Flux and Processes of Deposition of Atmospheric Sea Salt to the Earth Surface

Dry deposition, rainfall and atmospheric particulate samples collected at a coastal Rhode Island site, and aboard RV/Trident over the North West Atlantic Ocean were analysed for Na and Mg in order to understand the rate and mechanisms of removal of atmospheric sea salt. Perkin El mer atomic absorption spectrophotometer models 303 and 360 were used for the analyses. Samples col l ~cted at the R.I. coastal site were divided into 3 different categories, namely, "marine", "land" and "mixed" samples depending on the local surface wind direction during sampling. Based upon the Mg/Na rat i c..1 , it appears that most of the Na and Mg in "marine" samples were derived from the ocean while "land" samples essentially consist of Na and Mg of crustal origin and "mixed" samples were a combination of the two. The rate of wet removal of atmospheric Na and Mg, calculated from "marine" and "land" samples and the rate of dry depos iti t·r, of sea . salt calculated only from "marine" samples, appear to be proportional to the wind speed. Consideration of the Mg/Na ratios in rainfall and dry fallout samples observed from the ''marine", "mixed" and "land" samples categories, suggest that rainfall scavenges sea salt Na and Mg more efficiently than crustal particles. The annual global removal rate of atmospheric sea salt was estimated at '\J3.0x1015 g/yr.60% of this is removed by rainfall and the rest by dry deposition. Of the total cyclic sea salt removed annually over the globe, less than 7% is deposited over land. 90% of this is removed by rainfall and the rest by dry deposition.

It appears that a Hungarian qeologist named Posephny was the first person to visualize the geochemical implication of sea salt transport between the ocean, the atmosphere, and land in 187 7 (Eriksson 1959).
Posephny used geological evidence to conclude that chloride in river water and saline deposits must .have originated in the sea and been carried inland by the atmosphere . Posephny's observation went largely without notice until 1959 when Erik Eriksson, in Sweden, made the first quantitative estimations for the removal of atmospheric sea salt by rainfall and dry deposition over the ocean and land. This will be discussed in full deta i1 1 a ter.

A. Removal Mechanisms of Atmospheric Sea Salt
Jun~e (1963) classified the removal mechanisms for atmospheric aerosols as wet removal by precipitation, dry removal by sedimentation, and dry removal by impaction on obstacles on the earth surface. The third mechanism can be important in land covered with forests (Junge 1963;Madgwick and Ovington, 1959;Georgii 1962).
Wet removal comprises two distinct processes l) rainout, which is the removal due to processes within the clouds, and 2) washout, which is the process of removal by rainfall below the clouds. During the journey of the raindrops from the cloud base downwards, in addition to the washout processes, evaporation of the raindrops also occurs . The net effect of these three processes, namely rainout, washout and evaporation, determine the concentration of any substance in rain water .
During the rainout process, atmospheric aerosols enter the cloud droplets in the following ways: 2 1 ) consumption of condensation nuclei, 2 ) The attachment of aerosol particles to the cloud elements by Brownian motion, J) The attachment of aerosol particles by the water vapour gradient ( Facy-effect). For atmospheric sea salt and other aerosol particles of radii >O.l um, the consumption of condensation nuclei is considered the primary mechanism of rainout. The other two mechanisms, 2 and 3 are important for Aitken particles (radii < O.lu111), (Junge 1963). Martens, et al. (1973), observed a chemical change during the transformation of aerosols to cloud droplets on the eastern end of the island of Puerto Rico. Comparison of elemental ratios in particles and rain and of the elemental particle size distribution in and out of the clouds showed an apparent efficiency of 100% for the marine particles (Na, I, Br, Cl-) compared to less than 25 % scavenging efficiency for the terrestrial fraction of the aerosol (Al, Mn, V).
\4hen the rain drops leave the cloud base, they scavenge aerosol particles as they fall and they begin to evaporate. Both these processes lead to an increase in the rain water concentration of dissolved substances. \'1ashout and evaporation are both inversely proportional to the radius of the rain drop (Junge, 1963).
where K is the concentration of any dissolved substance in rain water due to washout and evaporation and r is the rain drop radius.
c Although the mass of sea salt collected by a rain drop falling through air containing sea salt depends upon the volume of air swept out by l · drop, i.e., proportional to the cross-sectional area of Uie fa l mg l · drop or rc 2 , the concentration of salt in the rain drop the f al mg is inversely proportional to the volume of the rain drop or l/ re 3 .
The net salinity change would be proportional to l/ re. In addition small rain drops become more saline than larger ones due to evaporation.
Again, although evaporation is proportional to the surface area of the drop or re~ the salinity is proportional to l/rc 3 (Turner 1955;Duce,et al., + 1969). Duce et al., 1969 measured the Na and Cl content of rain as a function of rain intensity. Large raindrops were associated with higher rainfall intensity. They found that high intensity rains were associated + with 101>1 concentrations of Na and Cl while low intensity rain showers + -+ results in rains with higher amounts of Na and Cl . The Na /Cl ratio, however, was found to be constant, independent of eva poration, washout or the intensity of rain.
13 . Review of the History of Sea Salt .iQ_ Rain vlater Although the history of research on chemical compounds in rain and snow date back to 1750 (Eriksson, 1959), for a long time ammonia and nitrate nitrogen were the only compounds of interest. The first determination of sea salt in rain water was made at the beginning of the nineteenth century, when several investigators around the world measured Cl in rain water collected at coastal and inland locations, (Eriksson, 1952). Although the geochemical significance of atmospheric transport of sea salt between the ocean and land was suggested by Posephny as early as 1877, little was done to evaluate the processes of Production and removal of atmospheric sea salt until the middle of the 20th century. During this early period in the history of atmospheric sea salt chemistry, the majority of the studies done were concerned with measuring chloride and/or other sea salt components in raint ., collected over specified periods of time. This was undertaken \·Ja e, in several localities around the world, including inland and marine Eriksson (1952) extensively reviewed the various studies areas.
during this period. Some of this pioneer work, as quoted from Eriksson's review (1952), merits mentioning . Two points, however, should be made before proceeding further. Firstly all these reported values of sea salt in rain and in snow include dry deposition as well, because it appears that the total deposition was measured, not rainfall alone.
secondly, most of these investiqations were not done uncer meteorologically controlled or monitored conditions, and hence the results cannot be taken as strictly representative of marine or continental environments.
Although the first determination of chloride in rain water was made at the beginning of the nineteenth century, systematic work was not published until 1851 when Arago in Paris reported monthly amounts of Cl in rain water for a half year period. Another systematic investigation was carried out by Smith in 1872, He analysed rain water collected at different places in the British Isles. His reported values for the chloride content of rain water showed a considerable variation with a maximum of 54.6 mg/l on the west coast of England and a minimum of l . 20 mg/l in London. Smith, however, could not determine the cause of this considerable variation (Eriksson, 1952).
In Eriksson's review tables of the Cl content in mg/l of rain fall collected in Europe, America, Asia, Africa, New Zealand and Australia were presented. From these data, Eriksson found that the Cl content of rain water is always higher in coastal areas compared to inland places. He also noticed that the concentrations of Cl in rain water do not vary much in inland places whereas coastal locations show a great variation. Remarkably, high concentrations have been observed at Mt . Vernon, Iowa in the United States, although it is located inland far from both the Atlantic and Pacific oceans, (Trieschmann, l919;Knight, 1920;Shaffer, 1921;Fries, 1923;Woehlk, 1923;Knight, l924;Kynett, 1929;Krehl et. al., 1935). At this location one of the longest series of analyses of Cl in rain was carried out, covering the period 1919-1935 . As there is no other likely sauce for the Cl in the rain, it must be assumed that most of it originated from the ocean and was transported inland.
Extremely high amounts of salt falling with rain and snow were also reported in central Russia (see for example Krivopalova, 1971Krivopalova, ~ushkevich, 1971Krivopalova, K h 1968) Th d N + rA ++ K+, c ++ c1-so --; as tanov, . ey measure a , ·1g , a , , 4 , CO) . Although it is possible that a fraction of these ions in the rain and snow originated in the sea and was carried inland by the atmosphere, it appears that most of them were crustal since the ionic ratios in these samples approached the crustal values. Zverev, et. al., (1973) determined the average chemical composition of atmospheric precipitation for 10 climatic zones of the USSR. He found that the mean concentrations for the USSR were : Na = 3.14 mg/l Mg = 0.70 II Ca = 3.37 II contrary to the previous Russian data, these values suggest that much of the Na+and Cl-in rainfall in USSR may be of oceanic origin, as the Na/Cl ratio of 0.81 approaches that of sea water, 0.55. This data, hoviever does not give typical sea salt ratios for cations. It is likely that the deviations are . due to the inclusion of crustal material containing these metals .
Francesco Palmieri (1966) studied the salt content of rainfall in Caserta, on the southern coast of Italy. He determined a total depth of rainfall of 63 cm/year. He calculated a yearly total deposition for Na of 347 J..g/cm 2 ; Mg, 61.9 µg/cm 2 ; K, 121 µg/cm 2 , and Ca, 33.8 µg/cm 2 .
Although his data gives a Mg/Na ratio of 0.18 compared with a sen salt ratio of 0.12, the K/Na and Ca/Na ratios are very high compared to their sea water ratios, which are: K/Na -0.038, Ca/Na= 0.038 (Riley and Chester, 1971). Since Caserta is on the coast it seems that the deviations in the K/Na and Ca/Na ratios (as well as the Mg/Na ratio) are due to the addition of crustal K and Ca to the samples. Zhavoronkina,(1958) in Russia observed that the salt content of rain increased in winter and decreased in summer in coastal areas. He explained the salt increase in winter as being due to winter storms over the ocean. Girenko, (1959) made a mere detailed investigation of the effect of weather on the total mineral content of rain and snow. He concluded from his study of an annual cycle that the maximum minerC'.l conte nt in rain corresponds to the lowest amounts . of precipitation, the largest period of drought between rainfall, and the most frequent strong winds during the corresponding month.  reviewed and summarized results of chloride analyses in rain water f d"ff l d h ld rom i erent p aces aroun t e wor . Based on Mather 1 s [rl .ksson's reviews (1952) the following can be deduced. and 1. The ocean is the major souce of airborne salt. This is clearly shown by the higher salt content of rain in coastal areas compared to inland zones.
2. The concentration of sea salt in precipitation shows an annual and geographic variation.
3. The composition of salt in precipitation deviates from the characteristic sea water ratios as one goes inland.

