JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 86, NO. C9, PAGES 8085-8092, SEPTEMBER 20, 1981
`
`Bubble and Aerosol Spectra Produced by
`a Laboratory 'Breaking Wave'
`
`RAMON J. CIPRIANO AND DUNCAN C. BLANCHARD
`
`Atmospheric Sciences Research Center, State University of New York at Albany, Albany, New York 12222
`
`The relative contribution of jet and film drops from bursting bubbles to the sea-salt component of the
`marine aerosol is poorly understood. An analysis of the bubble and aerosol spectra produced by a labora-
`tory model of a breaking wave or whitecap shows that film drops may play a much more important role
`than previously accorded. The model strongly suggests that most of the droplets smaller than 5-10/(cid:127)m in
`diameter originate as film drops, derived from bubbles larger than I mm. The water-to-air flux of such
`droplets is adequate to account for the majority of maritime cloud condensation nuclei. The model also
`suggests that droplets larger than 20-25/an originate as jet drops, derived from bubbles smaller than 1
`mm. The model breaking wave produces an upwelling plume of bubbles whose concentration for all
`bubble sizes vastly exceeds the steady state or background bubble population observed at sea at depths
`greater than I m. Bubbles of up to 10 mm diameter were produced, and the bubble flux reached 200
`cm -2 s -(cid:127). Whitecap bubble spectra, presently unavailable, are therefore essential in making more accu-
`rate assessments of marine aerosol production.
`
`INTRODUCTION
`
`A major source of the sea-salt component of the marine
`aerosol is the bursting of whitecap-produced bubbles [Blan-
`chard and Woodcock, 1980]. Bursting bubbles produce two
`types of droplets: film drops, from the rupture of the bubble
`film, and jet drops, by the breakup of the vertically rising jet
`of water from the collapsing bubble cavity [Blanchard, 1975].
`Jet drops are of the order of one tenth the bubble diameter
`[Blanchard and Woodcock, 1957]. The number of jet drops per
`bubble decreases from as many as five or six for a 300/(cid:127)m di-
`ameter bubble to only one for bubbles larger than about 3
`ram. Conversely, the number of film drops per bubble in-
`creases with increasing bubble diameter: bubbles smaller than
`~300/an produce no film drops, a 2 mm bubble produces up
`to 100, and a 6 mm bubble up to 1000 [Blanchard, 1963; Day,
`1963]. Film-drop size distributions have been obtained by
`Blanchard and $yzdek [1975], M. Tomaides and K. T. Whitby
`(unpublished data, 1975), and Cipriano [1979], although the
`data are not as complete as those for jet drops.
`It is clear from the above that the relative contribution of
`jet and film drops to aerosol production depends critically on
`the shape of the bubble spectrum. Moore and Mason [1954]
`noted that if the majority of bubbles produced by breaking
`waves are larger than about 500/an, relative jet drop contri-
`bution would be minimal, since the terminal velocity of jet
`drops from such larger bubbles prevents their remaining air-
`borne. Partially on the basis of their laboratory simulation of
`wave breaking, they suggest that airborne salt nuclei < 10 -9 g
`(i.e., seawater drops < 40 /(cid:127)m diameter) originate as film
`drops. Blanchard and Woodcock [1957] measured the bubble
`spectrum near a small whitecap and found the population of
`bubbles <500/.tin to be far greater than those larger, implying
`a jet drop mode of origin. However, Blanchard [1963] and
`Blanchard and Woodcock [1980] note that estimates of the rel-
`ative contribution of the two types of droplets, which are
`based on their observed whitecap bubble spectrum, must be
`regarded with caution, since experimental difficulties pre-
`vented the measurement of bubbles >500 tma.
`The experiments discussed here were undertaken as a first
`
`Copyright ¸ 1981 by the American Geophysical Union.
`
`step in addressing this problem. The main approach was to
`make observations of both bubble and aerosol spectra pro-
`duced by a laboratory simulation of a breaking wave. The ob-
`served bubble spectra, in conjunction with extant laboratory
`data on the numbers and sizes of jet and film drops produced
`by individually bursting bubbles, can be used to calculate the
`resultant aerosol spectrum. The estimated spectrum can then
`be compared to that actually observed.
