THE OCEANS AS SOURCE OF CLOUD-FOI~M[NG NUCLEI

Transcription

1 THE OCEANS AS SOURCE OF CLOUD-FOI~M[NG NUCLEI by B. J. MASO~ (*) Summary -- The number and size of salt particles produced by the bursting of air bubbles in sea water has been measured. Bubbles of diameters varying between 1/2 and 2 mm each produced about 300 nuclei the sizes of which, under the electron microscope, were mainly between 0.1 ~. and 0.5 ~ diameter. They appeared to consist mainly of sodium chloride, the smallest ones containing only 10-15g of salt. These results, together with measurements of the size distribution of salt nuclel collected over the oceans in areas of spray formation, indicate that the total ~oncentrations of salt nuclei over the oceans in winds of up to 15 m sec -1 probably do not exceed 100 cm -3. The corresponding rate of production of salt nuclei at the sea surface is estimated to be 1000 cm -2 sec -1. It is therefore inferred that sea spray contributes perhaps only one-fifth of the nuclei involved in cloud formation, the majority being the products of combustion, either natural or man-made. 1. Introduction -- In this paper I propose to re-open the old, but still unsettled controversy as to whether or not salt particles produced by sea spray are the main source of the condensation nuclei involved in cloud formation. AITKE~ ( ), I believe, was the first to suggest that the c<salt dust ~ resulting from the evaporation of sea spray was an active source of nuclei and this view has been amply confirmed by later workers. Sea-salt particles of mass > g have been found by WOODCOCK up to heights of 3,000 m over the sea, their numbers increasing steadily with increasing surface wind speed but rarely exceeding a few particles per cm ~. That these particles may be carried for considerable distances over land was also established by WOODCOCK (i952), who found that ii0 km inland, the concentration and size distribution of the particles were much the same at all levels from 150 to 1,500 m and were not appreciably different from similar measurements made over the sea. Much the same result was obtained by TWOMEY (1955) from his measurements of the concentrations of salt nuclei of mass > i0 -l~ g during a series of aircraft flights over south-eastern Australia on occasions when the air was undisturbed by precipitation or convective activity. Also, nuclei of sea salt, and in particular of the NaC1 component, have been identified in the larger haze droplets and in some cloud and fog droplets forming over land. So, the question is not whether sea-sah particles can, or do act as condensation nuclei, for clearly, by virtue of their size and hygroscopicity they are very effective ones, (*) Dr. B. J. MAson-, Department of Meteorology, Imperial College, London S. IF.7.

2 - Some but whether they are produced in sufficient numbers to be the main source of cloud-forming nuclei. A direct attack on the problem would involve the capture of large numbers of cloud droplets and determination of the nature of their nuclei by microscopic and chemical methods. Although a little work has been done along these lines {see w 5), it would be quite impracticable to obtain data which would be geographically and meteorologically representative, even were it possible to identify the smaller nuclei of mass only l0 -~ g. We have, therefore, to adopt a more indirect approach. The problem may be posed as follows: i) Are the sea-salt particles occurring in the atmosphere present in sufficient concentrations to account for the formation of observed cloud-droplet populations? ii) Are sea-salt nuclei produced at a rate sufficient to replace those lost from the atmosphere by precipitation? It is with these two questions that this paper is mainly concerned. 2. The concentration and size distribution of large salt nuclei over the~sea - of the large aerosol particles (r > 0.2!~.) occurring over both continents and oceans J IO T ~., \ z I I -I I I \t I0 "( I0 -~ 10 -;'I I0 -'~" I0-" to -I~ lo -~ IO "B NUCLEAR MASS (~1 Fig. 1 - The size distribution of large and giant nuclei over the oceans, a) MooRI~ & -MAson, Type I nuclei, ocean, b) WOODCOCK, Florida, tropical storm, c) D'ALBE, Bay of _Monaco. wind 9 m,~sec, d) 5'IooRE & MASON, Type II nuclei, ocean. are composed of sodium chloride or of sea salt, and there is little doubt that these are produced as spray at the sea surface. However, the concentrations of those nuclei large enough to be positively identified are low. As shown in Fig. 1, MOORE & M~so~ (i954) found nuclei of mass > 2 X 10-14g and dry radius > 10 -~ cm in concentrations of only 10 cm -3 over the oceans, even in winds of up to t5 m sec -~, while WooDcocK (1950) found that the concentrations of nuclei with m > 10 -~2 g did not exceed 4 cm -3 even in winds of gale force. Clearly these large nuclei are

