containers/vials - Transport Isolated transport container 2

Table B.2: Requirements for sampling different species such as major ions, trace elements, water isotopes, black carbon and Microplastic. The minimum sample volume represents the absolute minimum value of the volume at which a sample can be analyzed with current analysis methods such as plasma mass spectrometry (ions, trace element) and ringdown spectrometer (isotopes). Cat. Major Ions Trace Elements Water Isotopes Black Carbon Microplastic Material Sampling PP or HDPE PP or HDPE PP or HDPE PP or HDPE glas or metal Tools/Containers Recommended 2000 ml jar glass 50 ml (PP) vials 50 ml (PP) vials 50 ml (PP) vials 50 ml (PP) vials sample storage containers Minimal sample 5 mL 3 mL 3 ml 10 ml 500 ml Vol. (water) Recommended 25 mL 25 mL 25 ml 30 ml 2000 ml Vol. (water) 3 x dish washer 5 x 24h bath in (VE-water, no Pre-cleaning of 5 x 24h bath in ultrapure water 5 x 24h bath in 3 x 24h bath in soap) vials applied ultrapure water and once with ultrapure water ultrapure water rinsing in situ with 0.2 M HNO 3 snow Contamination - clean-overall - clean-overall - clean-overall - clean-overall - cotton lab coat prevention during - respir. mask - respir. mask - respir. mask - respir. mask - no synthetics sampling - clean gloves - clean gloves - clean gloves - clean gloves clothes if possible Importance of melt- prevention during moderate high moderate high low transport Overall risk of contamination high high moderate moderate moderate during sampling

B.2.3 Choice of field site The profile location should be selected in such a way, that the following criteria are met: - Protected from avalanches - Undisturbed and as homogeneous as possible snow cover (possibly probing) - Not in ridge locations as there may be snow drift impact - Far away from local disturbance (roads, huts) that affect the seasonal chemical deposition - Take in account wind direction and solar radiation.

160

B.2 Sampling

B.2.4 Snow pit preparation The excavation of the snow profile (approx. 1.5 x 1.5 m) can be carried out analogously to a traditional profile. For an efficient excavation it is recommended to create snow blocks with the shovel or the saw and extract them. The following points should be considered: - The pit wall on which the snow profile is created (profile wall) must be shielded from direct sunlight. - After digging the pit, use a shovel or saw to prepare the profile wall as vertically and flat as possible. A straight and flat profile wall supports the collection of high-quality samples. - In strong winds, the snow shoveled out of the pit should be leeward from the pit. - If something gets exposed to sunlight, snow blocks can be stacked to build a protective wall. The snow pit has to be prepared for efficient sampling with defined spots for the tools (protected from direct sunlight), for filled containers (precooling in the snowpack recommended) and the protocol writing (Fig. B.1a).

Figure B.1: a) Two team members, dressed in clean overalls, are collecting samples inside the snow pit. Writing the protocols and storing of the filled samples is maintained by a third member outside the snow pit. b) Temporary storage of collected samples protected from sunlight.

B.2.5 Sample collection If besides the impurity profile a traditional snow profile has to be recorded, this can be done both, before and after sampling. A recording prior to the sampling has the advantage of revealing layer boundaries. Additional SnowMicroPen measurements are highly recommended, as they give a more precise localization than traditional profile (Pielmeier and Schneebeli, 2003b). A near-infrared photography before and during sampling can also be useful to localize the exact depositional features afterwards (Matzl and Schneebeli, 2006). Simultaneous recording should be avoided in any case. As soon as everything is prepared (and not before), all team members should be dressing up with clean room overalls, respirator facemasks, and ultra-clean plastic gloves, to avoid contamination during the sampling.

161

B.2 Sampling

Figure B.2: Recommended sampling procedure using cylindrical vials: A) Sampler is pushed into wall. B) Vials are vertically pushed towards the sampler. Depending on the sampler area, several vials can be filled without removing the sampler. C) With help of a spatula the vials are turned back upside-up. D) Vials are sealed and stored in a sun-protected place. E) When all samples for the specific height are collected (six inhere) the sampler is removed from the snow wall and cleaned from snow rests. F) The sampler is aligned for next collection.