`How does one produce a bubble spectrum? Many investiga-
`tors (including the authors) have attempted to simulate the
`vigorous splashing and bubbling that occurs when a wave
`breaks at sea by forcing air through a glass frit or sparget im-
`mersed in seawater. However, the bubble spectrum depends
`on frit pore size and airflow rate and is thus rather arbitrary.
`In nature, the whitecap bubble spectrum results from the
`breakup of large volumes of air entrained by the breaking
`wave. We believe this process is analogous to what occurs
`when falling water breaks up to produce a raindrop spectrum:
`above a certain rain intensity, the shape of the raindrop spec-
`trum remains constant, although the total number of drops
`continues to increase [Blanchard and Spencer, 1970; $rivas-
`tava, 1978]. In like manner, it is not unreasonable to expect
`that as the volume of entrained air increases, its breakup pro-
`duces bubble spectra that approach a characteristic shape.
`Therefore, we believe that in the laboratory we can model a
`bubble spectrum whose shape approaches that produced by
`whitecaps at sea. Thus, we should be able to get meaningful
`data on the relative importance of large and small bubbles
`and of jet and film drops.
`
`EXPERIMENTAL
`
`PROCEDURE
`
`Figure 1 shows the basic experimental arrangement. It con-
`sisted of a small tank, a centrifugal pump, and a wier-like de-
`vice. The tank was filled with seawater at 26 ø + 2øC. The
`pump was used to fill the wier, which, perched about 33 cm
`above the water surface, produced a continuous waterfall.
`This falling water (~410 cm 3 s-') entrained air in a manner
`similar to a type of wave known as a plunging breaker, found
`in both shallow and deep water [Cokelet, 1977]. The entrained
`air produced a plume of bubbles that rose to the surface,
`
`Paper number !C07!3.
`0148-0227/81/001 C-0713501.00
`
`8085
`
`

`

`8086
`
`ClPRIANO AND BLANCHARD: BUBBLE AND AEROSOL SPECTRA
`
`FILTERED
`AIR
`
`(cid:127) INLET
`
`PLEXIGLASS ENCLOSURE
`
`I
`
`-'--- 0___
`
`(cid:127) o oo o
`o ø
`o
`o
`o
`o
`.O0o o00o0 .o BUBBLE
`o(cid:127) o 2 o (cid:127), (cid:127) o PLUME
`o
`o
`o o o
`øoooø øo o
`øOOo%OoOø
`
`compared to the field of view (B), the bubbles were confined
`to the focal plane of the camera (Rollei SL66), eliminating
`depth-of-field error. Two collectors were necessary to cover
`the broad range of bubble sizes encountered (-(cid:127)50/(cid:127)m
`to -(cid:127)8
`mm diameter). The spectrum of bubbles reaching the tank
`surface of course depends on how close to the center of the
`upwelling bubble plume the collector is placed. Calling this
`distance a (Figure 2b), photographs were taken at a = 0, 2, 7,
`and 12 cm. Approximately 6 x 103 bubbles were thus counted
`and classed by size.
`It must be emphasized that this laboratory simulation
`avoids a number of difficulties inherent in measuring the
`whitecap bubble production rate at sea, not the least of which
`relates to sampling time. When a wave breaks, the largest
`bubbles surface first, and thus the bubble spectrum is a func-
`tion of time as well as position in the whitecap. To obtain the
`total number of bubbles produced, a sequence of spectra in
`both space and time must be secured, an extremely difficult if
`not impossible task. In the present case, the time dependence
`is eliminated by operating the system in a steady state. The
`bubble population at a given position in the upwelling plume
`is constant; bubbles rise continuously through the collectors
`and can be photographed at leisure.
`To obtain the aerosol spectra, filtered air was passed into the
`Plexiglas enclosure at a rate Fd. This airstream served to mix
`with and dilute (cid:127)he aerosol within. By varying F d (i.e., mean
`particle residence time in the enclosure) the sedimentation loss
`could be evaluated. Relative humidity (RH) was monitored
`with an Assman psychrometer. The large end of the aerosol
`spectrum was measured with a Royco © model 225 optical par-
`ticle counter (particle diameter from 0.5 to 15 /an, in five
`channels). Gardner counters were used to measure con-
`densation nuclei (CN) concentrations. Although the count
`was too low to obtain size discrimination
`of the submicron
`portion with an electrical aerosol analyzer [Liu and Pui, 1975],
`some useful results were obtained with a Sinclair diffusion
`battery [Sinclair et al., 1976]. Particle loss by diffusion and im-
`paction was taken into consideration.
