Aerosol deposition to coastal : a wind tunnel approach

Linnaeus University Dissertations

No 43/2011

AEROSOL DEPOSITION TO COASTAL FORESTS: A WIND TUNNEL APPROACH

AUSRA REINAP

LINNAEUS UNIVERSITY PRESS

AEROSOL DEPOSITION TO COASTAL FORESTS: A WIND TUNNEL APPROACH Doctoral dissertation, School of Natural Sciences, Linnaeus University 2011.

Cover photo: Ausra Reinap ISBN: 978-91-86491-71-0 Printed by: Intellecta Infolog, Gothenburg

Abstract

Reinap, Ausra. (2011). Aerosol deposition to coastal forests: a wind tunnel approach. Linnaeus University Dissertations No 43/2011. ISBN: 978-91-86491-71-0. Written in English.

Aerodynamically rough surfaces of forests provide for efficient air/ exchange of mass, heat and momentum. In that context, the effects of edges come into focus, and therefore, coastal-zone forests constitute a particular concern. Aerosol-sink modelling is of importance to the global-scale context, because sink strengths influence the concentration of aerosol particles in the atmosphere, and that concentration, in turn, influences climate. Dry deposition models are insufficient due to a lack of semi-empirical data and because of difficulties in parameterization of the efficiency (E) with which capture aerosols. Quantifications of such parameters promote possibilities for modelling aerosol-sink processes within various canopy layers. This thesis focuses on studies of sea-salt aerosol dry deposition within models of canopies exposed to artificially generated aerosols in a wind tunnel. The overall goal is to advance the understanding of deposition processes in forest ecosystems. Aims are to determine capture efficiencies and deposition velocities (Vd) for oak (Quercus robur L.), to investigate E and Vd dependence on aerosol particle size, wind velocity and vegetation structural elements such as Area Index (LAI), to explore edge effects on deposition, to relate my results to natural situations in the field, and to address modelling applications. Results demonstrate that forest ecosystems would experience substantially increased deposition at edges. The findings suggest that field measurements of deposition in the interior of a forest “island” in an otherwise open landscape would underestimate the deposition to the entire forest. The obtained capture efficiencies can be translated into deposition velocities for trees with a specific leaf area. An increase of Vd with increasing wind speed is found, and is consistent with other studies. Results confirm advantages of the wind tunnel approach, including its ability to enable experiments under controlled conditions. However, several problems require that explicit sub-models be developed of wind-speed dependent effects on leaf posture in the aerosol flow field and that gradients in relative close to leaf surfaces need further attention. The results also propose needs for a range of further experimental investigations regarding aerosol deposition across the complete sea-to-land aerodynamic transition.

Keywords: NaCl aerosol, dry deposition, climate change, coastal-zone forest, Quercus robur, wind tunnel, edge effect

Table of Contents

ABSTRACT...... IV SAMMANFATTNING ...... V ACKNOWLEDGMENTS...... VII LIST OF PAPERS...... IX ABBREVIATIONS AND SYMBOLS ...... XI 1 INTRODUCTION ...... 1 1.1 A CHANGING CLIMATE FOR AEROSOL DEPOSITION ...... 2 1.2 FOREST-ATMOSPHERE INTERACTIONS ...... 5 1.3 SINK AND SOURCE PROCESSES IN FOREST ECOSYSTEMS ...... 6 1.4 SCOPE AND AIMS OF MY STUDY...... 9 1.5 THESIS OUTLINE...... 11 2 APPROACHES ON AEROSOL DRY DEPOSITION...... 15 2.1 THROUGHFALL STUDIES...... 16 2.2 MICRO-METEOROLOGICAL METHODS...... 16 2.3 WIND TUNNEL EXPERIMENTS ...... 17 2.4 MODELS...... 18 3 WIND TUNNEL APPROACH...... 19 3.1 PRINCIPLES ...... 19 3.2 PARAMETERS ...... 24 3.3 WIND TUNNEL STUDIES VERSUS REALITY IN THE FIELD...... 25 4 RESULTS AND IMPLICATIONS ...... 27 4.1 VEGETATION PROPERTIES VERSUS DEPOSITION VELOCITY ...... 27 4.2 WIND PATTERNS AND TURBULENCE...... 28 4.3 CAPTURE EFFICIENCY AND DEPOSITION VELOCITY...... 37 4.4 APPLICATIONS - EXPERIMENTAL VERSUS MODEL ...... 40 5 CONCLUSIONS AND IMPLICATIONS ...... 45 5.1 MAIN CONCLUSIONS ...... 45 5.2 IMPLICATIONS ...... 46 5.3 FUTURE CONCERNS ...... 48 REFERENCES...... 49

iii Abstract Aerodynamically rough surfaces of forests provide for efficient air/canopy exchange of mass, heat and momentum. In that context, the effects of forest edges come into focus, and therefore, coastal-zone forests constitute a particular concern. Aerosol-sink modelling is of importance to the global-scale context, because sink strengths influence the concentration of aerosol particles in the atmosphere, and that concentration, in turn, influences climate. Dry deposition models are insufficient due to a lack of semi-empirical data and because of difficulties in parameterization of the efficiency (E) with which leaves capture aerosols. Quantifications of such parameters promote possibilities for modelling aerosol-sink processes within various canopy layers. This thesis focuses on studies of sea-salt aerosol dry deposition within models of oak canopies exposed to artificially generated aerosols in a wind tunnel. The overall goal is to advance the understanding of deposition processes in forest ecosystems. Aims are to determine capture efficiencies and deposition velocities (Vd) for oak (Quercus robur L.), to investigate E and Vd dependence on aerosol particle size, wind velocity and vegetation structural elements such as Leaf Area Index (LAI), to explore edge effects on deposition, to relate my results to natural situations in the field, and to address modelling applications. This thesis is a result of five studies. The first study is based on developing a wind tunnel approach with a main focus on establishing reference conditions. The next step is to quantify E and provide estimates of how E, with respect to a well defined mass-vs-particle-size distribution, varies with wind speed. To that end, a special wash-off technique is developed. Finally, edge effects on deposition processes are investigated. Results demonstrate that forest ecosystems would experience substantially increased deposition at edges. The findings suggest that field measurements of deposition in the interior of a forest “island” in an otherwise open landscape would underestimate the deposition to the entire forest. Results clearly indicate needs for further research on the effects of LAI on capture efficiency and deposition velocity. The obtained capture efficiencies can be translated into deposition velocities for trees with a specific leaf area. An increase of Vd with increasing wind speed is found, and is consistent with other studies. Results confirm advantages of the wind tunnel approach, including its ability to enable experiments under controlled conditions. However, several problems require that explicit sub-models be developed of wind-speed dependent effects on leaf posture in the aerosol flow field and that gradients in relative humidity close to leaf surfaces need further attention. The results also propose needs for a range of further experimental investigations regarding aerosol deposition across the complete sea-to-land aerodynamic transition.

iv Sammanfattning Den övergripande ramen för projektet utgörs av de komplexa ömsesidiga samspelen mellan aerosolers sänkprocesser, klimatets påverkan på dessa processer, och i sin tur effekten av aerosoler på klimatet. En aerosol är en ansamling av små partiklar (från några nanometer till några mikrometer) som i fast form eller i droppform svävar in en gas. Depositionsmekanismer (även benämnda sänkprocesser) för aerosoler bestämmer tillsammans med styrkan hos olika aerosolkällor koncentrationen av aerosolpartiklar i atmosfären. Denna koncentration påverkar klimatet via direkta (ljusspridning) och indirekta (molnbildning) processer, och klimatförändringar i sin tur påverkar på flera sätt sänkstyrkor för aerosoler via förändringar i meteorologiska faktorer av betydelse för aerosolers depositionshastigheter. Depositionen påverkar biogeokemin hos ekologiska system och därmed också flera aspekter inom naturresurshushållningen. Kustskogar är av särskilt intresse eftersom de påverkar aerodynamiken (luftens flödesmönster) i övergången från/till hav (”aerodynamiskt jämna” ytor) till/från land (”aerodynamiskt grova” ytor). Sådana övergångar skapar en potential för särskilda effekter i skogsbryn (”kanteffekter”) på aerosolers sänkmekanismer. Aerodynamiskt grova skogsytor ger ett effektivt utbyte av massa, värme och rörelsemängd då atmosfär och lövverk växelverkar. Kustskogar utsätts dessutom för särskild vindcirkulation och för havssalter i det marina gränsskiktet, vilket kan leda till högre deposition av aerosolburna föroreningar såväl som substanser (exempelvis baskatjoner) som kan motverka föroreningseffekter. Kanteffekter är alltså av särskilt intresse vid studier av aerosoldeposition (partikelburet nedfall) till kustnära skogar. Matematiska modeller för aerosoldeposition är otillräckliga på grund av brist på semi-empiriska data och på grund av svårigheter med att fysikaliskt beskriva de partikeluppfångande egenskaperna hos blad och barr. Experimentella bestämningar av denna uppfångningsförmåga kan därför främja möjligheterna att modellera aerosoldepositionen till olika skikt i lövverket, och därmed till skogen som helhet. Uppfångningsförmågan kan beskrivas på olika sätt, till exempel i termer av uppfångningseffektivitet (”aerosol capture efficiency”, ”E”) eller som en depositionshastighet (”aerosol deposition velocity”, ”Vd”). Min avhandling fokuserar på studier av depositionen av havssaltsaerosoler till modeller av ekskog utsatta för artificiellt genererade aerosoler i en vindtunnel. Det övergripande målet är att utveckla förståelsen av aerosoldepositionsprocesser i skogekosystem. Syftet är att fastställa den aerosoluppfångande förmågan hos ekblad (Quercus robur L.), att undersöka hur E och Vd beror av aerosolpartikelstorlek, vindhastighet och vegetationsstruktur

v (bland annat ”Leaf Area Index”, ”LAI”, dvs mängden bladyta över viss markyta), att utforska kanteffekter på depositionen, att relatera mina resultat till naturliga situationer i fält, samt att diskutera möjligheterna att tillämpa de experimentella resultaten i matematisk modellering. Denna avhandling är ett resultat av fem studier. I den första studien utvecklades en vindtunnelbaserad metod och vissa referensförhållanden fastställdes. Nästa steg var att kvantifiera E och ge uppskattningar av hur E, för aerosoler med vissa fördelningar av massan med avseende på ingående partikelstorlekar, varierar med vindhastigheten. I samband med det utvecklades en särskild teknik för att tvätta bort infångade partiklar från bladen. Med hjälp av den tekniken utfördes ytterligare studier av E och Vd. Slutligen undersöktes kanteffekter på depositionen. Resultaten visar att skogsekosystem skulle få väsentligt ökad deposition i skogsbrynen. Resultaten tyder på att mätningar av aerosoldepositionen enbart i det inre av en ”skogsö" i ett annars öppet landskap skulle underskatta nedfallet till hela skogen. Resultaten visar också på behov av ytterligare forskning om effekterna av LAI på E och Vd. En ökning av E och Vd med ökande vindhastighet kunde fastställas i mina experiment, och är förenlig med andra studier. Resultaten bekräftar fördelarna med vindtunnelexperimentella metoder, inbegripet deras förmåga att möjliggöra experiment under kontrollerade förhållanden. Men flera problem kräver att man utvecklar explicita sub-modeller för vindhastighetens effekter på bladens sätt att rikta sig i luftflödet, och för effekten av förändringar i relativ fuktighet nära bladens ytor. Resultaten pekar också på behov av en rad ytterligare experimentella undersökningar av aerosoldeposition över den kompletta aerodynamiska övergången från hav till land.

vi Acknowledgments I will never forget the day I saw a wind- tunnel for the first time: some unrealistic complex system with plenty of tubes and instruments meant to control the atmosphere. I was confused but at the same time excited and curious about how it works. Today, I am happy to have become a friend with aerosols and the tunnel. For that I am most grateful to my supervisor Bo Wiman for introducing me to aerosol science, for guidance, always taking time for inspiring discussions, always supporting and being kind. I am very grateful to my co-supervisor Birgitta Svenningsson, for sharing her excellent knowledge, great ideas, for support, all help and joyful discussions. I am thankful to Bo Bergbäck for support, Anneli and Göran Göransson for help with GIS and Sara Gunnarsson for being the best teacher in my laboratory world. Thanks to Agneta Häggerud for helping me in the laboratory and introducing me to laboratory instruments. Thanks to Anders Gudmundsson for lending me an impactor for my introductory experimental campaigns. Comments by Bo Carlsson and Tomas Öberg are greatly appreciated. Special thanks to Birgitta Svenningsson and Per Kristiansson for giving me an opportunity to have a work place at Lund University. I am grateful for the funding by the Kalmar University Faculty Board of Science and Engineering. As this trip was rather long there are many people who no longer work in the department but whose support are greatly appreciated. I would like to thank my little research group Per and Eva for discussions, support, all fun and nice time together. I have missed that since you left! Special thanks go to Lena Pettersson for always being helpful with any kind of practical matters. Other Barack gang members – Solomia, Renè, Johanna, and Marianne – contributed to making tea/coffee conversations in our cosy library room so warm and enjoyable. Thanks to: Margaretha, Anna, Boel for all help with administrative matters; Sven, Anders and George Gleffe for helping with special technical matters and construction tasks; Nina, David, Anna, Pernilla, Ulf, Christian, Sofie, Fabian for being there and for all fun discussions; Monika – for taking time to read Kappan and comments; other Kalmar colleagues and colleagues at Department of Physics at Lund University for creating a warm working atmosphere! I am grateful to Hans and Jūrate for giving me shelter in Kalmar, hospitality, and all care to my children. Giedre and Per have always kept their home doors open. Lina, Lars, Kalle, Kerstin, Jannet – you made my days in Kalmar much happier. Pille – for caring about me, all fun and joyful moments. Silvija, for critical discussion about science in general and for believing in me. Special

vii thanks to Lund families for babysitting any time I needed. To Hanna, for sauna afternoons and keeping my mood up! My families at another side of the Baltic Sea for all support! Vanaisa – sending me lots of kalli-kalli! I felt much stronger. Tėčiui – rūpestingai klausinėjant apie šio darbo naudą kasdieniniame gyvenime. Mamai, you would have been proud and happy! Avo – your love and support are invaluable. You have been always there - being a great friend and my tough critic! My wonderful girls Elze-Triine, Ulla-Liise and Ellen-Grete – you are everything, most real! It has been a fantastic trip! Ačiū! Lund, 2011/Aušra

viii List of Papers I. Reinap A. (2004) Aerosol deposition at the sea/land interface: pilot experiments using a wind tunnel. The ESS Bulletin 2, 1, 47-73. [ISSN1651-5382]1 I planned, prepared and performed the experiments. I analyzed the data and wrote the paper. II. Reinap A., Wiman B.L.B. and Svenningsson B. (2008) An introductory wind tunnel study of aerosol-borne chloride capture by Quercus robur leaves. (Manuscript) I was responsible for preparing and performing the experiments. I carried out the chemical analysis, data evaluation and analysis. I prepared the figures and wrote the major part of the manuscript. III. Reinap A., Wiman B.L.B., Gunnarsson S. and Svenningsson B. (2010) Dry deposition of NaCl aerosols: theory and method for a modified leaf-washing technique. Atmospheric Measurement Techniques Discussions 3, 4, 3851 – 3876. [DOI 10.5194/amtd-3-3851-2010] 2 I initiated the paper, prepared the experiments and performed the measurements, analyzed the data and wrote the major part of the paper. IV. Reinap A., Wiman B.L.B., Svenningsson B. and Gunnarsson S. (2009) Oak leaves as aerosol collectors: relationships with wind velocity and particle size distribution. Trees 23, 1263-1274 [DOI 10.1007/ s00468- 3 009-0366-4] I contributed to planning and preparing the measurements, and performed the experiments, data evaluation and analysis. I was responsible for planning and writing parts of the paper. V. Reinap A., Wiman B.L.B., Svenningsson B., Gunnarsson S. (2010) Forest-edge effects on sea-salt aerosol deposition: an experimental study. (Submitted to Boreal Environment Research) I contributed to planning and designing the experiments. I was responsible for preparing the instruments, performing the experiments, evaluating and analyzing the data and writing the manuscript.

