CONN -6--l'"-

~04 y

!

, 'L!

c

The Climate and HealthImplications

of Bubble-Mediated

Sea-Air Exchange

Edward C. Monahanand

Margaret A. Van Patten

Editors

Proceedings of an international symposium by the sametitle held at the University of Connecticut et Avery Point.Groton, Connecticut. on October 7 9, !988, sponsoredby the Connectkvt Sea Grant Colleoo Program,

A pob4cation of the ConnectCvtSea Grant College ProgramCT-SG-89-08

Library of Congress Catalog Card Number: 89-0S110lSBN: 1-878301464

Copyright ! 1989 by the Connecticut Sea Grant College ProgramAll rights reserved. Printed in the United States of America No part of this bookmay be used or reproduced in any form or by any means, or stored in a databaseor retrieval system, without prior written permission of the publisher, except inthe case of brief quotations embodied in critical articles and reviews.

Publication of this book is supported by the Natiorutl Oceanic and AtmosphericAdministration via Sea Grant No. NA85AA-D-SG103.

Production Editor. Margaret A. Van PattenEditorial Assistant: Patricia A, BeelhamBook Design: Margaret A. Van PattenCover Design: Betsy J. Suprenant

Distributed by the Connecticut Sea Grant College Pngp3rn, Marin: SciencesInstitute, University of Connecticut at Avery Point, Groton, Cf 06340

Contents

Preface .

�

65

147

Chapter 1Bacteria and Other Materials tn Drops From BUrsting BubblesDuncan C. Blanchard..........................................., ............... l

Chapter 2Me Aerosolizatian of MycobacteriaJoseph O. Falkinham, Ill

Chapter 3Marine Toxins in Bubble-Generated Aerosol

Richard H. Pierce et al 27

Chapter 4From the Laboratory Tank to the Global OceanEdward C. Monahan.. 43

Chapter 5'%e Occurrence of Large Salt Water Droplets at LowElevations Over the Open Ckxan

Gerri dc Lccuw

Chapter 5Relationships Between Marine Aerosols,Oe~ic Whitecaps,and Low-Elevation Winds Observed Duringthe HEXMAX

Experiment in the North SeaRoman hfarks and Edward C. Monahan. .83

Chapter 7Enhancement of Sea-to-Air Moisty Hux by Spray DropletsAssociated With Breaking Waves and Bubble Entrainment:Simulations in INST Wind Tunnel

Parricc hfcstaycr . ~ e ~ Iota ~ t ~ ~ ~ w ~ ~ e s t 101Chapter 8

Modeling the Droplet Contribution to the Sea-to-AirMoisture Flux

C.W. Foirall and X J. Edson 121

Chapter 9Abstracts From Poster Presentations

EL Andrcas; D. K,R'os T. Torgcrscn cr a/.; R. Ciprianoand DX. woolf�.

Chapter 10Panel Discussion: identification of Critical Rest~h TopicsS. E. Larscn . 165

BVBBLE-MEg]gygg g&l AIR EXCHANGE

Preface

The papers included in this volume were all presentedduring the Sea Grant Syrnposiurn on "Qimate and Health Implica-tions of Bubble Mediated Sea-Air Exchange" held at the Univer-sity of Connecticut, Avery Point, on 6 and 7 October 1988. Thissyniposium was one of several events scheduled to celebrate thedesignation, by the U, S. Secretary of Commerce, of the Universityof Connecticut as a fuH-fledged Sea Grant College.

The topic of this syrnposiurn was felt to be particularly ap-propriate at this time, when an improved understanding of climate,and of those fa cacao which influence global climate change, is thegoal of the National Oceanic «nd Atmospheric Administration, thefo'Leoil agency which sponsors the several dozen Sea Grant pro-grams located in the various coastal and Great Lakes states.

This scientific meeting also coincided with the formal dedi-cation, on this same campus, of the Laboratory for the Study of thePhysics and Chemistry of the Sea Surface, and of the new whitecapsimulation tank housed in this facility. The research into a varietyof bubble-mediated sea surface processes carried out in this labora-tory, whose outfitting was made possible by a grant from theUniversity of Connecticut Research Foundation, is supported bysuch funding agencies as the ONce of Naval Research and theNational Science Foundation. In the six months that have elapsedsince its dedication, scientists from France, the Netherlands,Denmark, and the United Kingdom, and from states as distant asCalifornia, have traveled to the Marine Sciences Institute at AveryPrin and participated in a series of co-operative experiments

centered on the new whitecap simulation tank. Involved in thework in this laboratory during this same interval have been Graduate Students and Post-Doctoral Scholars from Ireland, Poland, thePeople's Republic of China, and the United Kingdom, as well asfmrn the United States. This bringing together of scientists fromall over, to pool their insights, talents, and equipment in pursuit ofthe answers to pressing questions relevant to our marine envimn-merit, is, we believe, an ideal, and at the same time and efficient,way to conduct rese~h. Certainly this practical approach to thescientific endeavor is one fully consonant with the responsibility oeach Sea Grant College to foster research, education, and advisoryservices, in the fields of applied marine science.

We are particularly pleased to be able to include in this vol-ume a number of papers which axe truly multi-discipliriary innature, spanning within a single paper the range of disciplines fmrrmicrobiology/medicine to physics/meteorology, We est that thereading of these contributions may encourage others to engage inthese sorts of fruitful, cross-disciplinary, studies.

This symposiurri, and the associated poster session, benefitedimmeasurably from the dedicated efforts of aH of the staff of theConnecticut Sea Grant Office.

This volume was brought to its final form thanks in largemeasure to the work and perserverance of my co-editor, Ms. PegVan Patten, who was assisted in this effort by Ms. PatriciaBeetham.

Neither the symposium nor this volume would have beenpossible without the funding provided to the ConnecticUt SeaGrant College Program by the National Sea Grant Office, NationalOceanic and Atmospheric Administration, U. S. Department ofCommerce, via Grant NA85AA-D-SG101. Additional support forthe symposium was provided by the University of ConnecticutResearch Foundation. 7

Ed.ward C. Monahan

Avery Point, April 1989

BUBBLE-MEDI4TEO SEA-4IR EXCH4HGE

The new NMecap SimIation Tank /V an its podium in the Laboratory forthe Study of the Physics and Chemistry of the Sea Surface at the Univer-sity of Connecticut Marine Sciences Institute. Visible near the bottom ofthe tank is an array of frits used to produce a continuous plume ofbubbles as required in the series of Petit-CL USE experiments, carriedout with the support of N.S.F. and C.N.R.S. France!, to assess thecontribution of bubble-generated droplets to the sea-ai r moisture fly.Beneath the movable gantry, the hood that serves to enclose the spaceabove the tank can be seen.

BUBBLE-MEOIA TED SEA-A IR EXCHANGE

Bacteria and Other Materialsin Drops from Bursting Bubbles

Duncan C. Blanchard

Atmospheric Sciences Research CenterState University of New York at AlbanyAlbany, NY 12222

Abstract

The ubiquitous breaking waves on lakes, rivers, and at sea areprobably the most important source of air bubbles, but on a localscale the falling of precipitation into the water may also be impor-tant. Bubbles can collect bacteria as they rise through the water,and when they burst at the surface many of the bacteria areskimmed off the bubble and into the jet drops. Depending upon anumber of factors, including drop size and the distance the bubblesmove through the water, the bacterial concentration in the jet dropscan be several hundred times that in the ~ater where the bubblesburst. Film drops enriched in bacteria have also been found. Undercertain conditions, the enrichment of jet and film drops with bac-teria, viruses, and toxins may be a health hazard to people livingalong the shore.

1. Introduction

Over 2,300 years ago, Hippocrates wrote about the influence of~cather and the environment on human health. His ideas, andthose of others down through the centuries, have been discussed bySargent �982!. Much of our environment consists of the lakes,

O.C BLANCHARD

rivers, an~ and oceans that cover over 70% of the surface of the earth-F~m these waters an aerosol is produced and carried by the windsto great distances and heights.Since all natural waters contain bacteria, it would only becommon sense to believe that the concentrations of bacteria in thedroplets rising from these waters are unchanged in their journeyfrom the water into the atmosphere. Common sense, however, inthis case can be wrong- Numerous laboratory experiments haveshown that the concentration of some species of bacteria in thedroplets is hundreds of times that in the water from whence theycame. Many of us are exposed to these droplets, since our citiesand towns are usually located near rivers, lakes, or the sea. Wemust understand how bacteria are enriched in the droplets, for theymay occasionally pose a hazard to human health.

This paper, a review of the work on the problem, begins with ashort account of the experiments that showed the importance ofbursting bubbles in producing an aerosol from natural bodies ofwater. Much of this review, especially that dealing with the bac-teria, dews heavily from one of my earlier papers Blanchard,1983!.

2. The Production of Jet and Filtn Drops

Over 30 years ago, marine meteorologists studying the for-mation of rain-drops in marine clouds Woodcock and Blanchard,1955! began to investigate the mechanism by which giant conden-sation nuclei are produced at the surface of the sea Moore andMason, 1954; Blanchard and Woodcock, 1957!. This work stillgoes on and, consequently, most of what we know about aerosolproduction from natural waters is only for seawater. Relativelylittle work has been done with freshwater Monahan and Zietlow,1969; Monahan, 1969!.

Most of the aerosol generated from natural waters is in theform of jet and film drops from the bursting of air bubbles. Thebubbles can be produced by a variety of ways: by rain and snowfalling into the water Fig. 1!, by waterfalls and rapids, and by thetmaking of wind-induced waves Blanchard and Woodcock,195~! 'Ae coverage of the sea by breaking waves or whitecaps

BUBBLE-MEOIA TED S&l-4IR EVCHANGE

increases with more than the third power of the wind speed Monahan and O'Muircheartaigh, 1980!, and on average the worldocean whitecap coverage is about 1 per cent. Bubble spectra withinwhitecaps at sea have not been obtained, but in a laboratory modelbreaking wave, Cipriano and Blanchard �981! found bubbles from<1 mirn to about 10 mm diameter. The bubble flux to the surfacewas about 2 x 10' m' s'.

When a bubble bursts at the surface of the water Fig. 2!, sameof its surface free energy is converted into kinetic energy of a jet ofwater that rises rapidly from the bubble cavity {Blanchard, 1963;MacIntyre, 1972!. The jet, first postulated by Stuhlman {1932! andexperimentally confirmed by Woodcock et aI. {1953!, breaks upinto 1 to 10 drops, the more the smaller the bubble. The first jetdrop, usually called the top drop, is about one-tenth the bubblediameter, but the lower drops can be larger or smaller. Most arelarger. The maximum in the ejection height of the top drop isnearly 20 cm for drops from bubbles of 2 mrn diameter. No jetdrops are produced from bubbles � or 8 mm.

Fig. 1. Foor ways to produce bub6les in the sea The first three have beenoxpenrnentally guoven, buf the fourth is still only conjecture.

Fig. Z. Jet dropsmoving rap'dfyppwald from the j etof a collapsing t-rnmdiameter bubb!e.Some of the dropsmay have escapedfiom the fiekf of view.

Bursting bubbles also produce film drops that arise from thedisintegration of the thin film of water that separates the air in thebubble from the atmosphere Fig. 3!. Unlike jet drops, film dropsare hard to measure, most being � pm diameter Cipriano et al.,1987}, but sorte are >100 pm Resch et al., 1986!. It has long beenthought that film dmp production increased steadily with bubblesize, but Blanchard and Syzdek �988! recently reported a maxi-mum of about 75 film drops fmm bubbles of 2 to 2.5 rnm diameter.

3. The Knrichrrl.nt ef Sacteria in Jet and Film Drops

Horrocks �907! appears to have been the first to show thatbacteria could be ejected into the atmosphere by bubbles burstingat the surface of bacterial-laden waters. Nearly half a centurypassed before Woodcock �955!, unaware of the earlier work,suggested that jet drops could carry bacteria. In 1970, Blanchardand Syzdek found thai the concentration of the bacterium, S.morcescens, in jet drops could far exceed that in the water wherethe bubbles burst, A measure of the bacterial enrichment is givenby the enrichment factor EF, the ratio of the concentration of bac-teria in the drop at the moment of production before evaporationoccurs! to the concentration in the water from which the drop wasproduced. High EFs depend on 1! a bubble arriving at the surfacewith many bacteria attached to it, and 2! by having these bacteriaskimmed off the surface of the bubble and into the jet drops.

O.C. BLhN ttevu

Fig 4 Schematic outline of anexperiment fo obtain the numberof bacteria carried by the top jetdip or the top two or more! as afunction of the distance the bubblehas risen through the bacterialsuspension.

dramatically from near-neg!igible values to about 400 as the BRDincreased to only 3 cm, but then increased more slowly ta 600 at a3RD of 10 crn. Ihe marked decrease in the rate of increase of EFat about 3 cm appears to be because the bubble changes from aAuid to a solid sphere. During the first few centimeters of rise, abubble of this size moves as a fluid sphere clean bubble! with aclean surface Taiesco and Blanchard, 1979!. In addition tobacteria, it collects surface-active material during its ascent. Themobility of the bubble surface, therefore, decreases until enoughsurfactant has accumulated to render the surface immobile Clift etal., l978!. From then on it rises as a solid sphere dirty bubble!.The rate of collection of bacteria by a bubble in the early period ofrise, when its surface is clean, is much higher than the rate later on,when its surface is dirty %eber et al., 1983!,

Very few bacteria make contact with the bubble by diffusion orinertial fart:es. Most do so by interception, a collection mechanism

at depends on the finite size of the bacteria %'cber, $98 i!. Sinceinertial effects are negligible, bacteria follow the fluid strearnlinesaround the bube u ble. Due to their finite size, however, those bacterifollowin sg trearnlines very close to the bubble make contact and

se teria

st@ scavenged.

Fig. 5. The enrr'chmentfactor EF} for bacteriain the topjet drop as afvre6on of the distance

the bubbie rises througha bacteria] suspension ofS. marcescens in poofwater. Three drfferentexperiment are shown. From Blanchard et al.198f!

Drop size.Having seen how the EF increases with BRD for a given

bubble size, we' ll now look at what happens when the BRD is heldconstant and bubble size is varied. This is how Blanchard and

Syzdek �970! did their first bacterial experiments. Their results,presented in a graph of the tap drop EF as a function of drop size or, if multiplied by 10, the bubble size, since the top drop is aboutone tenth the bubble size!, showed that the EF reached a maximumfor drops of about 50 pm diameter. The maximum was also foundby Bezdek and Carlucci �972! with seawater suspensions of S.mannorubra, by Blanchard and Syzdek �972!, again using dis-ti!led water suspensions of S. marr.escens, and by Hejkal et al.�9SO! with suspensions of a number of species of marine bacteriain NaCl solutions.

In the center of Fig. 6 is a graph showing a composite of all thedata. The surrounding sketches suggest a possible explanation forthe EF maximum. Let us assume, following MacIntyre �972!, thatthe microlayer thickness, kR, skimmed off the surface of the burst-ing bubble to produce the jet drops, is proportional to the bubbleradius R. For the smallest bubbles it is probable that kR is less thanthe size of the bacteria attached to the bubble. Most of these bacte-ria, therefore, will be left behind when the microlayer is trans-formed into jet drops. In fact, the EF might be � for drops 20p.m from bubbles less than about 200 pm diameter!.

As bubble size increases from small to intermediate, an optimal

sol, but Pierce �986! believes they might be and has experimentsunderway to detect it.

Legionella pnemnophila, the cause of the respiratory diseaselegionellosis, commonly known as the Legionnaires' disease, is apa.thogen that can cause disease after being ejected into the atmos-phere by bubbling and splashing processes. Although it has beenisolated from many lakes Wiermans er al., 1979!, the outbreak ofthe disease appears to be singularly associated with the aerosolgenerated by a.ir conditioning cooling towers Dondero et al.,1980; Friedman er al., 1987!. L. pnesunophi la has been foundenriched in the foam in cooling towers Colbourne et al., 1987!and in jet drops from bursting bubbles Parker et al., 1983!.

Gruft et aL �975!, aware that Mycobacteriurn ietracell~lareinfections in humans occur more frequently along the southeasternshares of the United States, suggested that the source of the bacte-rium was in estuaries and other coastal waters. Since their experi-rnents showed that the bacterium could be enriched in jet drops,they postulated that the infectious aerosol was produced by the seaand carried inland by onshore winds. Later, Parker er al. �983!indicated that the source of infection might also be an aerosol pro-duced by fresh and brackish waters.

Viruses deliberately mixed with seawater in the surf zone Baylor er al., 1977! were found to be highly enriched in thebubble-produced aerosol that passed over the shore. This couldhave public health implications since viruses exist in the sewagethat is often dumped into the coastal regions of the sea.

Some pollutants produced on land are carried to the sea byrivers, where a portion of them are ejected into the atmosphere injet and film drops to be carried by onshore winds back to the landagain. Fraizier er aL �977! have used this scenario to explain thesharp falloff of radioactivity in plants with distance inland from theshore along sections of the French coast.In certain regions along the shore in Italy, pine trees have beenaffected by a bubble-produced aerosol. Gellini er al. �985! believethat surfactants carried to the sea by rivers become airborne aboardjet and film drops. During onshore winds the drops stick to thepine needles where a synergistic reaction between the sodium

towers containing the bacterium responsible for Legionnaires'disease. Under some conditions, trees and other vegetation alongthe shore can be damaged by pollutants that are scavenged fromthe water by bubbles and then ejected into the air on an aerosol thatis carried inland by the wind.

Acknowgedgrnents

I am grateful to my colleague, Lawrence Syzdek, whose exper-tise and persistence over the years has made much of this workpossible. The present work was supported by the National ScienceFoundation under ATM-8514211,

References

Art, H. W., F. H. Borrnann, G. K. Voigt, and G. M. Woodwell, 1974: Barrierisland forest ecosystem. role of meteorologic inputs. Science. !84:60-62.

Asai, S., J. J. Krzanowski, W. H. Anderson, D. F. Martin. J. B. Poison, R. F.Lockey, S. C. Bukantz, and A. Szentivanyi, 1982: Effects of the toxin ofred tide, PtychoChscus brevis, on canine tracheal smooth muscle: a possiblenew asthma-triggering mechanism. I. AQcrgy Cion. Immunol. 69:418-428.

Baylor, E. R., M. B. Baylor, D. C. Blanchard, L. D. Syzdek, and C. Appel, 1977:Virus transfer from surf to wind. Science. k9$:575-580.

Bezdek, H, F�and A. F. Carlucci, 1972: Surface concentration of marinebacteria. Litnnol Oceanogr. 17'.566-569.

Blanchard, D. C., 1963: The electrificadon of the atmosphere by particles frombubbles in the setL Prog. Oceanog. 1.71-202.