`Further details on experimental technique are discussed at
`length elsewhere [Cipriano, 1979].
`
`RESULTS
`
`CENTRIFUGAL
`PUMP
`Fig. 1. Simplified scale drawing of model breaking wave. Seawater
`tank is 0.5 m in diameter at top, 26.5 cm deep.
`
`SEAWATER TANK'
`
`burst, and ejected an aerosol that was confined by the Plexi-
`glas enclosure.
`The bubble spectra were measured photographically, with
`the aid of two 'bubble collectors' shown schematically in Fig-
`ure 2a. These each consisted of two parallel glass plates,
`sealed at the top and sides. The bottom of a collector was sub-
`merged 1-2 cm in the region of the tank surface to be exam-
`ined. Then, by controlling the pressure drop inside the collec-
`tor with a vacuum pump, a slab of bubble-containing water
`could be drawn up to nearly the top and held in position in-
`definitely. Since the distance between the plates (A) was small
`
`
`
`(cid:127)-TO VACUUM PUMP--._(cid:127)
`0 o o 0 o
`o
`o
`o
`
`o
`
`o
`
`o
`
`o
`
`0
`
`0
`
`o 0
`0 o 0
`
`o
`
`0
`
`Oøo ø
`
`øø o
`
`SIDE VIEW
`
`O o
`
`o
`
`o
`
`o 0
`
`0
`
`0
`
`o
`
`0
`
`0
`
`o
`
`o 0
`
`0 0
`o
`
`o
`0 ø o oo
`
`FRONT VIEW
`
`(cid:127)(cid:127)
`
`Bubble Spectra
`
`
`
`SMALL' A: 0.60cm, B: 8.6cm l
`Bubble histograms for a = 0 and a = 7 cm are shown in
`
`BUBBLE COLLECTORS LARGE A 3.43cm, B:19.7cmJ
`Figure 3. The spectra are very similar up to bubble diameters
`of about 1 mm; these smaller bubbles are more easily dis-
`persed horizontally owing to their slower rise speeds. As
`FILTERED
`bubble size increases from 1 to 10 mm, the population at a = 7
`AIR
`cm falls increasingly short of that at a = 0 cm, the plume cen-
`NLET
`ter. However, the most important thing to note here is the
`rather substantial number of bubbles larger than 1 mm, in ei-
`ther case. For example, at a = 0, the ,number in the 100-300
`/(cid:127)m interval is only about 2 orders of magnitude greater than
`those in the 4-5.6 mm interval, and about 1 order of magni-
`tude greater than those in the 2-2.8 mm interval. Recalling
`that each 2-mm bubble may produce ~ 100 film drops, that
`each 100-300/an bubble can produce --5 jet drops, and allow-
`ing for the fact that the larger bubbles have about a 5 fold
`greater rise speed, one notes that film drop production may be
`10 times as efficient as that of jet drops in this simple example.
`The spectrum at a = 12 cm, nearly at the tank edge, is
`shown in Figure 4 (solid line). Smaller bandwidths have been
`
`TOP VIEW OF SEAWATER TANK
`Fig. 2. Bubble collectors used for the photographic determination
`of bubble spectra, and view of seawater tank surface. The parame-
`ter a denotes the distance between the center of the upwelling plume
`of bubbles and the collector.
`
`

`

`CIPRIANO AND BLANCHARD: BUBBLE AND AEROSOL SPECTRA
`
`8087
`
`the number of bubbles reaching the surface of the tank per
`unit time (in given size intervals), since the production rate in
`the steady state must equal the rate at which they arrive at the
`surface and burst. The bubble production rate per unit area
`(bubble flux) is simply equal to the product of bubble concen-
`tration and rise speed. The bubble flux is a function of a and
`when integrated over the whole tank surface gives the bubble
`production rate.
`The result of this integration is shown in Figure 5. Note that
`the relative production of bubbles in the 1.0-1.4 mm interval
`versus the 100-300 (cid:127)/zm
`interval is actually greater than the
`relative concentration of bubbles in these two classes near the
`plume center (Figure 4). In other words, total bubble produc-
`tion falls off less rapidly with increasing size than bubble con-
`centration, even at a = 0 where the large bubble population is
`a maximum. This is because the much greater rise speed of
`the larger bubbles more than compensates for their more lim-
`ited horizontal dispersion.
