LEVELI
rCORPORATION
DTICAftELECTEI MAR 26 1981
i
D
I
I
DISTRIBUTION STAtEMNT AT
Approved for public releaSe
J
S81 3 05 008I/
I13
T CORPORATION
Accession For P. 0. Box 2019-iTIS GRA&I 2600 Garden Rd.DTIC TAB 0 Monterey, California 93940Unannounced 0 Phone (408) 649-3880Just ificaticOrL__
::~-1:~-11 February 1981
Dist ribution/ BDM/M-TR-0002-81
Av'i4jjb±1,,t Codes
Avail an/orDiut Si,ccial
~DTIC
S ELECTEDl 'MA, 2 6 i981
D
An Analysis of Surface Production
ofSea-Salt AerosolsIC. W. Fairall
BDM/NPS
Monterey, California 93940
Ir3!T!BU' 1ON :,&A ,/.i;T ,l
Approved for public el~e(so;
Di:,tribution Unlimitod
I
j V~tAV CLASSIFICATION Olp TuNS PACC (0% D-00 tC.rn...dj p.
iL4 OT umb ft . G VT C~g _ 7 R PJ 9tMN@* CA O G u~ m NUM611 U
_____________________________________ 81DM/M-TR-06092-81
4~~~~~a qotmc oh - GmieI FREO T PEIODMUE.
) C..Farl Project Enine NI00014-78-C-0204
9 pgAkOEmwI.GORGANIZATION dAimi-N AND OONI91 to. PNOORAM ELEMENT. PROJECT. TASKA 9A & WORK UN IT NUM411NS1
The BDM Corporation Wo rk Order 08642 22600 Garden RoadMonterey,_CA__93940 _______________
it comTROLLINO OFF)CI NAME &NO ADDRESS * a'
Dean of Research Code 012 11( Febmp 8Naval Post Graduate School a. *4w"FPSox
Monterey, CA 93940 214 MONITONING AGEN"cv NME &R&OO*Es""ed" 7II 1. Cantoiidmg 1fte)j #11. ICURITY' CLASS. (at thi le ,mr
Mr. Robin Simpson NR Representative UNCLASSIFIEDStanford University, Room 16,.,Durond Aeronautics Building' Q J al. OE[CLASIPICATiONIOOWN6GNtACING
Stanford, CA 94305 I16 0DiTa0i STATEMENT (of this *spot$)
~ ~ r~. :~:- "-,DISTIRIBUTION ST-ATE-MENT A-
Approved for public ieeoae;Distribution Unlimited-
17 OINuINSAEET(1**AI~ ~8eS2.d Ilcn
10 SUPPLEMENTARY NOTES
is K IV WORDS fc~tmp 801 Ite* eed ame and #*it 67 6J0011 "WIe)
30 ADS? ACT1 (Covo*1* It-** 6140 of ne0.00m ad IAI1fr 6P 610641 minb4e
Pagqe X
DD . ',0:',,1473 to 1 O FYO 01 -11 4111 OhS OUSOLITIrI NCLMS I F I IFD
5ECUN01TV CLASSIFICATION OF T"1% PAGE (6 0 1.000
I THE BDM CORPORATION
I FOREWORD
This report was prepared under Work Order Number 422 of Contract
N00014-78-C-0204 in support of the U.S. Naval Postgraduate School researchproject supported by the Naval Air Systems Command (Air 370) and the Naval
J Material Command (EO/MET). The work was performed in support of the Envi-
ronmental Physics Group at NPS under the direction of Professors G. E.
Schacher and K. L. Davidson.
I
I iiiL ' ..
I THE BDM CORPORATION
ITABLE OF CONTENTS
FOREWORD ii
TABLE OF CONTENTS iii
LIST OF FIGURES iv
LIST OF TABLES v
ABSTRACT vi
A. INTRODUCTION 1
B. AEROSOL SPECTRUM 2
C. FLUX SPECTRUM ANALYSIS 5
D. FLUX SPECTRUM AND WIND SPEED 10
E. NON-EQUILIBRIUM 16
INITIAL DISTRIBUTION LIST 19
V
_ _ _ _ !