1 Available at http://www.bom.hik.se/ess/pdf/bullenNr1_2004tillhemsidan040526.pdf 2 Reprinted with kind permission of Copernicus Publications on behalf of the European Geosciences Union 3 Reprinted with kind permission of Springer

ix Related publications not included in the thesis Conference papers (peer-reviewed) VI. Reinap A. (2003) Some characteristics of a wind tunnel for studying aerosol deposition at the sea/land interface. Proceedings of the Nordic Society for Aerosol Research (NOSA) Aerosol Symposium 2003; Copenhagen, November 13-14, 2003; pp. 79-80. VII. Reinap A. (2005) Wind tunnel study of aerosols in an oak canopy micro-cosm: estimates of capture efficiency and deposition velocity. Proceedings of the Nordic Society for Aerosol Research (NOSA) Aerosol Symposium 2005; Göteborg, November 3-4, 2005; pp. 93-94. VIII. Reinap A., Wiman B.L.B. and Svenningsson B. (2008) Aerosol capture by oak leaves – an experimental investigation. Proceedings of the 2008 European Aerosol Conference, Thessaloniki, August 24-29, 2008; Abstract T03A058P. IX. Reinap A., Wiman B.L.B., Svenningsson B. and Gunnarsson S. (2009) Oak leaves as aerosol collectors. Proceedings of the Nordic Society for Aerosol Research (NOSA) Symposium 2009; Lund, November 12-13, 2009; p. 70.

x Abbreviations and symbols A, constant in hyperbolic wind profile formula [dimensionless]

2 Atot , total single-sided leaf area [m ] BVOC, biogenic volatile organic compounds C, concentration [μ-equivalents of tracers per m3] CCN, Cloud Condensation Nuclei CNC, Condensation Nuclei Counter DMPS, differential mobility particle sizer D, deposition rate [μ-equivalents m-2 s-1], also known as “vertical flux”

Dp, aerosol particle diameter [μm]

Dpg, aerosol particle geometric diameter [μm] d, zero plane displacement [m] E, capture efficiency of volume containing collectors [dimensionless] F, flux approaching air-volume containing leaves [μ-equivalents of tracers per m2 and s] I, turbulent intensity [m s-1] (if normalized, then dimensionless) K, turbulent eddy diffusivity [m2s-1] k, von Karman’s constant (≈ 0.41) LAI, leaf area index - m2 of single-sided leaf area per m2 of ground surface area [m2 m-2]

Lmix, mixing length [m]

Ma, the amount of aerosol-borne tracers available for capture [μ-equivalents]

Mleaf, amount actually collected by the leaves [μ-equivalents] MMD, Mass Median Diameter, average particle diameter to the right of which 50% of the aerosol mass resides.

σg, geometric standard deviation [dimensionless ] RH, relative humidity [%] S, function that describes effects of air stability conditions [dimensionless] t, exposure time [s], also used as general symbol for time

xi u, wind velocity [ms-1] ū, mean wind velocity [ms-1] u , the friction velocity [ms-1], represents the turbulent velocity fluctuations in the air

Vd, the deposition velocity related to ground surface area [m s-1]

-1 VL, the deposition velocity related to leaf surface area [m s ] zc, canopy height [m] zo, the roughness length [m] such that u = 0 when z = d + zo u’, w’, horizontal and vertical velocity fluctuations [m s-1]

xii 1 INTRODUCTION

Aerosols – suspensions of tiny particles in the air (Baron & Willeke, 2001) – have a considerable impact on climate, biogeochemistry of ecosystems and our health. The atmospheric aerosol can consist of liquid or solid particles. They arise from a variety of sources: natural, anthropogenic, and miscellaneous sources that might have natural as well as anthropogenic origin. In terms of mass generated, and on a global scale, natural sources of particles considerably exceed anthropogenic emissions. The aerosol particles can be emitted to the atmosphere directly (primary particles) or be formed as a result of gas-to- particle conversion by various chemical processes (secondary particles). The global emission of aerosols from natural sources accounts for up to 88.7% (by weight) while from anthropogenic sources the fraction is up to 17.2%. The residence time of an aerosol particle in the atmosphere ranges from a few seconds in the troposphere, and up to a few years in the stratosphere. Atmospheric aerosols occur in the size range from nano-metres to a few tens of micrometers. Whitby (1978) identified three modes: a nucleation mode (with particle diameter Dp 0.005 - 0.1 μm), an accumulation mode (Dp 0.1 - 2 μm) and a coarse mode (Dp > 2 μm) (cf. also Warneck, 1988; Baron & Willeke, 2001). Two classes are often present in the size range Dp 0.005-1 μm: Aitken and accumulation-type particles (e.g., Hoppel & Frick, 1990; Quinn et al., 1995). Aitken-size particles have diameters in the range 0.03- 0.08 μm. They consist mainly of primary particles and particles formed by for instance coagulation or condensation of secondary aerosol material. For off- shore marine cases with wind speeds above 7-11 ms-1 ultra-large particles (Dp > 300 μm) are injected into the marine boundary layer (O'Dowd et al., 2001). Coarse-mode particles originate mainly from primary particles.

The chemical content of the atmospheric aerosol varies considerably between different sites and points in time. For instance, close to the sea, sea spray components such as chlorine, sodium, and magnesium dominate, while crustal material, for example silicon, aluminium, iron and calcium, and biological

1 material are found over land (cf. Torseth et al., 2000; Schlesinger, 1997). Coarse-particle oriented aerosols are in main focus in my experiments.

Various physical and chemical processes determine atmospheric deposition that takes place all the time over a range of temporal and spatial scales. Dry deposition and wet deposition are responsible for the loads of many various compounds deposited in the biosphere: base cations (Na+, Ca2+, K+, Mg2+) that influence cycling in forest ecosystem, marine aerosols that contain base cations, but also sulphates, nitrate and ammonia that contribute to acidification and eutrophication, and toxic metals such as Zn, Pb, Cd. Dry deposition of atmospheric particles contributes significantly to global biogeochemical cycling because it accounts for a large fraction, around 60%, of the total deposition of many chemical compounds in the atmosphere (Lovett et al., 1984). Therefore, there is a need for a deeper understanding of atmospheric deposition processes in relation to a wide variety of vegetative surfaces (with differing surface morphology) as well as in regard to increasing sources of fine particles in particular. During the recent decades certain progress has been made in modelling of dry deposition processes (Pryor, 2006; Petroff et al., 2007). Models, however, suffer from a lack of semi-empirical data and because of difficulties in the parameterization of the particle-capture efficiency, E. Many models cannot be validated due to lack of empirical data. Thus, more wind-tunnel studies are needed (Fowler et al., 2009) to improve model development and to validate multi-layer deposition models. Often, important parameters (e.g. Leaf Area Index – LAI, leaf size, E) are not provided in scientific studies. Insufficient data on the morphology and structure of rough vegetative collecting elements lead to knowledge gaps while comparing the deposition results between various studies.

In the following, a short background is given concerning: aerosols’ interactions with climate and forest biogeochemistry; deposition mechanisms; and my particular interest in deposition to coastal-forest ecosystems, with a focus on forest edge effects as well as atmosphere-canopy interplay. Finally, this introductory chapter summarizes the aims and scope of this thesis.

1.1 A changing climate for aerosol deposition The climate system is a huge complex of positive and negative feedbacks (cf. e.g. Watson and Lovelock, 1983; May, 1973; Holling, 1973) that still remains poorly understood. Aerosols constitute one of the major uncertainties in climate-change research.

The global climate is a complex system that varies in space and time. Aerosols represent large uncertainties in our ability to predict climate change (IPPC,

2 2007) and deposition of pollutants to sensitive terrestrial and aquatic ecosystems. Why are aerosols and their effects important to study in the context of climate scenarios? There are a few reasons for that: aerosols originate from a variety of sources; they are distributed across a wide spectrum of particle sizes; aerosols have atmospheric lifetimes that are much shorter than those of most greenhouse gases; their concentration and composition have great spatial and temporal variability. Most aerosols are regional in nature owing to their short lifetime, the regional distribution of sources, and the variability in their properties. Seasonal meteorological conditions determine how far aerosols are transported from their sources as well as how they are vertically distributed through the atmosphere. Elevated aerosol layers can be picked up by strong winds and transported from Africa or Asia; examples are Saharan dust plumes (Kaufman et al., 2002).

It is known that humans contribute to climate change in many various ways, but naturally produced aerosols also have a great importance. Fig. 1 shows natural components of the global climate system related to aerosols where focus in this thesis is on the link between forests and marine aerosols.

Volcanoes Atmosphere

Geosphere Aerosols, clouds, radiation

Direct & Biogenic Dust Elemental & Diffuse gases organic C radiation Deposition of light Deposition absorbing aerosol BVOCs Sea salt Ice albedo feedback

Cryosphere Dust Wildfires Forests Marine systems

Edge-effect Veg cover

Sea ice loss

Figure 1. Schematic view of the natural components of the global climate system related with aerosols (slightly revised from Carslaw et al., 2010). The boxes indicate natural components of the global system and arrows indicate aerosol induced interactions among various components. For

3 instance, if deposition to forests increases, that would affect the concentration in the atmosphere that in turn would affect climate because the deposition will change the balance between inputs and outputs to the atmosphere and therefore particle concentration in the atmosphere.

There is a problem in defining what a natural source is. There is a “grey zone” where natural sources are influenced by human activities such as land use. In for instance megacities, anthropogenic sources dominate, while on the global scale there is a distinct dominance of natural sources. For instance, aerosols from wind erosion of deserts might increase because of anthropogenic effects like desertification.

Various biospheric sources, such as marine systems (Fig.1), and transformation processes, modify properties of aerosol particles and trace gases in the atmosphere. An constitutes a source not only of dimethyl-sulphide (DMS) emissions but also of sea spray emissions containing sulphate, base cations and other elements. Forests act as a sink as they have a high efficiency in scavenging and retaining atmospheric inputs due to their ability to generate substantial turbulence, their high surface area and aerodynamic resistance (Wiman et al., 1990; Marcos et al., 2002). The retention of elements in canopies is a vital ecological mechanism, enriching and enabling interspecies competition. Topography and canopy structure are important for the amount of atmospheric deposition. In particular, coastal forest edges are likely to receive increasing amounts of deposits of marine aerosols and pollutants (cf. e.g. Gustafsson and Franzén, 2000; Wiman et al., 2003) distressing nutrient turn-over as well as posing risks. Forests also emit terpenes and related compounds that affect climate in various ways (Seinfeld and Pandis, 1998). Dust emissions vary substantially in space and time and may fertilize with iron that may in turn stimulate DMS emissions (Andreae and Crutzen, 1997).

Anthropogenic sources (even though not included in Fig. 1) contribute to the inflow of various constituents to the global climate system. Environmental- policy based actions, such as emission regulations, forest management and land use change, have an impact on the aerosol sink-source balance.

The concentration of aerosols in the atmosphere is influenced by sink processes in the atmosphere itself and also at the earth’s surface – as deposition to forest ecosystems that affect climate (cf. Wiman et al., 1990). Aerosols are strong agents for changing the biogeochemistry of ecological systems, not in the least forests and related natural resources. Effects of aerosols on biogeochemical fluxes are thus direct (deposition) as well as indirect (via effects on climate). The feedbacks between climate and biogeochemical fluxes are not well known.

4

Direct link The direct aerosol-biogeochemistry link comes from the fact that aerosol sinks (such as forests, lakes, or oceans) involve deposition of nutrients as well as acids, base cations and metals. Aerosol properties are modified during the transport by dry or wet deposition, in-cloud processes, and atmospheric chemical reactions.

Long-term changes in meteorological conditions affect atmospheric transport patterns and dry deposition processes, modify phenology and also influence physiological activity. The characteristics of vegetation cover and land use may be altered on various scales because of human activities, effects of climate on species composition and ecosystem function. The rising temperature and changes in distribution, elevated CO2 and ozone concentrations may act as significant modifiers to vegetation stomatal exchange (Cape, 2008). Indirect link The indirect aerosol-climate-biogeochemistry connections arise because atmospheric aerosols in themselves affect atmospheric optics and radiation balances, and also influence cloud formation, acting as CCN (Baltensperger, 2010). Aerosols’ effects on the radiation balance of the atmosphere in some cases cause warming and in other cases cooling. Particles that absorb solar radiation, such as soot particles from combustion, are responsible for warming, while for instance dust or sulphate (SO42-) aerosol particles tend to cool the earth’s surface by scattering short wave radiation (Kaufman et al., 2002; IPPC, 2007). Climate, in turn, affects biogeochemical flows of ecologically active substances and can cause some of these substances – for example some metals – to become mobilized; thus new characteristics in biogeochemical fluxes might appear.

1.2 Forest-atmosphere interactions Forest vegetation is able to assimilate various elements, among them trace metals, and facilitates cycling through canopy leaching, or through decomposition of leaf and litter (e.g. Likens, 1977). Trees can also retain significant amounts of some elements through foliar absorption that depends on many factors such as the species, nutrient status, and the age of a leaf, the thickness of the cuticle, relative humidity in the leaf boundary layer and the nature of the substance or solute (Marschner, 1986). Forests present efficient capture surfaces for many aerosol borne substances. Intercepted substances

5 might be absorbed by foliage or microbes living on it, or they may be washed off leaves in turn enriching soils.

During the recent decades lots of efforts have been devoted to discussions on forest ecosystems in the context of climate change mainly focusing on the fact that forests can act as carbon sinks. However, forests act as sinks not only for carbon but also for many other elements. Because the vegetation contains dynamic exchange surfaces, it does play a crucial role in biogeochemical cycling as a source by emitting biogenic volatile organic compounds (BVOC) that account for 800 Tg C y-1 of the global VOC emissions. Thus, biogenic sources contribute more than 85 % of global total VOC emissions (Guenther et al., 2006). The largest contributors of total BVOC emissions are isoprene and monoterpenes. Coastal forests at risk The vitality of trees and entire forest ecosystems defines the capacity to cope with stress, resulting from the occurrence of and natural stresses caused by changes in environmental conditions, such as diseases, drought, frost or windthrow (Erisman et al., 1998). Coastal-zone forests in many ways could be seen as vulnerable ecosystem due to prevailing strong winds in the marine boundary layer (Zielinski, 2004), high humidity and many transitional zones that promote edge effects on deposition processes. Forest stands along coasts are subject to deposition of coarse particles, in particular sea-salt particles (Gustafsson and Franzén, 2000). Sea-salt concentration is highest over oceanic regions up to 15 μgm-3 (for dry particle radius of 5 μm) and may extend long distances inland resulting in levels around 3 μgm-3 (Foltescu et al., 2005). Quercus communities constitute a good representative of coastal-zone forests (cf. Skjøth et al., 2008). Forest ecosystems might benefit due to increased deposition of sea-salt nutrients but might also suffer due to deposition of pollutants and toxic substances. In particular, deposition modelling suggests that at forest edges, inputs of constituents can be high. Due to increased deposition of air pollutants such as sulphur, nitrogen compounds, ozone and base cations, forests would experience a number of harmful changes such as damage to epicuticular waxes (Bargel et al., 2006), decrease of photosynthesis and respiration, enhanced transpiration and drought, and increased parasite injury (Erisman et al., 1998; Karnosky et al., 2007; Cape, 2008).

1.3 Sink and source processes in forest ecosystems Dry and wet deposition are the two major removal processes for atmospheric aerosol particles. Very close to the ground, the main mechanisms for particle removal are settling due to gravity, and other dry deposition processes at the

6 earth’s surfaces; whereas at altitudes about 100 m above ground level, precipitation scavenging is the predominant removal mechanism (Seinfeld and Pandis, 1998, pp. 50). Wet deposition thus occurs via and snow and also through fog and, in mountainous areas, cloud-droplet deposition (although fog and cloud deposition may also be categorized as aerosol deposition rather than precipitation-borne deposition). Wet deposition is a process that appears episodically, through nucleation scavenging (e.g. of cloud condensation nuclei, CCN) or “rain-out”, or through impaction scavenging (particles captured by falling raindrops or snow), also termed “wash-out”. Dry deposition, on the other hand, is a continuous process that occurs all the time. It is highly dependent on the nature of the particle-collecting surface, physical and chemical properties of particles, and meteorological parameters. Surface properties (e.g. LAI, morphology of collectors) constitute a major determining factor for dry deposition. Generally, rough surfaces such as forests have higher deposition velocities (Vd) than do less rough surfaces such as a grassland or shore. This is because the friction velocity (a measure of turbulent velocity fluctuations) is higher over rough surfaces. Canopies with a more complicated structure and with small leaves, such as coniferous forest canopies, tend to collect more particles than do deciduous forests (Freer-Smith et al., 2005; Wuyts et al., 2008). An enhancement in deposition can be observed for vegetative canopies where relative humidity increases due to evaporation and transpiration (Draaijers et al., 1993). The dry deposition process depends on the (aerodynamic4) aerosol particle size. Generally, the dry deposition velocity decreases with an increase of particle size for particles smaller than 0.1 μm, and increases with particle size for particles larger than 2 μm. Most of the common aerosol-borne pollutants, such as organic substances, toxic metals, and sulphur compounds, occur in the size range 0.1 – 1.0 μm. According to models by Slinn (1982), particles of around 0.1 to 0.5 μm have a minimum in the deposition-velocity versus particle-size curve (where interception5 is the almost only capture mechanism) and for larger particles deposition increases strongly with particle size as impaction and sedimentation become significant. However, deposition of < 0.1 μm particles increases with decreasing particle size as Brownian diffusion takes place.

Aerosol-particle deposition to vegetation is thus highly variable, depending on interactions at vegetative surfaces, on the meteorological and morphological characteristics of the site (e.g. a coastal forest site) which regulate turbulent

4 Aerodynamic diameter is the diameter of a spherical particle with the density 1 gcm-3 (i.e. water) that exhibits the same mechanics (such as settling velocity in tranquil air) as the particle in question. 5 Interception implies the fact that particles do have a spatial extension. This means that even if the mass point of a particle follows a streamline that passes outside a collector, the particle can still be captured by the collector.