c. Mechanisms of Production of Atmospheric Sea Salt Particles:
It is generally believed that sea salt particle with atmospheric residence time longer than a few minutes are produced by bubbles bursting at the sea surface. Blanchard and Woodcock (1957) investigated the various processes for bubble production at the sea surface. They found that breaking waves, or white caps are the most important mechanism for bubble formation. Sea salt particles are produced and injected in the atmosphere when these bubbles burst. (Boyce, 1951). Kientzler et al. (1954) showed that irrmediately after a bubble bursts on the ocean surface, it forms a bubble jet which ejects 2 to 5 droplets into the air . The diameter of the jet droplet is usually ~10% of the bubble diameter from which they form (Blanchard, 1963). Mason (1957) and Blanchard (1963) found that when the bubbles burst, other droplets 1-iere also formed due to the shattering of the bubble film cap. Little is known about the size and number distribution of these film drops.
In addition little is known about the size and number of bubbles in the open ocean under various wind speed conditions. Blanchard and \foodcock (1957) (1970 1977) used acoustic attenuation measurements to study Me win ' · bubbles size distribution and densities in coastal waters. He investigated bubbles of radius of 15-300 um at depth of 3-36 meters. He found relationships between the bubble size or densities and the season, time of day, wind speed and presence or absence of sea slicks. He postulated that micro bubbles are entrained by aerosols as they fall into the ocean as well as produced by biological activity. Blanchard (1963) estimated that at any moment ~3 % of the ocean surface is covered by white caps.