`The data in Figure 5 can be viewed as a reasonable first ap-
`proximation of the relative numbers of bubbles of various
`sizes produced by the breakup of entrained air in seawater at
`26 øC. The total rate of air entrainment can be found either by
`direct measurement (i.e., by capturing the air contributed by
`all the bursting bubbles) or by calculation from Figure 5. Both
`methods gave a value of 125 _+ 17 cm 3 s -l, nearly a third of the
`volume flow of water, illustrating the efficiency with which air
`is entrained by falling water. Less than 5% of the entrained air
`is converted
`into bubbles smaller than 1 mm in diameter.
`The aerosol production rate is similarly defined as the total
`number of droplets of various sizes produced by the model
`wave per unit time. The production rate of droplets sampled
`by the Royco counter is shown in Figure 6, reduced to show
`original or unevaporated droplet diameter (for seawater, the
`diameter of the salt nucleus, assuming sphericity, is approxi-
`mately one fourth the unevaporated drop diameter). The error
`
`n
`I--
`Z
`IJ_l
`0
`Z
`0
`0
`
`IJ_l
`'-J 01
`m
`ß
`
`BU8BLE DIAMETER
`
`(mm)
`
`Fig. 3. Bubble frequency distributions for a -- 0 and 7 cm. Size
`thresholds are 0.05, 0.1, 0.3, 0.5, 0.7, 1.0, 1.4, 2.0, 2.8, 4.0, 5.6, and 8.0
`
`used to provide greater resolution. Bubbles larger than 1 mm
`are almost completely absent here, whereas the number be-
`tween 100 and 300/fin ((cid:127)4 + 2.3 -- 6.3/cm 3) is nearly the
`same as at a = 0 or 7 cm ((cid:127)7.2/cm3). This again reflects the
`rapid depletion of larger bubbles with increasing distance
`from the center of the upwelling bubble plume, in contrast to
`the rather uniform dispersion of the smaller bubbles.
`Three observed oceanic spectra, reduced to equivalent
`bandwidth, are also presented in Figure 4. Blanchard and
`Woodcock [1957] obtained their data (dashed line) at a depth
`of about 10 cm, a few seconds after a breaking wave had
`passed. They had to wait several seconds to allow the largest
`bubbles (several millimeters) to rise first to the surface, to
`avoid bubble interference in their collector (a Plexiglas box).
`Their spectrum agrees well with that of the model wave for
`the smallest size interval (100-200/,an), but shows increasingly
`fewer bubbles as size increases to 1 mm. This is easily attribut-
`able to the delay before sampling. In 3 s, bubbles of 600
`diameter rise about 24 cm. Thus, if any 600-/zm bubbles are
`sampled, they must be those from the portion of the whitecap-
`produced plume that extends below 24 cm. This is substantial
`distance, since the penetration depth of the bubble plume is of
`the order of the waterfall height (Figure 1).
`The results of Johnson and Cooke [1979] and Kolovayev
`[1976], who employed photographic techniques, are also
`shown in Figure 4, at depths of 0.7 m (sawtooth) and 1.5 m
`(dot-dash), respectively. Their data represent a steady state or
`background bubble spectrum, since no specific attempt was
`made to sample in a region immediately after a wave had bro-
`ken. Comparison of all spectra in Figure 4 shows clearly that
`the population of larger bubbles becomes increasingly de-
`pleted as sampling depth increases and as the time elapsed be-
`tween bubble production and sampling increases. Even at a
`depth of only 70 cm, few bubbles larger than about 400/(cid:127)m di-
`ameter are observed.
`
`Bubble and Aerosol Production Rates
`
`A comparison between estimated and observed droplet pro-
`duction via bursting bubbles requires a knowledge of the total
`bubble production rate of the model wave. This is the same as
`
`I
`
`[ IlllllJ
`
`I IllIlL
`I
`Present work ((cid:127):12cm)-
`Blanchard
`(cid:127)
`Woodcock (1957)
`Nv(cid:127)A dohnson & Cooke(1979).
`
`Jj ---- Kolovoyev (1976)
`
`
`!