7 exchange between vegetation and the atmosphere (cf. Fowler, 1999). The process that governs the deposition of submicron aerosol particles to vegetative surfaces is turbulent transport through the surface layer followed by transport through the viscous sub-layer surrounding the vegetation elements (Fig. 2). Increased turbulence increases transport and decreases the effective thickness of the viscous sub-layer and increases the drag coefficient. As a result, Vd increases with increasing wind speed (Paper II, IV). In order to cross the viscous layer several mechanisms are involved. Brownian diffusion is most efficient for very small particles with diameters less than 0.1 μm. For larger particles near 1 μm inertial impaction6 and interception become important deposition mechanisms. Furthermore, phoretic processes (such as thermophoresis and diffusiophoresis) or electrical forces might cause deposition (Slinn, 1982; Tammet et al., 2001).

The collection of particles by vegetation can be influenced by particle rebound or re-suspension. Re-suspension is most effective for super-micron particles (Ould-Dada et al., 2002). Nemitz et al. (2000) presented initial measurements of chemically resolved aerosol fluxes at a coastal site and showed that chloride was deposited at low wind speeds, presumably reflecting deposition of sea salt, while it was emitted during windy periods, probably reflecting wind-driven re- suspension of previously deposited material. This, in combination with the existence of coastal wind circulations systems (cf. Skakalova et al., 2003), suggests one of several mechanisms through which an aerosol can circulate many times towards/from land, thus resulting in high deposition rates at forest edges along coasts.

Two layers can be distinguished in the boundary layer for transport of pollutants to the surface: the turbulent layer and the quasi-laminar layer, where the transport to the surface is dominated by molecular and Brownian diffusion for gases and particles, respectively (Fig. 2). The physical, biological and chemical nature of the surface determines the capture or absorbency of the gases and particles. These processes can be described by resistance analogues (Monteith and Unsworth, 1990) that are used in parameterization of deposition processes in field studies (Pryor et al., 2002) and models (Pryor et al., 2000).

6 Impaction implies that, because of inertia, particles will fail to follow the streamline on which they move initially as they approach an obstacle, and instead will be captured by that obstacle. The process can be described by what is known as the Stoke’s parameter.

8

Figure 2. The dry deposition process. Ra is the aerodynamic resistance, Rb is the quasi-laminar layer resistance, Rc is the surface resistance, C(t) is concentration as a function of time, zo is the roughness length (length such that u = 0 when z = d + zo) and d is the displacement height (from Erisman & Draaijers, 2003). The dry deposition process determines the concentration gradient above the vegetation surface. If the surface acts both as a source and sink, for instance for carbon dioxide or aerosol particles, the shape of the gradient will be determined by balance between the source and sink flux. Particles smaller than 0.1 μm are transported through the boundary layer by Brownian motion.

The deposition process is thus subject to controls by rates of turbulent diffusion above and within the vegetative canopy and to chemical reactions with a range of surfaces, including epicuticular wax, leaf surface water, senescent vegetation and (Fowler, 1999). The conventional theory for particle deposition to vegetation is mainly based on aerosol mechanics (cf. Fuchs, 1964), in combination with crop- (cf. Slinn, 1982) and wind tunnel studies (Chamberlain, 1967; Chamberlain, 1975).

1.4 Scope and aims of my study The framing of my project, which is part of a broader research theme on forest management and coastal-zone biogeochemistry (see e.g. Bodin, 2007; Wiman et al., 2007), emanates from my interest in interdisciplinary science (including symbiosis among environmental physics, fluid dynamics, plant ecology, meteorology and climate). In a broader context the questions of interest are:

9 ▪ To which extent are coastal-zone forests subjected to influx of (often long-range-transported) aerosol-borne heavy metals and acidifying substances in relation to (often fairly short-distance-transported) influx of substances (in particular base cations) that tend to counteract negative effects of forest/atmosphere interaction? ▪ How will climatic change in terms of meteorological parameters impose shifts in aerosol-borne inputs to the forest system? This thesis focuses on studying sea-salt aerosol dry deposition within models (or “micro-cosms”) of oak canopies exposed to artificially generated aerosols in a wind tunnel. This work is based on developing a wind tunnel approach to investigating dry deposition to these oak-canopy models. The main goals of the thesis are to advance the understanding of deposition processes in forest ecosystems as well as to investigate the impact of change in meteorological parameters on these processes within canopies. In particular, edge effects appearing due to transitions from smooth surfaces to aerodynamically rough surfaces of forest are of interest. The term “collection”, or “capture”, refers to interactions between the particles and the vegetation. These interactions involve mechanisms such as diffusion, sedimentation, impaction and interception, and lead to deposition to the vegetative surfaces. A particular focus is on the problem of whether aerodynamic edge effects in the sea/land-transition zone will influence the deposition rate of ecologically active substances (nutrients as well potentially harmful trace elements) in coastal areas. The main more specific aims are: ▪ To explore and understand physical principles and processes in the wind tunnel ▪ To develop and improve methods for quantifying deposits of artificially generated sea-salt aerosol particles on vegetation; such methods are crucial in estimating the aerosol-capture efficiency (E) of vegetation components such as leaves ▪ To experimentally determine capture efficiencies and deposition velocities (Vd), and uncertainties involved, for oak (Quercus robur L.)

▪ To investigate E and Vd dependence on aerosol particle size and wind velocity

▪ To investigate E and Vd dependence on vegetation structural elements (leaf area, LAI) and various configurations of the model canopy ▪ To explore edge effects on deposition processes in terms of deposition rates and deposition velocity

10 ▪ To relate my results to natural situations in the field and to address some modelling applications

1.5 Thesis outline The first paper addresses a wind-water tunnel approach as one of the ways to study meteorological and biogeochemical effects of atmospheric aerosols. Paper I focuses on the understanding of atmospheric aerosol physical properties (e.g. particle size distributions) in an experimental situation and on studying basic characteristics of the wind tunnel such as aerodynamics and wind patterns. It provides important characteristics of the wind tunnel as well as measuring equipments, and establishes reference conditions that are necessary for further experiments regarding deposition processes involving oak seedlings and sea-to-forest aerodynamic roughness transitions. The focus is on the following characteristics: aerosol concentration steady state; concentration build-up and decline processes within the wind tunnel; aerosol particle distributions: number-vs-size distribution and volume-vs-size distribution; wind profiles and relative humidity. The paper also briefly reviews the context of aerosol deposition in relation to meteorological-change effects on the deposition process, some biogeochemistry perspectives on aerosols in coastal forest ecosystems, and risk aspects in terms of vulnerability of forest ecosystems to stress factors emanating from changing atmospheric aerosol physical and chemical properties. As it was mentioned before, the mathematical models of deposition processes to forests require information on certain parameters that need investigation under experimental conditions. Among such parameters are capture efficiencies for vegetative surfaces under various conditions with respect to aerosol characteristics and wind velocities. Quantifications of such parameters thus promote possibilities for modelling aerosol-sink processes within various canopy layers. Paper II provides quantitative estimates of capture efficiencies, at various wind velocities, for oak leaves. In particular, my interest was on leaves capturing a poly-disperse aerosol carrying a tracer substance (chloride) with ca 80% of its mass borne by particles smaller than ca 3 μm. I tested various configurations (“models”) of oak-seedling arrangements in the wind tunnel; representing biological variation in terms of leaf area, leaf age and leaf- morphology characteristics. Finding suitable durations for exposing the leaves to the aerosol, under differing aerosol fluxes, in relation to possibilities for analyzing deposits on the leaves, was an important part of the study (Paper II). The focus is on exploring factors that might influence capture efficiencies, such as variation in the biological material used, and effects of relative

11 humidity on particle size. Thus, factors potentially underpinning the spread in capture efficiencies are discussed. Paper II also enabled me to map a range of needs for further methodology developments. Based on those experiences, Paper III focuses on principles and theories that underpin a wash-off technique that I and my colleagues developed for the further wind-tunnel studies of aerosol-particle collection efficiencies of oak leaves (Paper IV, V). The challenge was to establish the extent to which NaCl, an inexpensive and easy-to-handle (and essentially harmless) substance, could be used as a tracer also in further and more advanced laboratory-based deposition studies. I found that the chloride ion and to some extent also the sodium ion were usable, but that some improvements would be appropriate with respect to the leaf-washing technique. Paper III therefore helped establish a method for the next phase of studies of our particular system. The principles of the method should be applicable also on some other combinations of tracers and vegetative particle collectors, in laboratory-based studies. Paper IV extends the experimental investigation of capture efficiencies by giving quantitative estimates of capture efficiencies for homogeneous plant material, in this case oak leaves. Meteorological factors such as wind velocity, relative humidity and temperature were observed. In particular, the dependence of capture efficiency on wind velocity and on the distribution of aerosol-mass versus particle-size distribution is of interest, and for these experiments I had access to an Anderson-type cascade impactor. Paper IV therefore provides estimates of how capture efficiency, with respect to a well defined mass-vs-size distribution, varies with wind speed. The results have implications for estimating the deposition of aerosol-borne substances to forests, in particular in coastal regions, for which wind regimes may shift as a result of climate changes. The obtained capture efficiencies can be translated to deposition velocities (Vd) for trees with a specific leaf area. An increase of Vd with increasing wind speed was found, and is consistent with other studies. However, several problems require much further research. Among examples are: the needs to develop explicit sub-models of wind-speed dependent effects on leaf posture in the aerosol flow field and to study gradients in relative humidity close to leaf surfaces. Paper V investigates edge effects on deposition mechanism in terms of horizontal deposition profiles within an oak-forest model placed in the wind tunnel. The study provides quantitative estimates of deposition rates within an emulated beach-to-forest transition. Two aspects of the deposition rate around the edge are addressed: first in terms of the amounts deposited per unit time and unit ground surface area, and then with respect to amounts deposited per unit time and unit leaf surface area. The results assist in estimating capture efficiencies and deposition velocities at the model-forest edge and for the

12 entire model-forest. The findings confirm the existence of strong edge effects on deposition processes as suggested by mathematical models. The observations of enhanced deposition of aerosol particles at the model-forest wall are also in qualitative agreement with empirical findings by several other researchers.

13

14 2 APPROACHES ON AEROSOL DRY DEPOSITION

Experiments are based on observations and they rely on relevant facts. Preferably, facts need an explanation based on theory. Also, for a hypothesis to mature into a theory, experiments are needed to support (or disprove) a hypothesis. However, not every fact can be explained by theory (Chalmers, 1999). Approaches in interdisciplinary science integrate theories, laboratory studies, field measurements, modelling, and policy analysis and evaluation to help propose solutions to complex environmental challenges. Field research is used for various purposes so as to obtain data about aerosol chemistry and physics processes and their effects on atmospheric dynamics and interaction with ecosystems. In turn, models must be validated in order to be used as operational tools through observations and accurate measurements. Thus, empirical studies based on laboratory experiments can provide data to help test mathematical models of for instance aerosol deposition. Most models also need empirically determined parameter values for model applications, such as models used to estimate aerosol deposition to some specified forest system. Naturally, empirical data for model testing need to be completely unrelated to the data used for quantifying parameters in the model (otherwise “in- breeding” would occur). Models, if well corroborated and if conditions for applying them are satisfied, may in turn have a capability to predict some behaviour of a system (such as a forest), or a capacity to provide estimates of some quantities (such as the deposition rate of some substance). In interdisciplinary contexts, models can therefore be of importance to policy makers. However, “prediction” is a very strong and often misleading term. In environmental science, the concept “scenario” is to be preferred (cf. e.g. the terminology used by the Intergovernmental Panel on Climate Change). Research conducted through field studies, laboratory experiments and modelling studies can promote a better understanding and can improve “predictive” capabilities (i.e., capabilities for scenario-making) that would lead to better policy making. Below a short overview is given of methods used to measure aerosol dry deposition.

15 2.1 Throughfall studies Throughfall measurements are widely used (Lindberg et al., 1988; Rea et al., 2001; Erisman & Draaijers, 2003; Wuyts et al., 2008). The principle of the method is that the input of some substance to a forest canopy by dry deposition is assumed to be possible to determine from the difference between the amount collected in throughfall below the canopy and the input to the canopy at rain events. Among limitations in throughfall method are difficulties to distinguish between in-canopy and atmospheric sources of chemical substances. Another limitation is related to the larger spatial variability of throughfall fluxes within and between forest stands that causes difficulties in providing area representative deposition estimates (Draaijers et al., 1992). Discrepancies in deposition estimates might occur due to litterfall, insects, pollen and airborne substances deposited directly to througfall collecting devices (cf. Ferm, 1993). Moreover, deposition velocity might be overestimated due to the contribution of large particles that settle into the collector. Thus, sedimentation of coarse particles complicates forest throughfall methods (cf. Parker, 1983). Also, the throughfall chemistry of stands could be affected by and algae that are responsible for part of the uptake (such as with respect to N) attributed to the canopy (Lovett, 1992). On the other hand, throughfall methods have a number of advantages. Among them are: they are well suited to monitor deposition in a complex terrain or at forest edges; they give estimates of the total deposition; can be used for all deposited gases and particles, also coarse particles; enable continuous monitoring; and are relatively inexpensive (e.g. Erisman et al., 1995; Pryor et al., 2005).

2.2 Micro-meteorological methods Micro-meteorological methods (Nemitz et al., 2004; Pryor, 2006; Grönholm et al., 2007) are used to measure the vertical turbulent flux above the canopy, where conditions are assumed to be stable; the surface smooth, homogeneous and horizontal; and sources inside the canopy negligible. For instance, eddy correlation methods are based on the simultaneous sampling of the instantaneous (direct) wind velocity and particle concentration, and the direct calculation of the eddy correlation between their fluctuations. Among uncertainties that can lead to under- or over-estimations of the turbulent flux are that measurements of the local flux at the measurement height may be correct but may differ from the actual surface/atmosphere exchange because of flux divergence and the effects of chemical interactions (Fowler et al., 2009). Other discrepancies might appear due to an influence of the relative humidity inducing a variation of the particle size or a modification of the turbulent flux (Fairall, 1984), non-isokinetic sampling, and differences between the aerosol

16 size distribution that relates to the above-canopy sampling point and that relating to the within-canopy point. Such differences can be generated by for instance coagulation among particles and evaporation of BVOCs. Partly similar considerations pertain to gradient methods (Monteith and Unsworth, 1990). On the other hand, micro-meteorological methods have a number of advantages. Among them are: they allow for continuous and high time resolution measurements, provide area-integrated averages of the exchange rates between surface and the atmosphere (Baldocchi et al., 1988), and can indicate new particle formation events (Nilsson et al., 2001). For instance, improved instrumentation (e.g. time-of-flight aerosol mass spectrometry) enables eddy covariance flux measurements of aerosol components such as organics and sulphate above vegetation, providing deposition rates of the accumulation mode. Size-segregated particle flux measurements enable studies of the sub-100 nm size range (Fowler et al., 2009).

2.3 Wind tunnel experiments Wind tunnels of various types are used for dry deposition studies and investigations of aerodynamic properties of vegetative surfaces (e.g. Wiman, 1981; Shaw et al., 1994; Kinnersley et al., 1994; Ould-Dada, 2002). Like the above-mentioned methods, wind tunnel studies have a number of weaknesses as well as a range of advantages. Wind tunnels get some critics, as it is noted that aerodynamic conditions generated in wind tunnels are not representative, or representative only to a degree, of field conditions. This can imply that turbulence around the surface for which deposition is to be estimated has properties that deviate from field situations. Generalisations of wind-tunnel based results therefore require a certain amount of caution. Naturally, analogous caution has to be exercised when using for instance throughfall or micro-meteorological data for estimating deposition to other forests, and for other situations, than those for which measurements were made. However, wind-tunnel based experiments may give valuable insights into specific problems, such as pertaining to collection efficiency parameters for various vegetative structures. In particular, wind-tunnel experiments enable measurements of specific parameters such as capture efficiencies of specific plant components (such as leaves or needles) under controlled experimental conditions that in turn give possibilities to include these parameters in models. This thesis is based on a wind tunnel approach (see Chapter 3).