o. Geochemical eycle of Atmospheric Sea Salt
The size distribution of sea salt particles in the marine atmosphere was investigated by Woodcock and coworkers. Woodcock (1953) determined the size distribution of sea salt below the cloud base (500 m) over Hawaii and other marine and continental areas. He did this for several wind forces (Fig. 1). He found that the number and size of particles or airborne sea salt increase with increasing wind force and with decreasing altitude above the sea.
Using the size distribution of Woodcock (1953, Fig. l), and assuming a mean wind speed of 12 knots, Eriksson (1959), utilized Stokes Law to calculate the flux of sea salt particles that would be expected to fall through the surface laminar boundary layer into the sea. In the steady state this output sho~ld be equal to the total rate of production of sea salt particles in the area of the whitecap, from the ocean surface.
Eriksson calculated the rate of dry fallout over Hawaii, the Caribbean and Florida. He deduced an average dry deposition rate of 5.5 x io-12 gm -2 1 cm sec-of sea salt. The rate of removal over the entire ocean surface is then 5.4 x lo 14 g per year. He used rain and river runoff Cl data water relative to Cl He found that the ratio of Cl concentration in r ain concentration in river runoff was 0.3. He coneluded that over land dry deposition was twice as important as rain in transporting sea salt from the atmosphere to land. This led him to believe that over the ocean probably as much sea salt is removed by precipitation as by dry fa 11 out. Thus he suggested the to ta 1 amount of sea sa 1 t removed annually was approximately l0 15 g. He estimated that 90% of this amount is deposited over the ocean while the rest is removed over land as the cyclic salt subsequently carried by rivers, streams, and groundwater.
In 1963 Blanchard made a similar calculation but arrived at a 16 yearly removal of 10 g/yr of sea salt. Blanchard's calculation seems to be more refined because he made more realistic estimates of the mean wind speed over various parts .~f the world's ocean, not relying on a global average speed of 12 Kts. From an assumed annual oceanic rainfall of 89 cm for the Hawaiian area, and an estimated salinity of 3.4 mg/l in the rain, he calculated the flux of sea salt by rain as 9.60 x l0-12 -2 -1 gm cm sec . He compared this figure to the amount of sea salt fallout by sedimentation, which he computed as 5.1 x lo-12 gm cm-2 sec-l at a wind speed of 12 Kts, the average wind speed in Hawaii.
The ratio of the two fluxes led Blanchard to conclude that rainfall is tvJice as important as dry deposition as a removal mechanism of atmospheric sea salt to the ocean surface. If one accepts Eriksson's figure of 10% of the total sea salt deposited on land, Blanchard's estimate of total sea salt production of 10 16 gm/yr results in an estimated amount of cyclic sea salt deposited over land of ~10 15 9/yr. Livingstone (1963) calculated that the annual global input of cyclic sea salt from river runoff into the ocean is approximately 3xlo 14 g/yr, in between the estimates of Blanchard andEriksson. cawse et. al. (1972, 1974) made a two year analytical study of trace elements in the atmospheric environment in the United Kingdom.
TheY measured the concentration of 30 trace elements in airborne dust, rain water, and dry deposition at several localities. From these measurements they calculated a .dry dep0sition velocity for each element.
The deposition velocity, in cm/sec, v1as defined as the rate of dry -2 -1 3 deposition in ug cm sec divided by the concentration in air in i1g/cm .
They calculated the ratio of dry deposition to total deposit1on for Na+ and for Cl as 5%, i.e. rainfall was by far the major removal pathway.
Tsunogai (1975) investigated sea salt transport to land in Japan.
He found that the share of dry sedimentation was only 20% of the total removal over the various parts of Japan. Most of this 20% was transported during discrete events such as Summer typhoons.

E. Research Objectives
It can be seen that rainfall and dry fallout are the two main routes by which atmospheric sea salt is transported back to the ocean and to the land. As mentioned earlier, the total amount of atmospheric sea salt particles produced by the ocean is apparently in the range of 10 15 g/yr -1 o 1 6 g/yr. h · b · · T e primary o Ject1ves of this present investigation are to: 1. Test the validity of the two estimates of annual production of atmospheric sea salt particle.
2. Investigate and determine by chemical analyses the rate of dry and wet removal (rain) of itmospheric sea salt (Na and Mg) from the marine and near shore Rhode Island atmosphere to the earth surface, and to establish empirically the relationship between the rate of dry r emoval (rain) and wind speed and direction on the Rhode Island cind v1et coast.
Assess the role of dry fallout in comparison to wet removal 3.
(rain) of sea salt from the marine and coastal Rhode Island atmosphere.
4 . Calculate atmospheric deposition velocities to the earth surface in the coastal Rhode Island environment for particulate Na and Mg.
Dry deposition velocity v 9 has been defined (Chamberlain, 1960) as:

EXPERIMENTAL METHODS
Sarnp l in~ Procedures: A.

13
Rainfall, dry fallout, total deposition and atmospheric particulate samples were collected at the Narragansett Bay campus of the University of Rhode Island in southern R.I. The roof of a mobile trailer, which was approximately 2.5 meters above the ground, was used as an atmospheric sampling station. The stat ion is approx inw tely 150 mctrrs west of thr.
west coast of Narra0ansett Bay and 15 meters above sea level. A field recording wind system (model R.M. Young, 1973), was installed on the sampling station to provide a continuous record of wind speed and direction. Also a standard plastic rain gauge was used to record the amounts of rainfall.
Seventeen rainfall samples, 55 dry fallout samples and 28 atmospheric particulate samples were collected. The majority of these samples were collected during 24 hour periods continually from April 15 to June 28, 1976.
A few rainfall and dry fallout samples were collected during December 1975.
Duplicates of seven dry fallout samples were collected aboard RV/Trident, Cruise #169, Narragansett to Bermuda to Narragansett during July, 1975.
These samples were collected manually on top of the we t · laboratory.
The majority of the rainfall and dry fallout samples were collected in duplicate and a few in triplicate. Acid cleaned llxl0x4 cm and/or 42x29x25 cm plastic containers were used for this purpose. The atmospheric samples were collected on double vJhatman 41 filters using a high volume pump. These filters are reported to be 90% efficient for removing particles ~0.26 um in diameter for air (Stafford and Ettinger, 1972). Also the blank content of Na and Mg in these filters is low ( Cawse et al . , 1972;Dams et al . ' 1972) .
Out of the total numb er of dry fallout and rainfall samples , 34 dry fallout and 7 rain samples were collected using the automatic rain-dry fallout sampler (Fig. 2), while the rest were collected manually. Manual rain samples were obtained by sampling the rain shower beqinning immediately after the start of the shower and ending as soon as the rain stopped.
Dry fallout samples were collected during dry periods. Hhenever rainfall was collected with dry fallout samples, as was the case sometimes at inconvenient hours, the collected samples were considered as total fallout samples.

The Automatic Rain-Dry Fallout Collector (AEC)
This automatic collector is a modification of a device designed and built by Dr. Herbert Volchok of the Health and Safety Laboratory, ERDA, N.Y., N.Y. The system as shown (Fig. 2) consists of a two polyethylene bucket collector controlled by a printed circuit board rain sensor. Each collector was desinged to house precisely a -polyethylene bucket of a size 42x29x25 cm.
This way only single samples could be collected by the collector. In order to collect dry fallout replicate samples using the automatic collector, the system was modified as follows: a wooden platform wrapped v1ith polyethylene sheets and havin g the same size as the plastic buckets was placed in the collector instead of the dry fallout bucket. This platform was designed to house as many as 4 polyethylene buckets of a size llxl0x4 cm. During rainfall one vessel is exposed and collects rain while the dry fallout vessel (or vessels) is covered. \· J hen it is not raining the dry fallout collector(s) is exposed and the rain collector is covered . A heater is attached to the sensor to assure activation during cold periods and to dry the sensor so that a quick shut-off will occur at the end of precipitation. a: The automatic rain-dry fallout collector. Only the dry fallout half is shown. (The rain half is identical).
l 6 3. sample Handling ory fallout samples were obtained by washing the bottom and sides of the plastic buckets with distilled deionized water after sampling. Samples were stored in acid cleaned polyethylene bottles and kept in a freezer.
Blanks were obtained twice for every sample. The first blanks were taken from the clean plastic collection container before sampling using distilled deionized water. The second blanks were taken from the distilled deionized viater used to wash out the dry fallout samples.
Total fallout and rainfall samples were also stored in acid cleaned polyethylene bottles, kept in a freezer. For those samples of large volumes only aliquots were retained and the rest of the sample was disposed of after determining the total volume.
Uhatman 41 filter blanks were obtained periodically for the atmospheric particulate samples by treating filter blanks exactly as sample filters, except no air was drawn through them. Sample filters and blanks were sealed in polyethylene bags and stored in a freezer prior to analysis.