3 not present in sufficient numbers to constitute a major source of cloud-forming nuclei. It is, of course, possible that there are much larger concentrations of salt nuclei too small to be detected by the techniques employed by the last-named authors ; if ~o, these should be detectable in an Aitken counter. That the concentrations of Aitken nuclei over the oceans are generally much lower than over the land, and on occasion may be as low as 2 cm -3 (LA~DSBEI~r 1938), does not speak in favour of the sea being the main source of such nuclei. But the important question is whether nuclei are, on average, produced in sufficient quantities at the sea surface to account for the observed droplet concentrations in natural clouds, which are generally of the order a few hundreds per cm 3. Two-thirds of the Aitken counts made on board the Carnegie were less than 400 cm-~; MooRE (i952) has obtained a mean value of 650 cm -3 in the North Atlantic, while LA~DSBERG (1938) gives an average value of 940 cm -~ for all observations made over the oceans. If similar concentrations of nuclei existed at cloud base, and if all were involved in condensation, they might just account for the observed droplet populations, but one has to remember that the rate of cooling of the air in an Aitken counter is much more rapid than in a cloud, where only the large r Aitken nuclei may be activated. It may also be questioned whether all the Aitken nuclei which exist over the sea are produced at the sea surface, in view of the fact that small nuclei produced over the land (and also, perhaps, in the atmosphere) can remain in suspension for days or weeks-see Mool~E & MASO~ (i954). We also recall that both MooRE (1952) and OnTA (i951) found no significant increase in Aitken counts taken over the ocean as the wind increased, a result which is difficult to reconcile with the assumption that the majority of these nuclei were being produced by sea spray. 3. The rate of production of salt nuclei at the ocean smface -- In assessing whether any given source of nuclei can be of major importance in cloud formation, it is necessary to compare the rate at which the nuclei are supplied with the rate at which they are lost from the atmosphere by precipitation. From figures on the annual global rainfall and K6nLER'S mean value of 8.8 I~ for the radius of a cloud droplet, SIMPSON (i941b) estimated that if sea-salt nuclei were entirely responsible for condensation and that if every spray particle produced a cloud droplet, the average rate of production of nuclei over the whole ocean surface would have to be i,250 cm -2 sec -~. Making the further assumptions that only one-tenth of the ocean's surface is producing spray at any one time, and that only one-quarter of the droplets are carried aloft and remain suspended long enough to become involved in cloud formation, he estimated that spray droplets would have to be produced at the rate of 50,000 em -2 see -1. SIMPsoN believed that such a rate of production was impossible, and therefore, that sea salt could make only a minor contribution to the supply- of cloud-forming nuclei. Several factors may be mentioned which suggest that SIMPsoN's figure is an over-estimate. 1. He assumed that all the precipitation was formed by the coagulation of cloud droplets; but a large proportion of raindrops are formed by the melting of snowflakes which grow primarily by diffusion and so involve no appreciable consumption of nuclei, apart from those swept up by the falling precipitation. 2. Measurements suggest that the mean radius of droplets in precipitating clouds is greater than K/SHLER'S value of 8.8 ~.