17 2.4 Models Several mathematical models exist for aerosol deposition to forests. There are two major categories. One deals with process oriented models that describe the dry deposition velocity as a function of particle size based on mathematical parameterizations of removal processes such as Brownian diffusion, impaction, gravitational settling and interception (see Chapter 1) (e.g. Slinn, 1982; Davidson et al., 1982; Wiman and Ågren, 1985; Peters and Eiden, 1992; Ruijgrok et al., 1997; Petroff et al., 2007). Another is based on bulk resistance models where deposition velocity calculations are based on resistance-analogue schemes without distinguishing between particle sizes and removal processes (e.g. Pryor and Sørensen, 2000). As in all modelling, there are strengths and weaknesses. Among pros: models can be seen as a “boil-down” of current knowledge into “mathematical code”; help science identify knowledge gaps and uncertainties and are thus efficient research tools; since we cannot measure everything, everywhere, we have to generalize with the use of models. Among cons: if a model is too simplistic, it does not capture fundamentals of the system and therefore is not trustworthy; if a model is too complex, it cannot be corroborated and will also run the risk of incorrect representation of the system’s “real” structure; models require advanced mathematical skills and tools (such as highly sophisticated computers and computer software) rarely understandable to policy-makers who might therefore hold an over-belief in the model outputs. The modelling approaches show that limitations and simplifications cannot be avoided, and at the same time an operational model requires physically well defined and measurable parameters. The unavailability of input data and the complexity of parameters (cf. Wiman and Ågren, 1985; Ruijgrok et al., 1997) involved in deposition modelling cause difficulties for models being tested against measurements. In the following (Chapter 4, Paper II) I further address deposition models as well as to some extent the modelling of deposition processes. Also, Paper III contains a description of the principles through which capture-efficiency data for various wind speeds can be used to estimate deposition in a forest. However, in-depth discussion of deposition models is beyond the scope of this thesis.

18 3 WIND TUNNEL APPROACH

3.1 Principles The wind tunnel facility is a physical model where sink mechanisms in the aerosol/sea/land interface can be studied under laboratory conditions. The wind tunnel is designed for the experiments to emulate basic transition properties in the coastal-zone sink mechanisms from sea to land under controlled conditions. The sink mechanisms in themselves are primarily dependent on the aerosol particle size. If some substance is used as a tracer to emulate, under laboratory conditions, the size distribution of aerosols that pertain to real field situations at laboratory conditions, then experiments can give information on the relationships between particle size and sink strengths even though the real aerosols themselves cannot be used in the experiments due to leaching, retention7, etc. (see Paper III). Therefore, sink strengths in relation to particle size will implicitly provide basic information about the sink strengths of compounds of particular interest here – NaCl and sea salt aerosols (Paper III, V). The wind-water tunnel used in the experiments (Fig. 3) is a closed-circuit type. The tunnel per se is 6.5 m long (excluding various support constructions connected to the facility); for a brief account see Paper I, II, IV. Its working section is divided into two parts, each 2.1 m long and with a cross section of 0.43 m x 0.30 m (height by width), where, for the experiments reported here, the lower 0.13 m that constitute the water basin were covered to form a solid tunnel floor, or were used to host the leaf-arrangement support boxes (Fig. 4).

7 A substance can leach from interior of the leaf, or can be retained by the surface of the leaf due to various adhesive forces.

19

Figure 3. Schematic overview of the wind tunnel. 1: aerosol generator 1 (based on a modified bubble bursting process), 2: retort containing a 36% sea-salt solution from which aerosols are produced, 3: aerosol-conditioning container (allows very large particles to settle out, and also mixes the virgin aerosol from the generator), 4: wind generator, 5: honeycomb for turbulence damping, 6: position for wind-profile conditioning rods or perforated plate (not used for experiments reported here), 7: water tanks (for experiments reported here, emptied and covered with plates; the down- wind section instead holds a leaf arrangement, “big leaf” or “forest”), 8: relative-humidity and temperature probe, 9: 8-stage (plus back-up filter) Andersen Impactor, iso-kinetic inlet cone and aerosol sampling pump, 10: position of the wind-speed probe connected to an air-velocity transducer (TSI, 8455-300), 11: laser particle counters for qualitative aerosol-data recording (MetOne, HHPC-6 and LPC 227), 12: logger, 13: PC, 14: position of leaf arrangements.

The aerosol was generated by a modified bubble-bursting process in a retort (Fig. 3) outside (Paper II, III, IV) or inside the tunnel (Paper V). Mainly, the retort contained a saturated 36% (by volume) sea-salt-like solution8, whereof ≥ 99.8 % is NaCl. However, in some experiments (Paper II) purified sodium chloride was used. The substance is relatively uncomplicated to work with in laboratory environments, and in particular chloride – which is in focus in this contribution – is suitable for canopy exposure studies (e.g. Hansen and Nielsen, 1998). Sodium, although not inert at leaf surfaces and thus potentially posing problems in exposure-and-wash-off studies (Paper III), is also of interest as a tracer (Paper V).

8 Sodium chloride extracted from a natural Mediterranean sea salt.

20

Figure 4. The wind tunnel during an experimental run. From the left are seen the impactor connected to the wind tunnel, and the conditioning container as a part of the aerosol generation system. The picture to the right illustrates the oak model forest inside the tunnel, RH, temperature and wind velocity probes connected at the side of the tunnel.

It is beyond the scope of this thesis to present and discuss the mathematical details and partial differential equations that underpin process-oriented deposition modelling; the reader is referred to for example Wiman and Ågren (1985) and Pryor et al. (2008). However, in brief: the time derivative of particle concentration is essentially a function of turbulent transport of particle mass, wind velocity, and sink and source sub-functions that are particle size dependent. Clearly, for the many biogeochemically and biogeophysically important applications on laboratory and field situations, understanding and quantification of the sink and source functions are paramount to progress. For vegetation elements with not too complicated geometry (such as pine or spruce needles), and forests with not too complicated architecture (with respect to for instance tree spacing and canopy structure), sub-models for the sink function can make fair progress if based on theories for aerosol mechanics (cf. Fuchs, 1964; Wiman and Ågren, 1985; Wiman, 1985; Wiman, 1986; Wiman, 1988). However, the sink function can be strongly dependent on tree species and other forest characteristics, and for aerosol-collecting elements such as leaves, laboratory (Beckett et al., 2000; Ould-Dada et al., 2002) and field (Lindberg et al., 1986; Lindberg et al., 1988; Fowler et al., 2004) experiments are necessary. The source function is no less complicated and involves, inter alia, the (variable) strength of bubble-bursting processes, generation of aerosol precursors (such as hydrocarbons), pollen, and micro- fragments from vegetation; see for instance Bowen (1982), Duce (1983), Hardy et al. (1985), Wiman and Lannefors (1985), Warneck (1988), Andreae and Crutzen (1997), Seinfeld and Pandis (1998), Sander et al. (2003). Also includable are sources and sinks in the sense that particles residing within one

21 size range might exhibit a flux from it (sink process) into another size range (source process), driven by spatial and temporal changes in ambient relative humidity and by coagulation processes (e.g. Fairall, 1984; Cohen et al., 1987; Wiman, 2000; Svenningsson et al., 2006). The design of my experiments involved needs to keep aerosol concentration (C) reasonably stationary, so that, with t denoting time, ∂C/dt ≈ 0. Investigations such as those presented in Paper I showed that these needs could be satisfied. In regard to experimental situation in focus here, it can be noted that the source function can be separated into two sub-functions, one due to the artificial aerosol generation in the experimental system, and one due to potential biogenic emissions (wherein, for simplicity, it is included also potential re-suspension) from the leaves themselves. Analogously, the sink function can be separated into two parts, one denoting the aerosol depletion that occurs in the wind-tunnel as such (deposition to tunnel walls and other construction structures), and one denoting aerosol collection by leaves. Fig. 5 shows a typical leaf arrangement. Leaves were exposed to the NaCl aerosol flux for four hours. After many preparatory test experiments a suitable duration for exposure was found to optimize needs for striking a balance between several requirements, including sufficient precision in analyzing wash-offs of chloride and sodium from the leaves, maintenance of aerosol- concentration equilibrium in the wind tunnel, and suitable mass loads on the respective impactor (or in some case filter) stages (Fig. 6).

top view right side view

Figure 5. An example of a leaf arrangement inserted into the wind tunnel to form a “canopy model” or a “big-leaf model”.

For cases with poly-disperse aerosols sampled with size-selective instruments – which typically provide the concentration in particle classes Dp,n < Dp < Dp,n+1, the geometric mean diameter of the class is a reasonable proxy for Dp. The continuous function C(Dp) – which is typically log-normally distributed, in one or more modes (Whitby, 1978; Baron and Willeke, 2001) with respect to particle size – is then represented by the sum of class concentrations. In our experimental case, it was not possible to use mono-disperse aerosols. In addition, the distribution of our tracers (aerosol-borne chloride and sodium

22 mass) could be measured in eight size fractions by using an Andersen cascade impactor (Fig. 3; Fig. 4). An example of an exposed steel plate from the impactor is given in Fig. 6:

0.01 m

Figure 6. Sea salt cones on an impactor steel plate after impaction (Experiment date 20081001; effective cut-off diameter 1.1μm).

Based on the results and experiences presented in Paper II, methodological needs were identified, and, among other improvements, a refined leaf-wash off technique was developed (Paper III). The wash-off technique is based on finding out how many pre and post-exposure steps are needed to ensure that all washable deposits (of chloride and sodium) on the leaves, and on the impactor plates, are taken into account. In addition, many experiments were performed to study the extent to which the equivalent ratios between chloride and sodium in the aerosol solution were preserved in the overall measurement process (the “fingerprint” from the aerosol solution to the tunnel aerosol to the leaf wash-offs). Among objectives was thus to establish relevant time durations for pre-exposure and post-exposure wash-off steps. On the average, the 1st (2nd) wash-off step – with duration 5 (10) minutes – provided 96% (99%) of the amount of Cl- deposited on the leaves. For sodium, the general dynamics resembled that of chloride, but the sodium amounts washed off were on the average somewhat below (on equivalent basis) the chloride amounts. This is consistent with the tendency of Na+ to become adsorbed/absorbed at the leaf

23 surface. Chemical analysis pertaining to experiments in Paper II was performed with ion chromatography whereas for all subsequent experiments I had access to a high-precision ISE (Ion Selective Electrodes) technique.

The focus of Paper II and Paper IV is on E, the particle-capture efficiency of the leaves. E includes the potential propensity of leaves of a given tree species (here: Q. r.) to adopt a streamline posture when exposed to wind-flow (this approach is based on the convention established by Chamberlain, 1975), and thus it was chosen to see E as a species-specific function also in this respect. Many other properties of living collectors profoundly complicate the development of theory-based sub-models for the capture efficiency. The need for experimental approaches towards quantifying E is thus strong.

3.2 Parameters For a given chemical compound carried by aerosol particles the deposition rate can be expressed as (e.g. Monteith and Unsworth, 1990):

D (Dp)=Vd(Dp) C(Dp) (1) Here, D denotes the deposition rate [μ-equivalents m-2s-1], also known as -1 “vertical flux”, Vd, the deposition velocity per ground surface area[ms ], and C the aerosol mass concentration [μ-equivalents of Cl- or Na+ per m3]. Invoking the capture efficiency E, and applying certain approximations (Paper IV), the deposition rate can also be expressed as follows (cf. Wiman and Ågren, 1985): D=Eu[ ∫ ρdz]C (2) where u is wind velocity ms-1, ρ (dimension m2 of leaf area per m3 of air occupied by the leaves) is the leaf area density distribution (which in general depends on the vertical height z up through the canopy from the ground surface (z = 0)). The integral of ρ over z results in LAI, the leaf area index (m2 of total single-sided leaf surface area per m2 of ground surface area). The product Eu is termed “deposition velocity per unit leaf area” (related to a specified particle size distribution), VL (cf. Wiman and Ågren, 1985). Hence, VL = Eu. Letting VL operate on the total leaf area per unit ground area through VL times the integral over ρ (which gives LAI) times concentration C gives the amount collected by the leaf arrangement per unit ground area and unit time, i.e. the deposition rate D as defined by Eq. 1. The capture efficiency E [dimensionless] is a characteristic of the leaf surface, under the specified conditions, and is given by the ratio of "actually captured" to "available for capture",

E=Mleaf /Ma (3)

24 where Mleaf denotes the amount actually collected by the leaves under these conditions, and based on the washing-off procedure, Ma is the amount available for capture and, in principle, can be expressed as Ma= AtotCut, as described in more detail in Paper IV. Here, C is the concentration (in this case, μ-equivalents of the tracer per m3), u is the mean wind speed (ms-1) of the airflow approaching the leaves (so that Cu equates the flux F towards the leaves), t is the exposure duration (in our case, 4 x 3600s), and Atot is a total projected single-sided area, irrespective of leaf orientation. Atot is determined from high-resolution photos of the leaves, followed by digital surface-area quantification.

3.3 Wind tunnel studies versus reality in the field Wind tunnels are physical models where deposition processes can be studied under controlled conditions. Models, be they physical constructions or systems of formulations based on mathematics and physics, are only very limited representations of the real world. As discussed above (Section 2.4) this circumstance involves advantages as well as disadvantages. For instance, pressure gradients are different in field situations in comparison to a wind tunnel atmosphere where air is forced through a limited space. Many additional real-world phenomena often have to be excluded or approximated, or in some cases lead to studies of atmospheric processes instead simulated in water tunnels. On the other hand, experiments in for instance a real forest face problems with a vast range of uncontrollable ambient factors that complicate interpretations of the results, and, in addition, require logistics and resources rarely available. A wind-tunnel approach of the kind used here provides ability and flexibility to control source and sink (deposition) processes, and to simulate and control meteorological parameters to some extent, while complexity in nature is difficult to deal with in reality. As observed above (Chapter 2), interaction is needed between field studies, laboratory experiments, mathematical modelling, and careful policy-making. Some of the reviews on these matters (e.g. Cape, 2008) note the need for studies on the large-scale level where the complexity of a forest community represents real conditions. For example, the life-long response of forest trees is difficult to predict on the basis of short-term studies. Studies are also needed under realistic forest conditions where trees are subject to exposure in competitive situations, under naturally occurring stresses that act over the lifetime of the stand (Karnosky, 2003). On the other hand, small-scale studies provide valuable information that is necessary for understanding processes at the large-scale level. Therefore, small-scale studies such as laboratory-based measurements – including wind-tunnel studies – are of importance. Again, the advantage of laboratory experiments is that one can investigate phenomena

25 and processes under controlled conditions while field conditions present a range of complexities. It is generally known that canopy turbulence is dominated by intermittent and energetic coherent structures that are responsible for most of the momentum and scalar fluxes, and whose length scales are of the order of the canopy height in forests (Raupach et al., 1986; Brunet et al., 1994). The turbulence structure above and within the canopies has been reviewed by, among others, Raupach et al. (1981, 1991), Gao et al. (1989) and Finnigan (2000). However, not only are turbulence fields difficult to simulate accurately in wind tunnels, but turbulence is also difficult to measure in reality. Often, turbulent flow within and above the canopy (Finningan et al., 2009) implies many uncertainties. Modelling attempts are often based on statistical descriptions of turbulence (Massman et al., 1999; Kruijt et al., 2000) that in turn also imply simplifications, uncertainties and imperfections. My measurements of wind-velocity profiles, and the turbulence-intensity functions calculated from these profiles, within the canopy in the wind tunnel (see Chapter 4) gave results similar to those found for real forest canopies and other agricultural canopies. That implies that a characteristic boundary layer developed in the wind tunnel and that flow patterns adequately emulate those in a real forest. However, in order to achieve a complete simulation of atmospheric dynamics one needs a tunnel with much larger dimensions, and preferably also of open-ended type. Moreover, wind profiles within and above the model forest canopy that were observed in this project are similar to those presented by Kinnersley et al. (1994), who used a wind tunnel with dimensions 6 m long, 0.8 m wide and 1.2 m high. Wind direction and velocity affect the deposition patterns of aerosols to forest canopies. Under natural conditions, wind direction and average speed might change frequently, while in a wind tunnel air flow is in one direction. In my experiments, average wind speeds were kept at certain constant levels so as to enable studies of particle-capture efficiencies as a function of wind speed. From one perspective, one could argue that one-directional wind flow and constant wind speeds indeed deviate fundamentally from natural conditions. But, on the other hand, the capability to control wind direction and velocity is a major advantage of wind tunnels, in that a selection of important factors can be studied and their effects separated from a multitude of other factors. In addition, several situations exist in the field wherein wind directions and velocities stay fairly stable over long periods; coastal wind systems constitute a notable example.