Testing Programs for the Sampling Systems
To determine the optimum conditions for sample collection, several tests were conducted prior to the start of the sampling program for this project. These tests, discussed in full below, included the following: determination of the collection efficiency of two types of plastic buckets for rainfall and dry fallout, collected manually and automatically; investigation of the collection efficiency of 3 type of surfaces for dry fallout collection; and determination of the collection reliability of the automatic r .
a. Determination of rainfall collection efficiency of 42x29x24 cm and llxl0x4 cm plastic containers.
The total volume of 13 rain showers collected in 42x29x25 cm (1233cm 2 bottom area) and llxl0x4 cm (110cm 2 bottom area) plastic buckets was measured.
seven of these rain shower sample~ were collected manually and 6 of them were collected using the automatic rain dry fallout sampler. Manual samples were collected in duplicate and triplicate. During rain shmver number 5, two sets of samples, 51 and SI! were collected.
The collection efficiency of the two types of plastic containers and collection methods were calculated relative to the total rainfall measurements obtained from the standard plastic rain gauge. The results, shown in Table (1) indicate that the two types of collectors and collection efficiency for each of the two sized of plastic containers was 98%. The mean collection efficiency for the automatic collector which used the larger bucket, was also 98%, with sample number 12 not included. The overall efficiency for all collectors was also 98%. -amp es ~ollected usinq the automatic rain -dry fallout sampler. The rest of the ., samples were collected manually. : mean collection efficiency for 110 cm 2 plastic buckets = 98 + 7 •mean collection efficiency for 1233 cm2 plastic buckets = 98-t 7 experiment a 20 mesh plastic screen was placed over one of the In one buckets. The purpose of this screen was to determine the extent of contam-;nation, if any, of the samples by insects.
The results of these studies, shown in Table 2, indicate that dry and wet bucket surfaces collect abo~t the same quantities of either Na or Mg.
The amount of sea salt fallout collected by these two surfaces are always higher than the amount collected by the Whatrnan 41 filter. In the ex periment where a plastic screen was used, the results shO\.v that the bucket with the screen cover collected almost twice as much as either the dry or wet buckets. These findinqs may be explained as follows: the wet surface was chosen in attempt to simulate the ocean surface. The fact that dry and wet plastic bucket surfaces have approximately the same collection efficiency for sea salt may be due to the hygroscopic nature of sea salt particles which exist as droplets at relative humidity over 70%. It is possible that sea salt particles falling on a dry bucket surface adhere to the side and bottom of the bucket just as well as they when they strike on a wet surface.
Although there is only one datum point for the experiment where a screen cover was used, the higher collection efficiency of the screen surface may be real. This may be because the screen itself acts as an impaction surface or as a filter. As a result of these experiments the choice of a collection surface for dry deposition was limited to a wet or dry plastic bucket surface.
However, due to convenience and generally lower blanks, the dry plastic bucket surface was chosen over the wet surface. Also it was noticed that cont · · am1nat1on by insects was far less for the dry surface compared to the Wet surface. 20 94 t47 Mg a =Total fallout (rain and dry fallout). b =Manually collected dry fallout. c = Wha1Jnan 41 filter. d = Samples collected in plastic buckets were obtained by washing the bottom and sides of these buckets.
Dissolution of Na and Mg in the dry fallout samples from the bottom and sides of the plastic buckets This experiment was conducted to determine the number of washes required Sh O ut Na and Mg in the dry fallout samples collected by the plastic to wa buckets. The experiment was also intended to estimate the amounts of Na and Mg collected on the sides of the plastic buckets as compared to the bottoms.
Distilled deionized water was used to wash out 4 sampl es of dry fallout collected in 42x29x25 cm plastic buckets. The bottoms and the sides of these buckets were washed 3 times each using a wash bottle, only the bottoms of the buckets were carefully washed at first. To avoid washing out some of the Na and Mg collec ted on the sides of the buckets, an acid cleaned pipette was used to transfe r the 3 washes from the bottoms of the buckets.
After that, sea salt collected on the sides of the buckets was washed down to the bottoms and transferred from the buckets. Again 3 washes were used.
The amounts of Na and Mg i n each wash was determined (Table 3a and 3b). Of the total Na and Mg collected on the bottoms of the buckets, 97% and 89% respectively were obtained in the first washes. 86% of the total Na and 66% of the total Mg collected on the sides of the buckets were obtained in the first washes, (Table 3b). Essentially all the rest of the Na on the bottoms and sides of the buckets was removed in the second washes of the buckets. For Mg, three washes were generally required.
Of the t~al amounts of Na and Mg in the fallout samples, 88±7% of the Na and 78±5% of the Mg were collected on the bottoms of the plastic buckets (Table 3a and  T 5 = s 1 +s 2 +s 3 Ts/Tot. = ratio of T 8 to (T 8 +T 5 1 l< 100 Gpercentagel s 1 ;r 5 = Ratio of the amount of Na or Mg obtained from the first wash of the sides of the buckets relative to the total amount of Na or Mg collected by the sides of the buckets .
Mg compared with those falling on the bottoms of the buckets. It is possible that some segregation of the falling aerosol particles occurs when they reach the sampling container. Perhaps a higher proportion of the aerosol particles of crustal origin, with their smaller size and higher Mg/Na ratio, are preferentially .deposited on the sides of the buckets.
Aerosol particles of crustal origin are known to be enriched in Mg relative to Na as compared to marine aerosols, (Junge, 1963;Riley and Chester, 1971 A description of the instrument and the collection efficiency of the containers used in it for rainfall were given earlier in this chapter. The efficiency of the automatic rain-dry fallout collector for total deposition is defined in percent as equal to (total amount of Na or Mg in an automatic collector dry fallout sample + total amount of Na or Mg in an automatic collector rainfall sample) divided by the total amount of Na or Mg in a manual total fallout sample, all times 100. Four experiments were conducted for this purpose . Replicates of rainfall and dry fallout samples were collected usin g the automatic rain-dry fallout collector while, during the same period of time, total fallout samples were collected manually using the plastic buck~ts. The collection efficiency of the automatic samples for total deposition was calculated for Na and Mg, (Table   4). In experiment 3 and 4, collected durin9 the same roin sho1ver, dry fallout and total fallout samples were collected over two different intervals .
The mean total deposition efficiency was 92±6% for Na and 96±3% for Mg. This high efficiency combined with the high collection efficiency for rainfall samples (Table 1), make the automatic rain-dry fallout collector an acceptable instrument for convenient and reliable sampling of rainfall and dry fallout samples. Galloway et al. (1976), investigated the collector reliability for rain and dry fallout of several bulk and automated collectors.
They found that the automatic rain-dry fallout collector, better known as the U.S. Atomic Energy Commission Collector, was the most reliable of all the automatic collectors investigated.   The results of the blank filter study, shown in As seen from the data in Table 7  was carried out for 2 minutes using an ultrasonic cleaner. The volume of extract recovered was determined by weight. Then another 10 ml of distilled deionized water was used to re-extract the filters in order to determine the efficiency of the first extraction. Again the volume recovered was determined by weight. The results (Table 10), were presented as percent recovery of the volume of solvent used and of the Na and Mg recovered from the first extraction. As shown, the percentage of