Blandraid, D. C., 1983: The lrroduction, distribution, and bacterial enrichmentof the sea-salt aerosol. 1983: pp. 407454. In P. S. Liss and %. G. N.Slinn eds.!, Thc Air-Sca Exchange of Gases and Particles. Reidel Pub. Co.

Blandrard, D. C. and L. D. Syzdek, 1970: Mechanism for the water-to-airtransfer and concentration of bacteria. Science. 170..626-628.

Blanchard, D. C. and L. D. Syzdek, 1972: Concentration of bacteria in jetdrops from bursting bubbles. J. Geophys, Res. 77:5087-5099.

Blanchard, D.C. and L. D. Syzdek, 1975: Electrostatic collection of jet andfilm drops. Lirnnol. Oceatrogr. 20:762-774.

BLrrrchard, D.C., and L. D. Syzdek, 1977: Production of air bubbles of aspecified size. Chirr. Engr. Sci. 32:1109-1112.

Blanchard, D.C., and L. D. Syzdek, 1978: Seven problems in bubble and jetdrop resrmches. Lb neo , Oceanogr. 23:389400.

Blanchrrrd. D.C., and L. D. Syzdek. 1982: Water-to-air transfer and enrich-ment of bacteria in drops from burstmg bubbles. Apped Environ.

O,G. HLANC nAttv

43:1001-1005.B~~ D C and L.D. Syzdek, 1988: Filrn4rop production as a function. J. Geophys Res. 93;3649-3654.B~~ D C L. D. Syzdek, and M. E. Weber, 1981; Bobble scavenging of

;n freshwater quickly produces bacterial enrichment in airborneL;trt50 , Oceanogr. 26:961-964.Bl ~~ D C., and A. H. Woodcock. 1957: Bubble formation and rnodifi-the sea and its meteorological significance. Te lus. 9: 145-158.

Burger S. R., and J. W- Bennett, 1985: Droplet enrichment factors of pig-mented and nonpigmented Scrratia marcescens: possible selective functionfor prodig josin. Apped. Environ. Mi crobioL 50: 487490.CNbell; y J., 8, Kennedy, and M. A. Levin, 1976: Pseudomonas aerrrginosa-fecal coll fortIt relationships in estuarine and fresh recreational waters. J.gl graf pofl. Control Fed. 48:367-376.Cipriano, R. J-, 1979: Bubble and aerosol spectra produced by a laboratory'breaking wave.' Ph.D. thesis, Staae University of New York at Albany.Cipriano, R. J. and D. C. Blanchard, 1981: Bubble and aerosol spectra producedby a laboratory 'breaking wave.' J. Geophys. Res. 86:8085-8092.Cipriano, R. J., E- C- Monahan, P. A. Bowyer, and D. K. Woolf, 1987: MarinecondenstLtion nucleus generation iaferred from whitecap simulation tankresults. 1, Gcophys. Res. 92:6569-6576.Clift, R., J. R. Grace, and M, E. Weber, 1978: Bubbles, Drops, and Particles.Academic Press. 380 pp.Colhougae, J. S., P, J. Dennis, J. V. Lee, and M. R, Bailey, 1987: Legiortrtaires'diraase: reduction in risks associated with foaming in evaporative coolirtgtowers. Lancet. 1 ¹8534!:684,Crow, S. A., D. 6. Ahern, W. L. Coot. and A.W. Bourquin, 1975: Densitiesof bacteria and fungi in coastal surface filtns as determined by a membrane-adsori~ prteolt0e. Limno' Oceanogr. 20:644-646.

Dehlbwck, 8. ~ M. Hermansson, S. KjeUeberg, and B. Norkrans, 1981: Thehydropbcbicity of bacteria - an import;mt factor in their initial adhesion atthe air-water interface. Arch. Microbiol. 128:267-270.

Dee$ero, T J., Jr., R. C. Rendtorff.G. F, Mallison, R. M. Weeks, J. S. Levy,E.W. Wong, and W. Schaffner, 1980: An outbreak of Legionnaires'Disease associated with a contaminated air~tiorting cooling tower. N.angl. J. hfed. 302:365-370.

Ferguson.R. L., and A. V. palumbo, 1979: Distribution of suspended bac.~in neritic waters South of Leag Island during stratified conditions. Li~.Oceanogr, 24;697-705.

Fliermans, C. B., W. B. Cherry, L. H. Orri~en and L. Thacker, 1979: <solati~of Mgio~Ba pnc unrophila from nonepidemic-related aquatic habitats-Appld. Eevt'r. Microb. 37;1239-1242.

Fraizier, A., M, Ma!+w, J, C, Guary, 1977: Recherches preliminaites sur Ierole des aerosols dans le transfert de certains radioelements du milieu mariaau milieu terrestrc. J. Rich. Aoeos. 1 l:49~.

Friedman, S., K. Spitalny, J. Barbaree, Y. Faur, and R. McKinney, l987:Pontiac fever outbreak associated with a cooling tower. Amer. I. Publi cHealth. 77:568-572.

Gellini, R., F. Pantani, P. Grossoni, F. Bussotti. E. Barbolani, and C. Rinallo,1985: Further investigation on the causes of disorder of the coastalvegetation in the part of San Rossore central Italy!. Eur. J, For. Path.I5: 145-157

Gruft, H., J. Katz, and D. C. Blanchard, 1975: Postulated source of Myco-bacteriumintracellulare Battey! infection. Amer. I. Epidemiology.102: 311-318.

Hejkal, T. W., P. A. LaRock, and J. W. Winchester, 1980: Water-to-air frac-tionation of bacteria. Apped. Envir. Microbiology. 39:335-338.

Horrocks, W.H., 1907: Experiments made to determine the conditions underwhich "specific" bacteria derived from sewage may be present in the air ofventilating pipes, drains, inspection chambers and sewers. Proc. Roy. Soc.London 79B.255-266,

Maclntyre, F., 196$: Bubbles: a boundary-layer "microtome" for micron-thicksamples of a liquid surface. I. Phys. Chem. 72:589-592.

Maclntyre, F., 1972: Flow pattemsin breakingbubbles. J. Geophys. Res.77:52 1 l -5228.

Monahan, E.C., 1969: Fresh water whitecaps. J. Atmos. Sci. 26:1026-1029.Monahan, E. C., and I. O'Muircheartaigh, 1980: Optimal power-law descrip-

tion of oceanic whitecap coverage de pendence on wind speed. J, Phys.Oceanagr. 10.'2094-2099.

Monahan, E.C., and C. R. Zietlow, 1969: Laboratory comparisons of fresh-water and salt-water whitecaps. J. Geophys. Res. 74:6961-6966.

Moore, D. J., and B. J. Mason, 1954: The concentration, size distribution andprodoction rate of large salt nuclei over the oceans. Q.J. Roy. Met. Soc.80;583-590.

Parker, B. C., M. A. Ford, H. Gruft, and J. O. FaHunham III., 1983: Epidemiol-ogy of infection by nontuberculous mycobacteria. Am. Rev. Respir. Di s.I28:652-656.

Pierce, R. H., 1986: Red tide Ptychodiscus brevis! toxin aerosols: a review.Toxi con. 24:955-965.

Resch, F. J., J. S. Darrozes, and G. M. Afeti, 1986: Marine liquid aerosolproduction from bursting of air bubbles. I. Geophys. Res. 91: 1019-1029.

Sargent, F.S. II, 1982: Hippocratic Heritage: A Hi story of Ideas about Weatherand Hrunan Health. Pergamon Press. 581 pp.

Smith, B. J., 1968: A study of the mechanism by which biaaerosols are gener-ated when liquids containing microorganisms are aerated. Ph.D. thesis,Georgia Institute of Technology.

StuMman, O., 1932: The mechanics of effervescence. Physi cs 2:457-466.Syzdek, L. 13., 1985: Influence of Serrati a rnarcescens pigmentation an cell

concentrations in aerosols produced by bursting bubbles. Apptd. Environ.MicmbioL 49:173-178.

15

D.C, 8LANGHAHU

R and D. C, Blanchard, 1979: Dynamics of small bubble motion and~ng in freshwater. J. Rech. Atmos. 13: 215-2%.

198]. CoHision efficiencies for small particles with a sphericalat intermediate Reynolds numbers. J. Sepnration Process Tech-

~fggy 2. 29 33W~ ~ F., D.C. Blanchard, and L. D. Syzdek, 1983. The mechanism of~�enging of water-borne bacteria by a rising bubble. Limnot. Oceanogr-2g 101-105.~;Iutttts, R, P. 1973. Biosynthesis of Prodigiosin, a secondary metabolite of@err+I'a Irrarcescens. Appld. Microb. 25: 3% 832.W~~~k A. H., 1948: Note concerning human respiratory irritation associ-ated with high concentrations of plankton and mass mortality of maineorptnLsms, J. hfarim Res. 7: 56-62.

W~~ock, A, H., 1955'- Bursting bubbles and air pollution, Sewage andjrtdgstrial Wastes. 27: 1189-1192.

WoobAxk, A- H-, 19>2: Stnaller salt particles in oceanic air and bubblebehavior in the sea. J. Geophys. Res. 77: 5316-5321.

woodcock, A. H., and D. C. Blanchard, 1955: Tests of the salt-nuclei hypothe-sis of rain formation. TeHtrs. 7; 437-448.

.g~xk, A.H., C.F. Kientzler, A.B. Arons, and D,C. Blanchard, 1953: Giantcondensation nuclei from bursting bubbles. Nature l72: l 144.

16

enric inchment factors EF! were significantly different P 0.05,SMent's T-test! between some representatives of a single species i.e. M avium strains 13S and 18S aiid M. scrofulaceum strains14S ~d ]6S, Table 1!. However, in spite of the significance of ~

ferences there was great variability in the enrichment factors Table 1! and we sought to identify the sources of that variabihtjj-Iy~~g of Medium Composition on Aerosotization.

~e enrichment factor of cells of M. aviurn strain 13S grown ~Middlebrow 789 broth containing 1% vol/vol! glycerol and 10% viol! OADC enrichment EF = 230 4 160! was significantlylower P < 0.05, Student's T-test! than the value for cells grown iAtilter-sterilize James River water EF = 1,100+ 470!. This differ-ence was observed in spite of the fact that cells of both strains meresuspended in a portion of the same sample of James River waterfof aerosollzabon.

Epee'r of Ce0 Ag gregafion on Aerosolization.Earlier data had suggested that cell aggregation contributed ~

variation in enrichment factors Parker, et aL, 1983!. To determirawwhether aggregation of cells coiild influence measurement ofdroplet enrichrnerit, two approaches were chosen. Table 2 showsthe results of the effect of spreading ejected droplets following

20

bubble impact on the mediurri. The values for enrichment factorsof M. aviwn strain W227 and M. scrofidaceum strain 14S weresigniflicantly higher P 0.05, Student's T-test! when the dropletswere spread over the surface of the medium Table 2!. Though notstatistically significant, EF values for both M. serofidaceum strains16S and W220 were higher when droplets were spread Table 2!.

The second approach exarniaed the effect of 3-minute-vortex-ing a culture of M. avium strain 13S before suspending cells inJames River water for aerosolization. The EF value for an un-treated suspension of M. aviunt strain 13S was 10,500 + 2,900. TheEF value for the same culture vortexed 3 minutes before suspen-sion in James River water was 510 k 130. Vortexing, to produce a

less aggregated suspension, led to a significant drop in the enrich-rnent factor P 0.05, Student's T-test!.

Qec j of Sa6nity on Aerosolization.Because MAIS organisms are recovered Falkinharii, et al.,

1980! and can grow George, et al., 1980! in natural waters ofmoderate salinity i.e. 1 and 2% sea salts!, the possibility that saltconcentration influenced aerosolization of mycobacteria wasinvestigated. The data in Table 3 illustrate that mean enrichmentfactor values changed as the coeicentratioa of artificial sea saltsadded to James River water increa.:M. At low sea salts concentra-tions i.e. 1-3%!, the EF values for M. avium strain 18S rosesignificantly P < .05, Student's T-test!, while at the highest

21

Cell aggregation also affected enrichment in droplets Table 2!.!ncreased EF values for droplets which were spread probablyreflect the fact that droplets contained aggregated mycobacterialcells and spreading disrupted the aggregates. Likewise, the fall inEF values following vortexing was probably due to the disruptionof aggregates, so fewer aggregates were present in ejected dropletsyielding high cell numbers when spread. Because the extent of ag-gregation of mycobacterial cells changes with their growth stage i.e. late log and early stationary phase cultures are heavily aggre-gated; Falkinham, unpubHshed!, the growth stage of M. avium andM. scrofulaceum cells is likely to influence enrichment in ejecteddroplets as well.

Although low concentrations of artificial sea salts increased theEF values for rnycobacterial strains, this was accompanied bychanges in the volume of ejected droplets Table 3!. Because theincreases in EF values were offset by decTeases in droplet volumeejected, measured within the same time period for James Riverwater with and without 1% artificial sea salts, approximately thesame number of rnycobacterial cells were transferred except at thehighest concentration tested i.e. 4%, Table 3!.

The data demonstrate that the differences in droplet enrichmentvalues of M. aviv' and M. scrofulaceum strains are not associated

23

epidemiologicai markers for clinical and environmental isolates of Myco-bacterirun avium, M. inrracegulare and hf. scrofutaceam. J. Itifect Dis.153: 325-331.

Parker, B.C., M.A. Ford, H. Gruft and 3.0. Falkinham, Hl, l983: Epidemiologyof infection by nontuberculous mycobacteria. IV, Preferential aerosolimtionof hf ycobacteri um intracellularc from natural waters. Am. Rev. Respir. Dis.128: 652-656.

Weber, M.E., D.C. Blanchard and L.D. Syzdek, 1983: The mechanism ofscavenging of waterborne bacteria by a rising bubble. Li mnol. Oceamgr.28:101-105.

Wendt, S.L., K.L. George, B.C. Parker, H. Gru.R and J.O. Falkinham, $0, 1980:Epidemiology of infection by nontubercttlous mycobacteria. Ill. Isolation ofpotentially pathogenic mycobacteria from aerosols. Am. Rev. Respir. Dis.122: 259-263

Wolinsky, E., 1979: Nontuberculous mycobacteria and associated diseases. Am,Rev. Respir. &is. 119:107-159.

25

While investigating a red tide episode at Venice, Florida, in1947, Woodcock �948! surmised the relationship betweenrespiratory irritation and aerosolized dinoflagellate products. Atthat time he suggested that the irritant-containing aerosols wereprobably formed by bursting bubbles from breaking waves withinthe dinoflagellate bloom area.

Since this early insight, very little new information about bio-synthesized marine toxin aerosols has been provided. Steidingerand Joyce, 1973; Baden et al., 1982; Pierce, 1986!. Recent ad-vances in knowledge about red tide toxins and improved analyticalmethods have allowed a more in-depth investigation of this phe-nomenon. The chemical structures of two of the primary neurotox-ins biosynthesized by P. brevis were elucidated by Lin et aL�981! and Shimizu �982!. These compounds consist of a hetero-cyclic oxygenated fused ring system, with a terminal aldehyde oralcohol, designated brevetoxin-B BTX-8! and GB-3, respectively.To date, as many as eight chemical toxins have been reportedunder various nomenclatures Padilla et aL, 1975; Baden andMende, 1982; Chou et al., 1985; Nakanishi, 1985; Pierce er a!.,l985; Shirnizu et al., 1986!. Poli er al., 1986, proposed a. revisednomenclature to help eliminate confusion fmm the various systemsin use. The chemical structure of six of these toxins identified inour samples is given in Figure 1.

Figure f. Chemical structures for P. brevis toxins

29

~stallized toxin extracts. Sufficient quantities of all othertoxins were not available to prepare standard HPLC curves forquantitative analysis. For this report, all toxins, other than PbTx-2,+ere assigns uv response factors identical to pbTx-3 for quanti6-cation. Toxins are listed in the figures in order of elution duringHpLC ana}ysis; PbTx-3, 6, -2, -7, -5, -1. The other two possibletoxins, PbTx-4 and PbTx-8, have not been identified in ourggnples,

gesults and Discu<mion

Fin'kt SarnptesResults of toxin analysis for the Florida red tides Figure 4!

show PbTx-2 to be the most abundant toxin average of 200 Pg/I!.about 4 to 8 times mare abundant than PbTx-3 and PbTx-5 av~-aging 30 p.g/l!, resulting in a ratio of PbTx-2/3 near 7. The aerosolsamples revealed similar patterns except that the relative amoont ofPbTx-3 ui~scd with respect to PbTx-2 �/3 ratio = ca. 3!, sug-gesting an enrichment of PbTx-3 in aerosol relative to PbTx-2.

Samples from the North Carolina bloom contained the samethree tnost abundant toxins Figure 5!; however, PbTx-2 was m~abundant relative to PbTx-3 in comparison to the Florida sampleswith a PbTx-2/-3 ratio of 8 to 12. The aerosol sample also exhiWitcd an increase of PbTx-3 over PbTx-2 as compared to the water

Figure 4. Tcvrinconcentration iaFIarida red tideMooms, hlay4987, water ancfaerosol sa~4ea.

Fgure 7, Toxinennchmenf inbubM~unfucedfoam and p!t dIUps P. brevis cujt0re,25 x 70 ' cells',30-minute samplern bobber chamberat W'C,94'/

entire mass balance, however, it becomes evident that an increasein PbTx-3 was also observed in the remaining culture after removalof toxins by bubbling, which is contrary to selective removal ofPbTx-3 by bubbles. Also, the find mass balance showed a de-crease in PbTx-2 in proportion to the increase in PbTx-3, providingstrong evidence for the conversion of PbTx-2 to PbTx-3, as wasindicated in the toxin degradation stud.y above.

The other two toxins found in sufficient quantity to evaluate

37

Figure S. Toxinenrichment ibubble-pmtuc adaerosol P. brevisculture,40 x 10'ce/i@1! 4-hr samplein lab aerosol

chamber ei th

10 kt air velocity,24'C, 34'/oe

2. Toxins were enriched in surface foam at the sea-air inter-

face by bubble-mediated transport. For PbTx-2 and PbTx-3, foamenrichment factors ranged from 2.4 to 3.6 times the concentrationsin the original sample.

3. Toxins were highly enriched in bubble-produced jet dropsand aerosol. For PbTx-2 and PbTx-3, enrichment factors rangedfrom 35 to 45 for jet drops and from 20 to 50 for aerosol.

4. Selective enrichment of one toxin over another was not

verified; however, the apparent enrichment of PbTx-3 over PbTx-2 was found to result from the conversion of PbTx-2 to PbTx-3

during the aerosolization process.

5. Larger quantities of xed tide-containing seawater and/or labcultures are needed to investigate bubble-mediated enrichment ofother P. brevis toxins as well as bacteria and bacteria toxins asso-

ciated with red tides.

39

Gershey, R.M, 1983: Characterization of seawater organic matter carried bybubble-gerierated ~msols. Limnol. Oceanogr. M 309.