26 4 RESULTS AND IMPLICATIONS

4.1 Vegetation properties versus deposition velocity It is a challenging task to “enclose” an entire forest stand and measure its entire uptake of for instance CO2, ground-near ozone, and aerosols, although some such studies exist (e.g. Lamaud et al., 2002). Therefore, an often necessary approach is to experimentally determine uptake capacities in the laboratory or through measuring in the field with special equipment for just a few leaves enclosed in “chambers” of various types and then find a way to scale up. In this thesis focus is on a “big-leaf approach”, using leaf ensembles consisting of several individual leaves with a known total leaf surface area. Exposure of one single leaf per run is possible, in principle; in practice, however, the exposure durations needed to obtain quantifiable amounts of aerosol mass on the leaf would be too large. The interaction between vegetation surface and the atmosphere, such as interception of solar irradiation, precipitation interception, energy conversion, mass, momentum and gas exchange, is substantially determined by the vegetation surface. Every individual surface that contributes to the aerosol collection is characterized by specific geometrical properties, such as its shape, its size and its orientation in space. Artificial or surrogate collectors (e.g. Lindberg et al., 1988; Dambrine et al., 1998; Ferm et al., 1999) may not be representative of the uptake by vegetative surfaces because they lack morphological variability. Aerosol-capture efficiencies determined for leaves with known leaf area characteristics give possibilities for scaling-up for various canopy layers within the tree and forest. Thus, leaf characteristic parameters (LAI, etc.) and their relation with the entire forest stand structure as well as deposition velocities are of concern. During the vegetation period of deciduous forests, total vegetation surface itself is mainly composed of leaf area, and by a lesser part of twigs, branches and stem surface. LAI is a key structural characteristic of vegetation and land cover because of the role of green leaves in a wide range of biological and physical processes. For example, information on typical values of LAI is

27 required for scaling-up from the leaf level to the canopy level, with respect to measurements of water vapour, CO2 conductance and flux (Scurlock et al., 2001). The typical single-sided LAI value for oak forests is around 2 to 5 (cf. Gaydarova, 2003). Results from surface wash experiments (Lindberg et al., 1988, Draaijers et al., 1993; Finkelstein, 2001) and deposition modelling (Lovett and Reiners, 1986) suggest a linear increase between deposition velocity of various constituents and LAI. However, some studies based on wind tunnel experiments (e.g. Ould-Dada, 2002) and detailed process modelling (Wiman and Ågren, 1985) suggest that deposition velocity can be higher for vegetation with smaller LAI. Studies by Shaw et al. (1994) and Ould-Dada (2002) note higher deposition velocities for lower LAI-values (one sided) (for example, for LAI = 0.07, Vd was found to be 0.19 cms-1 while for LAI= 0.62 a Vd of 0.05 cms-1 was found) which lead to the conclusion that the expected increase in deposition velocity with increasing LAI could not be established. Many exploratory experiments were performed for Paper II, including experiments regarding LAI influence on E and Vd (Reinap, unpublished). The wind profiles obtained in these experiments show that where LAI was highest wind speed was lowest. Even if there is much leaf area potentially capable of capturing particles, the wind speed is low, so that in particular coarse particles would be subjected to low capture efficiencies. Contrariwise, at low LAI each leaf might experience a higher wind speed; thus higher deposition velocities could be expected. Further research on the effects of vegetation canopy structure on capture efficiency and deposition velocity is clearly desirable.

4.2 Wind patterns and turbulence Particle deposition mechanisms are highly dependent on wind velocity; therefore the effects of wind speed on deposition velocity are of major importance in interpreting capture-efficiency data. In addition, the turbulence conditions are of concern in several ways, inasmuch turbulence processes transport aerosol particles into and out of canopies, influence leaf fluttering, and affect the extension of leaf-boundary layers (through which the final step of deposition mechanisms such as impaction and Brownian diffusion takes place). In order to gain insight into wind and turbulence conditions in the wind tunnel used in my experiments separate studies had to be carried out, because measurements of for instance wind profiles inside the “forest models” could not be carried out concomitantly with runs during which the “forests” were exposed to an aerosol flux (inside-“forest” wind measurements would have disturbed and potentially contaminated such runs). Hence, preparatory studies with “forest-model” arrangements similar to those described in Paper V were performed. Whereas vertical wind-profiles (from the tunnel floor upwards towards the central axis of the tunnel) could be measured,

28 instrumentation for direct turbulence measurements (needing, for example, sonic anemometers able to measure wind fluctuations in the x, y, and z directions) were not available. However, theory is available for estimating turbulent eddy diffusivity (or, alternatively, turbulent intensity) from wind profiles; I will return to this later. Fig. 7a shows representative cases of vertical wind profiles without and with a “forest model” present. Here, wind velocity was 5 ms-1 at the tunnel central axis and at the position for the iso-kinetic aerosol-sampling probe inlet. The height of the “forest model” was about 0.12 m, and it’s LAI (single-sided) about 2.8.

0.16

0.14 r 0.12 floo

0.10 tunnel

0.08 above

0.06 height

z, 0.04

0.02

0.00 0123456 u, wind velocity [m s‐1]

wi thout canopy wi th canopy

Figure 7a. Wind profiles measured with and without ”forest model” in the tunnel.

Fig. 7a shows that in the reference case (no "mini-forest") the wind speed near the surface is close to 0, increasing with height z (given in m) above the ground surface (here constituted by the tunnel floor). At z ≈ 0.15 (half the height of the tunnel working section) wind speed stabilizes, and then begins to decline upwards towards the tunnel ceiling. The profile is approximately logarithmic up to z ≈ 0.15 m. As expected, the profile adopts a more complicated shape inside the "mini-forest". This profile shape, with a velocity increase (a “bulge“) below the lower canopy limit and the ground (the “trunk space”), is not uncommon in field situations (cf. e.g. Allen, 1968; Gardiner, 1994; Finnigan and Brunet, 1995). In the “trunk region” there is thus a counter-gradient momentum flux presenting a resistance to downward particle movement. This means that deposition rather occurs in the canopy than to the tunnel floor. It merits mention that although a few studies show that the

29 aerodynamics of full-scale forest canopies can be reasonably well simulated in wind tunnels, several uncertainties and difficulties still remain (Kinnersley et al., 1994), including gusts that can dominate momentum transfer from the air to the canopy (Raupach, 1989). Also, wind profiles depend on details of leaf and canopy structure, and forest architecture (cf. e.g. Wiman and Ågren, 1985). For further analysis of the relationship between the wind profiles and turbulence it may first be noted that turbulent intensity I can be defined as (cf. e.g. Monteith and Unsworth, 1990; Kinnersley et al., 1994) I = (u´w´)0.5 /ū (4) where (u´ w´) is the average (over a length of time) of the products of the horizontal (u´) and vertical (w´) velocity fluctuations and ū is the mean horizontal wind speed at a reference level (such as at the canopy top, or at some level above the canopy). In a region where the logarithmic wind profile holds, and if the turbulent intensity is the same in horizontal and vertical directions (isotropic turbulence), u´ = w´ and the friction velocity u* = (u´2)0.5 is then a measure of turbulence intensity through I = u*/ū. The friction velocity can be estimated from wind-profile measurements above the canopy as well as from simultaneous measurements of u´ and w´. Inside canopies, however, direct application of Eq. 4 necessitates special measuring equipment which, as observed above, was not available for my studies. Instead, turbulence conditions for these experiments can be addressed as detailed in Paper V; below is a brief overview. A common approach is to interpret the relation between mean wind profile and turbulence as a set of eddies with extensions given by a so-called mixing length, Lmix (cf. e.g. Monteith and Unsworth, 1990). Under neutral conditions with respect to temperature as a function of height, the eddies can be seen as circular. Under unstable conditions, eddies tend towards ellipses with increasing vertical extensions and decreasing horizontal extensions; under stable conditions towards ellipses with increasing horizontal extensions and decreasing vertical extensions. One can then define the turbulent eddy diffusivity K(z) as a function of Lmix(z), the derivative of the wind-velocity profile (du(z)/dz), and a function S(z) that describes effects of air stability conditions; S(z) = 1 under neutral conditions, which were at hand in my experiments. Above the canopy, Lmix can be related to the zero-plane displacement (d), which in turn can be related to canopy height (zc); inside the forest, Lmix can be related to LAI and zc. The turbulent eddy-diffusivity profile corresponding to Fig. 7a can therefore be estimated from my data, resulting in Fig. 7b. Among other things, Fig. 7b highlights the counter- gradient momentum flux in the “trunk region”.

30 K, turbulent diffusivity [m2 s‐1] ‐0.01 0 0.01 0.02 0.03 0.16 0.16

0.14 0.14

[m]

0.12 0.12 floor

0.10 0.1

0.08 0.08 tunnel

0.06 0.06 above 0.04 0.04 height

0.02 0.02 z, 0.00 0 0 0.1 0.2 0.3 0.4 I, turbulent intensity (normalised to tunnel axis level)

Figure 7b. Turbulence inside (< 0.12 m) and above the ”mini-forest” in terms of eddy diffusivity (bold line) and turbulent intensity (dashed line), calculated from data in Fig. 7a (see main text and Paper V for details).

Also, it merits mention that the mixing length is related to velocity fluctuations u´ and w´, and in the isotropic case one can write u´ = w´ = Lmix du(z)/dz. Eq. 4a can then be used, and with ū representing the mean wind speed at the tunnel axis, results in Fig. 7b (turbulent intensity). The occurrence of a maximum in turbulence intensity (I) within the "mini-forest" suggests, again, that significant parts of the deposition are located to the canopy level. The intensity profile is in reasonable agreement with profiles obtained for “forest models” in much larger wind tunnels (see e.g. Meroney, 1968; Kinnersley et al., 1994). It is now of obvious interest to compare my measured wind profiles with common mathematical profile models and with literature data. Over many types of uniform surfaces, mean wind speed increases logarithmically with height above the canopy, giving the standard wind-profile equation:

* u = (u /k)ln [(z-d)/z0] (5) with u*, k and d defined in the previous paragraph; z0 is the aerodynamic roughness length. Eq. 5 is valid in the air above (but not too far from) the surface elements where the assumption that shearing stress is independent of height is satisfied. Because of the absorption of momentum below the tops of canopy elements, the equation is not valid in this region, nor is it valid immediately above very

31 tall vegetation where the relation between flux and gradient breaks down (Monteith & Unsworth, 1990). Wind profiles within canopies can be approximately described by various existing models, whereof a hyperbolic wind profile model (Cowan, 1968) has the advantage of giving wind speed zero at ground-surface level (z = 0), as would be physically expected. This is a feature not inherent in for instance exponential profile models. The hyperbolic profile model can be written as

0.5 u = uc[sinh(Az/zc)/sinh(A)] (6) where uc = u(zc), i.e. the wind speed at the top (z = zc) of the forest. A is a factor that can be related to forest structure in terms of LAI and zc (cf. Wiman, 1985). Thus, connecting both profiles – i.e., sub-canopy hyperbolic (Eq. 6) and above-canopy logarithmic (Eq. 5) profiles – offers a means to simulate the effect of LAI and canopy height (zc) on the overall wind profile. Examples are given in Fig. 8a (with zc kept constant and LAI is varied) and Fig. 8b (where LAI is kept constant and zc is varied). The diagrams below thus illustrate principles of the wind profiles’ dependence on LAI and canopy height zc. It is obvious, however, that the model does not well represent the experimentally observed influence of LAI on wind velocity (cf. Fig. 7a). This is because the hyperbolic model for mathematical reasons cannot represent the “bulge”. Nevertheless, the profile-modelling approach exemplified below is common and in many cases a useful approximation in deposition modelling, inasmuch it does at least indicate some properties of the wind flow close to the ground.

32 0.20 0.18

0.16 [m]

z 0.14 LAI 0.5 0.12 1.5 ground, 0.10 2.5 0.08 3.5 above 4.5 0.06 5.5 height 0.04 6.5 0.02

0.00 024681012

wind speed, u [ms‐1]

Figure 8a. Illustration of the effect forest structural parameters (LAI and canopy height) on the shape of a wind-profile generated from combining an inside-canopy hyperbolic model with an above-forest logarithmic model. Here, canophy height is kept constant (at zc = 0.12 m), and LAI (single sided) is varied (between 0.5 and 6.5 m2 m-2; as shown by the curve-colours and the LAI- values they pertain to in the right-hand legends). Wind speed at canopy top is set to 5 ms-1.

0.20

0.18

0.16 [m]

z 0.14 zc 0.04 0.12 0.06 ground,

0.10 0.08 0.10

above 0.08

0.12 0.06 0.14 height 0.04 0.16

0.02

0.00 0481216

wind speed, u [ms‐1]

Figure 8b. Illustration of the effect forest structural parameters (LAI and canopy height) on the shape of a wind-profile generated from combining an inside-canopy hyperbolic model with an above-forest logarithmic model. Here, canopy height is varied (between zc = 0.04 and 0.16 m; as shown by the curve-colours and the zc -values they pertain to in the right-hand legends), and LAI (single sided) is kept constant (at 1.2 m2m-2). Wind speed at canopy top is set to 5 ms-1.

33

My profile data, however, suggest that caution has to be exercised when applying generalized profiles in such contexts, not in the least because the measured wind velocity profile for the oak-canopy model inserted into the tunnel is in fair agreement with those observed in other studies (cf. e.g. Raupach and Thom, 1981; Grace, 1983, Kinnersley et al., 1994; Ould-Dada, 2002). It is interesting to note that wind-speed profiles through a canopy of Holm oak (Kinnersley et al., 1994) and in this case oak (Q. robur) are similar to wind profiles often found for spruce canopies (e.g. Ould-Dada, 2002). The decrease in wind velocity downwards through the canopy is related to aerodynamic drag imposed by leaves, branches and stems and hence to the vertical distribution of biomass. Obviously, the hyperbolic model is not a very good representative of profile data measured in the field and in the laboratory. As observed above, this is because the hyperbolic function for mathematical reasons simply can only indicate the small but existing “tunnel” in the trunk space below the canopy, where air can flow more freely. On the other hand, the hyperbolic profile better describes in-canopy wind flow than does for instance an exponential profile. In conclusion, the combined profile measured inside and above the canopy presented here compares well with results from other wind-tunnel and field studies. Thus, my wind tunnel measurements, addressing particle-capture efficiencies, represent reality in the field to a fair extent. For certain types of calculations (see Paper IV and V) it was of interest to investigate whether the aerosol flux approaching the main part of the canopy, i.e. from the upwind smooth-surface direction, could be reasonably well approximated as constant. Fig. 9 below shows the 2-dimensional wind- velocity profile just upwind (ca 0.35 m) of the exposed leaf arrangement that was determined through measurements in the wind tunnel space along vertical as well as horizontal coordinates for the three wind velocities 2, 5 and 10 ms-1. In order to provide additional detail, Fig. 10 shows the data in a different manner. Measurements were confined to the cross-section area extending a distance Δz from the tunnel floor and a distance Δy from the walls and upwards to the level 14 cm above the tunnel floor; the remaining area is not accessible to measurements. For the wind-speed measurements, the sensor dimensions allowed for Δz ≈ 0.5 cm and Δy ≈ 1 cm; for the concentration measurements probe dimensions enabled Δz ≈ 1.5 cm and Δy ≈ 1.5 cm. Measurements with and without leaf arrangements showed that downwind presence of oak seedlings did not influence upwind flow. The diagrams show that aerosol particles were carried towards the main part of the canopy by essentially the same wind velocity along vertical as well as horizontal coordinates.

34

u=10 ms-1 u=5 ms-1 u=2 ms-1 10

5

0

-5 height, z [cm] height, -10

0 -10 -5 0 5 5 10 10 wind speed, u [ms-1] width, y [cm] Figure 9. Wind speed as a function of y (width) and z (height) coordinates in the tunnel upwind the model oak forest for wind velocities 2, 5 and 10 ms-1.

u = 2 ms-1 u = 5 ms-1 u = 10 ms-1

10

5

0 1.510.5 2 0 0 4 321 50 08 2641537 9 08

8 5 7

5 4 9 2

6

1

5 4 0 4

.

5 1.5

5 3

3

1 .

1 2

3 1 7

1 9 0.5

0

2 5 6 2 1 3

4

2 5

1 height, z [cm] z height, -5 2

8 8 -10 2 7 6

4 4 1

2 5

. 4

5 4 3 1. 1 5 5 3 3 6 7 015 01. 21 21 534 12 -10 -5 0 5 10 -10 -5 0 5 10 -10 -5 0 5 10 width, y [cm] width, y [cm] width, y [cm] Figure 10. Wind speed as a function of y (width) and z (height) coordinates in the tunnel for wind velocities 2, 5 and 10 ms-1

Fig. 11 below shows the normalised particle-concentration fields for wind velocities 2, 5 and 10 ms-1 for various aerosol particle size ranges. The concentration profiles were measured across the cross-sectional area of the tunnel using a probe (calibrated with respect to the isokinetic sampling probe) connected to a laser particle counter (HHPC-6) sampling six size classes (see Fig. 3).

35 u = 2 ms-1 Dp=0.3-0.5 m u = 2 ms-1 Dp=1.0-2.0 m u = 2 ms-1 Dp=2.0-5.0 m

10

5

0 0 0 . . .95 0.95 9 9 0 5 0 .9 . 0.9 .9 0 9 5 5 0 5 9 5 .85 . 8 5 0 0 0 0.9 0. . 0. .9 0.8 9 0 0 85 .95 0 0 .8 .95 . 9 0. 0 8 . height, z [cm] z height, 95 9 -5 0. 5 0 5 .8 .8 8 0 0 0 .85 . 0 0 0. .8 0.95 85 5 8 0. 0 0.8 0. 8 -10 . 9 9 .

0

.8 0 -10 -5 0 5 10 -10 -5 0 5 10 -10 -5 0 5 10 width, y [cm] width, y [cm] width, y [cm]

u = 5 ms-1 Dp=0.3-0.5 m u = 5 ms-1 Dp=1.0-2.0 m u = 5 ms-1 Dp=2.0-5.0 m

10

5

0 .

9 9 0.95 0.9 9 . 5 . 0 0 0 5 . 5 0 9 9 5 . 0 . 8 5 8 . 5 5 0 0

0 9 .