42
The results of the analyses of dry fallout, rain and atmospheric particulate samples collected in this study are given in Table 12-22. Except for samples 3,4,5,6 and 10 and samples collected aboard RV/Trident (Table 18), all samples were collected over a 24 hour period continually from April 15 to June 28, 1976. Samples collection information is shown on Table II.

B. Classification of Samples
After studying the meteorological data compiled during sampling, dry fallout, rainfall and atmospheric samples were divided into 3 different categories depending on the local surface wind direction during the sampling period (Fi~. 5). Samples collected while the wind direction was between 0900. and 230° were considered "marine" samples and were designated s.samples collected within the sector 230°-010° were considered "land" samples and were designated L. Samples collected while the wind was blowing from near the boundary between the land and marine sectors and/or where the wind direction shifted from the marine to the land sector or vice versa were considered "mixed" samples and were designated S/l. To this category, samples collected within the sector 010°-090° were also included.     Tables 12-14. Least-squares linear regression analysis was used, for each category of samples, to evaluate the relationship between the wind speed and both the rate of dry fallout and the atmospheric concentrations of particulate Na and Mg. Intercepts, slopes and correlation coefficients calculated from these regression analyses were obtained and are also presented in Table 12-14. ii. "Corrected to Marine" samples Since over 50% of the total number of dry fallout and atmospheric samples were "mixed" samples, an attempt was made to recalculate the rates of dry fallout and atmospheric concentrations for the marine fraction of Na and Mg in these samples. Only "mixed" samples where the wind shifted from marine to land sectors or vice versa were chosen for this calculation. These constituted the majority of the "mixed'' samples category. Calculations were made using the following parameters determined for each samples: a) the time intervals during which the local surface wind was blowing off the ocean and off land; b) the rate of dry fallout for the land fraction of the samples. This rate was assumed to be a constant, not varying with wind speed and equal to the mean rate of dry fa 11 out calculated from the "land 11 subenvi ronment of the R.I. c~astal environment. (0.260xl0-12 gNa/cm?sec and 0.215xl0-12 2 gMg/ cm. sec). The Na and Mg rate of dry fa 11 out for the marine fraction was then calculated as follows: ( 2)    corrected rate of dry fallout in g x lo-12 cm-2 sec-1 for Na or Mg. rate of dry fallout for a "mixed" sample in gxlO-l~ cm-2 sec-l for Na or Mg. FL = assumed rate of dry fallout for "land" fraction = 0.260xlo-12 gNa/cm 2 /sec -12 2 = 0.215xl0 gMg/cm /sec = total sampling period in seconds for a "mixed" sample.  (Table 12) .
The "corrected-to-marine 1 ' rates of dry fallout, and atmospheric concentrations for particulate Na and Mg, the corrected Mg/Na ratios, and the wind speeds during sampling of the marine fractions of the "mixed" samples are given in Table 15 . Correlation between these values and wind speed as well as sl_ opes and intercepts from linear least square regression analysis are also tabulated in Table 15. These same parameters were recalculated for the total sample population after the "marine" samples were combined with the "corrected to marine" samples (Table 16) .
Since "mixed" samples collected within the wind sector 010-090° showed a different trend from the rest of "rnixed' 1 samples, they were l isted separately in Table 17 . Since samp le 14 has a continental fraction, it was corrected to the 010-090° sector.

b. Samples collected aboard RV/Trident
Dry fallout samples were collected during cruise #169 from Narragansett to Bermuda to Narragansett in June-July 1975 . The rates of Na and Mg dry fallout were calculated for the samples and are reported along with surface wind speed, in Table 18. Fig . 6 shows the stations where these samples were collected.