Hoffman, E J. and R.A. Duce. 1974: The organic carbon content of marineaerosols collected in Bermuda. J. Geophys. Res. 79:4474.

1Qentzler, CZ., A, Arons, D.C. Blanchard and A.H. Woodcock. 1954: Photo-graphic invesogation of the projection of droplets by bubbles bimsting at awatn surface. Telhts 6:1.

Lin, Y.Y., M. Risk, S.M, Ray, D. van Engen, J. Clardy, J. Golzk, JL. James andK. Nakanishi. 1981: Brevetoxin B from the red tide dirtoflagellate,Ptychodiscus brevis Gymnodini un breve!. J. A.m. CheIn. Soc. 103:6773.

Maclntyre, F. 1974: The top millimeter of the ocean. Sci. Am. 230:62.Monahan, E.C. 1986: The ocean as a source for atmospheric particles. ln. P.

Buat-Menard ed.!, The Role of Air-Sea Exchange in Geochemical Cycling.D. Reidel, Hingharn, MA, pp, 129-163.

Nakanishi, K. 1985: Thechemistryofbrevetoxins: areview. In: S.Ellis ed.!Brevetoxins: Chemistry and Pharmacology of "Red Tide" Toxins fromPtychodiscus brevis formerly Gymnodinimn breve!. Toxicon 23:473-479.

Narita, H., S. Matsubara, N. Miwa, S. Akahane, M. Muralrami, T. Goto, M.Nara, T. Noguchi, T. Saito, Y. Shida and K. Hashimoto. 1987: Vibrio aigi-noiytims, a TTX-producing bacterium isolated from the starfish As-tropecten polyocanthus. ¹ppon Sirisan Gakrraishi. 53�!:617Ml.

Padilla, G.M., Y.S. Kim and DK Martin. 1975: Separation and analysis oftoxins isolated from a red tide sample of Gymnodinium breve. In: V.R. LoCicero ed.!, Proceedings of the First International Conference of ToxicDinoflagellate Blooms, Wakefield, Ma~machusetts Science and TechnologyFoundation, p. 299.

Pierce, R.H. 1986: Red tide Prychodiscus brevis! toxin aerosols: a review.To@icon 24:955-965.

Pierce, R.H. 1987: Cooperative scientific effort identifies red tide toxin.Environs X�!:7-12.

Pierce, R.H.. R,C. Brown and J3t. Kucklick. 1985: Analysis of Ptychodiscssbrcvts toxins by reverse phase HPLC. In: D.M. Anderson, A.W. Whiteand D.G. Baden eds.!, Toxic Dinoflagetlates, Eisevier, NY, p.309.

Pierce, R.H., R.C. Brown and KX. Hardrrian. 1987: Mote Marine LaboratoryRed Tide Program, 1986-1987. Final Report to the H Dept. of NaturalResides, Contract K'4149. Septrmber ! I, 1987. 15 pp.

Poli, M.A., TJ. Mende and D.G. Baden. 1986. Brevetoxins, unique activatorsof voltage-sensitive Na chanriels bind to specific sites in rmt brain synap-tosornes. Molec. Phrrmac. 30:129-135.

Shimizu, Y. 1978'. Dinoflagellate toxins. In: H', Scheuer ed.!.MarineNatural Products, Chemical and Biological Perspectives, Academic Press,NY, p. 1.

Shirnizu, Y. 1982: Recent progress in Marine toxin retie mh. PNre AppLChem. 54: 1973-1980.

41

Shirnizu.Y., KN. Chou.H. Banda, G. van Duyne and J.C. Clardy. 1986-Structtle of Brevetoxin A GB-1 Toxin!, the most potent toxin in theRortidrt rtrd @de organim G. breve Ptychodiscus brevis!. J. Am. Clretrr-Soe. 1OI:515-516.

Simidu.V..T. Roguchi, D-F. Hwang. Y. Shida and K. Hashimoto. 1987:Mane bacteria which pmduce Letrodotoxin. Appl. Eevirorl. Micro.539!:1714-1715.

Steidinyr, K.A. 1983: A reevaluation of toxin dinoflagellate biology andecology. Prig. PAycoL Res. 2: 148.

Steidinger, K.A. and KW Joyce. 1973. Florida red tides. Rorlh ~PrltrrMof Natrrrrk Resources Ettucational Series No. 17.

Tester, P5'., P3L Fow ler snd J.T. Turner. 1989: Gulfstxearn trmnsport of thetoxrc r0d tide dinofhrgeHate, Ptychodiscrrs brevis, from Rorida to NorthCarolina. ln. -E.M. Copper, E J. Carpenter and V.M. Bricelj eds.!,Ho'mfNytoploehoa Bhoorrl: Cartses and Impacts of Recrv rent Broerr Tuk's ~Other Unusrrul Blooms, Springer Verlag, Berlin gn press!.

WoodccrJc, A.H. 1948: Note coezrning human respiaexy irritation associ3¹'<with high concenlnrtions of plankton and mass mortality of marine xgrlr-isms. Sears 1'ousia.J. Mtmr. Res. $6.

Woodcock, A.H., C. Kientzler, A. Aronus and D. Blanchard. 1953: Giant coderrsatr'on nuclei Ace burlting bubbles. Nature 172:1144.

Woolf. D.K., P.A. Bowyer, and E.C. Morurhan, 1987: Discriminanng Be~m~the «lrn Dlops and Jet Drops Produced by a Simulabxf Whitecap, J.Gcaphys. Res., 92, 5 l42-5150,

From the Laboratory Tankto the Global Ocean

Edward C. MonahanMaririe Sciences InstituteUniversity of ConnecticutAvery Point, Groton, CT 06340

Abstract

A series of models have been developed to predict the aerosolflux up from the sea surface, the bubble flux up to the sea surface,and the rate of gas transfer through the sea surface. In each case,the process is described explicitly in terrris of oceanic whitecapcoverage. The laboratory measurements of the contribution of anindividual whitecap are combined with assessments of global orregional oceanic whitecap coverage to make the desired estimatesof the fluxes of droplets from, bubbles to, and gases through, theocean surface. While the utility of these models has already beendemonstrated, the value of these models, and of this modelingapproach, will be enhanced if answers can be obtained for the 24questions set out in the text.

l. Introduction

When we began our study of sea spray several decades ago,with an initial investigation of the role of spray droplets in thedownward transport of forward momentum Monahan, l968!, wequickly learned that, under the high wind conditions which were ofthe most geophysical interest, it was extremely difficult to measurethe concentration of spray droplets in the lowest meter over the

43

measure in the field the "crop" of spray droplets injected into theair as a consequence of the decay of a specific whitecap, we re-sorted to producing "captive" transient whitecaps in hooded white-cap simulation tanks and counting the aerosol particles thatappeared in the air within the hoods as a result of the decay of indi-vidual whitecaps Monahan, et al., 1982; l986}. A further refine-ment involved the recasting of this spray generation model in termsof the rate of reduction of the volume of the aerated plume associ-ated with each whitecap, which resulted in a conceptually moresatisfying model than the one relating aerosol flux to the rate ofdecay of the surface area of a whitecap Monahan, 1986; 1988!.

It is insightful to review these models, arid the models beingformulated to define the influence of whitecaps on the rates of sea-air gas exchange e.g., Monahan and Spi0ane, 1984!, and to makeexplicit the assumptions incorporated in, or "hidden within," theresulting mathematical expressions. In this fashion, we hope tomake clear some of the critical questions stiR to be answered aboutthese several air-sea exchange processes.

But before we begin our review of the models themselves, weneed to consider what is meant by the term, "fraction of the seasurface covered by whitecaps," and how this quantity may berelated to increases in sea surface rrucrowave emissivity, a propertyof the ocean that can be routinely, and remotely, monitored byexisting and proposed satellite systems.

2. Measuriag Oceanic %'hitecap Coverage

By projecting oblique 35mm photographic images of the seasurface and analyzing them by the standard "manual" technique Monahan, 1969!, it has been possible to assess the fraction of thesea surface covered by mature whitecaps Stage B of Figure 1!, aswell as by active whitecaps Stage A! and aerated spilling waves.As a consequence of their greater horizontal extent, and relativelylonger duration, the fraction of the sea surface that is at any instantcovered by Stage B whitecaps far exceeds the fraction of the seasurface covered by the higher albedo Stage A whitecaps.

This is demonstrated by the curves plotted on Figure 2, where

HYPOTHETiCAL EVOLUTION OF BUBBLE PLUME/CLOUD

OPT/GAL/PMYSICAL MODEL

~ 4

lp ~atvaaLE POENATTIPNVIA Ala PIPIII~lr F.EA s!IFTFPOElLPEDP Ie .

DL

'DF'I 5t!D

15'

e HSSLE papawFttaf FQtNRg DTSPILLHID OIT Pl.uttauavs5 alaafwd wa rEltaCPO ~ I . ~ . ~ffv

AC ' !E

!1rt T EPE

$5ID

5 AGHDROUHD SIISSLIE PTHPINUAI.OHV llrTER IAOST ~ CNHITI Dt AL PHSAH SIC NAVE SHE%acl LSEDDI 5 Ol ~ 5 5".

le'

105E< T

I 5' le 55 'IXI tae Nete DMINEE IIACNVI. St Hat

F4~ ~ 4+~< wftN~ 4+ay and W%9e cloud evolutfon. Candidate~+%p+~ fot'teach tt~ ortebbk pnme devekprmnt shown cmkeft aidesof paoek SpMme7 80 te~~ed fern Rgure 9.

'L 92 5 I I1

ELIAFICE5IGHATVtlttwl lltplrlOt TLCTA ~ I T

ltaalGLOW-f'LF tlIIDH

I OVF Hfatkl FIPHF ILV Oa VIP F 0

C AIVHA5

'IHAFACtFEATUIIT5

TCCIIHFFUPFFOR HF VO TE

~ PTKALDETIC'IUH

FHOH AAPvaDF IHCITIDULI

~ Uael I cl OJP5

~ UaeLP CIDHCENTTSASKHHDLFTE UHFDRNDIFNFPDE PTVI OF E IFOFIEPVTAAA&wATTPHUAIFHO FLIVVER.~ I OL DHTD Oa Plw rtt54PLUNa CAONHSEOTTESPAEAat A - ~ !VlALPEDO S.t � E.P

ev Seta CTHTCAHTFIV rtpvaVEGHEAEES Vawtaa 5' l&TPCPlrw W~IHIFDL~~~ USSLE. HiJIESEPS SF»DalFDHTAH SAIOV- NHDAFEICN ICLOUD PEKFIAVPE Sr r NOTTDNOl' ~ t Ht a 'f -5 5 Htctoua CRoss EECTSDNSEAREA DEDFIEASEN ~ ~atff AKIT IEEPOFIEFHFPFII ' 'e.ALSEDD 0 ~ .Dl

Line H J. corresponds to the whitecap - LU m elevation wina speed;

i.e., W, U!, expression of Monahan and O'Mumheartaigh �980!,which they obtained by applying the technique af robust Bi-weightfitting to the photographic whitecap data sets found in Monahan�971! and Toba and Chaen �973!, while Line Al is a representa-tion of the W� U. AT! expression that arose from the analysis,with a Hamamatsu Area Analyzer, of oblique U-matic videorecordings of the sea surface taken during the MIZEX 1983 Manahan, et al., 1985!, MIZEX 84 Monahan and Woolf, 1986!,HEXPILOT Monahan, et al., 1985!, and HEXMAX Monahan, etal., 1988b! experiments. The video-based expression, given herefor near neutral atmospheric stability, i.e., for AT equals zero, onlytakes into consideration the high albedo Stage A whitecaps andspilling breakers Monahan, Wilson, and Woolf, in Monahan. etal., 1988a!, effectively excluding the relatively low albedo Stage 8whitecaps which of necessity fall below the discrimination bright-ness level routinely adopted for use with the Hamamatsu AreaAnalyzer. Line B2, also included on Figun: 2, is the W, U, AT!expression obtained by Monahan and O'Muircheartaigh �986!from the analysis of the composite whitecap data base comprisedof the photographic results from BOMEX plus other warm seas Monahan, 1971!, the East China Sea and adjacent waters Tabaand Chaen, 1973!, JASIN Monahan, et al., 1981!, STREX Doyle,1984!, and MIZEX 83 Monahan, et aI., 1984!. In this instance,AT, the surface water temperature minus the deck height air tem-perature, has again been set equal to zero; i.e., Line 82 representsthe W, U! expression for near neutrd stability.

Only in the Soviet literature can be found the msults obtainedfrom the analysis of specific sets of photographic images for boththe areas of the sea surface associated with crests and active StageA whitecaps, and for direct comparison the areas occupied on thesarrie images by foam and passive Stage B whitecaps.

The fraction of the sea surface covered by breaking wave crestsfound by Bondur and Sharkov �982! is illustrated by Curve A2,while the fraction of the sea surface they deduced from the samephotographs to be covered by striplike foam is presented by curveB3. Bortkovskii's �987! results for active whitecaps, based on

47

Fjp. 2. Wind dependence0 Stage A and&age 8whit~ coverage. Seetext for id'errtlfy o eachlabeled curve. Note thatboth axes bear logarith-mic sca/es.

Monahan and O'Muircheartaigh, 1986!, what is the wind speedabove which these power-law expressions likewise do not apply?

Q6 - For a constant 10 m e!evation wind speed and uniformatmospheric stability, what is the quantitative effect of changes inthe kinematic viscosity, i.e., of changes in sea surface temperature,on whitecap coverage?

Q7 � Can we quantify, via water wave theory, the influence offinite fetch and limited wind duratio~ on W, and %�?

Q8 - WhHe the two-dimensional surface divergence associatedwith the idealized, buoyant, whitecap bubble plume should miti-gate the effect of any surface rnonolayers on the persistence of

49

Figure 3. Near-surfaceresident bubble spectra forwinds of 11-13 ms-' Reg+�W x B!; range of spectrapredW~by Equation 7forU10of 13ms',i.e., for yequalto00241 CurveGoceanic spectrum, 1,5 depth11- f3 ms' winds, Ko/ovary�976!. Curve H coastalspectrum, 0.7m depth, 11.13ms' winds, Johnson anrfCooke �979!. Curve i, fr9sh.water wind-wave flumespectrum, 0.05 m depth,13ms' winds, 26.2 m fetch,Baldy and Bovrguel � 985!.Region B -CJ at topcorresponds to range ofspectra model predicts forimmediately beneath awhitecap, or for LV, equal to1.00.

."'"Ps can the actual role of organic material in the oceanic< hve a dof sutfactant6lms on the sea surface, 4 stabiliz-

ing whitecap bubbles, and hence in extending the ]i fetimes ofindividual whitecaps, be explicated?3. Whifecap-Dependent Sea Surface Aerosol

Generabon Models

Our initial expression, incarcerating quantities derive fr ~ +study of srnaB whitecaps in laboratorjj ~ and quantities obtaihorn recording whitecaps at sea Monahan, ei N., 19S3; Mona".'1986! is re rodp uced, with sight change in notation, as Equation .

W =3.86x10 U

The near surface bubble population spectra predicted byEquation 7 to be associated with winds of l3 ms ' is represented bythe lower hatched band depicted on Figure 3, The upper bound ofthis range of bubble populations, given by the curve W x B,, wasdetermined assuming that each bubble produces only one jetdroplet; i.e., that J equals one, and that the bubbles are hydro-dynarnically dirty, and thus rise with the terminal velocity appro-priate for a sphere with no tangential relative flow at its irrImediatesurface. Conversely, the lower bound of this range, illustrated bycurve W x C�was calculated assuming that each bubble gives riseto five jet droplets and that the bubbles are hydrodynarnicallyclean. By way of reference, the band of bubble population spectrapredicted by our model for 1D0% whitecap coverage is also in-cluded Region B,-C,! on Figure 3.

Near-surface oceanic bubble spectra have been obtained byKolovayev �976!, and by Johnson and Cooke �979!, when thewind was blowing 11-13ms '. It is apparent from FiguN: 3 that forall bubble radii, R, greater than 120 pm Kolovayev's spectrum curve G! falls essentially within the band of spectra predicted byEquation 7 when the tlFP r tenn was evaluated for winds of13 ms '. Johnson and Cooke's spectrum Curve H!, on the otherhand, is consistent with the model predictions for all R larger than50 pre. The discrepancy in the 5D to 120 pro radius intervalbetween Kolovayev's results for 1.5 m depth, obtained usingNeuimin's semi-automatic cylindrical bubble trap Glotov, er al.,19'61!. equipped with a camera and three lamps to illuminate thebubbles, and the results obtained by Johnson and Cooke using at adepth of 0.7 I a camera in a waterproof housing accompanied bythree external strobe lamps, may, as was suggested by Johnson andCooke �979! be due to some dissolution, and perhaps also somecoalescence, of bubbles in this size range during their residencewithin the Neuimin bubble trap. ft may be assumed that Glotov,et al., would not agree with this interpretation, as they state in their1962 paper that the bubbles rested against the underside of the

55

7 to predict the approximate shape and amplitude of the largeradius portion of the near-surface, in-situ, bubble spectrum, givessupport to our contention that the bubble flux modeling approachset forth in this section, and the aerosol flux modeling schemedescribed in Section 3, are fundamentally valid, and thus meritfurther development.

Q16 - Is the relationship found by Blanchard �963! to pertainbetween the radius, R, of an isolated bubble breaking at a quietwater surface, and the radius, r, of the daughter jet dmplets, stillfully valid in the case of whitecap bubbles, which are not isolatedand often burst at an agitated sea-air interface?

Q17 - Is the transition between the jet droplet dominatedportion of the aerosol spectrum and the film droplet dominatedportion adequately described by the laboratory findings of Woolf,et al �987!?

Q18 - Will the function, J, representing the number of jet drop-lets produced per bursting bubble, upon further laboratory studiesprove to be dependent upon R; i,e�shouki we expect ultimately tosubstitute a J R! term in Equation 7?

Q19 - Is it rteasonable to expect that the J, or J R!, expressiondetermined from laboratory studies of the bursting of individualbubbles, be appropriate for use in modelling the relationship be-tween the sea surface aerosol Aux and the whitecap bubble flux,given that many whitecap bubbles do not burst in isolation nor on arelatively calm sea surface?

Q20 - Need we consider the possible influence of protractedbubble-surface-lifetime on tlFPr, ter/ itR, and J Ri, in perfectingour model?

Q21 - Can we, with any confidence, predict when bubbles, ofRadius R, in a specified part of the world ocean, will typicallyprove to be hydmdynamically dirty, and thus behave accordingly?

Q22 - Given the variety of fates that Thorpe �982! has docu-mented await small bubbles in the open ocean, is there any reasonto hope that a model such as the one described above can ever beused to predict the resident population of these smaller bubbles?