THE BDM CORPORATION
ILIST OF FIGURES
Figure Page
1 Ensemble average total aerosol volume 4spectra from JASIN. The number to theright of the spectrum is the wind speedcategory in meter/second.
2 Time series of atmospheric 222Rn activity 7(solid line, Larsen, Kasemir and Bressan,1979) ai.: continental aerosol coefficient(dashed line), A'.
3 Time series of the Lerms of Equation 9 8for 2 pm radius and 15 pm radiusparticles. The dashed line is theaverage contribution of the surface pro-duction term and should be the sum ofthe other three terms.
4 The ensemble average surface flux spectrum, 9F (r), for the CEWCOM-78 analysis period(verage wind speed, 8 m/sec).
5 Ensemble average contribution of entrain- 11ment (W^), Stokes fallout (W ) and surfaceflux W F1 F/V s assuming a slate of dynamicequii ri u m.
6 Ensemble average equilibrium sea salt volume 12spectra from JASIN. The number to the rightof the spectrum is the wind speed category inmeters/second.
7 Ensemble average surface flux spectra de- 14duced from Equation 11 and Table II. Thenumber to the right of the spectrum is thewind speed in meters/second,
8 Changes in aerosol spectral density (dV /dt) 18versus changes in mixing volume (-dh/dti.This graph illustrates thr dominance ofthese terms in Equation 9 for short timeperiods.
vii
w - - - - - - -- " - : -'"=i-- '"'' m, - "
I THE BDM CORPORATION
II LIST OF TABLES
Table Page
I CEWCOM-78 meteorological and aerosol data 6for the time period analyzed for thisreport. The aerosol volume spectra densityV(R) is given in each radius, R, in pm.
II Equilibrium sea-salt aerosol-spectra in 13pm2/cm 3 as a function of radius (pm) andwind speed (m/sec).
III Surface sea-salt aerosol flux, F (r), 15(m:m 2/cm/sec) as a function of p~rticleradius (pm) and wind speed (m/sec).
IV The equilibrium time constant for aerosols 16at S 0.8, h 400 m and We = 0.4 cm/sec.
IIII
I
iI
ix
L7I
I THE BDM CORPORATION
i ABSTRACT
iThe rate of production of sea water droplets by bubble bursting at
the ocean surface is virtually impossible to measure directly. Laboratory
simulations are also difficult. However, the surface flux can be inferred
from measurements of the rate of change of aerosol densities and changes in
the height of the marine boundary mixed layer. Data from CEWCOM-78 has
been analyzed to produce the aerosol surface flux volume spectrum from 0.5
to 15 pro radius at a wind speed of 8 m/sec. Using this flux spectrum and
equilibrium aerosol spectra from JASIN, similar flux spectra are calculated
for sind speeds from 5 to 15 m/sec.
II
I
ITHE BDM CORPORATION
IA. INTRODUCTION
There are two compone nts to aerosols in the marine boundary layer:
1) continental (background) and 2) locally generated sea-spray droplets.
The sea-spray droplets are generated primarily by the bursting of bubbles
Iat the sea surface. The bubble are produced by biological activity,
chemical reaction and breaking waves (white caps). At wind speeds greater
than about 3 meters per second, white caps are the primary contributor
to the bursting bubbles. Thus, the sea surface is a continuous source of
sea-salt aerosols in the marine boundary layer. These surface produced
aerosols can be characterized by a surface flux spectrum, Fs(r), which
represents the volume of aerosol produced per square centimeter of ocean
I surface each second as a function of aerosol particle radius. This quantity
is a function of wind speed.