9 9

. 8 . 0 5 0

0 0.9 height, z [cm] z height, -5 0.9 85 0.8 0. 8 0. 5 .8 0 .8 0 0

0 0 . 0 -10 8 . . . 9 9 5 8 5 0 5 . 9 5 0 0.8 .8 0.9 0.8 -10 -5 0 5 10 -10 -5 0 5 10 -10 -5 0 5 10 width, y [cm] width, y [cm] width, y [cm]

u = 10 ms-1 Dp=0.3-0.5 m u = 10 ms-1 Dp=1.0-2.0 m u = 10 ms-1 Dp=2.0-5.0 m

10

5

0 0

5 0. . . 0 8 8 8 . . 85 85 0 0

. 0 0 8 0

5 . 5 . 8 8 . 9 0

0 5 0.9 0.9 0.85 .8 0

9 0 . 0.8 . 0 9 0 0 . 9 . height, z [cm] z height, 0 . 8 8 -5 . 5 .9 9 8 . 0 5 5 0

5

0 9 9 . 0 . .

8 0 . 0 9 .95 9 8 0 0. . 5 8 0 .

5 9 0 0 . 5 . 0 -10 9

5 8 . 0 .85 0 . 0 5 9 8 9 . 9 . 5 9 . 0 . 0 0 0 95 0 85 0 -10 -5 0 5 10 -10 -5 0 5 10 -10 -5 0 5 10 width, y [cm] width, y [cm] width, y [cm] Figure 11. The normalised concentration field for wind velocities 2, 5 and 10 ms-1 for various aerosol particle size ranges.

Thus, data (Fig. 10; Fig. 11) showed that the aerosol flux approaching the leaf-arrangements was essentially constant over the major part of the wind tunnel’s cross section.

36 4.3 Capture efficiency and deposition velocity One of the main aims within this project is to provide estimates of capture efficiency that in turn can serve as an important data input into models that provide deposition velocities and deposition rates. In this section results of obtained values are compared between different studies (Paper IV and Paper V) and with field data obtained by other researchers. As outlined in Section 1.5, a series of experiments was devoted to investigations of longitudinal deposition patterns along a transect from an aerodynamically smooth “beach” and further downwind across a model-forest edge and further into the forest. Paper V reveals that average estimates of capture efficiency at the edge were 0.029 % for Cl- and 0.024 % for Na+. Deposition velocities, Vd (related to ground surface area) for the model forest, with an average total LAI of 1.2 ± 0.3, were 0.05 ± 0.02 cms-1 and 0.04 ± 0.02 -1 - + cms for Cl and Na respectively. The mean geometric particle size (Dpg) of the size distribution was on the average 1.55 μm; the geometric standard deviation (σg) was approximately 1.85. In Paper IV the obtained average oak-leaf capture efficiencies ranged from 0.005 % of the approaching aerosol mass flux at wind speed 2 ms-1 to 0.01 % of the flux at wind speeds 10 ms-1, respectively. In these experiments, the aerosol mass versus aerodynamic particle size distribution was characterized by a geometric mean aerodynamic particle diameter around 1.2 μm and a geometric standard deviation around 1.8. Corresponding deposition velocities (Vd) for a leaf arrangement with a given leaf area index of 2 at wind speeds 2 ms-1 ranged from 0.02 to 0.05 cms-1. The difference between the mean values of deposition velocities among the studies is not statistically significant at 95% confidence level (p<0.05). Among the explanations are relatively identical experimental conditions: fine-mode dominated sea-salt aerosol particles (emulated by the artificially generated NaCl aerosol), homogeneous plant material, and the same wind velocity (the experiments detailed in Paper IV used 2, 5 and 10 ms-1; for comparison with Paper V, the data referring to 2 m s-1 in Paper IV are selected here). This again confirms advantages of the wind tunnel approach, including its ability to enable experiments under controlled conditions. Interestingly, however, the experiments underpinning Paper II resulted in substantially higher average values for the capture efficiencies than those that pertain to the experiments behind Paper IV and Paper V. The spread of data around the mean values presented in Paper II is large, however, which makes some caution necessary when comparing them with Paper IV and Paper V data. Nevertheless, a few observations on the differences with respect to the experimental conditions merit discussion. They include: a different “seedling bank” was used for the experiments in Paper II where temperature and light

37 varied considerably and leaves were of differing ages; the aerosol conditioner in the Paper II experimental set up was smaller thus resulting in higher magnitude of coarse particles; the aerosol generation system set-up was different using a smaller (1 litre) container for ultra-pure NaCl aerosol- solution, whereas in Papers IV and V a 2 litre container and a less pure NaCl salt were used; 2-stage Nuclepore filter systems were used for the Paper II experiments, and thus just coarse-versus-total mass ratios could be noted; the analytical technique for the Paper II experiments was different and more difficult to control. It is known that even fairly small changes to experimental and analytical designs can have rather large effects. However, these introductory studies gave good insights regarding the order of magnitude pertaining to capture efficiency, and also enabled identification of several needs for acquiring better control of the experimental set-ups. Quantitative comparisons between studies by different researchers often face some difficulties. Among them are that information on certain specifics pertaining to experiments and field studies is often lacking or is incomplete, such as with respect to LAI, wind speed, and particle-size distribution. However, my results can be compared to some extent with laboratory and field data obtained by other researchers. There are a few results on deposition velocities for spruce and oak canopies available based on wind-tunnel studies (see Table 1). Freer-Smith (2004), while studying effects of species and wind- speed on deposition velocities in a wind-tunnel set-up, found Vd-values for oak leaves (Q. petraea) at around 0.8 cms-1 when mean wind speed was 3 ms-1; however, no data were given to relate this value to leaf surface area or LAI (see Table 1). Ould-Dada (2002) found 0.5 cms-1 for the deposition velocity pertaining to the total canopy of spruce, whereas for the lowest layers of canopy where LAI-values were highest (but wind-speed lowest) the deposition velocity was much lower (0.05 cms-1) for aerosols with mass median [aerodynamic] diameter (MM[A]D) of 0.82 μm. The overall range of relative Vd values, from about 0.02 cms-1 up to about 1.7 cms-1, reported here (Paper II, IV, V) is essentially consistent with the range comprising information from other studies.

38

Table 1. Deposition velocities and capture efficiencies for oak and pine canopies presented by other studies, in comparison with my data. In some of my studies I used natural sea salt that contains ≥ 99.8 % of NaCl.

Characteristic Canopy u MMD9 Mean E Mean Vd Method LAI Reference type [ms-1] [μm], type [%] [cms-1] 2- Lindberg et al. ≤ 0.5 (SO4 ) 0.06 Oak Through-fall - - 2- - (1982) 3.4 (SO4 ) 6.4 0.4-2.4 (Cd, Höfken et al. SO42- , NH4+; Spruce Through-fall - - - 0.6-2.5 (1983) Pb, Mn, Cl-, Fe, - NO3 ) 2- Lindberg et al. Fine(SO4 ) 0.2-0.7 Oak Through-fall - - - (1986) Coarse 0.4 Belot et al. Norway Wind tunnel, 0.2-1.4 5 - - 0.107 (1994) spruce model (r. isotopes) 1 0.13 0.13 Beckett et Pine 3 0.38 1.15 Wind tunnel - 1.2 (NaCl) al. (2000) 8 2.41 19.24 10 2.8 28.05 0.5 Ould-Dada Norway 0.82 Wind tunnel - 0.07 - 0.19 (2002) (spruce) spruce (Uranyl acetate) 0.62 0.05

Freer-Smith et 3 0.28 0.8 Oak Wind tunnel - 0.8 (NaCl) al. (2004) 6 0.29 1.7 9 0.35 3.1 PM10 Mixed 0.57-3.64 Freer-Smith et PM2 (poplar, Through-fall - - - 0.81-9.22 al. (2005) PM1 maple) 25.43-31.72

Fowler et al. Deciduous Micromet. - - 0.6-0.8 (Pb) - 0.7-1.07 (2004) (Q. robur) Micromet. Pryor (2006) Deciduous - - 0.07 0.15 (EC) 2 0.35 0.7±0.6 Oak (Q. < 3 (NaCl) Paper II Wind tunnel 5 2 0.23 1.1±0.8 robur) See Note 1 9 0.19 1.7±0.9 2 < 2 (Sea salt; ≥ 0.02-0.05 Oak (Q. Paper IV Wind tunnel 5 2 99.8 % NaCl) 0.02-0.13 robur) 10 See Note 2 0.2-0.6 < 2 (Sea salt; ≥ Oak (Q. 0.029(Cl) 0.03-0.07 Paper V Wind tunnel 2 0.9-1.7 99.8 % NaCl) robur) 0.024(Na) 0.02-0.06 See Note 3

9 MMD - mass median particle diameter is the particle diameter that divides the frequency distribution in half; fifty percent of the aerosol mass is carried by particles with a larger diameter, and fifty percent of the aerosol mass is carried by particles with a smaller diameter.

39 Note 1: details on mass-versus-size distribution are unknown; 21% of the mass was borne by particles larger than 3 μm in aerodynamic diameter (see Paper II). Note 2: mass-versus-size distribution defined by Dpg = 1.2 μm and σg = 1.8. Note 3: mass-versus-size distribution defined by Dpg = 1.55 μm and σg = 1.85.

On the basis of field measurements (throughfall methods and studies of various aerosol-borne compounds), Lindberg et al. (1982) reported values from 0.06 cms-1 (compounds in particles with MMD ≤ 0.5 μm) to 6.4 cms-1 (MMD 3.4 μm).

4.4 Applications - experimental versus model Often, the aim of experimental work is to help formulate empirical or semi- empirical models useful in a broader modelling framework, in its turn often part of an even wider inter-disciplinary and policy-relevant context. The obtained experimental estimates could be thus used in models. However, like all kind of models deposition models require some simplifications. Mathematical descriptions of the basic physical processes involved in particle deposition to vegetative surfaces have a fairly long-standing history and are essentially well established (Slinn, 1982; Wiman and Ågren, 1985; Pryor, 2006; Petroff et al., 2007). However, as emphasised above (e.g. Section 1.5), major research needs remain because the deposition problem goes beyond physics, into bio-physics. In order to illustrate how my results can be used, this section presents a simplified approach where experimental values (cf. Paper II, IV, V) serve as input parameters. Many types of aerosol-collecting vegetative elements, leaves in particular, do not lend themselves easily, or at all, to physics-based modelling of the above- mentioned processes, which, as emphasized above (Section 1.3), therefore need experimental quantification with means such as offered by the wind- tunnel design described in this work. Nevertheless, it is of interest to compare my experimental results with theories from aerosol mechanics. Such a comparison is best made on the basis of deposition velocity per unit leaf surface area, VL. Fig. 12 below, based on submodels for VL used by Wiman and Ågren (1985) illustrates the qualitative behaviour of VL versus wind speed u for a strongly idealized leaf-shaped collector, and for selected Dp-values. In this diagram, experimentally determined VL–values are also included.

40

Figure 12. Examples of deposition velocities (per unit collector surface area), VL, versus wind speed, based on theories for strongly idealized collectors, and for various illustrative combinations of particle size (Dp) and characteristic collector dimension (dc). Also included are VL –values (circles), together with curves (black lines) indicating spread in my experimental data (here given as one standard deviation) resulting from one of the studies (Paper II). The theory-based curves include sub-models for gravitational settling, impaction, interception, diffusion, in this case based on Wiman and Ågren (1985).

Although detailed information with respect to the mass versus particle-size distribution of the chloride aerosol is lacking for this test case, the theory- based curves indicate that the experimentally determined VL -function is related to the behaviour of particles around 1 μm and smaller, and to characteristic collector dimensions around 1 cm or smaller. This feature is consistent with the chloride-aerosol characteristics in Paper II, i.e., the clear dominance of mass on the 2nd filter stage. The comparison also indicates that leaf microstructures (such as hairs) are important in the aerosol capture process. However, with increasing wind speed particle bounce-off, and thus re-suspension into the air (e.g. Wiman and Ågren, 1985; Ould-Dada, 2001), might reduce net capture efficiency.

41 Although the above comparison between aerosol-mechanics theories and experimental data of course is not conclusive, it suggests that sub-models for particle deposition to oak leaves cannot be based on aerosol mechanics considerations alone. It also suggests a principle for further analysis in Paper IV and V where particle size distributions are known. The comparisons with other experimental data (see Section 4.3) suggest that obtained results will assist in formulating semi-empirical sub-models for VL as a function of wind speed (see Paper IV). The reasoning below illustrates the principle for this. Fig. 13 illustrates (in a strongly stylised manner) principles for using experimental data to estimate deposition rates in “scaling-up” operations (cf. Paper IV). The forest is divided into a number of strata, and each stratum is assigned a leaf area index, a wind velocity, and an aerosol-mass concentration distributed with particle size. Whereas the concentrations within the respective strata can be assigned different values, the size-distributions as such are assumed maintained by aerosol inflow to the forest. The experimentally determined particle-capture efficiency for oak leaves, as a function of wind velocity, can then be used for calculating the deposition rate to the forest unit in question.

LAI u1 1 E1

E E2 u LAI2 2

E3

Di ≈ Ei ui LAIi Ci

z LAI3 u3

distribution of aerosol‐mass concentration C vs particle size LAI Figure 13. An illustration of principles for estimating deposition rate for a forest unit. Here, the forest is “digitized” into a number (j) of strata; in this case j = 3, for simplicity. For each stratum (j) are assigned a leaf area index, a wind velocity, an aerosol-mass concentration, and a (normalised) distribution of concentrations with particle size (so that distribution characteristics Dpg and σg are maintained). These parameters (LAIi , ui, Ci) are assumed representative of the

42 more detailed behaviour in each stratum (the larger the number of strata the closer the parameters would represent those details). The size-distributions are assumed maintained by aerosol inflow to the forest. The diagram in the right-hand corner illustrates leaf area index by strata as a function of canopy height. The experimentally determined particle-capture efficiency for oak leaves, as a function of wind velocity, can then be used for calculating the deposition rate Dj to each stratum; the deposition rate to the forest aggregate in question is then the sum of the Dj -values.

The deposition rate to a given forest and for an aerosol with specified particle size distribution is thus a question of integration with respect to x, y and z coordinates and particle diameter. Deposition rates and concentration depletion in the forest depend on forest type and structure. For instance, as discussed in Section 4.1, a sparse forest is likely to experience higher deposition rates than would a dense forest, and in the edge region of a forest deposition will be higher than in its interior. Fig. 13 is intended to indicate opportunities as well as constraints with using experimental data in scaling-up and modelling. What does one need to know in order to find a deposition rate, D, (cf. Section 3.2) for a forest? One must provide a condition for aerosol particle concentration or flux. In this exemplifying case (Fig. 13), it is assumed that the mass concentration is known (or measured) for each stratum, and that its distribution with particle size is maintained at steady-state by sources compensating for sinks. One must know (or have access to sub-models for) the LAI for each stratum (obtainable through various measurement techniques); the wind velocity profile (wind speed u as a function of canopy height, z); the turbulence conditions (such as the eddy diffusivity as a function of z); the particle-size distribution data (obtainable from field measurements with for instance cascade impactors or differential mobility particle sizers (DMPS)); and E, capture efficiencies.

My data provide E-values for oak leaves, for various wind speeds and for specified mass-vs-size distributions. For some substances (cf. Paper IV), those distributions a representative of distributions measured in the field. The capture-efficiency values obtained are species specific but with some caution could serve as proxies for other deciduous species. To my knowledge, there are no other studies providing specific capture-efficiency estimates for deciduous species and very few studies providing capture efficiencies for other types of vegetative surfaces. Also, my experimental findings place deposition modelling on safer and more realistic ground than do sub-models for capture efficiency that are based on aerosol mechanics only. Thus, by knowing this specific capacity of leaves with certain characteristic features one can, in principle, calculate (or at least reasonably estimate) the sink capacity of the entire forest that is of importance from a biogeochemical point of view.

43 However, assumptions and simplifications have been made to illustrate the principles. For instance, there is turbulence in the field that is difficult to emulate in the wind tunnel (and in any wind tunnel, although the larger the tunnel the better the emulation). The criterion that concentration is stable within each stratum must be satisfied (over reasonably long periods of time, concentration averages can be used to approximately meet that requirement). Assuming stable particle-size distributions in the forest is more problematic, however, and needs to be addressed in further research that has access to mono-disperse aerosols (so that capture efficiencies can be measured also as functions of both wind speed and particle size, rather than as functions of wind speed and size-distribution characteristics).

Hence, bridging from experimental information to “real-world” systems pose interesting challenges. One cannot avoid simplifications in combining experimental and model thinking.

Generally, the estimation of aerosol deposition rates to forests face uncertainties due to limitations in micrometeorological methodologies, throughfall approaches, and deposition models (cf. Section 2). For instance, it is not possible to measure the flux at the surface itself and forest edges and complex terrain present fundamental problems. Multiple resistance-analogues frameworks apply best to a flat, spatially extensive, homogeneous surface. Therefore, there is a need for an estimate or approach that could provide aerosol deposition rates for the forest including forest edges and other vegetation inhomogeneities. Such estimates – focused on E, capture efficiency – have been obtained in my experimental work.