Q23 - Can the agreement between the wind dependence four}dby Farms and Lemon �984! for the population of bubbles in

through the viscous surface layer assuaged to exist everywhere botin the interior of whitecaps, and k, is the much higher pistonvelocity identified with the turbulent transfer of gases through awhitecap vent.

k~ = kM l � W~!+ kTWa

7n light of the detailed discussions of gas transfer throughsmooth and roughened, but not whitecap covered, sea surfacesavailable in the recent literature Liss and Merlivat, 1986, Coantic,1986!, it may be more appropriate to rewrite Equation 9, as hasbeen done in Equation 10, to make explicit the wind-dependentnature of k,

kE = k~ U + kM U !� � We!+ kTWB �0!

59

It should be noted that the last term, k, W� is still by far themost strongly wind-dependent term in this expression.

Recent, preliminary, gas exchange experiments carried out inWhitecap Simulation Tank HI in the Marine Sciences Institute atthe University of Connecticut have shown that the generation of asmall whitecap once every 210 seconds, in ari instance where thewater in the tank is already being continuously stirred, is enough. tomore than triple the evasion velocity of radon gas when comparedto its evasion velocity when the tank is simply being continuouslystimxi The effectiveness of whitecaps in stimulating gas ex-change can be appreciated when it is noted that the average frac-tion of the sea water surface in the tank covered by whitecapsduring each 210-second cycle of this experiment was 0.1%; i.e., Wwas just 0.001. Further measurements of this sort will make itpossible for us to evaluate k� the piston velocity associated withthe turbulent transfer of gases via a whitecap vent.

@24 - Is it reasonable to assume that k�will be independent ofthe atua of the individual whitecap, or is it more probable that kwill increase with increasing whitecap size7

5, COllcIEL%KNls

Voile the aerosol flux, bubble flux, and gas transfer modelsdescribed in the previous sections are at markedly different stagespf refinement, all show clear promise, and support the contentiothat a modeling approach in which the individual whitecap, orbubble plume, is taken as the key element enabling us to extrapo-late the laboratory tank results to global ocean is a fundamental<!'valid one.

It is obvious that even the most detailed of our models stan~ ~+need of refinement and revision, and it is to be hoped that theanswers to the two dozen questions interspersed in the foregoirtgtext will be Of particular utility in the further development of the~several models.

Adtrsomledgments

Our investigation of whitecaps in the laboratory and in the fieMhas been supported thmughout by the Office of Naval Rescarch-Cumntly our research on whitecaps, marine aerosol particles.bubb!es, bubble plumes, and gas exchange is supported by <NRContract N00014-87-K-0185 and by ONR Contract NOGOI4-87-K-0069. This papcz is Contribution No. 206 from the Marine Sci-ences Institute of the University of Connecticut.

References

Avcllan, F., and 1'. Reach. 1983: A Scattering Light probe for the Measuremetttof Oceanic Air Bubble Sizes. kt. Journal of Mtdtiphase Flo~, 9, 649~3-

Baldy, S., and M, Bourguel. 1985: Measurements of Bitbbles in a StationaryField of Breaking Waves by a Laser-Based Single-Particle ScatteringTechnique. J. Ge0phys. Res., 90, 1037-1047.

8lanchard, D.C., 1963: The electrifcation of the atmosphere by particles fmmbubbles in the sea. Prog, Occanog., 1, 71-202.

Bondur, V.G.. and E.A. Sharav, 1982: Statistical Propenies of Whitecaps otI aRough Sea. Oceanology, 1, 274-279.

Borlkovskii, R.S., 1987: Air-Sea Exchange of Heat and MoistureDurinfStorms, D. Reidet Pub. Co., Kkxdrecht, The Netherlands, 194 pp, revisedEnglish edition!

Bowyer, P.A., 1986: Electrostatic Phenomena Associated With the MarineAerosol Arising from the Bursting of Whitecap Bubbles, Ph.D. Thesis,National University of Ireland, 155 pp.

Burk, S.D.. 1984: The Generation, Turbulent Transfer, and Deposition of theSea-Salt Aerosol, J. Atmos. Sci. 41, 3041-3051.

Cipriano, R.J., and D.C. Blanchard, 1981: Bubble and Aerosol Spectra Pro-duced by a Labrxstory 'Breaking Wave.' J. Geophys. Res., 86, 8085-8092.

Coantic, M., 1986.' A Model of Gas Transfer Across Air-Water Interfaces withCapillary Waves, J. Geophys. Res. 91, 3925-3943,

deLeeuw, G., 1987: Near-Surface Particle Size Distribution Profiles Over theNorth Sea. J. Geophys. Res. 92, 14631-14635.

Doyle, D.M., 1984: Marine Aerosol Research in the Gulf of Alaska and an theIrish West Coast Inishmore!, M.Sc. Thesis, National University of Ireland,140 pp. also University College Galway Report No. 6 to the Office ofNaval Research!.

Farmer, D.M., and D. D. Lemon, 1984.' The Influence of Bubbles on AmbientNoise in the CVmn at High Wind Speeds. J. Phys. Oceanogr., 14, 1762-78.

Glotov, V.P., P.A. Kolovayev, and G.G. Neuiinin, 1962: Investigation of theScattering of Sound by Bubbles Generated by an Artificial Wind in SeaWater and the Statistical Distribution of Bubble Sizes. Sov Phys.,Acoustics, 7, 341-345.

Johnson, B., and R.C. Cooke, l979: Bubble Populations and Spectra in CoastalWaters: A Photographic Approach. J. Geophys. Res.. 84, 3761-3766.

Kanwisher, 7�1963: On the Exchange of Gases Between the Atinosphere andthe Sea. Deep Sea Res, 10, 195-207,

Kientzler, C.F., A.B. Ass, D.C. Blanchard, and A.H. Woodcock, 1954:Photographic Investigation of the Projection of Droplets by BubblesBursting at the Water Surface, Tellus, 6, 1-7.

Kolovayev, P.A�1976: Investigation of the Concentration and Statistical SizeDistribution of Wind-Produced Bubbles in the Near-Surface C~Layer. Oceanology, 15, 659-661.

Liss, P.S., and L. Merlivat, 1986: Air-Sea Gas Exchange Rates: Introductionand Synthesis, in The Role of Air-Sea Exchange in Geochemical CycliIt g, P,Buat.-Menard, Ed., D. Reidel Publishing Co., Dordrecht, 113-127.

Monahaa, EC., 1968: Sea Spray as a Function of Low Elevation %find Speed.J. Geophys. Res. 73, 1127-1137.

Monahatm, F C., 1969: Fresh Water Whitecaps, Jour' of the AtmosphericSciences, 26, 1026-1029.

Mcmahari, E.C., 1971: Oci"saic Whitecaps.J. Phys. Oceanogr. I, 139-144.llomshan.E.C., 1986: fhe Ocean as a Source for Atiwephmc Particles, in The

Role of Air-Sea Exchange in Geochemical Cycling, P. Buat-Menard, Ed., D.Reidel Publishing Co., Dordnecht, 129-163.

Mceahtri, E.C., 1988: Whitecap Coverage as a Remotely Monitorablelndica-tion of the Rate of Bubble Injection into the Oi~c Mixed Layer, in SeaSurface Sottnd, B.R. Kerrnan, Ed., Kluwer Academic Publishers, Dordrecht,

anadem, F C., M.B. Wilson, and D.K. Woolf, 1988b: HEXMAX WhitecapClimatology. Foam Crest Coverage in the North Sea, 16 October - 23November 1986, Proceedings, ~84AX Workshop, Univ. of WashingtonTech. Report in press!.

4oruthan, E.C., and D.K. Woolf, 1986: &~nic Whitecaps, Their Contributionto Air-Sea Exchange, and Their Influence on the MABL, Whitecap ReportNo. 1, to ONR from MSI, UConn, 135 pp.

4mahan, E C�and C.R. Zietlow, 1969: Laboratory Comparisons of Fresh-Water amd Salt-Water Whitecaps, J. Geophys. Res. 'N,6961-6966.

%under, C., 1986: Sodium Chloride and Temperature Effects on Bubbles, inOceanic Whitecaps and Their Role in Air-Sea Exchange Processes. E.C.Monahan and G. Mac Niocaill, Eds., D. Reidel Pub. and Galway U. Press,278.

kesch, F,, 1986; Oceanic Air Bubbles as Generators of Marine Aerosols, inOceanic Whitecaps and Their Role in Air-Sea Exchange Processes, E.C.Monahan and G. MacNiocaill, Eds., D. Reidel Pub. Co. and Galway U.Press, 101-1 12.

Scott, J,C., 1975a: The Role of Salt in Whitecap Persistence. Deep Sea Res.. 22,653457.

Scott.J.C., 1975b: The Preparation of Water for Surf~<lean FluidMechanics. J. Fluid Mech., 69, 339-351.

Spillane, M.C., E.C. Monahan, P.A. Bowyer, D.M. Doyle, and P.J. Stabeno,19&6: Whitecaps and Global Fluxes, in Oceanic % hitecaps and Their Rolein Air-Sea Exchange Processes, E.C. Monahan and G. MacNiocaill, Eds.,D, Reidel Pub. and Galway U, Press, 209-218.

Strarnska, M., 1987: Vertical Profiles of Sea Salt Aerosol in the AtmosphericSurface Layer. A Numerical Model. Acta Geophysica Polonica, 35, 87-! 00.

Thorpe.S.A., 1982; On the Clouds of Bubbles Farmed by Bn~g Wind-Waves in Deep Water, and Their Role in Air-Sea Gas Transfer. Phil. Trans.R. Soc. London, A304, 155-210.

Toba, Y., artd M. Chaen, 1973: Quantitative Expression of the Breakmg ofWind Waves on the Sea Surface. Records, Oceanogr. Works Japan, 1?�1-11.

Woolf, D.K., 1988: The Role of Oceanic Whitecaps in Geochemical Transportin the Upper Ocean and the Marine Environment, Ph.D. Thesis, NationalUniversity of Ireland, 330 pp.

Woolf, D.K., P.A. Bowyer, and E.C. Monahan, 1987: Discriminating Betweenthe Film-Drops and Jet-Drops Produced by a Simulated Whitecap, J.Geophys. Res. 92, 5142-5150.

Woolf, D.K., E.C. Manahan, and D.E. Spiel, 1988: Quantification of the MarineAerosol Produced by Whitecaps, Preprint Volume: Seventh Conference onOcean-Atmosphere Interaction, 31 January - 5 February 1988, Anaheim,California, American Meteorological Society, 182-185.

data at 0. 3 I were sampled. The data at 2.5 rn are typical for thoseobtained with the impactor.

Preobrazhenskii �973! measured particle size distributionsfrom a ship at 3 heights: 1.5-2 m, 4 m, and 7 m. Samples werecollected on oil-coated glass plates. The data were presented fortwo ranges of wind speeds: 7-12rns ' and 15-25 ms '. Those col-lected at 1.5-2 m ASL, converted to number concentrations perdiameter interval, are presented in Figure 1. At higher elevationsthe concentrations are lower.

Wu, et aI., �984! measured droplet concentrations from a raftin Delaware Bay at elevatians between 0.13 and 1.3 m. Windspeeds were 6-8 ms '. Droplets in the 30-400 pm diameter rangewere detected by the extinction of a laser beam when the dropletspassed through the sensor volume. Absolute concentrations werenot obtained, the data were presented as the normalized frequen-cies of occurrence fj of particles in certain size bands.

The normalized droplet spectra are described by two relations.ln the jet-droplet ejection zone Wu, er al. obtained:

f-D' 504,m<Dc150pm

f-D~' 1504.m<D <400prn z<20crn

Ds 400 pm D

Above the maximum ejection height of the jet droplets thesame relations apply, but for different size ranges:

f � D ' 10 pm < D 50 pm

f-D~' 50pm <D <200 pm z>20cm

f - D s 200 ~rn D

D is the particle diameter, and z is the- elevation above the in-stantaneous water surface. The size distributions were found to beindependent of both elevation and wind speed.

The relations between f and D for the smallest sizes in eqns. 1

67

OO Vlo vi

C!OO'VD

I

C!

C! C!CO

I

O O

LC4

Ch

O0 O

~0

o R. 5,! s

C4lo

4

X

m C4k

CO

~

g~5m

< ssW

5% .9 s':

~4

2 g

C b

~

different places and under different atmospheric and oceaniccoaditioas.

Chaen �973! concluded from his experiments that on theaverage the vertical distributions of the particle concentratiaiisapproach a power law dependence. Preobrazhenskii �973! con-cloded that the particle volume decreases exponentially withheight.

The profile published by Blanchard and Woodcock �980! for

Figure 2.Representativeexamp/es ofdroplet concen-tration profiles inthe atmosphericsurface /ayerover the ocean.

8 res ' wind speed has been reproduced in Figure 2, together withexamples of other profiles discussed below.

Wu, et al., �984! measured relative profiles of particle concen-trations. The 30-50 pm particles were observed to be rather homo-geneously distributed within the lowest meter above the instantane-ous water surface, with only a small negative gradient, cf. Figure 2.The gradient becomes stronger as particle size increases. Above1m the concentrations drop draxnatically, for all sizes.

From our experiments in the North Atlantic in 1983 de Leeuw,

71

1989!. Unfortunately direct comparisons were not possible sinceboth instruments could not be mounted on the wave follower si-rnultaneously. Nevertheless the profiles measured with thescatterorneter and with the Rotorod were often similar and thescattemmeter data confirm the previously observed occurrence of amaximum in the particle pmfiles de Leeuw, 1989a!.

On the other hand, the data collected with both the Rotorod andthe optical scatterorneter are often scattered in the vertica1, whichrrught hide the occurrence of a maximum in the profile.

4. Production of sea-salt aerosols and dispersal inthe surface layer.

The production and subsequent dispersion of aerosols in thesurface layer are discussed in the foGowing sections. Comprehen-sive documentation exists oa the production, cf. the reviews byBlanchard �983! and Monahan �986!. Some salient features ofproduction are reviewed below.

4.I Produmon.

Salt water droplets can be produced at the sea surface byvarious mechanisms. The bubble mechanism has been studiedextensively in the laboratory, cf. Blanchard �983! for a review.When an air bubble in salt water reaches the surface it stays therefor a second or more. Surface tension forces the bubble to reach ancquihbriurn position before it bursts, Upon bursting the thin film atthe surface breaks up into typically 10-1000 particles the so-caGedfihn dmps!. Mis applies to bubbles larger than about 2 mm indiam'. SmaHer bubbles produce less film drops. The film dropspec+urn peaks at 5 pm, and may extend hem smaller than 1 pmto larger than 30 pan Blanchard, 1983!.

When the bubble bursts, the surface free energy is convertedinto hnetic energy of a jet of water which rises from the bottom ofthe bubble cavity. The water column overshoots the surface andheals up into typically 1- 10 droplets, d.epending on bubble size.The maximum ejection heights of these ~called jet drops havebeen observed to be nearly 20 crn for the top drop and only a

73

due to the somewhat enhanced pressure that forces the flow backinto the trough. Berate the whole process may be repeated.

During this process gravitation and turbulence are also active.If W!xu, the particles are dragged downward and are likely de-posited into the sea. In the other case, W<xu., they might escapethe eddy into the mixed-layer. Since these processes are statisticalin nature, this distinction is not sharp. Only a fraction of the par-ticles is expected to reach the mixed layer, which depends on size,density and wind speed.

Qnce in the mixed layer the particles are subject to gravitationand turbulent transport, while entrainment, subsidence andstability also affect the concentrations cf. Fairall, et al., 1984!.

The exchange of particles between the surface layer and themixed layer is not limited to freshly produced droplets. 'Aged'particles that are trapped in the eddies are subject to the sameprocesses described above. Hence they might be resuspended ordeposited. Another interesting possibility is the interaction throughcoagulation between aged particles and fresh droplets, which mightenhance the deposition rate.

In view of the above, the concept of a viscous sub-layer, whichis successfully used in the description of air-sea exchange of, e.g.,humidity, might be less suitable to aerosol modelling. It is notonly the escape of the droplets through the sub-layer that has to beconsidered, but also the transport of the droplets in the zone ofinfluence of eave-induced eddies. Previously we called this theturbulent buffer zone, since the droplets may be temporarilytrapped in the eddies. In effect, this buffer zone may be considerecias an elevated and distribute sauxee.

6. Discussion.

Particle size distributions and particle concentration profileshave been measured near the air-sea interface with methods basaion different physical principles, Optical techniques employed w~photographic Monahan, 1968!, laser extinction measurements Wu, et al., 1984! and measurements of near-forward laserscattering de Leeuw, 1989a, 1!. Passive impaction methods were

BUBBLE-MEDIATED SEA-AIR EXCH4NGE

used by Chaen �973! and Preobrazhenskii �973!; active irnpac-tioa methods were used by Chaen �973! and de Leeuw �.986,1987, 1989a!. Various coatings were applied to retain the particles.This requires different processing and calibration techniques.The size distributions were measured under a wide range of mete-orological and oceanographic conditions, at different locations.

This is reflected in the results displayed in Figure 1. Due tochanging meteorological conditions, the concentrations observedduring our measurements in the North Atlantic de Leeuw, 1986!vary over one decade. These data are representative for the averageof the other measurements in the same size range. Largedifferences were observed, however, spanning almost four decadesat 10 pm to two decades around 10G pm,

The shapes of the surface layer profiles are strongly dependenton wind speed. The occurrence of non-logarithmic profiles underbreaking wave conditions was demonstrated by our experiments inthe North Atlantic de Leeuw, 1986!. The maximum in the profileswas confirmed by the HEXOS experiments in 1984 de ~uw,1987! and in 1986 de Leeuw, 1989a, b!. The analysis of theseprofiles led to new insights into the surface production rate deLeeuw and Davidson, 1989!. The application of a viscous sublayercoracept might need reconsideration. The use of an effective sourcestrength in mixed-layer models yields correct predictions of thetrends in mixed-layer aerosol profiles de Leeuw, 1989c!. Near thesurface, however, the predicted concentrations increase faster thanindicated by our data de Leeuw and Davidson, 1989!. Hence thesurface production rates are overestimated. Better results might beobtained by considering a distributed source function that extendsover the region where wave-induced eddies are active. The extentof this region will depend on the meteorological and oceano-graphic conditions. Empirical relations to estiinate the elevatedsource strengths were presented in de Leeuw �986!, This simplemodel is based on a single data set consisting of 17 profiles, Inclu-sioa of the HEXOS data is expected to lead to refinements yieldinga more rehable modeL The results are important to quantify theeffects of droplets on, e.g., the processes mentioned in theUitmduction. When they are trapped in the turbulent buffer zone

79

Particle Size Distributions and Relative Humidity Dur0tg HEX' .Submitted to Tellus.

de Leeuw, G., 1989b: investigations on Turbulent HIIctuations of ParticleConcentrations and Relative Humidity in the Marine Atmospheric SurfaceLayer. J. Geophys. Res 94, 3261-3269.

de Leeuw, G., 1989c: Modeling of Extinction aad Backscatter Profiles in theMarine Mixed-Layer. Appld. Opt. 28, 1356-1359.

de Leeuw, G., and K.L. Davidson, 1989: Mixed-Layer ProQing with Lidar andModeling of the Aerosol Vertical Structure. Submitted to J. Arm. 0ceaei cTectrn.

de Leeuw, G., J.A. Boden, L.H.Cohen, M. Deutekom, A 3, T, de Krijger, 6 J.Kunz, M.M. Moerman and C.W. Larnberts. 1989. An Angular ScatteringDevice for the DeterInination of Turbulent Huctnations of Aerosol Concen-trations, Submitted to Rev. Sci. Instr.