The sampling is carried out starting from the surface towards the ground: - Prior to collecting samples, a few centimeters of the profile wall must be cut off with a clean spatula to eliminate possible contamination from the preceding excavation. - The rectangular (15 × 24 cm) polycarbonate sampler is pushed horizontally into the profile wall (Fig. B.2 A) - The sampling containers (vials) are pushed vertically towards the sampler (Fig. B.2 B). - For each different species (ions, trace elements and stable water isotopes), separate vials must be used. In order to prevent possible cross contamination (e.g. trace element vials that are

prepared with 0.2 M HNO3) between the samples, the area above the sampler should be divided into different areas and the corresponding vials always punctured in the same area. - With a spatula the full vials should be turned back upside-up, sealed and placed in the snowpack for temporary cooling (Fig. B.2 C) - After collecting all samples of a layer, the sampler is pulled out of the profile wall and any remaining snow is removed (Fig. B.2 E). - The sampler is aligned to the next height (according to the applied sampling resolution e.g. 6 cm lower) and inserted again into the profile wall (Fig. B.2 F).

162

B.3 Conclusion

This procedure is repeated until the ground is reached. For the lowest layer, the sample containers can be filled directly to the ground. Finally, the snow profile must be backfilled with snow to re-establish a homogeneous snow surface. The samples are packed into an insulated box and transported promptly to the next cold laboratory. For the collection of microplastic samples, the equipment differs significantly (Tab. B.2). Furthermore, the quantity required for the analysis is higher. In the MiSo project, we took these samples only from the surface. For this purpose, snow was shoveled into jar glasses with a metal scoop (Fig B.3a). For a microplastic snow profile sampling we recommend an adapted proceeding as shown in “Sampling for BC and microorganisms” in Gallet et. al (2018, p.18). In this procedure, entire snow blocks are cut out of the profile wall layer by layer with a saw. Instead of being packed into plastic bags, the blocks must be filled into glass containers. If feasible, no synthetic clothes should worn during sampling and samples should be collected off-wind to avoid contamination.

Figure B.3: a) Metal spoon for collecting microplastic samples. b) Filled glass containers at the collection location ready for the transport to the lab.

B.3 Conclusion In this protocol, we presented the snow sampling procedures that have been designed and applied during the MiSo project. As internationally recognized protocols are currently lacking, it is even more important to provide a clear description of the methods applied and, if possible, to share and compare them within the cryosphere community. For the implementation of future campaigns, it will be essential that compulsory protocols are available and followed. This not only increases the quality of the collected data, but also allows the comparison of different measurement campaigns among each other. This chapter should be a step in the direction of a recognized sampling protocol.