44 5 CONCLUSIONS AND IMPLICATIONS

The intention of this thesis is to contribute to the physical understanding of deposition to deciduous species. Coastal-zone forests constitute a particular case of interest in that they are subject to aerosol-borne fluxes of various substances under specific meteorological conditions such as strong prevailing winds and shifts in relative humidity. My research process is mostly based on experimental work which is related to wind tunnel studies. In a broader scope, as outlined in Chapter 1, my empirical results can assist in formulating mathematical deposition models of relevance to inter-disciplinary problems that involve climate change and shifts in land-use and forestry practices.

5.1 Main conclusions My work has shown that wind-tunnel based studies offer a useful methodology for studying deposition processes with a focus on the aerosol- particle capture efficiency (or aerosol sink capacity) of certain vegetation surfaces. Deposition studies with respect to deciduous species have been underrepresented, which thus warrants further special research efforts. Capture efficiency is the most appropriate parameter to include in process- oriented models of forests as aerosol sinks. Also, it provides possibilities for simulating some of the natural conditions that influence aerosol deposition in coastal environments. Data on capture efficiencies obtained in this thesis (Paper II, IV, V) provide possibilities for detailed modelling of the sink capacity at various canopy layers. This illustrates how my results can assist in improving the sub-models of capture efficiency that are used in the modelling of fluxes of aerosol-borne substances to forests. The results in this Thesis (Paper II, IV) demonstrate that meteorological regimes that involve increasing wind velocities would lead to increasing aerosol-deposition velocities. For instance, if climate change, in terms of meteorological shift, would lead to increasing source strengths (e.g. sea spray

45 emissions from bubble bursting) deposition rates might increase substantially because of increasing concentrations and because of increasing deposition velocities. My results (Paper II) indicate that capture efficiency and deposition velocity increase with LAI; this is not the case for some other studies (Section 4.1). Data (Paper IV) reveal higher deposition velocity values at low LAI. Further research on the effects of vegetation canopy structure and morphology on capture efficiency and deposition velocity is clearly needed. My data demonstrate that forest ecosystems would experience substantially increased deposition at edges (Paper V). The deposition velocity is higher at the forest edge, higher wind speeds at the edge and higher impaction efficiency being the major explanation. The findings suggest that field measurements of deposition in the interior of a forest “island” in an otherwise open landscape would underestimate the deposition to the entire forest (cf. Erisman and Draaijers, 2003). Deposition models used at regional scales usually treat deposition as a one- dimensional (vertical) transfer to homogeneous surfaces with infinite extension. The landscape patchiness and inhomogeneities are rarely taken into account. This is partly due to lack of data regarding the spatial mosaic of subsystems within forest ecosystems and lack of data on land use. Moreover, edge effects and processes related to that are still scarcely studied (Erisman and Draaijers, 2003).

5.2 Implications What are the possibilities and difficulties for using the results obtained in these studies? For instance, capture efficiency estimates give opportunities for multi-layer deposition models (aerosol-sink processes within various canopy layers) where the leaves’ capacity to catch the aerosols particles can be scaled up to the entire forest. That is, by knowing this specific capacity of leaves with certain characteristic features one can, in principle, calculate the sink capacity of the entire forest. The resulting values for aerosol deposition to the forest are, in turn, important from a biogeochemical point of view. Aerosol deposition might be a significant source of nutrients to the forest ecosystem and part of the internal cycling. For instance, acidification of forests was a dominating environmental issue in the 1980s and 1990s (cf. Erisman et al., 1997), and remains a major concern in many parts of the world. The influx of sea salt particles containing Mg2+ and other base cations counteracts acidification and would lead to improved nutrient balance in the forest and thus better health of the forest (cf. e.g. Reynolds, 2000). On the other hand, it can also lead to overloads of macro-constituents, micro-constituents and trace elements that would have negative effects on the vitality of forests.

46 However, as mentioned before (Section 3.3), some difficulties could be expected when trying to apply and relate laboratory results at the large-scale level, i.e., the real forest ecosystem level. These include problems such as turbulence fields and wind direction. It would also be difficult or rather impossible to achieve an adequate description of the architectural composition of mature trees, as well to consider other vegetation types surrounding the trees. For improved simulations of atmospheric dynamics one needs a tunnel with much larger dimensions than the one I had access to. Although, in theory, some of the problems can be addressed through using other methods (such as field-based throughfall studies; cf. Section 2.1) each of them is, in practice, associated with its particular difficulties. Therefore, several methods – including those presented in this Thesis – need to be explored, and results from applying them need to be combined to help advance aerosol-deposition science. Considering the many uncertainties in field flux measurements, more wind tunnel studies are needed under better controlled conditions. These conditions include surface morphology of leaves, stability with respect to particle-size distributions, and possibilities for studying aerosols with a wide range of physical and chemical properties. In addition, the question is how detailed a model should be to describe adequately the surface interactions with aerosols. What information should be reported on surface morphology for future model development in order to attempt inclusion of these effects in the future? My data contribute in this context: estimations of capture efficiency (Papers II, IV, V) for oak leaves, for various wind speeds and specific mass-vs-size distribution will help modellers towards better mathematical descriptions of aerosol deposition processes. A complete description of the surface of a vegetative collector is difficult in particular for a forest that includes many species of varying geometry and structure. Both the macro-scale structure of for instance a leaf (size, geometrical dimensions, shape) and, in particular, the micro-scale structure – hairiness, micro-structures, wax layers and other structures that are “giant” in relation to an aerosol particle of for instance 1 μm size – are of concern. These structures can affect the strength of deposition mechanisms in very subtle ways. Differences in surface structure can occur between even seemingly similar . Theoretical developments demonstrate that models of dry deposition need to consider canopy structure and small-scale morphology (Fowler et al., 2009; Petroff et al., 2009). High-quality wind-tunnel experiments, coupled with improved measurement technology would help decrease remaining uncertainties.

47 5.3 Future concerns My experimental results suggest that some methodological improvements are desirable to enable further research on potentially important processes that were identified during my research but that could not be addressed in depth. In particular, methods need to be developed for more extensive studies on the effects of relative-humidity gradients near the vegetation surface, because the presence of water on leaf surfaces influences the surface chemistry and thereby the deposition of various compounds (Burkhardt, 2010). Relative humidity plays an important role in the deposition of various aerosol-borne substances mainly by affecting the aerosol-particle diameter due to hygroscopic growth (cf. Quinn and Ondov, 1998). Also, better control of turbulence fields will be needed to help narrow inevitable gaps between real field conditions and conditions obtainable in wind-tunnel based studies. Although aerosol-mass versus particle-size distributions generated in my experiments to a reasonable extent emulate distributions of some substances in aerosols occurring in the field, possibilities to better simulate certain particle size distributions also need to be addressed in further research. My results also propose needs for a range of further experimental investigations. Among such needs are studies that extend the framework of the beach-to-forest transition investigated in Paper V to include also the coastal water surface. That is, aerosol deposition across the complete sea-to-land aerodynamic transition needs laboratory-based as well as field-based research. Among several reasons for such research are needs to add more science to the ongoing debate on geoengineering manipulations such as deliberately increasing source strengths by spraying sea salt aerosols into the atmosphere (cf. Salter et al., 2008; Korhonen et al., 2010). (The underlying idea is to generate clouds that in turn affect radiation balances and thus climate.) As suggested by my experiments, such geoengineering schemes would strongly increase deposition in coastal-forest zones, where edge effects (Paper V) are of great importance. Not only sea salt injections but also an increase of wind speed in coastal zones as well as possible shift in land-use (terrestrial geoengineering) in coastal zones would affect deposition. Models that include advanced sink specific parameterization at various canopy levels would be beneficial at local as well as regional scales. Among reasons for that are landscape patchiness (which is related to practices in forestry and agriculture) which is of importance to edge effects on deposition, as discussed in Paper V. Aerosol-sink modelling is also of importance to the global-scale context. As was discussed in Chapter 1, this is because the concentration of aerosol particles in the atmosphere influences climate and the concentration levels are determined by the strengths of sources and sinks.

48 References Allen L.H. 1968. Turbulence and windspeed spectra within a Japanes Larch plantation. J. Applied Meteorol. 7: 73-78. Andreae M.O. & Crutzen P.J. 1997. Atmospheric aerosols: biogeochemical sources and role in atmospheric chemistry. Science 276: 1052-1059. Bargel H., Koch K., Cerman Z. & Neinhuis C. 2006. Structure–function relationships of the plant cuticle and cuticular waxes — a smart material? Functional Plant Biology 33: 893–910. Baron P. & Willeke K. 2001. Aerosol Measurements. Principles, Techniques and Applications. Wiley Interscience. John Willey & Sons, Inc. Publication, New York: 61-129, 667. Baldocchi D.D., Hicks B.B. & Meyers T.P. 1988. Measuring biosphere-atmosphere exchanges of biologically related gases with micrometeorological methods. Ecology 69(5): 1331-1340. Baldocchi D., Falge E. & Wilson K. 2001. A spectral analysis of biosphere-atmosphere trace gas flux densities and meteorological variables across hour to multi-year time scales. Agr. Forest Meteorol. 107 (1):1-27. Baltensperger U. 2010. Aerosols in clearer focus. Science 329 (5998): 1474-1475. Beckett K.P., Freer-Smith P.H. & Taylor G. 2000. Particulate pollution capture by urban trees: effect of species and windspeed. Glob. Chan. Biol. 6: 995-1003. Belot Y., Camus H., Gauthier D. & Caput C. 1994. Uptake of small particles by tree canopies. Sci. Total Environ. 157: 1-6. Bodin P. 2007. Multiple Goals and Multiple Stresses: the Role of Ecological and Biogeophysical Process Models in Managing the Forests of the Future. Doctoral Thesis [Kalmar University, ISBN: 978-91-89584-82-2]. Bowen H.J.M. (ed.) 1982. Environmental Chemistry, Volume 2. The Royal Society of Chemistry, Burlington House, London: 36. Brunet Y., Itier B., McAneney J. & Lagouarde J.P. 1994. Downwind evolution of scalar fluxes and surface resistance under conditions of local advection. Part II: measurements over barley. Agr. Forest Meteorol. 71 (3-4): 227-245. Burkhardt J. 2010. Hygroscopic particles on leaves: nutrients or desiccants? Ecological Monographs 80 (3): 369-399. Cape J.N. 2008. Interactions of forests with secondary air pollutants: Some challenges for future research. Environ. Pollut. 155 (3): 391-397. Chalmers A.F. 1999. What Is This Thing Called Science? Huckett Publishing Co., London. Chamberlain A.C. 1967. Transport of Lycopodium spores and other small particles to rough surfaces. Proc. R. Soc. A 296: 45-70. Chamberlain A.C. 1975. The movement of particles in plant communities. In: Monteith J.L. (ed.) Vegetation and the Atmosphere. Academic Press, London. Cohen M. D., Flagan R. C. & Seinfeld J. H. 1987. Studies of concentrated electrolyte solutions using the electrodynamic balance. 1. Water activities for single-electrolyte solutions. J. Phys. Chem. 91: 4563-4574. Carslaw K.S., Boucher O., Spracklen D.V., Mann G.W., Rae J.G.L., Woodward S. & Kulmala M. 2010. A review of natural aerosol interactions and feedbacks within the Earth system. Atmos. Chem. Phys. 10 (4): 1701-1737.

49 Cowan I.R. 1968. Mass heat and momentum exchange between stands of plants and their atmospheric environment. Q. Jl. R. Met. Soc. 94: 318-332. Dambrine E., Pollier B., Bonneau M. & Ignatova N. 1998. Use of artificial trees to assess dry deposition in spruce stands. Atmos. Environ. 32 (10): 1817-1824. Davidson C.I., Miller J.M. & Pleskow M.A. 1982. The influence of surface structure on predicted particle dry deposition to natural grass canopies. Water Air Soil Poll. 18: 25- 44. Draaijers G.P.J., van Ek R. & Meijers R. 1992. Research on the impact of forest stand structure on atmospheric deposition. Environ. Pollut. 75: 243-249. Draaijers G.P.J. & Erisman J.W. 1993. Atmospheric deposition onto forest stands: throughfall estimates compared to estimates from inference. Atmos. Environ. 27A: 43– 55. Duce R.A., Mohnen V.A., Zimmerman P.R., Chatfiels R., Grosjean D., Cautreels W., Jaenicke R., Ogren J.A., Pellizzari E.D. & Wallace G.T. 1983. Organic material in the global troposphere. Reviews of Geophysics and Space Physics (ISSN 0034-6853), 21: 921- 952. Erisman J.W. & Draaijers G. P. J. 1995. Atmospheric Deposition in Relation to Acidification and Eutrophication, Studies in Environmental Science, 63, Elsevier, Amsterdam, The Netherlands. Erisman J.W., Draaijers G., Duyzer J., Hofschreuder P., Van Leeuwen N., Römer F., Ruijgrok W., Wyers P. & Gallagher M. 1997. Particle deposition to forests – summary if results and applications. Atmos. Environ. 31: 321-332. Erisman J.W., Draaijers G.P.J, Steingrover E., Van Dijk H., Boxman A. & de Vries W. 1998. Assessment of the exposure and loads of acidifying and eutrophying pollutants and ozone, as well as their harmful influence on the vitality of the trees and the Speulder forest ecosystem as a whole. Water Air Soil Poll. 105, 33-4: 539-571. Erisman J.W. & Draaijers G. 2003. Deposition to forests in Europe: most important factors influencing dry deposition and models used for generalisation. Atmos. Poll. 124: 379-388. Fairall C.W. 1984. Interpretation of eddy-correlation measurements of particle deposition and aerosol flux. Atmos. Environ. 18: 1329-1337. Ferm M. & Hultberg H. 1999. Dry deposition and internal circulation of nitrogen, sulphur and base cations to a coniferous forest. Atmos. Environ. 33: 4421-4430. Finnigan J.J. & Brunet Y. 1995. Turbulent airflow in forests on flat and hilly terrain. In: Couts, M., and Grace, J. (ed.), Wind and Trees, Cambridge University Press, Cambridge, 3-38. Finnigan J.J. & Shaw R.H. 2000. A Wind-Tunnel Study of Airflow in Waving Wheat: An EOF Analysis of the Structure of the Large-Eddy Motion. Boundary-Layer Meteorol. 96(1-2): 211-255. Finnigan J.J., Shaw R.H. & Patton E.G. 2009. Turbulence structure above a vegetation canopy. Journal of Fluid Mechanics, 637. Finkelstein P.L. 2001. Deposition Velocities of SO2 and O3 over Agricultural and Forest Ecosystems. Water Air Soil Poll.: Focus 1 (5-6): 49-57. Foltescu V.L., Pryor S.C. & Bennet C. 2005. Sea salt generation, dispersion and removal on the regional scale. Atmos. Environ. 39: 2123-2133. Fowler D., Pilegaard K., Sutton M.A., Ambus P., Raivonen M., Duyzer J., Simpson D., Fagerli H., Fuzzi S., Schjoerring J.K., Granier C., Neftel A., Isaksen I.S.A., Laj P.,

50 Maione M., Monks P.S. & Burkhardt J. 2009. Atmospheric composition change: Ecosystems–Atmosphere interactions. Atmos. Environ. 43: 5193-5267. Fowler D., Skiba U., Nemitz E., Choubedar F., Branford D., Donovan R. & Rowland P. 2004. Measuring aerosol and heavy metal deposition on urban woodland and grass using inventories of 210 Pb and metal concentrations in soil. Water Air Soil Poll.: Focus 4: 483-499. Fowler D., Cape J.N., Coyle M., Flechard C., Kuylenstierna J., Hicks K., Derwent D., Johnson C. & Stevenson D. 1999. The global exposure of forests to air pollutants. Water Air Soil Poll. 116: 5-32. Freer-Smith P.H., El-Khatib A.A. & Taylor G. 2004. Capture of particulate pollution by trees: a comparison of species typical of semi-arid areas (Ficus nitida and Eucalyptus globulus) with European and North America species. Water Air Soil Poll. 1: 1-15. Freer-Smith P.H., Beckett K.P. & Taylor G. 2005. Deposition velocities to Sorbus aria, Acer campestre, Populus deltoides x trichocarpa 'Beaupre', Pinus nigra and x Cupressocyparis leylandii for coarse, fine and ultra-fine particles in the urban environment. Environ. Pollut. 133 (1): 157-167. Fuchs N.A. 1964. The Mechanics of Aerosols. Pergamon Press, Oxford. Gao W., Shaw R. H. & Paw U K.T. 1989. Observation of Organised Structure in Turbulent Flow within and above a Forest Canopy. Boundary-Layer Meteorol. 47: 349– 377. Gardiner B.A. 1994. Wind and wind forces in a plantation spruce forest. Boundary-Layer Meteorol. 67: 161–186. Gaydarova P.N. 2003. Deciduous forest communities in the Black Sea coastal Strandzha region: temporal and spatial characteristics of leaf are index and density. Trees 17: 237- 243. Grace J. 1983. Plant-Atmosphere Relationships. Chapman & Hall, London, pp. 92. Grönholm T., Aalto P.P., Hiltunen V., Rannink U., Rinne J., Laakso L., Hyvonen S., Vesala T. & Kulmala M. 2007. Measurements of aerosol particle dry deposition velocity using the relaxed eddy accumulation technique. Tellus B 59 (3): 381-386. Guenther A., Karl T., Harley P., Wiedinmyer C., Palmer P.I. & Geron C. 2006. Estimates of global terrestrial isoprene emissions using MEGAN (Model of Emissions of Gases and Aerosols from Nature). Atmos. Chem. Phys. 6: 3181–3210. Gustafsson M.E.R. & Franzén L.G. 2000. Inland transport of marine aerosol in southern Sweden. Atmos. Environ. 34: 313-325. Hansen B. & Nielsen K.E. 1998. Comparison of acidic deposition to semi-natural ecosystems in Denmark-coastal heath, inland heath and oak wood. Atmos. Environ. 32: 1075-1086. Hardy J.T., Apts C.W., Creclius E.A. & Fellingham G.W. 1985. The sea-surface microlayer: fate and residence times of atmospheric metals. Limnol. Oceanogr. 30: 93- 101. Holling C.S. 1973. Resilience and stability of ecological systems. Annual Review of Ecology and Systematics 4: 1-23. Hoppel W.A. & Frick G.M. 1990. Submicron aerosol size distributions measured over the tropical and south Pacific. Atmos. Environ. 24A: 645-659. Höfken K.D., Meixner F.X. & Ehhalt D.H. 1983. Deposition of atmospheric trace constituents onto different natural surfaces, in Pruppacher, H.R., Semonin, R.G.,