Edson, J.B., C.W. Fairail, SE. Larsen and P.G. Mestayer, 1988: A RandomWalk Simulation of the Turbulent Transport of Evaporating Jet Drops ia theAir-Sea Simulation Tunnel During HEX1ST, 7th Conf. on Air-Sea lnterac-tiorI, Jan. 31-Feb. 5, Anaheim, Calif.

Fairall, C,W., KL.. Davidson and G.E. Schacher, 1983: An Analysis of theSurftxA; Production of Sca-Salt Aerosols. TeÃus 358, 31-39.

FairaQ, C.W., KL.. Davidson and G.E. Schacher, 1984: Applicatioo of aMixed-Layer Model to Aerosols in the Marine Boundary Layer. Tellus 368,203-211.

Fairall, C.W. and SZ. Larsen, 1984: Dry Deposition, Surface Production andDynamics of Aerosols in the Marine Boundary Layer. Amos. Krtviron. 18,69-77.

KraLts, E.B., 1972: Atmosphere-Ocean Inter':tion, Clarendon Press, Oxford.Mestayer, P.G., C.W. Fairall, SZ. Larrup, DZ. Spiel and J. Edson, 1989:

Turbulent Transport and Evaporation of Droplets Generated at an Air-WaterInterface. Proc. Sixth Symp. Turb. Shear Rows, Treble, France, Sept. 7-9.

Mestaye', P., and C. Lefauconnier, 1988: Spray Droplet Generation, Transport,and Evaporation in a Wind Wave Tunnel During the Humidity Exchangeover the Sea Experiments in the Simulation Tunnel.J. Geophys. Res. 93,572-586.

Mestayer, P.G., C. Lefauconnier, M. Rouault, S. Larsen, C.W. Fairall, J.B.Edson, DX. Spiel, K.L. Davidson, E.C. Menalhan, D. Woolf, G. de Leeuw,H. Gucinski, K.B. Katsaros and J. DeCosmo, 1988: HEXIST NATOAdvancol Worh@cy on Humidity Exchange Over the Sea, De Bi@Epe,The Netherlands, April 25-29,

Mceah'In, E.C., 1968: Sea Spray as a Function of Low-Elevatioe Wind Speed.J.Geopkys. Res. 73, 1127-1137.

Montkan, E.C., 1986: The Gmm as a Seurce for Atmospheric Particles. In: P,Buat-Menard ed.!, The Role of Air-Sea Exchange in Geochemical Cychng.Reidel, 129-163.

81

G. de LE&JW

Monahan, E.C.. and LG. O'Muircheartaigh, 1986: Whitecaps and the PassiveRemote Sensing of the Ocean Surface. 1nr. J. Remote Sensing 7, 627-642-

Monahan, E.C., D.E. Spiel, and K3 . Davidson, l983: Model of Marine P~aszlgeneration Via Whitecaps and Wave Disruption. Ninth Conf. on Aerospsceand Aeronautical Meteor., June 6-9, Omaha, Nebr.

~'LY .13333 ' ff C fpp yp I I IINear-Water Layer of the Atmosphere. Fluid Mech.-Soviet Res. 2, 95-100.

Wu, J., 1979: Spray in the Atmospheric Surface Layer: Review and Aaalysis ofLaboratory and Oceanic Results. J. Geophys Res., 84, 1693-1704.

Wu, J., J3. Murray artd R J. Lai, 1984: Production and Distributions of SeaSpray. J. Gcophys. Res. 89, 8163-8169.

R. hNRKS and E.C. MONAkAN

when the wind increased to above 15 ms '. Correlation coefficientsbetween S-SAMC and wind speed were found to increase withwind speed and with increasing sizes of the sea-spray aggregatesdispersed in air-

1. Introduction

Sea-spray generation processes have been investigated andclosely correlated with wind speed e.g. Woodcock, 1953, Blan-chard, 1963, Garbalewski, 1980, and others!. In the initial stagethe turbulent components of the air flow over the sea surfacegenerate fine ripples and capillary waves, which create roughsurface elements. Through such rough elements in a high fre-quency range, the energy of air turbulence is transferred to thedeveloping wind waves. In a random wave field those waves thathave developed to the critical steepness wiH dissipate energy bybreaking and forming whitecaps. This is the main source ofbubbles at the sea surface. It is these bubbles which generate jetand film drops upon bursting and affect the turbulent diffusion ofheat and moisture. When the wind-wave-whitecap system reachesa high state of development the tearing of drops and water parcelsfrom the wave crests contributes further to the sea-spray flux to the

atmosphere.During heavy storms or hurricanes water panels from wave

crests as well as rough elements of the surface are mechanicallydislodged, creating an air/sea mixing zone. Furthermore, the sea-spray production processes may be modified by other peripheralconditions. These include wave breaking in shallow water, or atobstacles i,e. icebergs!, air bubble advection in the ocean surfacelayer, supersaturation of the sea water, wave-wave interaction, at-rnospheric precipitation, stability at the air/sea interface, and otherrnetcomlogical and oceanographic conditions. An attempt tocorrelate sea-salt concentration with wind speed and other mete-orological factors has been made by Lovett �978!. It was fouridthat total sea-salt concentration could be represented as a functionof the wind speed by the equation ln~.16u+1.45 Q = sea-saltconcentration in begin' and u = wind speed in ms '!. Exton et al..�983! conducted field experiments at a coastal site on the islarid

BUBBLED-aeeOIa TEO SEA-AIR ETC:HAA~

of South Uist in the Outer Hebrides. They suggest that th< sdistribution of the sea-salt could be sub-divided into tw> ~1 gseparated at r&.3 p.m, the lour corresponding to long raNgaerosol and the higher to locally-produced aerosol.

The complex interrelations between 10 m elevation wind>whitecaps and marine aerosols were investigated by Monaha"aL, �983!, They report the positive dependence of aerosol con«ntration upon whitecap cover, increasing with droplet radius- T"total mass of the sea-salt aerosols as a function of wind slteed nfl +and whitecap coverage were determinated by Marks �987!. In t»sstudy, the static stability at the air/sea interface was found toinhibit the sea-salt aerosol production. Vertical profiles of the sea-salt aerosols have been investigated by Chaen �973!, Pre-abrazhenskii �973!, Blanchard and Woodcock �980!, Blancharder al. �984}, de Leeuw �986! and Strmska �987!. Furthermore,some laboratory studies on the size distribution of particles inprofiles have been done by Wu �979!, Koga and Toba �981! andKoga �984!. Both the particle concentration and its variation ~ithheight are strongly dependent on wind speed, and the averageconcentration gradients decrease with decreasing particle diameterat a given wind speed de Leeuw, 1986!.

The present investigation was carried out to combine the 8-SAMC and whitecap coverage data obtained during the HEXMAXexperiment. This eKort is part of the intercornparison of the vari-Ous data sets collected to study the evaporation and spraydropletfluxes Rom the sea at the moderate- to high-wind speeds.

2. Measurements

Data were collected from the DUtch research platform Noord-wijk, located 9 km offshore in the North Sea.

The S-SAMC ObServatiOns COnSiSted Of rnl~suxerrients at threeheights; 4-5 m, 12 m and 18,3 m, and the measurement of sea-saltsize distribution at 12 m elevation. For profile measurements asenii-isokinetic air filtration unit with adjustable air speed at theinlet was used. A High Volume Cascade Impactor � stages!,Sierra 230, was used for the size distribution rrieasuxements.

R. A44RKS and E.C, MONAHAN

y@atman 41 filters were used for the actual collection of theaerosol sampled by both instruments. The air-filtration-unit inletswere oriented in the direction of the oncoming wind.

To collect a sufficient amount of aerosol, both saoipling unitswere operated in ambient air for 0.5-3 hours. After each run, eightexposed filters were unloaded and pressed to get pills of about 1crn diameter.

Neutron activation analysis was carried out on a total of 750pills. Irradiation was performed for 10 minutes at a thermal neu-tron flux of about 10" sec ' cm ' using the reactor at the GKSSResearch Centre Fed. Rep. of Germany!. The concentration of Nawas determined from the intensity of the 1368 kev Gamma-Rayemission of Na-24 as measured with a Ge Li! detector.

The whitecap coverage was recorded using a video camera, aSONY Trinicon Model DXC 1800 P, and a color portable videorecorder SONY Model V0-4800PS!, The line-of-sight of thecamera was set 20 below the horizontal. The camera shelter itselfwas fixed on the exterior of the platform at an elevation of about12 m above the surface of the North Sea. The fraction of the seasurface covered by young stage A! whitecaps has been determinedfor 159 periods during the experiment by analyzing the videoimages with a Harnaroatsu Area Analyzer Model C1143-00!.

Background meteorological and oceanographic data wereautomatically recorded at 10-minute intervals by the platform dataacquisition system. The wind speeds measured 28 m above themean sea level were adjusted to the 10 m elevation wind speedequivalent using the logarithmic wind profile law.

3 Sea-salt aerosol-total mass concentration data

The aerosol data obtained using the air filtration unit at 12 melevation were subdivided into five classes according to the difer-ent air advection patterns present during the experiment. Theseclasses were partitioned using 1000 mb air mass trajectories and»eluded air flow from a SW-NNW' sector, from ~iV-N sector,from the English Channel, from the coastal zone, and fmm land. InFiig 1 the five classes of S-SAMC are presented as a function of

BUBBLE-MEDIA TED SEA-AIR EXCHANGE Fy. 1. See-saltaerosol mass

concentration of the

fi5aten-unit versus10 m winct speed fordiferent advection airBows. h4easurernentswere taken at f2 rnelevation during dry~iris anct curves!arxf rainy pointsdenoted by Iettels!weather conditions.

Curves 3,4,7,$0aredeserked in TaMe $.

the wind speed u�0! for dry and rainy weather conditions. The S-SAMC data in dry weather, associated with air advection fromSW-NÃAf sector curves 3 and 4!, from a NNW-N sector curve10! as well as from the English Channel in spite of rainy weathercondition!, points denoted by E!, indicated a considerably highermass concentration than did the S-SAMC data associated with airmass advection from the coastal zone curve 7! or land curve 9!.In the latter case the air had to travel over the sea a distance of 10-

26 km.

Saxnples collected during rainy weather are indicated in Fig. Iby lemers which correspond to advection air flow over: the EnglishChannel E class for frequent rain, and EF class for fog and rain!,the North Sea, SW-~iVv N class! and the coastal zone C class!.Generally, the S-SAMC data collected during rainy weather at 12rn elevation showed lower concentrations and larger variations,

R. ANRKS and E.C. MOIVAHAN

identified with samples collected during foggy weather conditions see points denoted by EF on Fig. 1!.

To study the influence of wind speed u�0! on S-SAMC thesaxnples collected when the air was advecting from the SW-I'QPiVsector representing for the most part winds &om over the NorthSea! were partitioned into two categories. The first included the S-SAMC values measured when the wind speed were between l. and15 ms ' see curve 3 on Fig. 1}, and the second category includedthe S-SAMC values obtained when the wind was between 15 and24 ms ' see curve 4 on Fig. 1!. The differences in the slope ofthese two curves show that the S-SAMC undergoes a marked en-hancement when the wind speed exceeds 15ms ' see Table 1!.

This initial analysis shows some remarkable differences in S-SAMC associated with different air advection patterns over theNorth Sea. These differences support the hypothesis that sea-saltaerosol over the North Sea is of a mixed type, originating fromboth Atlantic and locally emitting sources. Nevertheless, the

BUBBLE-MEDIA TEO SEA-AIR EXCHANGE

S-SAMC can be related to the air/sea interaction coriditions overthe North Sea itself, especially when wind speeds exceed 10 ms '.Ia order to compare five earlier studies with our own, the datahave been expressed in the form in M!au+in b! and are pre-sented in Table 2. The associated value of the quantity a deter-mined from our studies is seen to agree well with those obtainedby Woodcock and Lovett, under sitnilar sampling conditions.

4. Five size cia~mes of the sea-salt aerosol

Recognizing that the pumping system of the multi-stage im-pactor was too weak to sample at the correct rate during heavywind conditions, the S-SAMC data obtained from the multi-stage

R. A44RKS and E.C. MONAHAN

itnpactor were corrected using the simultaneously measured datafrom the isokinetic air filtration unit. This was done by multiplyingthe, original results obtained from each of five impactor stagesQ j,x=1,2,3,4,5! by the ration R=M F!/ M I! where M F! and M l!are the total the S-SAMC obtained from the filtration unit and fromtlie tnulti-stage impactor respectively. Data processed in thistnanner for the five size classes of S-SAMC were plotted in Fig. 2as a function of wind speed u�0! for those samples collected whenthe air advection was fmm the SW-NTAV sector over the NorthSea. The results, delineated by the ten estimates fitted to the S-SAMC data as ln M!, show that giant particles see stage 1!, in thesize range 7,2-30 pm in diameter, increase in concentration withwind speed tnore rapidly than does the concentration of smallerparticles represented by the stages 2, 3, 4, and 5 which correspondto size ranges 3.0-7.2 N,m, 1,5-3.0 gin, 0.95-1.5 pm and 0.49-0.95ptn respectively. This is more obvious for wind speeds l-15 m ' r3nge A in Fig. 2!. The few data collected for wind speeds 15-24ms ' see range B in Fig. 2! are too scarce to give statistically

Fig 2 Five size classesof sea-salt aerosof massconcentration plotted as aln M! versus u�0! windspeed. Measurementstaken at f2 m elevation.Curves are describedinTable 3.

R. A44RKS errd E.C. MOWAHAN

Corr.coeff.

Curve fitCurveNo,

Size cut-off inp.m /stage no.

Data set/Wind range

0.690.550.460,440,16

7,2-30 /13.0-7.2 /21.5-3.0 /30.95-1.5 /4

0.49<.95/5

In M!~-0,1511+0.1143 uIn M!~1.1159+0.0695 uIn M! 0.3783+0.0467 uIn M!~-0.101 1+0.0480 uIn M! -2.0908+0,0620 u

North SeaSW-NNW

31/131/231/331/431/5

Range A1-15 ms'

In M! a-5.3906+0.4550 uIn M! ~-4.5049+0.4543 uIn M! -5.7850+0.4772 uIn M!a-6.3053+0.4774 uIn M! -2.2902+0.1552 u

North SeaSW-NNW

7.2-30 /13,0-7.2 /21.5-3.0 /30.95-1.5 /40,49-0. 95/5

0,940.950.960,970.71

31/131/231�31/4

31/5Range B15-24 ms'

In M!-1. 844+0.104 In W!In M! 2.521+0.079 In W!In M!- 1,295+0.059 In W!In M!-0.801+0.056 In W!In M! 0.043+0,109 In W!

7.2-30 /13.0-7.2 /21.5-3,0 /30.95-1.5 /40.45-0.95/5

North SeaSW-NNW

0.390.370,310.300.45

32/132/232/332/432/5

Range A1-15 ms'

North SeaSW-NNW

In M! 1.844+0.104 In W!In M! 2.521+0.079 in W!In M!~1.295+0.059 In W!In M!~0.801+0.056 In W!In M! 0.043+0.109 In W!

32/132/232/332/432/5

0.390.370.310.300.45

7.2-30 /13.0-7.2 /21.5-3,0 /30.95-1.5 /40.45-0.95/5

Range A1-15 ms'

Table 3. Least squares curve estimates fitted to the In M! data of tive sizeclasses of S-SAMC M figm s! versus 10 m wind speed u ms '! and In W! white-cap coverage W fraction!.

92

three elevations, 4.5 rn, 12 m and 18,3 m, show a marked enhance-rnent in concentration at 4.S rn elevation of the S-SAMC when thewind speed exceeds about 10 ms', and at 12 rn and 18.3 melevations when the wind speed exceeds 15 ms '. The averagevertical profiles of the S-SAMC were calculated using the curvefits 25, 26 and 27 described in Table 4! and are shown for differ-ent wind speed conditions in Fig 5. The profiles illustrate thatwhen the wind speed exceeded 10 ms ' the S-SAMC gradientbetween 4.5 m and 12 m elevations increased dramatically with in-creasing wind speed. At the same time the gradient of the S-SAMCbetween 12 m and 18.3 rn appears to increase slightly with increas-ing wind speed.

BUBBLE-MEDIATED SEA-AIR EXCHANGE

Fig. 4. Distribution ofsea-sa/t aerosol massconcentration at threeheights- 4.$ m, 12 m,$8.3 m above the meansea level plotted asIn M! against u�0! windspeed. Curves aredescribed in Table 4.

6. Variation of sea-salt aerosol mass concentration with white-

cap coverage

Values of S-SAMC M in ggrn'!, at 4.5 m, 12 m and 18.3 rnheight obtained almost simultaneously with whitecap coverage Wfraction! are plotted as ln M! against ln W! in Fig. 6. The S-SAMCdata were again partitioned into low-wind-speed range A! andhigh-wind-speed range B! data. For the 4.5 m S-SAMC data thesplit range A / range B! was made at 10 ms '. For the 12 rn and18.3 m S-SAMC data the split was made at 15 ms'. Curves fittedto the data f'rom both range A and range B and for all three heightsare included in Fig. 6. The results show a marked enhancement ofthe S-SAMC at 4.5 m elevation when whitecap cover exceedsabout 5.73 x 10 which corresponds to W u�0!=10 ms '!! andAat 12 rn and 18.3 m elevations when W�exceeds about 2.03 x 10 which corresponds to W� u�0!=15 ms '!.

To complete the presentation of the results, the S-SAMC dataobtained from the multi-stage irnpactor at 12 m elevation were

93

R, h44RKS and E.C, MONAHAN

Curve

No.Data set Carr.

coeff.Height Wind rangein m u in ms'!

Curve fit

25/A

25/8

North Sea

SW-NNW0.76

6.76In M!-1.2512+0.1793 uIn M!--5.3887+D.7917 u

4.5 1-10

f 0-16

North SeaSW-NNW

0.59

0.95In M! 1.8540+0.0729 uIn M! -3.7904+0.4573 u

1-15

15-24

27/A

27/8North SeaSW-NNW

In M! 1.9355+0.0591 uIn M! -3,6587+0.4403 u

0.45

0.97

18.3 1-15

15-24

28/A

28/BNorth SeaSW-NNW

In M!-3.00+0.048 In W!In M!-9.39+0.852 In W!