4 The original nuclei produced by spray may multiply by crystallization of the solution droplets at low humidities and produce a number of separate crystals as observed by DESSE~S (1949). But, as shattering of the nucleus during crystallization has been directly observed only with rare, giant nuclei (and under conditions when the presence of the support may have had some influence), it is doubtful whether this is an important factor in determining the concentration of sea-salt nuclei. This view receives some support from a recent laboratory investigation by LODCE & BAER (1954) which indicated that desiccation of sodium chloride droplets of dry radius 0.5 to 4,a in the free air was rarely accompanied by shattering of the nucleus. 4. Nuclei dep0sitedon the earth's surface by precipitation may be subsequently carried up into the atmosphere and become involved more than once in cloud formation. For these reasons, SIMPSO~-'s figure is probably too large, perhaps by one order of magnitude, but if the sea surface is to be capable of supplying enough l~ig. 2 - Stages in the bursting of an air bubble in water. (From KNELgIAN, DOMIltlOWSKI & NEWITT, Nature, 173, p. 261, 1954). condensation nuclei for cloud formation, it must still produce them at a rate of about 104 em -2 sec -1 in the regions where waves are breaking. The rate of produeti0n of the large and giant nuclei (m > 2 X I0-1~ g) by breaking waves has been investigated by MOORE & MASON (1954) in a wind-wave tunnel in which it was possible to simulate the conditions which obtain over the ocean when the wind speed at a height of i0 m ranges between 0 and 16 m see -1. In the strongest winds, the rate of production of nuclei with m > 2 X i0-13g " at the water surface was found to be 40 em -2 see -~ ; the concentrations of smaller salt nuclei could not be measured because of the large numbers of hygroscopic combustion nuclei in the air. An independent estimate of the rate of nucleus production was also made by the same authors from their size-distribution curves of large and giant nuclei

5 collected over the ocean. In winds of 15 m sec -1, they estimated the rate of production of nuclei with m > 10 -zz g to be 43 cm -~ sec -1, in excellent agreement with the value obtained in the wave tunnel, and that of nuclei with m > 2 >,/ g to be 86 cm -2 sec -1. These rates are very small compared with the rate of total nucleus production required by S~MPso~-, and again suggest that the large and giant salt nuclei are only a minor source of condensation nuclei. 4. Production of smaller salt nuclei by bursting foam bubbles -- There still remains the possibility that large numbers of salt nuclei of m < 2 X g may result from the bursting of foam bubbles which are produced by the entrainment of air into the crests of breaking waves. The mechanism of bubble bursting has recently been studied with the aid of high-speed photography by W.OODCOCK et al. (1953), KIE~TZLrR et al. (1954), and by K?CELMA~ et al. (1954). The photographs of Fig. 2 show that in the initial stage a protuberance.develops on the hemispherical cap of the bubble as it breaks the surface, and that this bursts at the summit where it is thinnest. The air, in escaping through the hole, probably shatters the film into many fragments, but its disruption, which occurs in a few microseconds, has not yet been photographed. After bursting of the cap, the liquid rushes into the crater formed at the site of the bubble and produces a narrow unstable" jet which rises rapidly from the bottom of the bubble cavity and breaks up to form between 1 and 5 drops which are projected vertically into the air as shown in the photographs. The relation between the size of these drops and that of the parent bubble has been investigated by MooRE & MAso~ (1954). The mean size of the drops produced by bubbles of diameter between 0.3 mm and 4.3 ram was roughly 15 per cent of the corresponding bubble diameter. These measurements suggest that, if the majority of air bubbles produced by breaking waves are larger than 1,/'2 mm in diameter, most of the drops formed by the breaking-jet mechanism will fall back quickly into the sea and not contribute to the supply of potential condensation nuclei, although the smallest drops of diameter about 100 ~, containing about 10-Sg of salt, and having fall-speeds of about 25 cm sec-, might be carried up in strong winds. If, however, large numbers of appreciably smaller bubbles are formed, the correspondingly smaller drops produced by disruption of their jets might make a significant contribution to the giant nuclei. To determine whether the disruption of the bubble film produces much larger numbers of particles too small to be detected by conventional methods, the present author has studied the bursting of bubbles in an expansion chamber. The apparatus is shown in Fig. 3. Small controlled quantities of clean, filtered air were allowed to pass through two identical glass nozzles immersed below the surface of water contained in two identical glass chambers C1, C2 of I litre capacity, one containing sea water, and the other, distilled water. The air was forced, a little at a time, through the nozzles nl, nz to produce single bubbles by compressing with screw adjustments the aneroid capsules A1, A~. Before commencing a n experiment, both chambers were flushed via a, b, c, d, e with air supplied from a cylinder and cleaned by passing through a long glass-wool filter f. After allowing an equal number of bubbles to burst in each chamber, both chambers were sealed and the air inside them expanded by a sudden extension of the metal bellows B, special precautions being taken to prevent the production of extra bubbles during the expansion.