`
`L(cid:127)
`
`I
`I
`I
`
`L_
`1
`J
`
`L
`
`I
`
`I
`
`_[[[lllJ
`-
`_
`
`_
`
`i0 ø
`
`-
`
`_
`
`i
`
`(cid:127)
`
`I-
`
`n
`I--
`Z
`IJ_l
`
`Z
`0
`
`-
`
`i0 -[
`:
`-
`-
`
`i0 -a
`-
`i
`
`J
`-'(cid:127). L.(cid:127)
`L..
`i
`
`1 i L.
`
`I
`
`I
`
`I IIIII!
`
`-
`
`i0 -s
`
`I
`
`IIIII
`IJlll
`IOO
`IOOO
`BUBBLE DIAMETER (p.m)
`Fig. 4. Bubble frequency distributions of model wave (solid line)
`compared with those observed at sea: dashed line, depth 10 cm, light
`winds; sawtooth line, depth 70 cm, winds 20-25 kn; dot-dash line,
`depth 150 cm, winds 22-26 kn. Size thresholds are 100, 200, 300, 400,
`600, 800, and 1000/(cid:127).
`
`

`

`8088
`
`ClPRIANO AND BLANCHARD: BUBBLE AND AEROSOL SPECTRA
`
`10 5
`
`z
`
`BUBBLE DIAMETER (mm)
`
`Fig. 5. Steady state total bubble production rate (frequency distri-
`bution) of the model breaking wave.
`
`Comparison of Bubble and Aerosol Production
`To obtain an estimate of the total jet-drop production rate
`we need only combine the observed bubble production rate
`(Figure 5) with the empirical relationship between bubble di-
`ameter and jet-drop diameter [Blanchard and Woodcock,
`1957]. The result is shown in Figure 7, assuming an average of
`five jet drops per bubble. The upper and lower bounds for
`each size interval correspond to the upper and lower bounds of
`the bubble production rate; these in turn resulted from the in-
`tegration process used to sum over the tank surface, which
`was partitioned into a finite number of regions (i.e., the four
`corresponding to the values of a at which bubble spectra were
`'measured).
`Comparison of Figure 7 with the observed droplet produc-
`tion (Figure 6) reveals that the estimated jet-drop production
`rate is certainly adequate to explain the production of droplets
`with a diameter > 20/an. Of course, most of the very large
`droplets (> 100/(cid:127)m) are unable to remain airborne and fall
`back Mto the water. For drops in the smallest interval seen by
`the Royco (i.e., 1.2 to 3.5/(cid:127)m unevaporated diameter), Figure
`6 suggests a bubble production rate of around (3 x l(P)/5 = 6
`x 103 s -(cid:127) (again assuming five jet drops/bubble), whose diam-
`eter must be between about 12 and 35/(cid:127)m. Although bubbles
`of this size were beyond the resolution of the camera lens,
`their presence in sufficient number is highly questionable.
`Bubbles this small are forced back into solution soon after
`they are formed [Blanchard and Woodcock, 1957; Johnson and
`Cooke, 1979], owing to theft high internal pressure. The ob-
`served bubble spectra (Figure 3) suggest that a maximum in
`the distribution is occuring somewhere between 100 and 300
`Ima. Johnson and Cooke [1979] and Kolovayev [1976] were ca-
`pable of resolving bubbles of 35/(cid:127)m diameter. Both found a
`definite peak in the size distribution, at 100-150/(cid:127)m, with con-
`centration falling off sharply with decreasing bubble diameter.
`In the sea the absence of these smaller bubbles cannot be at-
`tributed to the fact that the samples were not obtained at the
`surface directly in a whitecap, for their rise speed (<0.5 cm/s)
`is small compared with eddies and currents in the water. Such
`
`bars increase toward the largest drop sizes owing to the in-
`creasing difficulty in evaluating sedimentation loss. When
`summed over the five channels, the production rate is 1.4 x
`105 s -(cid:127) (+ 25%). The condensation nuclei production rate was
`found to be 2.7 x 105 s -(cid:127) (+ 15%), from which we infer that
`approximately half the droplets are of submicron size. Mea-
`surements with the diffusion battery showed the presence of
`particles as small as 0.03/ma diameter, at RH = 90%, or 0.014
`/(cid:127)m diameter as dry sea salt. This suggests the bubble bursting
`process can directly produce very small particles, without in-
`voking a shattering mechanism via phase change [Iribarne et
`al., 1977].