0.21

0.39

4.5 'I -10

10-16

29/A

29/BNorth Sea

SW-NNW12 In M! 3.24+0.078 In W!

In M!~9.8f +1.21 8 In W!0.4O

0.88

1-15

15-24

30/A

30/8North SeaSW-NNW

18,3 In M! 3.680+0.f 39 In W!In M!-9.27+ 1 .2314 In W!

0 66

0.95

1-15

15-24

Table 4. Least squares curve estimates fitted to the In M! data of five sizeclasses af S-SAMC M pgrn~! versus 10 m wind speed u ms '! and In W! whitcap coverage W {fraction!,

7. Cone}usion

The sea-salt aerosol mass concentrations over the North Seashow remarkable differences that depend on whether the air isadvected from the English Channel, frotn the S%-NNW sector,ham the AMER'-N sector, from coastal zone, or from land. Thisindicates that S-SA over the North Sea is of a mixed type, originat-ing from both Atlantic and locally emitting sources. Me sea-saltaerosol associated with air advected from the NNW-N sector, as

obtained from the multi-stage impactor at 12 m elevation wereplotted as ln M! versus 1n W! in Fig. 7. 'Ae results, described bythe ten least squares linear curve fits show that S-SAMC within themeas~ size ranges of 0.49-30 p.m stages 1, 2, 3, 4 and 5! under-goes a marked enhancement when W exceeds a value of2.03 x 10' which corresponds as above to W� u�0!=15 res'!!.

BUBBLE-MEDIA TED SEA-AIR EXCHANGE

Fig. 5. Comparisonof vertical profiles ofsea-st' aerosolmass concentration

for different wind

speeds,

aerosol associated with air advected from the NNW-N sector, aswell as Rom the English Channel, were found to have a considera-bly high mass concentration than that associated with air massesadvected fmm other sectors. In both of these high concentrationcases the wind fetch over the sea was not limited by contact withland. This supports the proposition that fetch is one of the pararae-ters influencing the aerosol emission from the sea surface and theaerosol loading of a given air mass. A marked enhancement of the8-SAMC was observed at 4.5 m elevation when the wind speedu�0! exceeded 10 ms' and at 12 m and 18,3 rn e1evations whenthe wind speed was greater than 15 ms '. Ia addition, the verticalgradients of S-SAMC between 4.5 rn, 12 m and 18,3 m were smalluader conditions of light to moderate winds i.e. u�0! less than 10ms '! When the wind speed exceeded 10 ms' the vertical gradientof the 5-SAMC between 4.5 m and 12 m elevation increased

dramatically with increasing wind speed. This could be interpretedas a reflection of the increasingly important contribution to the at-rnospheric salt budget of spume drops, see also Monahan et aL,

BUBBLE-MEDlA TED SEA-AIR EXCHANGE

Fig. 7. Five size cfassesof sea-sa/t aeneol mass

ooncenfration pbtted asJa M! verses whitecapcoverage as h W!.Curve fits are descnbedia TsMe 9.

coverage fraction exceeds a value of 5.73 x 10 for the S-SAMC datameasured at 4.5 m elevation and when %�exceeds about2.03 x 10' for the S-SAMC data measured at 12 I and 18.3 iri eleva-tions. %'ithin the high-wind-speed range, correlation coefficients be-tvmm S-SAMC and W�were found to be even better than between S-SAMC and u�0!, see M W�! and M u�0!! fits in Table 5!. In es-sence, this result is consistent with the recently published work byMonahan, �988!, which suggested that whitecap coverage should bemore directly related to bubble injection rate and sea surface aerosolQux than to either u�0! or wind &iction velocity.

Th Bt tlat Kh i pm flh HEXMAXE pand will be subject to further analysis.

R. hHRKS and E,C, QONAHAN

Sire cut-off Wind speedpoints in u�0! rangep.rn / keight in ms '

Correlation coefficients

Exp. fit Exp. fit Lin. filM U�0!! M Wj M U�0j!

Lin fit

M Wj

7.2-30

12m0.02

0.99

0.78

0.96

1-15

15-240.69 0.07

0.94 0.88

0.05

0.99

0.56

0.95

3 0-7.2

12m0.67

0.96

1-15

15-240.12

0.84

1.5-3.0

12m

0.95-1.5

12m

0.120.99

1-15

15-240.60

0.96

0.46 0,170.96 0.92

0.15

0.99

0.39

0.96

1-15

15-240.21

0.92

0,43

0.97

0.14

0.99

0.49-0.95

12m0.32

0.95

1-15

15-240.16 0.20

0.71 0.78

Totaj mass4.5 rn

0.01

0.35

1-10

10-16 0.59

0.69 0.04

0.66 0.42

Total mass

12 rn0,07

0.99

1-10

15-24013

0.89

0.72

0.96

0.65

0.95

1-15

15-24

Total mass

18.3 m0.47

0.99

0.74

0.96

0.59 0.48

0.96 0.91

Table 5. List of the correiation coefficients of exponentiai and linear ieaslsquares estimates fitted to the sea-sait aerosol mass concentration M! data trtfive size classes and vertical profiies, versus u� 0! wind speed and whitecapcoverage W! data

The research on sea-salt aerosol was supported by the GKSS Re-search Centre Federal Republic of Germany!. We especiallythank Frofessor H. Grassl and Dr. B. Schneider for their svpportand assistance in carrying out the aerosol measurements andanalysis.

The whitecap investigation at the Marine Sciences Lrtstitute,University of Connecticut, is supported by the U.S. OKce of NavalResearch via contract N00014-87-K-0185. We would like to thmkProfessor G.R. Rumney for beneficial comments and Miss M. A.Archer who gled this paper.

NJBBLE-MEDIA TED SEA-AtR EXCHAHGE

References

B Ianchard, D. C. 1963: We electrification of the atmosphere by particles &onbubbles in the sea Progress in Oceanogr. 1, 71-202.

Blanchard, D. C., and Woodcock, A. H., 1980: The production concentrationand vertical distribution of the sea-salt aerosol. Am.N. Y. Ac. Sci., 330-347.

Blanchard, D. C.. and Woodcock, A. H., and Cipriano, R.j., 1984: The verticaldistribution of the concentration of sea-salt in the marine atmosphere nearHawaii, Tellus, 36 B., 118-123.

Chaen, M., 1973: Studies on the production of sea-salt particles on the seasurface. Mew. Fac. Fish., Kagoshima Univ. 22., 49-107.

de Leeuw, G., 1986: Vertical profiles of giant particles close above the seasurface. Tellgs, 388, 51-61.

Exton, H3., and gotham, J., and Park, P.M., and Smith, M.H., and Allan, R R.,1983: TIM: production and dispersal of marine aerosol. Oceanic whi tecaps,E.C. Monahan and G. Mac Niocaill eds.!, 175- 193.

Garbalewskl, C., 19&0: Dynamika morskich ujdadow rozproszonych. Inst. ofG>~enology, Polish Academy of Sci.

Koga, M., and Taba, Y., 1981: Droplet distribution and dispersion processes onbreaking wind waves. Sci. Pep. Tohoks Uruv.. s. 5, Tohoku Geophys J.!2&, 1-25.

Koga, M., 1984: Dispersal of droplets over being wind waves under thedirect action of wind. J. Oceanogr. Soc. Japan, 40, 29-38.

Lover, R.F., 1987: Quantitative measurement of airbcmc sea-salt in the NorthAtlaatic. Tellus, 30, 350-364.

Marks, R., 1987: Marine aerosols and whites in the North Atlantic andgreenland sea regions. Dt. hydrogr. Z 40, 71-79.

Monah;m, E. C., FairalL, C. W., Davidson, K. L., Boyle, P.J., 1983: Observedinter-relations between 10 m winds, ocean whitecaps and marine aerosols.Quart.J. Roy. Meteor. Soc., 109, 379-392.

Monahan, E. C., 1988. 'Whitecap coverage as a remotely monitorable indicationof the rate of bubble injection into the oceanic mixed layer. Sea SurfaceSomsd, B. R. Kerman ed.!, Kluwer Academic Publisher, Dordrecht, 85-96.

Pmecbrxdnnskii, L. YtL, 1973: Estimate of the content of the spray-drops in thenear-water layer of the atmosphere. Fluid. Mech. Soviet Res., 2, 95-100.

Strarnska, M., 1987: Vertical profIles of sea-salt aerosol in the atmospheresurf~ hyer. a numerical model. Acta Geophys. Pol., 35, 87-100.

Woodccek, A. H., 1953: Salt nuclei in marine air as a function of altitude andwind force. J. Met. 10, 362-371.

Wu, J., 1979: Spray in the atmospheric surface layer. review and analysis oflabceatory and oceanic results.J. Geophys. Res. 84, 1693-1704.

P.G. MES TAYER

defining a fuHy difFerent domain of study of the air-sea interac-tions: most of the relationships that describe the surface transfersin calmer sea states are no longer, or not fully, valid. The dynae-ics of the wave be"king have a direct influence on the generationof currents and turbulence in the water body, hence on 61 phenom-ena depending on the mixing of oceanic upper layers, The surfacegeometry and its disruption probably have some influence on thedynamics of the atmospheric surface layer and on surface windstress. Moreover, breaking waves disrupt the chemical and organIcsurface films; they produce and they entrain air bubbles in thewater body. Due to their entrainment under the surface, theirphysic+chemical exchanges with water at various depths, andtheir bursting at the surface, these bubbles are responsible for sea-to-air exchanges of gases, moisture and dry nuclei that cannot bedescribed by the same relationships as the surface transfers.

The interconnections of phenomena of very different natures,e.g. the strong influence of the water and surface-film chemistry oagas «nd moisture exchanges, increase the complexity of sea-ta-airexchanges by about one order of magnitude. The scarcity afreliable observations in the open sea, in addition to the largenumber of variables known to be strongly interconnected tDakes itdangerous to deduce relationships from mere statistics.

Gn the other hand, it has often been claimed that this largenumber of inuwwnnected variables makes it useless to study anypart of the phenomena in isolation in the laboratory. Ihis point ofview is beed on the idea that the results from incomplete simula-tions are statistically nearly never representative of the "realworld" or that their representativeness is, at best, poorly known.Nevertheless, it is true that in the laboratory the purely statisticalapgeoach is quite handy but of little use when the laws of similaritywith "real worM" phenomena are either unknown or inconsistenL

-ooo-

The most usual approach in tunnel simulations can be caUedthe "scale model." The pvceesscs under scrutiny can be describedby relationships that can be written in a dimensionless form bymeans of a limited number of independent scaling parameters, that

P.G. A4ESTAYER

velocity k� is a normalized surface flux defined by:

L FS G Gw! �!

where F is the gas flux through the area S and c and c are the gasconcentrations in air just above surface and in the water just undersurface, respectively. For U 9 ms', they observe clear linearvariation of k�with wind speed for both gases. But for higherwind speeds U > 9 ms'! they put in evidence a sudden jurnp ofthe values of k�up to a factor of 3, due to the outset of breakingwaves and the entrainment of air bubbles through which extra gastransfer takes place.

Mernery and Merlivat's �983! work represents a typicalexample of the "process study" approach. Facing this similaritylimit, they gave up the similarity concept sustaining relation �!, todesign a model of the bubble contribution to the flux itself. Pro-vided relatively crude assumptions on the bubble dynamics in thesub-wave turbulent field, the bubble generation process and thebubble spectral distribution, this model focuses on the water-bubble gas transfer and the bubble transport to the surface. Itemphasizes the important role of gas solubility and the differenceof the role of concentration gradient in wave-wind tunnel and atthe ocean-atmosphere surface. Finally it introduces the effect ofsurfactants. The tunnel data, that were obviously dangerous todirectly "export" to the ocean conditions, provide a good test bedfor the model that is shown to be in good agreement with the rareobservations made at sea.

5. Various attempts to describe the near-surface bubble popu-lation by general similarity laws have been recently summarizedby Wu �988!: the bubble concentration would vary proportionallyto the exponential of depth z and to wind speed at power 3,5, itsspectral distribution being universally proportional to r' for radii rlarger than 50 ljrn. Unfortunately these relationships do not appearto be clearly supported by physical bases but result essentiallyfrom statistical combinations of experimental data, Moreover theexperimental observations are not numerous, their results are quitescattered and they differ in nature, e.g. measurements of global

BUBBLE-MEDIA TEO SEA-AIR EXCHANGE

concentrations, of bubble spectra, of acoustic scattering crasssection, so that they take a common statistical significance onlyinside the author's reducing and a-priori "model", The handinessof such relationships should not mask the fact that although theylook like similarity laws, they are not.

To explore the mechanisms of generation and dispersion of thebubbles and their significance for predicting bubble populationsclose under and at the surface, Baldy �987, 1988! and Baldy andBaurguel �985, 1987! adopted a "process study" approach inadequacy with IMST tunnel emulation capabihties, In a first stageBaldy and Bourguel �985! developed a technique to estimateseveral local statistical characteristics of bubbles produced bybreaking waves in the tank, such as concentration probability func-tions, crossing frequencies, arrival time intervals and speed distri-butions. This technique is based on the adaptation of a laser-basedsingle-particle probe previously studied by Avellan and Resch�982! to the flume experimental conditions of sampling time,bubble density and probe position. But an important part of thetechnique stands in the original software that they specially de-signed for IMST Luminy Laboratory numerical system to acquireand process on-line the probe peculiar signal, in order to mcard theonly data necessary to estimate the upper-mentioned statisticalcharacteristics. In particular this real-time software accounts forthe laser beam light nonuniformity, oblique crossing and multiplereflexion problems.

In a second stage Baldy and Bourguel �985 and 1987! fullydocumented a small number of breaking wave conditions, at alldepths from the deepest bubble observation to the surface, witheven measurements between wave troughs and crests. The highstatistical significance of the data thus obtained allowed Baldy�988! to ascertain the relative importance of the twa main mecha-nisms, production and dispersion, in two different depth zones.

Because the statistical characteristics obtained by the rneasure-ments are in principle exactly representative only of the docu-mented flume breaking-waves, the last stage consists of incor-porating understanding of the mechanism so deduced into a gen-eral model of the generation-dispersion process, and to test it

Bonmarin, P. and A. Ramamonjiari3oa, 1985: Defctrmation to btea4ng waves ofdeep water gravity waves, Eqerirrtena in Flsub.3. 11-16.

Coantic, M., A. Ramarnonjiarisaa, P. Mestayer, F, Rescb snd A. Favre, 19&1:Wmd-water tunnel svnuhtton of smaH-scale ocean=atmosphere tntcracttons.J. G~ophys Res., 86, 66074626.

Coupiac, C., 1979. Contribution a l'etude de 1'influence dcs vagues sur 1'evapo-ration. ~se de Docteur-lngcltieur, INST, Univ. Aix-Msrseillc ll, France.

Davidson, K.L. and A.J. Frank, 1973: Wave-related fluctuations in the air-Howabove natural wavcs.I. Phys. Ocean., 3, 102-119,

de Leeuw, G., 19&6: Vertical proAles of giant particles close above the seasurface. Tellar 388, 51-6 l.

dc Lceuw, G. 1987: Near-surface particle size distributicN pedes over theNorth Sea.J. Geophys. Res.,9R, 1463 l-14635.

Edson, J. B., 1987: Lagrmngian model simulations of the turbulent transport andevaporation of spnly droplets in a wind-wave tunnel. Unpublished M.S.'Oasis, Depl of Meteorology, Pennsylvania State University, UniversityPark, PA 16&02, U.S.A.

Fairall, C.W. and S.E. Larscn, 19&3: Dry deposition, surface production anddynamics of acrosals in thc marine boundtrry layer. Atrn0s. Envir, 1&,69-77.

Fairall, C.W., KL.. Davidson and GZ. Schachcr, 1983, An analysis of thesurface production of sca-salt a.molL Tellus. 358, 31-39.

Fairall, C.W., K J.. Davidson and G.E Schacher, 1984. Apphcation of a mixedlayer model to emeols m thc marine boundary layer. Telttrs, 36b, 203-211.

Fairall, C.W.. J.S. Edson and M.A, Miller, 1988: Heat fluxcs, whitecaps and seaspray. Reread sttLring in the air-sc'a interactions. Plant and Gcernhaert, cds. in press!.

Fairall, C.W., S.E. Lttrscn, P.G. Mestayer and D.E. Spiel, 19&6 Measurementsof spray droplet transport and cvttporadon during HEIST, Sixth AM.S.Conference of Ocean-Atmospl cre Interaction, Miami, Florida; Jan. 13-17,Bell. of American Meteorological Society.

Franccy, R.J. and JX. Garrat, 1978: Eddy flux mcasun~mts over the ocean andrelated transfer coefficients. Botuukw'y Luyer hfetcorot. l4, 1S3-166.

Friehe, C.A. and K.F. Schmitt, 1976: Par3mcterizatim of air-sea interface fluxesof sensible heat and moisture by the bulk aerodynamics formulas.J. Phys.Oceae. 6, 801-&09.

KaLsatos, K.B., S.D. Smith, and W.A. Oost, 1987: HEXOS � Humidity Ex-change Over the Sea, a prognam for research on water-vapor and dropletfluxes frocn sea to air at moderate lo high wisd speeds. Ball. Abort. Nereorol.Soc. 6$, 466476.

Kharif, C., 19&7: A comparing between three and two dimensional instabilitiesof very steep gravity waves.J. Theorist. AppL hfecA. 6, 843- &64.

Kharif, C. and B. Roux, 1984: Stability if steep gravity waves. J. Thcorcr.Appkf. Mech. 3, 717-723.

Kharif C and A. Ranuunonjiarisoa. 1988: Deep-water gravity wave instabilitiesat large sttxpnets. PAys. Fluids Xl, 12&6-128&.

BV88LE-MEDfA7KOMA-rhfR EW'CHNVQf

Tech. Rep., Dept. Atmos. Sci., Univ. of Washington, Seattk,WA,Viguier, P., ! 985: Urc etude des perturbations supt:rhartnceiqses d'onde de

Stokes. These de Doctorat, IMST, Univ. Aix-Marseiile II, France.Wu, 3,, l988: Bubbles in the near-surface ocean; a general description.J.

Gropes. Res. 93, 587-590.

119

BUBBLE-MELMA TEOSEA-AlR EXCHAhA3E

12 Bulk transfer perspectiveBy correlating simultaneous measurements of vertical velocity

and specific humidity fluctuations, the latent heat flux for ex-ample! can be determined near the surface

H�= pL,w'q'

whew p is the density and L the latent heat of vaporization ofC

water. Through similarity theory FairaU et al., 1987! the surfaceAux, H, can be expressed in terms of bulk atmospheric proper-ties and the transfer coefficient, C<

Hw � � PL, CFQ[f � g z!]

where 6 is the mean wind speed and q the specific hurmdity «theight z, and q, the surface specific humidity. C can be thoughtof as the efficiency of moisture transfer fmm the sea surface inresponse to the forcing by windspeed and air-sea humidity differ-ence.