C. Deposition velocity for particulate Na and Mg
The deposition velocity was defined earlier in Chapter I. Na and Mg deposition velocities in cm/sec calculated for samples collected in the three sampling categories ("marine", "land", and "mixed") and for the"corrected to marine" samples are given in Table 19 . The mean Na and Mg deposition velocities for each subenvironment and the overall mean for the R. I . coastal environment are also presented in Table 19 .  Hind = s l ope X (rate of dry fallout or atmospheric concentration) + intercept aNot included in calculating the mean because there is a possible error while adjusting the rate to the 010-090° sector.       T =Total fallout, R = Rainfall M =Manually collected sample; A= samples collected using the automatic rain dry fallout collector * Thunderstorm Sample 53 was not included in the calculation of the mean and regression analysis.
T =Total fallout, R =Rain M =Manually collected sample ; A= Sample collected using the automatic rain dry fallout collector * Thunders to rm a Thundershoweroccurredwithin one hour (15 :00-16 :00) on 5/21.  The means of the rates of dry fallout in gxlo-12 ;cm 2 /sec and atmospheric concentration in ug/m 3 for Na and Mg collected in the 3 subenvironments of coastal R.I. environment are shown in Table 23.
Valt,1es obtained from "corrected to marine" samples are also included. As expected,these data indicates clearly that the ocean is the major source of atmospheric Na and Mg in coastal Rhode Island.
Sea salt particles produced by bubbles bursting at the ocean surface are carried by the prevailing winds in the atmosphere. Since they have residence times of a few minutes to a few days depending upon their size, (Junge, 1963), they are ultimately removed from the atmosphere by dry fallout and rainfall to the ocean surface, coastal areas and some smaller fraction to in~and zones. The higher dry removal rates and atmospheric concentrations shown for "mixed" samples compared to "land" samples is due to the marine fractions in these samples. This is clearly demonstrated by the fact that "corrected-to-marine" samples show values comparable to marine samples.
The correlation coefficients for the rates of dry fallout or atmos pheric con~entration for Na or Mg versus wind speed are above 0.80 for the "marine''samples category (Table 12). These are significant correlations at the 1% level, i.e. a 99% probability that a real correlation existsbetween these variables (Fisher 1958). A plot of wind speed versus the atmospheric concentrations and wind speed versus the rate of dry fallout for Na and Mg are shown in Fig. 7-8 respectively.
The regression equations for the resulting lines obtained in Fig. 7 and The intercepts in these equations represent the wind speed in miles/ hour at which Na and Mg rate of dry fallout or atmospheric concentration are equal to zero. This zero Na or Mg intercept may be explained as the minimum wind speed above which white caps (which produce the atmos- at any time , at wind speed less than 8-12 miles/hour is residual sea salt produced previously at a higher wind speed. Blanchard (1963) discussed the problem of attaining a steady state of atmospheric sea salt particles at a given wind speed. A steady state for a given wind force will be attained only if the wind blows at the same speed for a period of time equal or exceeding the average residence time for the smallest particle in the air-borne sea salt size spectrum.
The correlation coefficients for the Na and Mg rate of dry fallout versus wind speed for "land" samples .are 0.75 and 0.60 respectively.
These correlation coefficients are significant at the 2% and 5% level respectively (Fisher, 1958). Fig. 9 shows the plot of the Na and Mg rate of dry fallout versus wind speed for 11 land 11 samples.
Since the Na and Mg in these samples are primarily of crustal origin (Mg/Na= 1.1; Mg/Na in the earth's crust= 1.0, (Riley and Chester 1971), it is likely that the higher the wind force over land the more crustal The zero Na and Mg intercepts in these two equations at 9 miles/hour may be interpreted as the minimum wind speed above which significant quantities of soil and crustal weathering particles are lifted into the atmosphere by the wind. Since we do not know the particle size and residence time of these particles, it is not possible to determine accurately whether the question of 11 lag time 11 discussed earlier is also involved here. However, since the Beaufort scale indicates that wind ·raises dust on land at speeds of 13-18 miles/hour, the low intercepts may also be due to 11 lag time'' and equilibria between production and/or removal of these particles and the wind speed.
As indicated in Table 13 no significant correlation was observed betv.1een the atmospheric Na or Mg in 11 land 11 samples and the wind speed. This is probably because there are no atmospheric concentration samples available for the sampling periods which have the two highest rates of dry fallout and wind speeds.
As expected, 11 mixed 11 samples showed intermediate values between "marine 11 and 11 land 11 samples for the mean rates of dry fallout and atmospheric concentrations of Na and Mg and for the Mg/Na ratios (Table 14). It is apparent that the bulk of Mg and Na in these samples is derived from the ocean with the rest primarily contributed by the crustal ·weathering products. Linear least square regression analysis for this category of samples showed that the only significant correlation (1 %) was between Na dry fallout and wind speed. The correlation coefficient between Mg dry fallout and wind speed is significant at only the 10% level i.e. (90% probability that a real correlation exists, Fisher 1958).
In an attempt to look at these marine fractions of "mixed" samples, the "mixed" samples were "corrected-to-marine" as described in Chapter III (Table 15). Linear least square regression analysis showed that correlations between the corrected rates of dry fallout, corrected atmospheric Na and Mg with corrected wind speeds were very significant (1 % level). The linear regression equations for wind speed versus the Na and Mg rate· of dry fallout and atmospheric concentrations can be derived from the slopes and intercepts gfven iri Table 15.

Hin~C)
WinfC) = 2. 7 x Na + 4. 9 If we compare these equations with equations 6,7,8 and 9 derived for "marine" samples, it seems that only the Na/wind equations are comparable to each other. It is interesting, however, to observe that for the combined "marine" and "corrected-to-marine 11 samp 1 es significant cotrel ati ons were observed between both dry fa 11 out and atmospheric concentrations versus wind speed for both Na and Mg (see Table 16).
This suggests that the assumptions behind the calculation of the II corrected-to-marine" rates of dry fa 11 out and atmospheric concentration for Na and perhaps Mg may be valid. A plot of wind speed versus the atmospheric concentrations and wind speed versus the rate of dry fallout for Na and Mg in the 11 marine 11 plus 11 corrected-to-marine 11 sample populations combined are shown in Figs. 10-11. b. 11 Mhed 11 samples collected within the 010-090° wind sector Although samples collected within the wind sector 010-090° are essentially 11 mixed 11 samples, they appear to show a different trend than the rest of the samples . Since they merit special discussion they are listed separately in Table 17. As shown, it seems that the rates of dry fallout and atmospheric concentrations of Na and Mg in these samples are not related to wind speed. Although the wind speeds recorded while sampling these samples were quite variable, the differences in the rate of dry removal and atmospheric concentrations are negligible. As can be seen from the map of sampling sectors (Fig. 5), wind blowing within the sector 010-090° can be considered to bring marine air masses which have travelled over land for ~150 kilometers (100 miles). The constancy of the rate of dry fallout and atmospheric concentration of Na and Mg and their non-dependence on wind speed may suggest that the decrease in the rate of dry removal and atmospheric concentrations of sea salt as we go inland is not continually linear.
It is likely that most of the sea salt produced by the ocean is removed very rapidly and is primarily deposited in the coastal zones.
Sea salt particles of smaller size a~d longer residence time are possibly carried further inland. As shown in Table 17, the mean rate of Na dry fallout for samples collected in this sector is ~ l.5xlo-12 g/cm 2 sec at a mean wind speed of 12.5 miles/hour. If  I  I  I  I  I  I  I  I  I  I  I  I  I  I   I   I 0  0 0.3 0.6 0.9 1.2 1.5

FALLOUT
{ g x 10-12 /cm 2 sec) -....J (Jl we use equation 6 to calculate the rate of Na dry fallout at the coast at this wind speed (~3xlo-12 g/cm 2 sec), it appears that the rate of Na dry fallout has dropped by a factor of ~2 within the 150 Km (100 miles) distance from the coast. If we assume that the wind did not change speed and the air masses were not modified meteorologically during their travel from the coast to the sampling site, the mean lifetime of the larger sea salt particles may be estimated very roughly as ~s hours or less.