51 Slinn, W.G.N. (ed.), Precipitation Scavenging , Dry Deposition and Ressuspension, Elsevier, New York: 825-833. IPCC 4th Assessment Report, 2007, Solomon S., Qin D., Manning M., Chen Z., Marquis M., Averyt K.B., Tignor M. & Miller H.L. (ed.). Cambridge University Press, Cambridge, United Kingdom and New York, NY, USA. Karnosky D.F., Skelly J.M., Percy K.E. & Chappelka A.H. 2007. Perspectives regarding 50years of research on effects of tropospheric ozone air pollution on US forests. Environ. Pollut. 147 (3): 489-506. Karnosky D.F., Percy K.E., Thakur R.C., Honrath R.E., 2003. Air pollution and global change: a double challenge to forest ecosystems. In: Karnosky, D.F., Percy, K.E., Chappelka, A.H., Simpson, C., Pikkarainen, J. (ed.). Air Pollution, Global Change and Forests in the New Millenium, Elsevier, Amsterdam. Kaufman Y. J., Tanre D. & Boucher O. 2002. A satellite view of aerosols in the climate system. Nature 419: 215-223. Kinnersley R.P., Farrington-Smith J.G., Shaw G. & Minski M.J. 1994. Aerodynamic characterisation of model tree canopies in a wind tunnel. Sci. Tot. Environ. 157: 29-33. Korhonen H., Carslaw K.S. & Romakkaniemi S. 2010. Enhancement of marine cloud albedo via controlled sea spray injections: a global model study of the influence of emission rates, microphysics and transport. Atmos. Chem. Phys. Discussions 10 (1): 735- 761. Kruijt B., Malhi Y., Lloyd J., Norbre A.D., Miranda A.C., Pereira M.G.P., Culf A. & Grace, J. 2000. Turbulence statistics above and within two Amazon rain forest canopies. Boundary-Layer Meteorol. 94 (2): 297-331. Lamaud E., Carrara A., Brunet Y., Lopez A. & Druilhet A. 2002. Ozone fluxes above and within a pine forest canopy in dry and wet conditions. Atmos. Environ. 36: 77-88. Likens G.E., Bormann F.H., Pierce R.S., Eaton J.S. & Johnson N.M. 1977. Biogeochemistry of a Forested Ecosystem, Springer-Verlag, New York. Lindberg S.E., Harris R.C. & Turner R.R. 1982. Atmospheric deposition of metals to forest vegetation. Science 215: 1609-1610. Lindberg S.E., Lovett G.M., Richter D.D. & Johnson D.W. 1986. Atmospheric deposition and canopy interactions of major ions in a forest. Science 231: 141-145. Lindberg S.E. & Garten C.T. 1988. Sources of sulphur in forest canopy throughfall. Nature 336: 148-151. Lovett G.M. & Lindberg S.E. 1984. Dry deposition and canopy exchange in a mixed oak forest as determined by analysis of throughfall. J. Appl. Ecol. 21: 1013-1027. Lovett G.M. & Reiners W.A. 1986. Canopy structure and cloud water deposition in subalpine coniferous forests. Tellus 38B: 319-327. Lovett G.M. 1992. Atmospheric deposition and and forest nutrient cycling, In D.W Johnson and S.E. Lindberg (ed.), Ecol. Studies, 91, Springer-Verlag, New York: 152- 166. May R.M. 1973. Stability and Complexity in Model Ecosystems. Princeton University Press, Princeton. Marcos G.M. & Gallardo Lancho J.F. 2002. Atmospheric deposition in oligotrophic Quercus pyrenaica forests: implications for forest nutrition. Forest Ecol. Manag. 171: 17- 29. Marschner H., Römheld V. & Kissel M. 1986. Different strategies in higher plants in mobilization and uptake of iron. J. of Plant Nutrition 9: 695-713.

52 Massman W. J. & Weil J. C. 1999. An Analytical one-Dimensional Second-Order Closure Model of Turbulence Statistics and the Lagrangian Time Scale Within and Above Plant Canopies of Arbitrary Structure. Boundary-Layer Meteorol. 91 (1): 81-108. Meroney R.N. 1968. Characteristics of wind and turbulence in and above model forests. J. App. Meteorol. 7: 780-788. Monteith J.L. & Unsworth M.H. 1990. Principles of Environmmental Physics, Edward Arnold, London. Nemitz E., Sutton M.A., Wyers G.P., Otjes R.P., Schjoerring J.K., Gallagher M.W., Parrington J., Fowler D. & Choularton T.W. 2000. Surface/atmosphere exchange and chemical interaction of gases and aerosols over oilseed rape. Agr. Forest Meteorol. 105 (4): 427-445. Nemitz E., Sutton M.A., Wyers G.P., Otjes R.P., Mennen M.G., van Putten E.M. & Gallagher M.W. 2004. Gas-particle interactions above a Dutch heathland: II. Concentrations and surface exchange fluxes of atmospheric particles. Atmos. Chem. Phys. 4(4): 1007-1024. Nilsson E.D., Rannik Ü., Buzoriu G., Kulmala M. & O`Dowd C. 2001. Effects on the continental boundary layer evolution, convection, turbulence and entrainment on aerosol formation. Tellus 53B: 441-461. Ould-Dada Z. 2002. Dry deposition profile of small particles within a model spruce canopy. Sci. Tot. Environ. 286: 83-96. Ould-Dada Z. 2001. Resuspension of small particles from tree surfaces. Atmos. Environ. 35: 3799-3809. Ould-Dada Z., Copplestone D., Toal M. & Shaw G. 2002. Effect of forest edges on deposition of radioactive aerosols. Atmos. Environ. 36: 5595–5606. O'Dowd C.D., Becker E. & Kulmala M. 2001. Mid-latitude North-Atlantic aerosol characteristics in clean and polluted air. Atmos. Research 58 (3): 167-185. Parker G.G. 1983. Throughfall and stemflow in the forest nutrient cycle. Adv Ecol Res. 13: 57-133. Peters K. & Eiden R. 1992. Modelling the dry deposition velocity of aerosol particles to a spruce forest. Atmos. Environ. 26A: 2555-2564. Petroff A., Mailliat A., Amielh M. & Anselmet F. 2007. Aerosol dry deposition on vegetative canopies. Part I: Review of present knowledge. Atmos. Environ. 42: 3625- 3653. Petroff A., Zhang L., Pryor S.C. & Belot Y. 2009. An extended dry depositon model for aerosols onto broadleaf canopies. Aerosol Science 40: 218-240. Pryor S.C. & Sørensen L.L. 2000. Nitric acid-sea salt reactions: Implications for nitrogen deposition to water surfaces. J. Appl. Meteorol. 39 (5): 725-732. Pryor S.C. & Barthelmie R.J. 2000. Particle Dry Deposition to Water Surfaces: Processes and Consequences. Mar. Pollut. Bull. 41 (1-6): 220-231. Pryor S.C., Barthelmie R.J., Jensen B., Jensen N.O. & Sørensen L.L. 2002. HNO3 fluxes to a deciduous forest derived using gradient and REA methods. Atmos. Environ. 36: 5993-5999. Pryor S.C. & Barthelmie R.J. 2005. Liquid and chemical fluxes in precipitation, throughfall, and stemflow: observations from a deciduous forest and a red pine plantation in the Midwestern U.S.A. Water Air Soil Poll. 163: 203-227. Pryor S.C. 2006. Size resolved particle deposition velocities of sub-100 nm diameter particles over a forest. Atmos. Environ. 40: 6192–6200.

53 Pryor S., Gallagher M., Sievering H., Larsen S.E., Barthelmie R.J., Birsan F., Nemitz E., Rinne J., Kulmala M., Grönholm T., Taipale R. & Vesala T. 2008. A review of measurement and modelling results of particle atmosphere-surface exchange. Tellus 60B: 42-75. Quinn P.K., Marshall S.F., Bates T.S., Covert D.S. & Kapustin V.N. 1995. Comparison of measured and calculated aerosol properties relevant to the direct radiative forcing of tropospheric sulfate aerosol in climate. J. of Geophysical Research 100: 8977-8991. Quinn T.L. & Ondov J.M. 1998. Influence of temporal changes in relative humidity on dry deposition velocities and fluxes of aerosol particles bearing trace elements. Atmos. Environ. 32: 3467-3479. Raupach M.R. & Thom A.S. 1981. Turbulence in and above plant canopies. Ann. Rev. Fluid Mech. 13: 97-129. Raupach M. R., Coppin P. A. & Legg B. J. 1986. Experiments on scalar dispersion within a modelplant canopy. Part 1. The turbulence structure. Boundary-Layer Meteorol. 35: 21–52. Raupach M.R., 1989. Turbulent transfer in plant canopies, in Russell G., Marshall B. and Jarvis P. G. (ed.) Plant Canopies: Their Growth, Form and Function. Cambridge University Press, Cambridge: 41-61. Raupach M.R., Antonia R.A. & Rajagopalan S. 1991. Rough-Wall Turbulent Boundary Layers. Appl. Mech. Rev. 44: 1–25. Rea A.W., Lindberg S.E. & Keeler G.J. 2001 Dry deposition and foliar leaching of mercury and selected trace elements in deciduous forest throughfall. Atmos. Environ. 35: 3453-3462. Reynolds B. 2000. An evaluation of critical loads of soil acidity in areas of high sea salt deposition. Sci. Tot. Environ. 253: 169-176. Ruijgrok W., Tieben H. & Eisinga P. 1997. The dry deposition of particles to a forest canopy: a comparison of model and experimental results. Atmos. Environ. 31: 399-415. Salter S., Sortino G. & Latham J. 2008. Sea-going hardware for the cloud albedo method of reversing global warming. Phil. Trans. R. Soc. A 366: 3989-4006. Sander R., Keene W.C., Pszenny A.A.P., Arimoto R., Ayers G.P., Baboukas E., Cainey J.M., Crutzen P.J., Duce R.A., Hönninger G., Huebert B.J., Maenhaut W., Mihalopoulos N., Turekian V.C. & Van Dingenen R. 2003. Inorganic bromine in the marine boundary layer: a critical review. Atmos. Chem. Phys. 3 (5): 1301-1336. Schlesinger W.H. 1997. Biogeochemistry. An analysis of Global Change. San Diego, London, Boston, New York, Sydney, Tokyo, Toronto: Academic Press: 82. Scurlock J.M.O., Asner G.P. & Gower S.T. 2001. Worldwide Historical Estimates and Bibliography of Leaf Area Index, 1932-2000. ORNL Technical Memorandum ORNL/TM-2001/268. Oak Ridge National Laboratory, Oak Ridge, Tenn. Seinfeld J.H. & Pandis S.N. 1998. Atmospheric Chemistry and Physics. From Air Pollution to Climate Change. John Willey & Sons, Inc., New York: 1000-1001. Shaw G., Farrington-Smith J.G., Kinnersley R. & Minski M.J. 1994. Dry deposition of aerosol particles within model spruce canopies. Sci. Tot. Environ. 157: 17-23. Skakalova T.S., Savov P.B., Grigorov I.Y. & Kolev I.N. 2003. Lidar observation of breeze structure during the transition periods at the Southern Bulgarian Black Sea coast. Atmos. Environ. 37: 299-311. Skjøth C.A., Geels C., Hvidberg M., Hertel O., Brandt J., Frohn L.M., Hansen K.M., Hedegård G.B., Christensen J.H. & Moseholm L. 2008. An inventory of tree species

54 in Europe – An essential data input for air pollution modelling. Ecol. Model. 217: 292- 304. Slinn W.G.N. 1982. Predictions for particle deposition to vegetative surfaces. Atmos. Environ. 16: 1785-1794. Svenningsson B., Rissler J., Swietlicki E., Mircea M., Bilde M., Facchini M.C., Decesari C., Fuzzi S., Zhou J., Mønster J. & Rosenørn T. 2006. Hygroscopic growth and critical supersaturations for mixed aerosol particles of inorganic and organic compounds of atmospheric relevance. Atmos. Chem. Phys. 6 (7): 1937-1952. Tammet H., Kimmel V. & Israelsson S. 2001. Effect of atmospheric electricity on dry deposition of airborne particles from atmosphere. Atmos. Environ. 35 (20): 3413-3419. Torseth K., Semb A., Schaug J., Hanssen J.E. & Aamlid D. 2000. Processes affecting deposition of oxidised nitrogen and associated species in the coastal areas of Norway. Atmos. Environ. 34: 207-217. Watson A.J. & Lovelock J.E. 1983. Biological homeostasis of the global environment—the parable of Daisyworld. Tellus B 35: 284. Warneck P. 1988. Chemistry of the Natural Atmosphere. Academic Press, Inc. Harcourt Brace Jovanovich, Publishers, San Diego, CA: 323-351. Whitby K.T. 1978. The physical characteristics of sulfur aerosols. Atmos. Environ. 12: 135- 161. Wiman B.L.B. 1981. Aerosol collection by Scots pine seedlings: design and application of a wind tunnel method. OIKOS 36: 83-92. Wiman B.L.B. & Lannefors H.O. 1985. Aerosol characteristics in a mature coniferous forest – methodology, composition, sources and spatial concentration variations. Atmos. Environ. 19: 342-362. Wiman B.L.B. 1985. A non-stationary model for sulphur aerosol depletion and deposition in a young spruce stand. Tellus 37B: 230-242. Wiman B.L.B. & Ågren G.I. 1985. Aerosol depletion and deposition in forests – a model analysis. Atmos. Environ. 19: 335-347. Wiman B.L.B. 1986. Diurnal variations of aerosol concentrations inside and above a young spruce stand: modelling and measurements. Water Air Soil Poll. 31: 343-348. Wiman B.L.B. 1988. Aerosol capture by complex forest architecture. In: Verhoeven, J.T.A., Heil, G.W., Werger, M.J.A. (ed.), Vegetation Structure in Relation to Carbon and Nutrient Economy. SPB Academic Publishing bv., The Hague: 157-183. Wiman B.L.B., Unsworth M.H., Lindberg S.E., Bergkvist B., Jaenicke R. & Hansson H.C. 1990. Perspectives on aerosol deposition to natural surfaces: interactions between aerosol residence times, removal processes, the biosphere and global environmental change. J. of Aerosol Sc. 21 (3): 313-338. Wiman B.L.B. 2000. Aerosols at air/water/land interfaces: modelling and measurements. In Gryning S.-E. and Batchvarova E. (ed.) (2000) Air Pollution Modelling and Its Application XIII, Kluwer Academic/Plenum Publishers, New York: 687-698. Wiman B.L.B., Velchev K., Gaydarova P.N., Donev E.H. & Yurukova L. 2003. A note on aerosol mass-versus-size distributions in the south-east Bulgarian Black Sea coastal region. Bulgarian Journal of Meteorology and 13 (1-2): 26-39. Wiman B.L.B., Donev E.H., Nikolova P.N., Yurukova L.D. 2007. Aerosol-borne trace elements in Bulgaria: concentration levels in coastal and mountainous regions. Bulgarian Geophysical Journal 33: 3-19.

55 Wuyts K., Verheyen K., De Schrijver A., Cornelis W.M. & Gabriels D. 2008. The impact of forest edge structure on longitudinal patterns of deposition, wind speed, and turbulence. Atmos. Environ. 42: 8651–8660. Zielinski T. 2005. Studies of Aerosol Physical Properties in Coastal Areas. Aerosol Sci. Tech. 38 (5): 513-524.

56