6 Observation of a brightly-illuminated volume of the air in each chamber, immediately after the expansion, revealed the presence of a dense cloud of tiny droplets above the sea water but no such cloud (apart from a few odd droplets which may have escaped the cleaning process) was observed above the distilled water. It appeared that the bubbles bursting in the sea water produced very small droplets too smallto be seen by the naked eye or through a low-power microscope, and that these evaporated to leave behind minute salt particles which could act as condensation nuclei during a subsequent expansion. Choosing an expansion ratio beyond which no observable increase in the number of droplets was produced, the concentration of droplets in the cloud above the sea water was determined by estimating their average distance apart. Multiplying this concentration (number of droplets/cm s) by the volume of the air space in the chamber and dividing by the number of bubbles gave an average figure for the number of nuclei produced by each bursting bubble. A run of about five experiments was carried out for each particular size of bubble, the individual values for the number of nuclei produced per bubble being usually within a factor of two of the mean values for all the runs. More precise values were not to be expected in view of the errors involved in estimating visually ~ 9 h II Fig. 3 - Cloud-chamber apparatus for studying the bursting of air bubbles in sea water. the average distance apart of droplets in a turbulent cloud. Observations made on bubbles ranging in diameter from 0.25 to 2.15 mm indicated that, within the limits of experimental error, the number of particles produced per bubble was almost independent of bubble size, the best overall estimate being nuclei per bubble. That the smaller bubbles disintegrated to produce much the same number of particles as the larger ones is not, perhaps, surprising since the smaller bubbles burst more violently. The sizes of the nuclei produced by the bursting bubbles were determined by