c. Samples collected aboard R/VTrident
Out of the total number of samples collected aboard R/VTrident over the northwest Atlantic, dry fallout rates calculated for 4 samples were ~soxl o-12 gNa/cm 2 sec and for 3 samples were ~3-6 x l o-12 gNa/ cm 2 sec (see Table 18). The former group of samples were collected at a mean wind speed of 24 miles/hour, while the latter set of samples were collected at a wind speed between 12-16 miles/hour. It appears, however, that the high flux values obtained from the first set of samples were due to contamination with bow generated sea spray. Since these samples were collected while the ship was moving, sea spray contamination was unavoidable, especially near the bow of the ship where the samples · were collected. During the collection of the second group of samples, the sea surface was not as rough. Since the rate of dry fallout calculated from these latter samples agree well with those calculated from "marine" samples collected at the coastal Rhode Island site, it appears that the relationship established between the rate of dry fallout and wind speed for "marine" and "corrected-to-marine" samples collected at the Narragansett Bay Campus of U.R.I. also holds true for the open North Atlantic ocean and possibly the world's oceans.
d. The Mg/Na ratios The mean Mg/Na ratio of 0.16 for "marine" samples approaches the sea water ratio of 0.12. The excess Mg is possibly due to vertical mixing of the air masses with consequent addition of crustal Mg from continental sources . This probably happens when the marine air masses enter coastal areas. This explanation is further supported by the Mg/Na ratios observed for samples collected aboard R/VTrident over the North Atlantic Ocean (Table 18). The mean Mg/Na ratio calculated from 3 sa mples where both Mg and Na were analysed was found to be 0.126, very close to the sea water ratio of 0.1 2. On Lhe 0L11er hand "l.:itit.l" samples showed a mean Mg/Na ratio of l .l for the dry fallout and 0.79 for the atmospheric samples (Table 13). This is almost identical to the 1.0 Mg/Na ratio in the earth crust (Riley and Chester 1971).
This suggests that the major fraction of the Mg and Na in these samples is of crustal origin.
As expected "mixed" samples showed an intermediate Mg/Na ratio (rc.0.2-0 . 3) between the "marine" and "land" samples. Since this is considerably closer to the sea water Mg/Na ratio than the crustal Mg/Na ratio, it suggests that the source of the main bulk of the Na and Mg in these samples is the ocean.
As described earlier the mean Mg/Na ratio observed for samples collected aboard R/VTrident was found to be almost identical to the sea salt ratio . Since these samples were collected at a mid-ocean location, where contamination of the marine aerosol with crustal particles is minimal, this suggests that there is little chemical fractionation of sea salt Mg in the marine environment. Several investigators have reported that the Mg and other alkali and alkaline earth metals are enriched relative to Na in sea salt aerosols due to chemical fractionation of these particles when they are produced by bubbles bursting at the sea surface (eg. see Sugaw2ra et al. 1949;Koyama and Sugawara 1953, Oddie 1960, Komabayasi 1962, Chesselet et al. ,1972, Buat,..Menard et al. ,1974. This subject was reviewed recently by Duce arid Hoffman ( 1976); Hoffman et al., (1977). Observed enrichment of Mg and other alkali and alkaline earth metals relative to Na in environmental samples was found to result from the addition of non-sea salt particles, particularly crustal particles to these samples.

e . Deposition velocities for particulate Na and Mg
The mean deposition velocities (V ) calculated from the 3 different g subenvironments of Rhode Island coast and "corrected-to-marine'' samples are 2.4 ± 1 .6 cm/sec for Na and 3.0 ± l .6 cm/sec for Mg (Table 19). Cawse et al. (1972 determined V for Na to range from were classified as "marine", 6 were "land" and 3 were 11 mixed 11 ,depending on the wind direction during the rainfall. The mean concentrations of Na and Mg in rain water and the mean amounts of Na and Mg deposited in ug/cm 2 for each sub.environement are given in Tables 20, 21, 22   and summary Table 23. As shown by these data most of the Na and Mg removed by rainfall originated in the ocean (see Tables 20,21 ,22).
When the Na concentration and the total amount of rainfall are considered, more sea salt is removed by "marine" and "mixed" rain than by rain occurring when the wind is blowing off land. As would be expected linear least square regression analysis between wind speeds and concentrations of Na and Mg in rainfall showed significant correlations for "marine" rains and no significant correlation for "land" and "mixed 11 rain showers. A plot of wind speed versus Na and Mg concentrations in 11 marine 11 rain is shown in Fig. 12 . The regression equations for the resulting lines are: Since these rains are marine in origin the 7 miles/hour intercepts may again be considered as the minimum wind speed above which sea salt production occurs at the ocean surface. Since white caps are formed at wind speed between 8-12 miles/hour, in the open ocean, the low intercept of 7 mil es/hour may be a result of the 11 1 ag time 11 between production and removal of atmospheric sea salt. This concept was discussed earlier for atmospheric particulate Na and Mg concentrations and dry fallout.
The Na or Mg concentration in 11 marine 11 and 11 mixed 11 rains in mg/l, showed an inverse relationship with the rain depth in cm ( Fig. 13-14).
Since this relationship appears to be logarithmic, it appears washout processes may occur at the very beginning of the rain shower and may be completed very rapidly. After the completion of washout, the effect of rain is essentially dilution. This relationship was not noticed for 11 l and 11 rain.
The depth of rainfall (Table 20-22), h, in cm, is measured by a standard plastic rain gauge. The amounts of rain represented by h are sometimes higher than the amounts collected by the plastic buckets. This is apparently due to evaporation of the rain water in the latter collectors. The mean depth of rainfall for 11 marine 11 and 11 mixed 11 rain  samples was about l .Ocm, while the mean was only 0.4 cm for land rainfall. This is likely due to the higher quantities of moisture associated with "marine" air.
b) Selective removal of sea salt aerosol by rainfall Table 23 presents the mean Mg/Na ratios ~alculated for rainfalls collected for each of the 3 meteorological subenvironments on the Rhode Island coast. Also shown are the mean Mg/Na ratios observed for dry fall out sa111p l es collected \vi thin these s ubenvi ronments. The mean Mg/Na ratio in "marine" rainfall samples is 0.15, similar to the Mg/Na ratio observed for "marine" dry fall out samples. These values approach the sea salt ratio of 0.12. The small quantity of excess Mg is possibly due to the inclusion of Mg of crustal origin. It is interesting, however, to note that the Mg/Na ratios in "mixed" and "land" rain samples, 0.19; 0.32, are lower than their counter parts in dry fallout samples, 0.30 and 1 .1 respectively. This may suggest preferential wet removal or washout of sea salt Na and Mg relative to crustal Na or Mg. However, if we assume that during rainout Mg and Na are selectively consumed in the sea salt ratio (Martens et al.,1973), and that dry fallout samples are equivalent only to the fraction of Na and Mg collected by raindrops ·during washout, then the net Mg/Na ratio in rain water (a result of rainout and washout processes) must be lower than its counterpart of dry fallout in the "mixed" and "land" subenvironments. [3.
Geochemistry _Qf_ Atmospheric Sea Salt l. Significance of Wet Removal (rain) of Sea Salt Relative · to Ory Deposition in the Coastal Environment' of Rhode Island The mean rate of dry fallout and the mean rate of rainfall removal for Na and Mg were calculated from the total population of samples collected at the coastal Rhode Island site between 4/15/76 and6/28/76 (Table 24). As shown, if we compare the mean rate of wet removal (rain) of Na and Mg to the mean rate of dry removal during that period of time, it appears that equivalent amounts of Mg were removed by rainfall and dry fallout and 1.3 as much Na was removed by rainfall. Since some of the manually collected rainfall samples, especially those collected at inconvenient times, possibly contained a small fraction of dry fallout, the ratio of wet removal (rain) to dry removal of atmospheric Na and perhaps sea salt is possibly close to one in the coastal environment of Rhode Island. Although this ratio was calculated from samples collected during only one quarter. of the year, we will assume for subsequent calculations that the same ratio holds for the entire year. Although only 10.6 cm of rain fell on the Rhode Island coast during the sampling period between 4/15/76 -6/28/76, lower by a factor of two than the amount expected from the mean annual rain depth of 107 cm (National Weather Service Station, Warwick), it is possible that these rains contained higher sea salt concentrations than the average rains. As shown in Fig. 13-14, the concentrations of Na or Mg in rain water are inversely proportional to the depth (or volume) of rain.