`
`10 5
`
`i,I
`u
`
`o
`
`oz
`
`o
`
`z
`
`i
`
`I
`
`ii
`
`I
`
`IO
`
`IOO
`DROP DIAMETER (p.m)
`Fig. 7. Calculated jet-drop production derived from observed
`model wave bubble production, assuming five drops per bubble. To-
`tals for all intervals: upper bound, 7.0 x 105 s-(cid:127); best estimate, 3.9 x
`105 s-(cid:127); lower bound, 1.7 x 10 (cid:127) s -(cid:127).
`
`i I i , , t i]
`i
`1.2 2 3 4 5 678 i0
`ORIGINAL DROPLET DIAMETER (p.m)
`Fig. 6. Observed model wave droplet production, for diameter >1
`/an. Size thresholds are 1.2, 3.5, 12, 36, and 56/(cid:127)m.
`
`I
`
`I
`
`I
`
`i
`
`2o 30 40 50
`
`

`

`i0 (cid:127)--
`
`..
`
`--
`
`--
`
`_
`
`-
`
`_
`
`_
`
`(cid:127),
`
`..
`
`..
`
`Cclz
`F-
`--
`
`-.
`
`i.
`
`_
`
`'
`
`_
`
`-
`
`=.
`
`_
`
`io 4
`o.i
`
`I
`
`I
`
`I I
`
`I
`
`I
`
`I I
`
`I I
`io
`
`BUBBLE DIAMETER(mm)
`Fig. 8. Calculated film-drop production derived from observed
`model wave bubble production, using the upper bound data of Blan-
`chard [1963] and Day [1963]. Totals for all intervals: upper bound, 4.0
`x 106 s-l; best estimate, 2.5 x 106 s-l; lower bound, 1.0 x 106 s -(cid:127).
`
`CIPRIANO AND BLANCHARD: BUBBLE AND AEROSOL SPECTRA
`
`8089
`
`10 6 --
`:
`_
`
`-,
`
`'I
`-, _(cid:127) i_ -1t-
`
`i
`
`_
`-
`_
`
`_
`
`-I I-
`
`....
`
`-. -' -11- _
`
`tionship between bubble diameter and film-drop size distribu-
`tion is unknown except for a few specific cases, bubble diame-
`ter (rather than drop diameter) appears on the abscissa. The
`size intervals are the same as those in Figures 3 and 5; upper
`and lower bounds were derived as in Figure 7.
`Casual comparison of Figures 7 and 8 shows that the calcu-
`lations yield far more film drops than jet drops; the ratio of
`the best-estimate production rates is 2.5 x 106/3.9 x 105 = 6.4.
`This
`is in marked contrast
`to the Blanchard and Woodcock
`[1980] estimate of 1/75, based on their observed [Blanchard
`and Woodcock, 1957] bubble spectrum. However, as they
`noted, this spectrum is deficient in the population of larger
`bubbles.
`Since about half of the droplets produced by the model
`wave were of submicron size, it is difficult to escape the con-
`clusion that these originated as film drops, for otherwise we
`would have to assume the existence of numerous bubbles of
`diameter less than 10/(cid:127)m, which can reach the surface and
`burst. Since the time required for a 10-/(cid:127)m bubble to be forced
`back into solution is less than 10 s, even for water that is 15%
`supersaturated with air [Blanchard and Woodcock, 1957], and
`since such bubbles have negligible rise speed, this scenario
`seems unlikely.
`Figure 8 shows that estimated film-drop production reaches
`a maximum for bubbles of roughly 2 mm diameter; bubbles
`smaller than 1 mm or larger than 5 mm are relatively in-
`currents would surely be able to disperse these bubbles to a
`efficient. Conversely, jet drop production falls off rapidly as
`depth of 70 cm.
`bubble diameter increases beyond I mm. Even if it did not,
`Oceanic bubble spectra obtained by Medwin [1977] and
`these jet drops would have little effect, as they would remain
`Lovik [1980] from acoustic methods do not display a relative
`airborne for only a few seconds at most.