The bulk moisture transfer coefficietit has a value on the order

of lxlo for measurements made at 10 rn, and is expected to de-crease slowly with increasing windsgeed over ice Joffre,1982! orover water Liu et al., 1979! because of' increased sheltering of thesurface by the roughness elements. The effects of sea spray onevaporation of the ocean can be expected to show up as an en-hancement of the 'drop free' CE with increasing windspecd. 1n thesurnrnaries of transfer coefficient measurements given by Ander-son and Smith �981! there are very few measurements for windsabove 10 ms ' and most of those are not from the open ocean biitfrom beach sites where there are probably surf effects. Franceyand Garratt �979! found both transfer coefficients to increase withincreasing wind speed; surprisingly, the sensible heat coefficientincreased faster than the moisture coefficient. In a recent survey ofbulk parameterizations by Blanc �985!, only one of ten schemesprojected scalar transfer coeNcients that decreased with increasingwind speed.

BUBBLE-MEDIA TEO SEA-AIR EXCHhMGE

typically 1/10th the size of the parent bubble those which producejet drops range between 0.1 to 2.0 mm diarnctcr range!. Onc tofive jet preps axe produced per bubble while the much sxnaller filmdrops axe produced in the hundaxis. The mtc of production of jctdrops on a microphysical scale is much better known than that forfilm drops, which is still the subject of debate c.g., Cipriano er aL,1987!. It is now known that the larger size droplets greaer than10 pm radius! doiriinate the liquid water production by sea spray Stramska, 1987; MiHer, 1987; Edson, l987!, so our poor under-standing of fiLm droplets will not handicap the analysis of theeffects on the heat fluxes the same is not true if one's interest is inAitken nuclei, optically relevant aerosols, or cloud condensationnuclei!.

There is also considerable evidence that bubbles axe not the

only source of droplets. At wind speeds in excess of 13 ms ' themis a rapid increase in the observed sea salt aerosol concentrations atlarge sizes Monahan er al., l983; FairaO et al., 1983!. It has beenpostulated that this increase is due to the additional production ofdroplets by the so-called spume' mechanism where the strongturbulence simply blows the foam patch right off the top of abreaker, This phenomenon can be easily observed at high windsand its appearance threshhold constitutes a criteria fm' sea state 7.

22 Oceanic droplet source strengthThe droplet surface source strength is crudely defined as the

number of drops of a given size interval produced by each squarecentimeter of the ocean per second. Clearly, this strength is afunction of sea state as chaxmcterized by whitecap fraction and/orwind spcxxL Additional information is require to define thesource function because at each size the particles are ejected witha distribution of initial vertical velocities Blanchaxd and Wood-cock, 1957!. The usual approach is to assume that the dropletsmagically appear at the top of their inost probable trajectory Edson, 1987!, In other words, the droplets are treated as beingcreated by a distribution of elevated sources. The typical ejectionheight is on the ceder of 5 cm. This apl~ach is justifiable becausethe time scale for this pa~ss is quite small compared to the

BUBBLE-AttEDIATEOSEA-A!R EXCHANGE

3. Droplet Microphysics

3.1 Evaporation

Consider a single droplet of mass, m, given byP

4M = � CP

P 3 P�!

We assume that the particle is a saltwater solution so that

Pp=P + Po P�!r-o/r �!

where p is the density of water, p, the density of the particle withall water reinoved to produce a dry rmdius, r,

If the water vapor pressure exerted by the droplet, e, is greaterP

than its surroundings, e, then the droplet will evaporate at a rategiven by Pruppacher and Klett, 1978!

r/mz///t = 4rr fp�r[ez � r,r~,T<! � e]/ R,T! �!

where f is a ventilation factor. The dmplet vapor pressure can beconvert@ to an effective saturation vapor density or specifichumidity, q, at the surface of the droplet to yield

Bmz/Bt = -4rr f g�rp qz-q! f7l

From Fitzgerald �975! we estimate q byp

127

as rneasuremefits of oceanic bubble spectra become more reliableand reproducible. A comparison of recerit results Monahan, 1988;MiBer and Fairali, 1988! hem all three methtxis suggests that, asthe methods are refined, the agreetnent is improving and a reason-able consensus is anticipated in the near future.

C.W. FAIRALL and J.B. EDSON

Figure 1. Sample@praydropletconcentrationparameters from theHEXIS T experiment upper panel!. Theconcentration variablesare expressed as thevolume pm'! per radiusincrement turn!. a! Droplet volumeconcentration votflurf'/cm! versus radius upper pane/!,

b! Droplet sourcestrength, s�, as afunction of droplet radiusfor the spray bubblersused in the HEXISTexperiment. The sourceis expressed in units ofdroplet volume ym~!per square centimeter ofocean per second perpm radius increment.

BUBBLE-MEON 1' SEA AIR KXCHNVGf

i> -At/i~- � e! r � � a!� � y/ r/ r~! � 1]! � 5!

whereA = f,D�p/p �1!

The factor a is necessary because the evaporating droplet is as-surncd to be at the 'wet bulb' temperature Pmppacher andKlctt, 1978!, T�,

a = L, / R�Ts! T � T ! �2!

Edson �987! has shown that thc time for even 100 pm droplets toreach the wet bulb temperature is small compared to the turbulenceintegral time scale at the particie ejection height.

32 The particle size spectral densifyR Section 4 we wiH discuss the transport of particles or dmp-

lcts in o~s of a particle concentration variable. A variety of

where q is the saturation value for pure water with no surfacecurvature at the droplet temperature, T, and y a parameter thatdepends on the chemistry of the dry componc:nt y = l for sea salt}.

A spray droplet pleased from the ocean wiD lose; water massby evaporation until it approaches a state of equilibrium. A purewater chaplet wil! completely evaporate but a salt water dropletwill usually retain a considerable amount of water in equilibrium.The equilibrium condition is defined by setdnll Bm /ck&. Fmm�! and 8! this defines an equilibrium particle size, r,

r,/r,=G, S!=[ +T/�-S!]'" 9!where we have assumed that T I in equilibrium and S is the

P

ambient water vapor saturation ratio.The rate of change of size of an evaporating droplet is obtained

by taking the derivative of �! and combining it with �! and �!;this can bc written

C.W. FAIRALK anct J.B. EDSON

rticle concentration variables are used in the literature, the sim-:st is the total number concentration, N{r!, which is the total mber of particles per unit volume of air with radius smaller thanThe size spectral density, n r!, is the number per unit volumeIth radius greater than rW/2 but less than r+dr/t2 and is directlylated to N r!

�3!N r! = I n r'!dr'These distributions describe the salt water solution droplets as

.ey appear under ambient conditions. We can define a quasi-!nservative variable in terms of the mass of salt in the particlest', equivalently, the dry radius of the particles. For example, wem define the size spectral density that would result if the particlesere dried out completely as n r !. The total number of droplets

0 0er unit volume with dry radius less thanr is N r ! sothat

0 0 0

N, r,! = I 'n r~! dr~ {14!

Ensemble Average Models

.l Budget equationsFollow ring Fairall et al. l 987!, we can write the simple

ane&mensional, ensemble average budget equations for thetankard. meteorological variables as

DQ/Dz = r/ w' e' !/ dz � � L,/czjE {15a!

�5b!Dq�/ Dz = d w'q�'!/ z/ z + E-

Dq,/ Dt = � z/ w'q>-< w,> q>j/8 z E� �5c!

where the primes denote turbulent fluctuations, e is the potentialemperature, q the specific humidity vapor!, q the specifictumidity {liquid!, cw > the volume averaged fall velocity, and E

8he total evaporation rate

BUBBA-MEOIATEO SEA-AIR EXiW4PAE

E = � �zzp� I p! f r irn r!+ rr'n' r!] dr �6!

�-nP.»!f " "-.!n .!d =�np�i3!fr n r!dr �!

The droplets interact with the basic thermodynamic variables �5aand 15b! through the droplet evaporation term E!.

Ir! the same simplified 1-dimensional form, the particle bodgetequations become

Dno I Dz= 8[w'n� '+wzn' w, n � ]Id z+z �8a!

Dn I Dz = r7 r'pn+ -r'zn'! / Br ol[w'n'+ -w'n' w,-n'! i g z+z��8b}

where w denotes the particle slip velocity relative to the fluid!.

Standard formulae are available for the mean value of w as a

function of particle size Pruppacher and Klett, 1978!. The con-centration-slip covariance term is responsible for the inertialimpaction deposition mechanism Slinn et al., 1978! and also leadsto reduced turbulent transport for large particles.Again, note that the terms believed to be negligible above themolecular sublayer have been dropped From these expressions.Also note that the basic liquid water conservation equation �5c!can be obtained from �8b! by converting number density to massdensity and integrating over all radii.

While n is a, quasi-conservative variable, n r! is not because0

the dmplets change their size whi1e evaporating. Thus, �8a! is asimple conservation equation, but the analogous equation for thenon-conservative variable, n r!, requires the additional radius timederivative terms on the right hand side. Hence, two budget equa-tions are required, one to keep track of the salt �8a! and one to

131

Note that the molecular diffusion terms have been dropped becausethey are only relevant in the diffusion sublaycr within a mm of theinterface!. The liquid water content is the integral over the dr!opletdistribution

C. LV. FAIR4LL and J.B. ED9OH

keep track of the salt plus water �8b!. Notice that in order to use IG!, we must be able to specify the value of r that is appropriMfor each value of r. This is done by finding the value of r such that

N r! = N0 r~! �9!

42 Madeli jig the covariance termsEnsetllble average models are often classified by the method

used to mode1 the covariance terms that appear in �5! and �8!.Wc could form budget equations for the covariances, but thesewould contain third-order covariances. lf we termnatc this infmitchierarchy of equations by approximating the third-oCer terms ascoinbinatioas of second and first-order variables, then this is re-ferred to as a second-order closure model. A simpler approach is touse the conventional eddy-diffusion coefficient or, first-orderclosure! model, often referred to as K-theory.

Following the aiialogy of the molecular diffusion flux expres-sion, thc turbulent Aux is written

'q�' = � ECp& q�/ 8 �0!

where K� is the scalar gr3dient diffusion coefficierit. 4'ithin tmierealm of Monin-Obukhov similarity theory Fairall et al., 1987!.wc can write

K],= rzu /4], z/L! �1!

Kz �. Nzu, �2!

132

where u. is the friction velocity, k the von Karman constant �.4!,L the Monin-Obvthov stability length scale, and N� the dimen-sionless scalar gradient function. Since the droplet effects on theheat fluxes will only be important under rather strong windconditions, we can be confident in using the neutral approximatioo

to �1!

C, W. FAIRALL ancf J.B, EDSON

using a surface layer first-order closure model developed speciA-cally for this application Ling and Kao, 1976; Ling et al.,1978;Ling er aI..1980!. The budget equations were non-dimensionalizedusing wave height, wind sped, and air-sea temperature and hu-midity differences and solved for equilibrium conditions zem timederivatives!. The earlier work used only a single, Axed dropletsize, but 1ater papers allowed 5 droplet sizes �, 20, 40, 70, and 150trmn radius!. The droplets were assumed to be pure water so �8a!was not used and the r'g' covariance term was neglected. Thesurface source function, based on laboratory measurements. wasassumed to have a quadratic windspeed dependence but a windspeed inchqendent shape. Since the surface source function ap-pears to bc much larger than those discussed in section 2, it is notsurprising that substantial effects of droplets on the surfaceevapmition were found.

Strarnska l 987! has developed a K-theory model to study seasalt aerosol profiles that falls somewhere between the Burk andLing models in philosophy. As did Burk, Strarnska used the sur-face source model for particles smaller than 15 p.in radius fromMonahan er al. ]982! and assumed that these particles are in anevaporative equilibrium state. However, the evaporation necessaryto maintain this equilibrium is allowed to feedback onto the rnois-turc and ternpcrature profiles. As it turns out, these particles pro-duce a negligible effect on thc incan scalar profiles this is consis-tent with conclusions of Ling er al., 1980!. The e6'ects of largerdroplets «t high wind speeds �0 ms '! were examined by introduc-ing an «d hoc droplet pmAle based on near surface data and anassuined denea!a in the vertical. An evaporation equation similarto �0! was used assuming the droplets were pure water. This ledto an increase in temperature of about 2' K and an increase inhumidity of about S%. Stramska did not assume dynamic equilib-rium but started with «n initial profile and integrated the budgetequations in time. 'This permitted a study of the equilibrium re-sponse time of the aetmxHs as a function of size. The results werevery similar to that of FairaH ct al. �983! with 10 pm radius par-ticles requiring a few hemrs to reach dynamic equilibriurri. Thelarger the particle, the shorter the response time because the

removal process gravitational fallout! inneaes with size.

5. Monte Carlo Models

dw w� = � + g r!dh �3!

where w is the particle's vertical velocity', t is a time scale for themaroon, and t; t! is a random forcing function due to turbulence.

The above authors assume that the motion of the particles ispassive, i.e., they assutne that the particles are of such size thatthey follow the turbulent motion of the atmosphere exactly. Withthis assumption, the tirnescale in �3! is the Lagrangian integraltime scale of the turbulence, and the random forcing function isderived observing the constraints that

w' t! = 0 �4a!

�4b!

where a is the standard deviation of the atmospheric verticalvelocities.

The 'passive contaminant' assumption is not valid for the jetdrops -10 � 100@m radius! that wc are interested in for the seaspray problem. Gravitational and inertial effects tnust be includedin order to realistically sitnulate their trajectories. This is

135

5.1 Lagrangian Markov chainThe most common form of Monte Carlo simulation applied to

transport of droplets is the Lagrangian i.e., a reference framemoving with the particle! Markov chain. Maxkov chain simulationswerc used by Reid �979!, Legg and Raupach �982!, «nd Ley andThomson f l983! to successfully model dispersion of neutrallybuoyant particles within the surface layer. The Markov chain is afinite difference form of the Langevin equation; the Langevinequation being given by

BUBBLE-NEON TEO SEA-AJR Bf CHNWr 8

Rgure 2. Lagrangian rnocfel simulation of $0 droplet trajectories for theHEXIST experiment at 9 nv's no&nal eind speed. The vertical axis is fhe~t height cm! and the horizontal axis ks the droplet distance cm!downwind of the source The upper pane/Is a famI/y of trajectories for a 50lute radius droplet, the lower panelis for a 10@m radius droplet Both arereleased from t 0 cm for co~erison.

139

Rpure 3. Drog&t volume corcentratbe spectre vs. he/At prof' s tor hvoselect& size droptets from EXIST. 7' lines denote smoothed fIts toIneaSured data, the Symteks denOR Vat~ Sknulated With the L,agranglanmode/ averaged over 6 e moW runs. The b~ Irpd<Ar the standard4evhtfons from the mean vabes. The cancNions for th s example are aeohthe humkNy near 700%, no&nal 4odspeecf of 9 ms' and lnitia! edtuset 15 and 40 yrn for the two draplst sizes.I'a! aownwlnd fetch of 5 m. bf dolewvlnd fetch of 9 rn.

141

C. W, FAIRALL and J.B. EDSON

R vM 4. ~ <n FigUre 8 ~ fof Overrating norY+7ai retat~ ~~~ near50%j concNjons. The initial radii at time of reiease ere again 15 end40 pm

142

BU88LE-MEDM TKO SEAM& BCCNhhAK

contribution to a total latent flux of 225 %/rn ' has been computedto be on thc otdcr of 1 W/m ~. For the HRXIST situation thedmplet liquid water mass was 10 of the- avqxx coeamration,Qt'hich is quite lnodcst Irt a subsequent exptvhlhc+t namedCLU$E!, the entire surface of thc water in the tunnel was turnedinto a whitecap. This yielded droplet concentrations and latent heatfiux contributions more than two orders of magnitude greater.

Thc potential droplet contributioa can be also crudely csti-rnated by integrating thc souse function Fig. lb! over radius toobtain the total liquid ~ater sea spray Aux. This implies that thesea spray would contribute 785 W/m' to thc total latent heat Aux ifthc ocean were 100% whitecap covered and all droplets evaporatedbefore impacting the sea surface. For the open ocean, this nurrtbcrmust be considered to be uncertain by at least onc half an order ofmagnitude. The HEXIST experiment which had 1.5% 'whitecap'coverage and, hence, an 11 W/m potential! suggests that about10% of the spray liquid water was converted to vapor by

Fgure S.Est/mate ol the

drop!et ee-trkvthe to The

total waletransfer 5'

dowmehd of

the eh~for the 0 re

HEX5TsknvAN'ore

vaiAg thesward term

frrprn PZ!. Thedf~ eyertransfer ts

estirnahpd at225 NM.

143

evaporation.While the EXIST and CLUSE measurements and model

studies represent idealizations fresh water evaporating in a windtunnel!, the implication is that droplets will make a significantcontribution to the total evaporation when the wind speed ex-cd approximately 15 ms'.

Acknowledgments

This work is supported by the Office of Naval Research contract N00014-85-K-0250!. The authors wish to acknowledgecoHaboration with Soren Larsen of RIQ National Laboratory inRoskilde, Denmark; Patrice Mestayer of IMST in Marseille,Fr3rtce; and Ken Davidson and Don Spiel of NPS in Monterey,California, USA.

References

A~ron, R.J. and S.D. Smith, 1981; Evaporation coefficient for the sea surfacefrom eddy-flm mmmernents. J. Gcophys. Rcs., 86, 449456.

Blanc, T.V.,1985: Variation of bulk-derived surface flux, stability, and rough-ness results due to the use of different transfer coefficient schemes. J. Phys.Ocean.. 15, 6%1469.

Blanchard, D.C. and A.H. Woodcock, 1957: Bubble formationin the sea and its meteorological sigmficance, Tclhas, 9, 145-158.

Bhnchard, D.C., 1971: White+ at sea. J. Arrrros. Sci. 28, 645.Blanchard, D.C� l 975: Bubble scavenging and the water-to-air transfer of

organic material in the sea. Adv.iver Cherrr.,145, American Chemical Society,Washington, DC.

Burlt, S.D., 1984: The generation, turbulent transfer and deposition of the sea-salt aerosol.J. Areos. Sci., 41, 3040-3051.

Cipriana, R J. and D.C. Blanchard, 1981: Bubble and aerotel spectra producedby a labor3axy breaking wave.J. Gropes. Rcs., 86, 8085-8092,

Cipmtno, R.j., E.C. Monahan, P.A. Sawyer, and D,K. Woolf, 1987: Marinecondensation nucleus geseration inferred from whitecap simulation tankresults, J. Giophys. Res., to appm.

de Leeuw, G. 1986: Vertical profiles of giant particles close above the seasurface. Tcl us, 38B, 5141.