Sea Salt Yearly Removal Over the World Oceans
In order to estimate the rates of wet and dry removal and the annual removal rates of sea salt over the world's oceans, the mean rates of dry fallout and rainfall were calculated from the "marine" and "corrected-to-marine 11 samples as well as samples collected aboard RV/Trident (Table 24). The mean wind speed at which all these samples were collected is approximately 13 miles/hour (5.5 m/sec.) (Table 16).
Assuming an annual oceanic rain depth of 100 cm (Baumgartner and Reichel, 1975) and a mean Na concentration of 1.6 mg/l, (Table 20), the -12 2 rate of Na wet (rain) removal from the atmosphere is 5.1 x 10 g/cm sec, -12 2 equivalent to 17 x 10 g sea salt/cm sec. If we compare this figure to the mean rate of dry fallout of 3.3xlo-12 gNa/cm 2 .sec or llxlo-12 g sea salt/cm 2 sec, it appears that, over the ocean, approximately two thirds of the atmospheric sea salt returns to the ocean in precipitation and approximately one third as dry fallout (Table 24).
If we use the sea salt dry and wet removal rates and integrate over the entire area of the world's oceans, then the total annual removal rate of cyclic sea salt at a mean wind speed of ~20 Km/hour is approximately 30xlo 14 g/yr. Approximately 2/3 of this is removed by rainfall and the rest by dry fallout (Table 24). The 30xlo 14 g sea salt/yr removal rate determined is higher by a factor of 3 than the estimates made by Eriksson (1959) and is approximately one third the value obtained by Blanchard (1963) (Table 24).

Significance of Atmospheric Removal of Sea Salt Over the Coast of Rhode Island
If we compare the mean rate of Na wet and dry removal on the coast of Rhode Island for all sampling conditions with those for the marine environment (i.e. "marine" samples only) ( 4 0% as much as the amounts, per unit area, removed over the ocean.

Atmospheric Removal of Na and Mg Over Land
The Na and Mg analysed in "land 11 samples collected in coastal Rhode Island are essentially of crustal origin as shown by their Mg/Na ratios discussed earli er. Although the wind direction was off land during the collection of these samples it is possible that a fraction of the Na and Mg in these samples originated in the sea. In this respect these samples resemble th e aerosol at inland and continenLal environments, where a small fraction of the aerosol Na and Mg is of oceanic origin, and the main bulk originated in the earth 1 s crust. This small fraction of sea salt particles reaching mid-continent area are usually of smaller size and have longer residence times than the salt particles ~ver the ocean.
Taking all this into consideration one can attempt to estimate tentatively the yearly removal rate of Na and Mg over the continents.
The rate of dry and wet removal of Na over land is given in Table 24.
The rate of Na wet removal was calculated from a mean 11 land" rain concentration of l .03 mg/l and an annual rain depth of 75 cm (Baumgartner and Reichel, 1975). If we compare these rates of removal to the rates of removal over the oceans and take into consideration the fact that land area constitute only 25% of the earth 1 s surface, it seems that only 2% of the total sea salt removed by dry fallout over the earth, is deposited on land and 12% of the total sea salt removed by rainfall is deposited over land. The total amount of salt.calculated from the Na data re~oved over land would be ~2.2 x 10 14 g/yr . This is ~8% of the yearly global removal rate. Since a fraction of the Na and Mg in "land" samples is of crustal origin, and because the Rhode Island V. CONCLUSIONS The rates and processes of removal of atmospheric sea salt from the marine and Rhode Island coastal atmosphe re to the earth surface have been investigated. Samples collected at a Rhode Island coastal site appear to represent 3 meteorologically different subenvironments depending on the surface wind direction during sampling. These 3 subenvironments, which when superimposed r ep resent the coastal environment of Rhode Island have been designated 11 marine 11 , 11 land 11 , and "mixed", the latter a mixture of the first two.
Dry fallout, rainfall and atmospheric particulate samples collected in this study were analysed for Na and Mg using atomic absorption spectrophotometry. The rate of dry fallout and wet (rain) removal of atmospheric sea salt from the coastal Rhode Island "marine" subenvironment, showed a direct relationship with wind speed. Also the rate of Na dry fallou t for samples collected while the wind was off land appears to be dependent on the magnitude of wind speed. Dry fallout samples collected aboard R/VTrident over the North West Atlantic Ocean gave Na and Mg rates of dry removal comparable to the values obtained from "marine" sampl es collected at the coastal R.I. site. Dry deposition velocities for particulate Na and Mg, calculated from samples collected on the Rhode Island coast,were 2.4±1.6 cm/sec for Na and 3.0±1 .6 cm/sec for Mg.
The Na or Mg concentration in "marine" and "mixed" rains showed an invers e logarithmic relationship with the rain depth in cm. Comparison between the mean Mg/Na ratios in dry fallout samples with rainfall samples collected in the 3 subenvironments of coastal R.I. indicate that rainfall more efficiently removes sea salt aerosols relative to crustal particles.
Rainfall and dry de position are the major processes for global atmospheric removal of cyclic sea salt. The annual global removal rate of atm ospheric sea salt was estimated as ~3.0xlo 15 g/yr, higher by a factor of 3 than _an esti mate made by Eriksson (1959), and lower by a factor of 3 than the value calculated by Blanchard (1963). It is estimated that ~2/3 of the total cyclic sea salt is removed by rainfall and the rest by dry de position. Less than 2. 2x lo 14 9/yr of cyclic sea salt (~7% of the total) is removed over land by rainfall and dry deposition. Rainfall is 9 ti mes as important as dry deposition over land.