7 drawing the air in the chamber containing the sea water through a thermal precipitator, the particles being deposited on a grid for examination under an electron microscope. The electron micrographs revealed particles mostly between 0.2 and 0.5 ~ in diameter, the larger ones being cubical in shape and cousisted of sodium chloride. The smallest particles, of 0.1 ~ diameter, contained about 10-15g of salt, and presumably originated as droplets of sea water of diameter about 0.4 ~. 5. Discussion of results-total concentrations and rate of production of salt nuclei -- Extrapolation of MOORE & M~so~'s graphs showing the size distribution of salt nuclei occurring over the oceans in winds stronger than 6 m sec -1 down to nuclear masses of g (see the dotted line in Fig. 1) suggests that, in winds up to 15 m sec -1, the total concentration of salt nuclei would not exced 100 cm-~, while extrapolation of WOOl)COCK'S curve for winds of gale force would indicate only a slightly greater concentration. If the concentration of salt nuclei over the oceans is unlikely to exceed 100 cm -a, i.e. only about one-fifth of the average total nucleus concentrations measured by Moore (i952) and by OI~TA (i95i) with an Aitken counter, it is not surprising that they failed to detect a marked increase in the latter with increasing wind speed. If we assume the concentration of salt nuclei in the lower atmosphere to be proportional to their rate of production at the underlying sea surface (this is reasonable for the smaller, more numerous particles), then taking MooRE & MASO_,N's figure of about i00 cm -2 sec -1 for the production rate of salt nuclei of m > 2 10-~4 g, which exist in concentrations of about 10 cm -3, we deduce the total rate of production of salt nuclei (m > g) to be about 1000 cm -~ sec -I over those areas of the ocean where waves are breaking in winds stronger than 6 m sec -1. If, on average, 300 nuclei are produced per bursting bubble, this would suggest a rate of bubble formation of about 3 cm--" sec -1, which appears very reasonable. The figures for both the total concentration and production rate of sea-salt particles, as estimated in this paper, appear to be too small by a factor of at least five to account for the quantities of condensation nuclei consumed in the formation of precipitating clouds. The implication is that the remaining four-fifths of the nuclei involved in cloud formation are either the products of natural and man-made combustion or consist of dust particles carried up from the earth's surface. Some support for this latter conclusion comes from electron-mlcroscope examinations of the nuclei of cloud and fog droplets by the Japanese workers K~ROIWA (1951, 1953), OGIWAaA & OKITA (1953), and YAMA}IOTO & 0HTAKE (1953), which are summarized by the present authm- elsewhere in this volume (pp. 9-19). Only about 20 per cent of the cloud and fog droplets collected a few kilometres from the coast contained sea-salt particles ; about 40 per cent of the nuclei were identified as combustion products and 20 per cent as soil particles. Although no single argument given here in favour of thesis that the ocean is not the main source of condensation nuclei is, perhaps, conclusive~ the evidence as a whole points very strongly in this direction. This does not necessarily imply, however, that the global rainfall pattern has been essentially altered by human habitation of the earth. If there were no other source of nuclei available, the numbers of nuclei produced at the sea surface would probably be sufficient to produce clouds having rather smaller concentrations of correspondingly- larger droplets than are typical at present.

8 Ackno~dedgements -- I am indebted to Mr. D. MA~SH~XLL for his help in making observations with the expansion chamber, to Mr. M. BAKER, and to Mr. ~_~_A_T- THEWS of the Chemical Engineering Department, Imperial College, for the preparation and examination of specimens under the electron microscope. REFERENCES _A_ITXEN J. ( ): Trans. Roy. Soc. Edin., 30, p DESSE~S H. (1949): Quart. J. R. Met. Soc., 75, p KIENTZLER C. F., _A_nor~-s A. B., BLANCHARD D. C. & WOODCOCK i. H. (1954): Tellus, 6, p KNELI~IiN }7., DO~B1ROWSKI N., -TN-E- WlTT D. M. (1954): in~ature, 173, p Ku~OlWA D. (1951): J. Met., 8, p Kr~RolwA D. (1953): Studies on Fogs (Iffokkaido Univ.), p LAr~-DSBERC I~. (1938): Ergebn. Kosm. Phys. (Akad. Verlag, Leipzig), 3, p LODGE J. P. & BAER }7. (1954): J. Met., 11, p MAsoN B. J. (1954): Nature, 174, p MOORE D. J. (1952): Quart. J. R. Metl Soc., 78, p MooRs D. J. & MAsoN B. J. (1954): ibid., 80, p OClWARA S. & 0XlTA T. (1953): Tellus, 4, p OHTA S. (1951): Bull. Amer. _Met. Soc., 32, p SI~vSO~ G. C. (1941): Quart. J. R. Met. Soc., 67, p TWO~EY S. (1955): J. Met., 12, p WOODCOCK A. H. (1950): J. Met., 7, p WOODCOCK A. If. (1952): J. Met., 9, p WOODCOC~ A. Iff., KIENI'Z- LER C. F.; ARONS A. B. & BLANCHARD D. C. (1953): Nature, 173, p YA~IA- ~OTO G.& OHTAKE T. (1953): Sci. 1Rep. Tghoku Univ. (Geophys.), 5, p. 141.