`maximum over the range measured: bubble diameters from 20
`Although film-drop size distributions are presently avail-
`able only for selected bubble sizes, the data are sufficient to
`to 630/(cid:127)m. Wu [1981] has compared Medwin's spectra with
`those of Johnson and Cooke and Kolovayev. For bubble di-
`show that not only submicron drops, but also drops of up to
`ameter > 120/(cid:127)m, all have similar shapes. Wu concludes that perhaps 10 pm diameter, can be accounted for more easily via
`bubbles <120/(cid:127)m diameter observed by Medwin must have the film drop mechanism. For example, the bursting of 4.5-
`been produced by mechanisms other than air entrainment at mm diameter bubbles in seawater was found [Cipriano, 1979]
`the surface, particularly since their concentration does not de- to produce an average of 170 film drops per bubble of 2-6/(cid:127)m
`crease with depth.
`diameter. The production rate of bubbles in the 4-5.6 mm in-
`The number of film drops produced by a bursting bubble is terval is 102 to 103 s-' (Figure 5), which implies a 2-6/(cid:127)m film-
`highly variable, being greatly affected by the presence of or- drop production rate of~2 x 10 (cid:127) to 2 x 105 s -(cid:127). To account for
`ganic surface films on the water surface [Blanchard, 1963; Day
`this via jet drops implies a production of at least 4 x 103 to 4
`1963; Paterson and Spillane, 1969]. If such films were not pres- (cid:127) 10 (cid:127) bubbles s-' of 20-60/zm diameter (assuming $ drops/
`ent, and if the bubbles were to burst immediately upon arrival bubble); for reasons already discussed the production of such
`at the surface then large bubbles produce the number of film bubbles at this rate is very doubtful. Note that in this corn-
`drops previously mentioned; even if they do not burst imme- parison the calculated film-drop production from only one
`diately, they sometimes produce the maximum number of bubble-size interval has been used, and in particular one
`film drops [Blanchard, 1963]. In these experiments, the great which is relatively inefficient; a fair comparison requires in-
`flux of bubbles near the plume center created an upwelling
`tegration over all bubble size intervals. Blanchard and $yzdek
`that was more than adequate to prevent the buildup of surfac-
`[1975] and M. Tomaides and K. T. Whitby (unpublished data,
`tants [Blanchard and Syzdek, 1974]. Measurements of bubble
`1975) found that most of the film drops they observed from a
`rise speed (via streak photography) showed that the larger
`1.4-mm bubble, which is close to the size for maximum pro-
`bubbles were rising as fluid rather than as rigid spheres, which
`duction efficiency (Figure 8), were 1-10/(cid:127)m in diameter.
`in turn suggests [Detwiler and Blanchard, 1978] that they did
`Further evidence shedding light on the origin of the 1-10
`not adsorb a surfactant coating while rising to the surface; this
`/(cid:127)m drops was obtained in experiments in which the seawater
`could conceivably have the same suppressive effect on film-
`was inoculated with a known concentration of bacteria. The
`drop production as a surfactant coating on the bulk water sur-
`aerosol thus produced was sampled with an Andersen cascade
`face. Therefore, the upper bound film-drop production ob-
`impactor [Andersen, 1958]. Droplets of 3-10/(cid:127)m were found to
`served by Blanchard [1963] and Day [1963] for immediately
`be highly enriched in the bacteria. On the other hand, a num-
`bursting bubbles is used to calculate film drop production.
`ber of laboratory experiments with single bubbles suggest
`(Jet drop production is also suppressed by a surfactant coat-
`that, although bacteria are enriched in jet drops of 30-60/(cid:127)m
`ing, although to a some what lesser extent.)
`diameter, little or no enrichment occurs for jet drops smaller
`The film-drop production rate calculated from the observed
`than about 20/an [Blanchard, 1978]. Finally, Blanchard [1963]
`bubble production rate is shown in Figure 8. Since the rela-
`and Day [1963] have shown that large (several m'tllimeter)
`
`

`

`8090
`
`ClPRIANO AND BLANCHARD: BUBBLE AND AEROSOL SPECTRA
`
`(cid:127)
`
`I00
`
`o
`
`(cid:127)
`j
`
`o
`(cid:127)-
`
`75
`
`50-
`
`(cid:127)
`
`2_5-
`
`Q-
`
`O'
`0
`
`5
`I0
`15
`2_0
`2_5
`ORIGINAL DROP DIAMETER((cid:127)m)
`
`Fig. 9. Relative jet and film drop contribution inferred from anal-
`ysis of the model wave. Diagram is a first approximation only and
`should not be interpreted in an exact sense.