163

References Appendix B

References

AAA (2016). Snow, Weather and Avalanches: Observation Guidelines for Avalanche Programs in the United States. 3rd ed. American Avalanche Association (AAA). Abbatt, J. (2013). Arctic snowpack bromine release. Nat. Geosci. 6, 331–332. doi:10.1038/ngeo1805. Bartels-Rausch, T., Jacobi, H. W., Kahan, T. F., Thomas, J. L., Thomson, E. S., Abbatt, J. P. D., et al. (2014). A review of air-ice chemical and physical interactions (AICI): Liquids, quasi-liquids, and solids in snow. Atmos. Chem. Phys. 14, 1587–1633. doi:10.5194/acp-14-1587-2014. Bartels-Rausch, T., Wren, S. N., Schreiber, S., Riche, F., Schneebeli, M., and Ammann, M. (2013). Diffusion of volatile organics through porous snow: Impact of surface adsorption and grain boundaries. Atmos. Chem. Phys. 13, 6727–6739. doi:10.5194/acp-13-6727-2013. Beine, H. J., Dominé, F., Simpson, W., Honrath, R. E., Sparapani, R., Zhou, X., et al. (2002). Snow-pile and chamber experiments during the Polar Sunrise Experiment ‘Alert 2000’: exploration of nitrogen chemistry. Atmos. Environ. 36, 2707–2719. doi:10.1016/S1352-2310(02)00120-6. CAA (2007). Observation guidelines and recording standards for weather, snowpack and avalanches. Revelstoke, BC: Canadian Avalanche Association (CAA). Colbeck, S. C. (1987). History of snow-cover research. Journal of Glaciolology 33, 60–65. doi:10.3189/S0022143000215839. Darms, G., Dürr, L., and Schweizer, J. (2014). Field book for snow profile observations. Military m. Davos: WSL Institute for Snow and Avalanche Research SLF. Dominé, F., and Shepson, P. B. (2002). Air-snow interactions and atmospheric chemistry. Science 297, 1506–1510. doi:10.1126/science.1074610. Dürr, L., Darms, G., and Stucki, T. (2016). SLF-Beobachterhandbuch. Davos Available at: https://www.slf.ch/fileadmin/user_upload/WSL/Publikationen/Sonderformate/pdf/SLF- Beobachterhandbuch.pdf [Accessed September 6, 2019]. Fierz, C., Armstrong, R. L., Durand, Y., Etchevers, P., Greene, E., McClung, D. M., et al. (2009). The international classification for seasonal snow on the ground (UNESCO, IHP (International Hydrological Programme)– VII , Technical Documents in Hydrology , No 83 ; IACS. Paris. Gallet, J., Bjorkman, M. P., Larose, C., Luks, B., Martma, T., and Zdanowicz, C. (2018). Protocols and recommendations for the measurement of snow physical properties, and sampling of snow for black carbon, water isotopes, major ions and microorganisms. Grannas, A. M., Jones, A. E., Dibb, J., Ammann, M., Anastasio, C., Beine, H. J., et al. (2007). An overview of snow photochemistry: Evidence, mechanisms and impacts. Atmos. Chem. Phys. 7, 4329– 4373. doi:10.5194/acp-7-4329-2007. Ingersoll, G., Campbell, D., Mast, A., Clow, D., Nanus Leora, and Brent, F. (2009). Snowpack Chemistry Monitoring Protocol for the Rocky Mountain Network; Narrative and Standard Operating Procedures. Reston Available at: https://irma.nps.gov/DataStore/Reference/Profile/2184216 [Accessed September 10, 2019]. Marty, C., and Meister, R. (2012). Long-term snow and weather observations at Weissfluhjoch and its relation to other high-altitude observatories in the Alps. Theor. Appl. Climatol. 110, 573–583. doi:10.1007/s00704-012-0584-3. Matzl, M., and Schneebeli, M. (2006). Measuring specific surface area of snow by near-infrared photography. J. Glaciol. 52, 558–564. doi:10.3189/172756506781828412.

References Appendix B

McNeill, V. F., Grannas, A. M., Abbatt, J. P. D., Ammann, M., Ariya, P., Bartels-Rausch, T., et al. (2012). Organics in environmental ices: sources, chemistry, and impacts. Atmos. Chem. Phys. 12, 9653– 9678. doi:10.5194/acp-12-9653-2012. NZMSC (2017). New Zealand Guidelines and Recordings Standards for Weather, Snowpack and Avalanche Observations. 6th ed. Wellington: New Zealand Mountain Safety Council. Pielmeier, C., and Schneebeli, M. (2003a). Developments in the Stratigraphy of Snow. Surv. Geophys. 24, 389–416. doi:10.1023/B:GEOP.0000006073.25155.b0. Pielmeier, C., and Schneebeli, M. (2003b). Stratigraphy and changes in hardness of snow measured by hand, ramsonde and snow micro penetrometer: a comparison with planar sections. Cold Reg. Sci. Technol. 37, 393–405. doi:10.1016/S0165-232X(03)00079-X.