Edson, J.B., 1987. Lagr3ngian model sirnuhtions of the turbulent transport and

BUBBLE-NEO' TED SSi~ Erma~

evaporation of spray droplets in a wind-water tunnel. Mg. thesis De~.ment of Meteorology, Pennsylvania Stale University, university park, pAIQN2, USA.

Edson, J.B., C.W. Fairall, S.E. Larsen, and P.G. Mcstayer, 19gg; A rajttkxn w~simulation of thc turbulent tnLnspxt of cvttportIing Jetsimulation tunnel during HEXIST. Proccal 7th Conf. on Ocean Atnu .phcre Interactions, AMS, Anaheim, CA, 9-13.

Fairall, C.W., KI.. Davidson, and GZ. Schachcr, }953: An analysis of thesurface production of sca-salt acrosoL Talbot, 358, 31-39,

Fairail, C.W., J.B. Edson, and M.A. Miller. 1988: Heat Iluxes, whitecaps, andseaspray. Tcchnical rIyart. Department of Meteorology, Pennsylvania StateUniversity, University Plk, PA, 16802,47 pp.

Fitzgendd, J.W., 1975: Approximation formulas for the equilibrium size of anmosol parhcle as a function of its dry size and composition and theambient relaove humidity. L Appf. Nct., I1. 1044-1049.

FrtLncey, R.J. and J.R. Garrett, 1989: Is an ol.served windspeed dependence ofAMEX '75 heat transfer coefficients real'P Bowed.-layer Nctcorof.,16,249-260,

JofIre, S.M.. 1982; Momentum and heat transfers in the surface hyer over afrozen sea. Bound.-Layer MetcerrN.,24, 211-229.

Legg, B.J., and M.R. Raupach, 1982: Markov~ simulation of particles ininhomogcneous flows; 'Tle mean drift velocity induced by a gradient inKulerian velocity variance. Bo&.-Layt.r N'eteorof.,?4, 3-13.

My, A J., and D.J. Thomas, 1983: A random walk model of dispersion ia thediabatic surface layer. Quart.J.R. MeteoroL So@., 109, 867-880.

MM,B.C.~T.W. K,f978~ fftransfer pm' over the ocean under whitecap sea states.J.Phys.Oceanogr., 6, 306-315.

Ling, S.C., T.W. Kao, M, Asce and A. Saad, 1980: Micl~lets and transiertof moisture from the ocean. J, Eng. Mech. Dtv.,6. 1327-! 339.

I ' .WT,KB K .~I WB I 797'9' 8 .KI~fair-sea exchanges of heat and water vapor inclgding the molecular con-straints at the interface.J. flees, Sci., 36, 1722-1735.

Meek, C.C., and B.G. Jones. 1973: Studies of the behavxr of hcavy particles ina turbulent fluid flow.J. Aoeos. Sci.,30, 239-?44.

Miller, M.A., 1987.' Modeling the prod':tion and transport of sca-salt aemsols.M.S. thesis, Department of Meteorology, Pennsylvania State University.University Park, PA. ]6802. USA.

WU M.B.~CW.B~I.f9888 I IK IWWIproduction by oceanic whitecaps. Frt:hei 7th Conf. on Ocean-Atmos-phere Intetmctions, AMS, Anaheim, CA, 174-177.

Monahan, EC., 1988: Whitecap Coverage as a Remotely MonitorabIe Indica-tion of the Rate of Bubble Injection into the Oeztnic Mixed Layer, in SeaSwface Sour', BR. Kerman. M, Kluwer Academic Publishers, Dor-drecht., 153, 85-96.

145

C.W. FAIRALL and 3.8, EDSON

Mceahan. E.C., K J.. Davidson and DX. Spiel. 1982: Whitecap aerosolproductivity deduced from simulation tank measurements. J. Gcophys.Res., 87, 8898-8904,

Mon'than, E.C., C.%'. Fair3LL.K.L. Davidson, and P.3, Boyle, 1983: Observedinter-relations between 10 m winds, ocean whitecaps and marine aerosols.0 mrt. J. R. Met. Soc., 109, 379-392.

Moruahau.E.C., 1988: Whitecap Coverage as a Remotely Monitorable Indica-tion of the Rate of Bubb!e Injection into the Oceanic Mixed Layer, in ScaSurface Sound, 83K Kerman, Ed., Kluwer Academic Publishers, Dor-drecht., 153, 85-96.

Nicholls, S. and C.j. Readings. 1979: Aircraft observations of the structure ofthe Lower boundary layer over the sea. Qzxrt. I, R. Mci. Soc., 105, 785402.

~,L..I973P fu fW |dw dwater layer of the atmosphere. F/kaid Afech. Sov. Rcs., 2, 95-100.

Pruppacher, Mt. and J.D. Klett, 1978: Microphysics of Clouds and Precipita-tioa, Reidel, Khxdrecht, Holland.

Reid.J. D., 1979: Markov chain simulations of vertical dispersion in the neutralLayer for surface and elevated releases. Bound.-Layer hfefzorol., 16, 3-22.

Slinn, W.G.N., L. Hasse. 8.8. Hicks, A.W. Hogan, D. Lai, P.S. Liss, K.O.Munrtich, G.A. Sehrnel, and O. Vittori, 197': Some aspects of the tr3nsferof atmospheric trace constituents past the air-sea interface. Atmos. Environ.,1X, 2055-2087.

Strarnska, M., 1986: Vertical prohles of sea-salt aerosol in the atmospheresurface layer. a nutnerical model. Acra Gtophys, Pol., to appear.

Woolf, D.K., E.C. Monahan, and D.E. Spiel, 1988: Quantification of the marineaerceol produced by whitecaps. Proceed. 7th Conf. on Ocean-AtmosphereInteractions, AMS, Anaheim, CA, 182-185,

Wu, J.,1974: Evaportttion due to spray.J. Geophys. Rcs., 79, 41074109.

formation. Figure 2 shows model calculations of xT and t', fortypicml high !atitudc, polar low conditions. From this and similarfigures not showa, we see that t is always at least three orden ofmagnitude smaller than t. Thus, for spray droplets, the sensibleand latent heat transfers are essentially decoupled. In effect, theambient relative humidity has negligible effect on the thermalevolution of spray droplets; and the air-sea temperature differencehas negligible eff'cct on their moisture size! evolution.

Ia Figurc 2 I have also plotted the time constant xz the timerequired for a dmplet of radius r, to fall l meter in still air. If x, ismaUcr than t, or x, a spray droplet will likely fall back into thesca before transferring its available heat or moisture into the air.

Figure 1. Conceptual model o/ Neat and moisture transfer associated withsea spray at blah latitude.

Fmm the figure, we see that because of the rapidity of the thermalexchange, virtually ajl sca spray droplets reach thermal equilibriumwith the air. Only droplets smaller than about 10-20 pm, however,effectively reach moisture equilibrium before falling back into thesea

BUBKES-AEDIA TEO SEAPN &UHAM5IE

Fauve 2 Thee constalrtsfor the thaneaf ee>~ Tp aAd kY St IOXSfQFOoxchlAge ff,! at tworrrlatke hvmJtNtiee RH~,95%! ance lhe tire requirerfter ~ dmptet ~ taN r Irretee still air fr,! .The ailemperahrre T, ie -10'C,ltre eea water tempormtwe7 ie O'C, the ee» aetaceeaikHy e 34 ego, end fheatme~herfe pressure is1000 hPa

Acknowledgments.

Thc office of Naval Research supported this werk thoughcontracts N0001986MP67847 «nd N0001488WM22012.

References

KcDogg, W.W. and PK TwedeH, 1986: Surnrnary of the Wortshop on ArctiCLows. 9-10 May 1985, SauMer, Coktek!. Balk'I&r of tM Arrrrriran Mc'te-oiotogical Socicry 4'l, 186-193.

Pral'~her, HX. and J.D. FJett, 1978: Micmphysics of C orrds ancf Prcci pira-tiorr. Dr.edrecht D. ReidcL

149

sueeLE-AFDN TEO SEA AN &fCHAhCE

Fyure 2. Pmtuc5on of aerosol partefes whose Nvneters wwebetween 0.5 and 5 p.Irr, per breakfny wave event, for a series Of95 such bnaaking wave eventa Mote that olek «ciid was added lbthe water surface 8 Ae tank after events 7 and 27.

References

Skoahan, E. C., DX. Spiel and K.L. Davidson, 1986: Madel of marine aerosolgeneration via white~s and wave disruption. In: Oceanic Whitecaps andTheir Ro tier hir-Sea Kxchoagc Processes, E.C. Mceahaa and G. MacNio-caN, Cds. D. Reidel Publishing Co., Dorralrecht, Holland, l67- l74.

%'aolf, D.K. 1988. 1' role of oceanic whitecaps in geochemical transport in theupper ocean «nd the marine onviamnenL PhD Di|ycrtation, NarionalUniversity of Irelnnd, Dublin, Ireland.

Woolf, D. K. and E.C. Monahan, 19&&: L'&oratory investigations of the influ-ence oa Naarine aerosol production of the interaction of oceanic whitecapsand surface-active material. In: Aerosols and Climate, P.V, Hobbs, ed. A.Deeps Publishing, Hampton, Virginia t,'in press!.

Woolf, D.K., E.C. Monahtn and D.E. Spiel, 1988: guantifications of themarine acnxol production produced by whitecaps. Preprint Volume:Scvcetk Coqfcrcnce oe Oct'an-Aonosphere Interaction, 31 January-5February 1988, Pmkeim, CA, 182-1&5.

The Role Gf Breaking Waves is the Conb'ol ofthe Gas ExchangeProperties of the Sea Surface

T. TorgerscnR.MasonD. WoolfJ. BeooitM P. DowlingM. Wihon

E.C. Monahan

Introduction

The gas exchange properties of the surface ocean arc a primarycontrol on the response time of the atmosphere and the surfacelayer to natural and anthropogenic perturbation. If we arc tosuccessfully integrate ocean-atrriosphere coupling in.to global- andlocal-scale models, a predictive pararoeterization of the gas ex-change process must be established.

There have been numerous 6cld «nd laboratory studies of thecontrolling parameters af gas exchange, The role of wind velocityhas been clearly defined but the measured rates of gas exchangehave little correlation with the instantaneous wind and the localwind velocity field. This is the result of the variability of thesurface wind and the physical memory of the process of surfaceocean mixing/gas exchange. 'De principal means of determiningoceanic gas exchange rates, Rn-222 rmeasurernents, has an inherentradiochemical rnernory that integrates the process over severaldays. Finally, there is a nonlinear relation between the wind shearacross the ocean surface and the boundary layer controlling gasexchange.

An initial series of experiments to determine the relationshipbetween braking waves and the gas exchange pmcess was

155

recently completed. Breaking waves greatly increase the surfacearea available for gas exchange in a highly nonlinear manner! andthe relative area of breaking waves is related to both the surfaceroughness and thc local wind field. Such an approach may providea pararrw:omzation and/or a real time satellite predictive data basefar the accurate determination of octmnic gas exchange rates whichcotIld be inexpxated into a coupled ocean-atmosphere model.

The gases used to study this process were Rn-222 and Hg'.These gases with C- l4! represent the methodology and data basefor most oceanic gas exchange work. Furtherrrwee, the developingtechnology in this Department allows for the collection of atmos-pheric Hg' and thus the possibility of quantifying the role of thebreaking event itself as distinct from the wave and turbulenceprocesses which accompany breaking waves.

Experirrsental Set-Up

Sources for Rn-222 Ra-226! and Hg' Hg metal! were estab-lished using gas permwblc membranes which isolate the gas andliq,uid phase. These sources werc placed in a quiescent tank forseveral days to initialize the concentrations at some reasonablyhigh value. The spike sources were then removed and the experi-ment was begun under several different conditions; in both sea-water and freshwater.

l! BUCKET TIPPING AT A FIXED RATE: a tippingbucket was employed to simulate the formation of oceanicwhitecaps and the accompanying bubble formation Fig. l!.

�! STIRRING ONLY: a submersible pump was used tohorizontally homogenize thc tank for thc breaking wave experi-ments and it was necessary to conduct a control experiment,

�! NULL EXPERIMENT; a null experiment was conducted no bucket tipping, no stirring! to determine any experimentalartifacts.

T. TGRGERSEN er Na

lQ CD7C

O "apg %

Ol

C7 g

e 8

Q

Sg

ra

'. 8O CD47

yC

C

tb

g RCO

e ZW. tO

CO

ability to measure whitecap cover and surface roughness by satel-lite, potentially make this approach a more usefully predictive thanthe wind velocity parameterization.

BUBBLE-MOll lED SEA-AIR EXCNiNQE

The IMST Frit Bubble Spectra:Characteristics and Comparisons withLaboratory and Oceanic Breaking %1ve Spectra

Ramon J. Cipriano and David K. WoolfMarine Sciences Institute

University of ConnecticutAvery PointGroton, Connecticut 06340

For similar operating mechanisms, different IMST frits havefairly reproducible bubbLe spectra The spectrum in thc &it-produced bubble plume reaching the water surface is a sensitivefunction of position in it. For a given airflow rate, as distance fromthe plume axis inness, thc spectrum narrows markedly, themaximum concentration decreases significantly, and the peak inthe frequency distribution shifts to smaller bubbles, For a givenposition in the plume, increasing the airflow rate broadens thespectral shape, and simultaneously lowers the amplitude of thcfluency peak while shifting it to larger bubble radii. %be spec-tral width also broadens markedly as thc orientation af the fritchanges from horizontal to vertical. Both examples of spectralbroadening ee almost certainly due to enhanc& bubble coales-cence. For similar frit parameters, the seawater spectrum will havea tnuch greater population of smaller bubbles, whereas both seawa-ter and freshwater spectra tend to converge at larger bubble radii.As compared to bubble spectra obtained in several laboratorywhitecap simulations, wherein thc bubbles are formed the breakupof air entrained by wave '[sealing, the frit spectra are much

163

narrower, and have much steeper negative slopes at the large end.These experiments do suggest that realistic whitecap bubblespectra can be synthesized using an array of different frits, or a setof similar frits of suitable geometry.

BUBBLE-hKDQ TED SEA-AIR EXCHANGE

Identification of Critical Research Topics

Pane! Discussion Summary

Moderator: S.E. Lar.en

Risy National LaboratoryDK-4000 Roskilde

Denmark

The symposium from which this report originates ended with apanel discussion. The purpose was to identify questions of irnpor-tance for irnpmvernent of oor insight into the ciceanic bubble-droplet processes with respect to both the mechanisms drivingthese processes and their irnpor~ce for air-sea exchange,

The outcome of this discussion may be of interest to a moregeneral public than those present during the meeting. Therefore,the summary presented here has been included in the symposiumproceedings,

Of the many unresolved questions, the discussion tended torevolve amund a number of themes that will also bc used to stratify

the summary.First, the discussion ccntcmi on the microdcscription of

individual bubbles and their droplet production. Here the generalconcern was that although much information is available fromlaboratory work, many uncertainties remain due to the multipar-arneter character of the problems. Notably the following itemswere raised:

~ When bubbles burst at the water surface, both jet drops andfihn drops are ejected. Ejection characteristics including sizerelations between bubbles and. dmps need to be more thoroughlydescribed, to facilitate the work on relating conditions just belowthe sea surface to those just above.

~ The lack of systematic knowledge about ejection

characteristics becomes especial!y marked when we considerborsting bubbles in a foam pad or bubbles bursting through asurface film. In both cases our knowledge is essentially limitedbecause the ejection characteristics are highly variable.

I't was erriphasized that more experiments on enrichmentfactors ee noxbM The enrichment factors are defined as the ratiobetween the concentration of a contaminant mostly bacteria havebeen considered! in a drop and the corresponding bulk concentra-tion in the water from which the drop originates. The enrichmentarises from a combination of bubble scavenging when it risesthrough the water, and the fact that most of the bubble fi!m isincluded in the dip ejection when the bubble bursts. Quite thor-ough !aboratory work has been done on the enrichment factors, butonly for rather limited bubble and contaminant sizes and contami-nant types.

~ In recent years the experimental techniques have been com-p!emented by detailed numerical rriodels. This technique may beparticularly useful in connection with many of the above-consid-ered prob!ems, e.g., the characteristics of bubble bursting, becausethese are associated with mu! titudes of governing parameters andtheir variation. Such prob!ems might more easily be systernatica!lystudied by numerical models.

Next, the prob!erns about assigning source functions to themarine aeroso!s were discussed. These questions were consideredbecause in order to accurm,te!y predict the evolution of the aerosolspectrum, one needs to have a re!iab!e parameterization of theaerosol source at the surface as function of the relevant parlmeters.such as wind speed, ocean surface characteristics and diameter.The fol!owing points were raised;

~ Thee is a need to establish concensus for the source func-tions. This wou!d involve description not only of the amount ofdroplets ejected fmrri characteristic surfaces in different size bins,but also of the ejectiori characteristics such as speed, direction andheight.

Also, the turbu!ence transport close to the waves needs to beconsidered mm c!osely. This tmbu!ent diffusion is the onlyprocess that can take the ejected droplets further up than their

8VB8LE-MDfATED $SiAIR EÃCHAhKhf

ejection height. Since rmmy source functions are estimatedbackwards from measurements of spectra at greater height, agenerally accepted method for describing the turbulence fluxesclose to the surface becomes irnpmet, both when comparingdifferent souse function estimates and when using these estimatesto determine aerosol spectra.

Finally, it was emphasized that although aerosols from wave-generated bubbles were probably dominating, one should notcompletely neglect the other rnechanisrns. Here the chop andspume droplets especially are well-known far high wind ~M.However, maybe it was worth considering other bubble sourcessuch as rain and snow. Rain and snow wouki not only scavenge thealmdy existing particles, but also could create new populationswith different compositions, reflecting the composition of theprecipitation.

In spite of the above considerations, the dominating importanceof breaking waves for the oceanic bubble/droplet populations iswell recognized. 'Therefore, the relations between whitecaps, foamstructure and bubble production were considered essential. Severalpoints were raised:

~ The possibility for linking the whitecap coverage and therebythe bubble production to the wave spectrum emmet a promisingappmach. It would provide a rational way of describing the white-cap coverage and bubbIe and aerosol spectra for inhomogeneous lijnited fetch! and non-staticemry situations, by use of wavespectral models. It was reahxed that these models were far fromperfect and also that a number of issues would have to be resolvedconcerning the detailed bubble and foam structure «round a wave.

~ The structure, size and distributions of different parts of thewhitecap are situated relative to a breaking wave. There wasgeneral agreemermt about the necessity to distinguish at least be-tween the foam situated directly on the actively spilling ar plung-ing part of the waves and the foam situated where the bubbles re-crr|efgc behind the wave front.

The participants in the discussion emphasized the need formeasurement of bubble spectra under the different parts of thewaves to build up the systematic knowledge needed. It was pointed

167