`
`bubbles eject most of theft film drops to a height of several
`centimeters, whereas jet drops of diameter <6/(cid:127)m are ejected
`to <0.3 cm height [Blanchard, 1963]; thus the film drops have
`a higher probability of remaining airborne.
`Although the jet drop mechanism was adequate to explain
`the production of droplets >20/(cid:127)m diameter, completeness re-
`quires consideration of film drops. The experiments with the
`4.5-mm bubble showed that only one or two film drops per
`bubble were produced of 20-40/(cid:127)m diameter. The production
`rate for this size bubble by the model wave is less than 103
`(Figure 5) and yet the observed 20-40/(cid:127)m droplet production
`rate is closer to 104 (Figure 6). Although more film drop data
`are required to give a definite answer, it seems unlikely that
`integration over all bubble size intervals can make up the dif-
`ference, particularly since the 'more efficient' 1.4-mm bubbles
`produced hardly any droplets this large. But to produce 104 jet
`drops s -(cid:127) of 20-40/(cid:127)m diameter requires only ~ 104 bubbles s-'
`of 200--400/(cid:127)m diameter, even allowing only one jet drop per
`bubble. Figure 5 shows that this is easily met.
`For droplets of 'intermediate' diameter (i.e., 10-20 /(cid:127)m)
`there is apparently a region of overlapping contribution, in
`which as droplet size increases the film-drop influence fades
`away and jet-drop influence becomes dominant. This is shown
`semiqualitatively in Figure 9. Woodcock [1972], from an anal-
`ysis of Hawaiian and Alaskan marine aerosols, suggested that
`a transition from jet to film drops occurs at 1 or 2/(cid:127)m droplet
`diameter.
`It should be noted that the best estimate of total film-drop
`production (2.5 X 106 s-') is about an order of magnitude
`greater than the observed total particle-production rate, for
`reasons that can only be speculated upon. Certainly, there are
`many complex events occurring in this mass-bubbling situa-
`tion. At the center of the upwelling plume, large bubble flux is
`very great, and interference effects may be important. The wa-
`ter surface is not quiescent (as in single bubble experiments),
`but extremely agitated, which must surely affect the bursting
`process. In the lowest few centimeters above the water surface,
`droplet coalescence may occur. Around the periphery of the
`region of strong upwelling, a fraction of the largest bubbles do
`not burst immediately. While floating on the surface, some
`coalesce into giant (several centimeters) hemispheres while
`others cling together in rafts, as is observed at sea near break-
`ing waves. This coalescence greatly reduces the bubble film
`area available for drop formation. Of course smaller bubbles,
`potential sources of both jet and film drops, continue to rise
`beneath these structures, which sometimes occupied a signifi-
`cant fraction of the water surface surrounding the region of
`strong upwelling. These bubbles might have been inhibited
`from bursting freely, suppressing drop production; even if not
`
`r
`
`-
`
`-
`
`30
`
`so inhibited, any droplets produced may have impacted onto
`the interiors of the floating bubbles.
`These considerations illustrate the danger of making a
`simple linear extrapolation from the results of single-bubble
`experiments to the mass-bubbling situation. This precaution
`applies to both jet and film drops, particularly the latter. It
`must be reemphasized that the upper bound film drop produc-
`tion observed by Blanchard [1963] has been used, which for
`reasons such as the above may be inappropriate. Even in a sit-
`uation where bubbles burst individually, all of the factors con-
`trolling film drop production are not understood. Therefore,
`the results of such extrapolation must be interpreted with cau-
`tion. Nevertheless, the calculated ratio of film- to jet-drop pro-
`duction is striking. If the upper bound film drop versus bubble
`diameter relationship is reduced to one sixth, and the five jet
`drop per bubble assumption is left intact, one is still left with
`an equal contribution of both types of drops.
`
`DISCUSSION
`
`There are a number of difficulties inherent in comparing
`aerosol production by the model wave to that due to white-
`caps at sea. Nevertheless, such a comparison is illuminating
`and has some interesting implications.
`Consider first the production of condensation nuclei. From
`a knowledge of the background concentration of oceanic CN
`in regions remote from the continents, their residence time in
`the atmosphere, and the height through which they are dis-
`tributed, one can calculate the sea-to-air flux required to
`maintain this steady state. Blanchard [1969] found this to be
`about 100 cm -2 s -(cid:127) for a background concentration of 200
`cm -3. Mason [1957] arrived at a similar result, though