165

Acknowledgements

Whether time during the doctoral thesis is a curse or a blessing is debatable - my rating depends primarily on the period considered. However, there are many highlights and it’s a fact, that my years of research would rarely have been so positive and desirable without the active support of many people around me. I am very happy to express my thanks in the following section. First of all, I would like to thank my supervisor Martin Schneebeli. He supported me at all times and in many ways. I am grateful for his creative ideas, the tireless interpretation of metamorphism effects and the uncomplicated and pleasant cooperation. And of course, for the many fascinating discussions. Despite this, or perhaps precisely because discussion could run into a completely different topic as prospected, I appreciate his way to communicate. I would like to thank Prof. Dr. Konrad Steffen for supervising my PhD thesis, for his confidence in my work and also for his willingness to support me in side-projects such a as protagonist in snow documentary for Swiss television. Many thanks to my co-examiners Dr. Ulrich Krieger and Dr. Franziska Scholder-Aemisegger for carefully reviewing this thesis. I gratefully acknowledge the financial support by the Swiss National Science Foundation (SNSF) under Grant 155999. Being part of a larger project team (MiSo) and working in close collaboration with two other PhD students was a privilege. I thank Sven Avak for his thorough preparations of the experiments and for always providing clean material support. The countless hours spent together with snow production or elution experiments are, despite all efforts, in best memory and offered many funny moments. In this context I would also like to thank Jacinta Edebeli for her great help with the laboratory experiments, snow sample collection and her support during the course of the project. A further team member is Anja Eichler, whom I would like to say thanks for the planning and active support for the elution experiments as well as for the sampling. I'm grateful for her support in evaluating and writing down the results. I thank Thorsten Bartels-Rausch for bringing in the additional perspectives on state-of-the- art impurity research and especially for his dedicated help and motivation at just the right time during the publication process. Sabina Brütsch I would like to thank for her help during the field campaigns at Weissfluhjoch and the analysis of an incredible number of samples in the PSI laboratory. Further I thank Dimitri Osmont for his help with analyzing samples. I thank Franziska Aemisegger-Scholder for her excellent assistance in planning the isotope measurements. I greatly appreciated the cooperation in evaluating the results and writing my third manuscript. Her enthusiasm and her commitment supported me a lot in finishing this thesis.

167

Without the help of Bettina Richter, the thesis would hardly have been accomplished and there would probably not be a single multi-axis plot. For the valuable suggestions, the physical explanations, the Python programming, the review work as well as the entire support besides work I can’t thank enough. My special thanks go to Pirmin Ebner, Achille Cappelli and Bastian Bergfeld. They were always there for me, as friends, as contact persons for any kind of questions, with their great expertise and goodwill, for Python support as well as for plundering containers. These four years at the SLF without my office colleague and Aargau buddy Matthias Jaggi would not have been the same. As head of the SLF cold laboratory he actively supported me in all experiments in the lab and gave me a lot of freedom there. And if necessary, he covered my back. Further I am grateful to my team leader Henning Löwe for his clever input and the tolerance towards my IT tools. Margret Matzl I like to thank for her assistance in scanning and evaluating CT samples. I enjoyed it a lot being part of the SLF family. Many thanks to Stephan Simioni, Matschek Heck, Lino Schmidt, Stephie Mayer, Gian Darms, Martin Proksch, Joe Veitinger, Janet Prevey, Clare Webster, Julian Fisch, Andi Egloff, Giulia Mazzotti, Johanna Malle, Luki Stoffel, Anselm Köhler, James Glover, Amy Macfarlane, Christian Sommer, Christoph Marty, Cesar Vera Valero, Walter Steinkogler, Martina Sättele, Franziska Gerber, Phillip Crivelli, Franziska Roth, Quirine Kruyt and Bert Kroll. Besides saying thank you to all of you for the many professional and non-professional discussions about snow, I will mainly remember the great time with you! Another special thanks goes to Selina Jäckle, without her influence I would never have started this dissertation. Many thanks to Adrian Winkler and Laurin Merz for their professional work during the filming days in the lab and in the field and for the opportunity to present my work to a broad public - one of my highlights – and thanks to Julia Wessel for her outstanding support during the whole production. Further I thank Jörg Schüppbach for reviewing my texts. And I am very happy to extend my thanks to my family and friends. They have always been there for me, supported me or endured me, reviewed my chapters and in addition to my work, they have created the decisive balance.

168