The continuous production of sea-salt aerosols is balanced by several
removal mechanisms. One obvious mechanism is the loss of particles as they
fall back to the surface. This settling under gravity is called "Stokes
fallout" and is characterized by the Stokes velocity. The particles are
transported vertically by turbulence in the marine boundary layer and main-
tained at a uniform density throughout the mixed layer. The growth of the
height of the layer constitutes another loss mechanism called "entrainment."
IThe final loss mechanism is "rainout," which occurs when the particles
become condensation nuclei in the formation of clouds.
IGiven the surface flux spectrum and a parameterization of the removalmechanisms, one could predict evolutions of the aerosol density spectrum.
Unfortunately, the surface flux spectrum is not known. However, we can
reverse the process--that is, use parameterizations of the removal processes
and evolutions of the aerosol spectrum--to obtain estimates of the surface
I flux spectrum. This-report describes the calculation of F s(r) from data
taken by the NPS Environmental Physics Group during CEWCOM-78 and JASIN.
I
THE BDM CORPORATION
B. AEROSOL SPECTRUM
The first step to analyzing the evolution of the aerosol spectrum is
to remove those variations due to changes in the ambient relative humidity.
This is done by transforming each spectrum to a reference relative humidity
(RH 80% or a saturation ratio S = 0.8). We make the usual assumption
that the particle radius a saturation S is given by:
r = r0 g(s) (1)
where r0 is the radius at S = 0.8 and g(s) = 0.81 exp [0.066S/(1.058 - S)].
If we have a measured aerosol number density spectrum, n(r), then the
transformed spectrum at standard saturation, n'(ro ), is:
dn' '( ) = n (r g(s)) g(s) (2)
0
Since we prefer to work with the volume spectrum V(r) r3 n(r), the
corresponding relationship is:
dV _V(r V(r g(s))/g(S) (3)dr 0o
We now must separate the measured spectrum into its two basic components:
1) the background continental aerosol of nonlocal origin, Vc, which
is present above and within the marine layer; and
2) the sea-salt aerosol locally generated at the sea surface, V5.
which is well-mixed throughout the marine layer (Z < h).
Thus, we assume:
V(r) Vc (r) + V s(r) Z < h (4a)
V(r) = Vc (r) Z > h (4b)
2
THE 8DM CORPORATION
The continental aerosol component is represented with a Junge typedistribution,
f V (r) = A/r (5)
where A is the continental coefficient. Since there is very little sea-
salt volume production for r < 0.3 pm, we can use the small size rarvge of
the spectrum to calculate the continental coefficient (A' = ro V'(r )).
This is docuwented in Figure 1 where ensemble averages of volume spectra
indicated that V'(r ) is essentially independent of wind speed at r = 0.1 pim.0 0
At low level wind speeds, the r = 0.1 inl and r = 0.3 pim ensemble average
spectra are described quite accurately by Equation 5. The claim that A'
is an accurate index of continental influence is nicely validated by com-
parison with atmospheric Radon activity from CEWCOM-78 (Figure 2). Thus,
locally generatod sea-salt component can be calculated using:
V'(ro) = V'(ro) - A'/r 0 (6)
Time remainder oi this report will deal exclusively with the volume
spectra transformed to standard humidity, V'(ro). However, to simplify
t!,e notation we will drop the prime and sub-zero with the understanding the
VS now refers to the spectrum at RH = 801, with radius, r, an implied
variable.
The evolution of the aerosol spectrum is described by the following
equation:
h dV-S F - (We + Ws) Vs (7)
3
II
THE BDM CORPORATION
102;RH=85%
JASIN20 min u ave
20
o 1 *13
4 10°L.L.
10 0- x7
h3
1 01 0, , ,.,,111 I I ,1 10.1 1 10
Radius (pm)Fi.,ire 1 Ensemble averacle total aerosolol lul;ie spectra from JASIN.
The number to the rigjht of the spectrum is the wind speedcatejory in weter/second.
I THE BDM CORPORATION
where h is the height of the boundary layer, W the entrairment velocityeand 1W is the Stokes velocity:S
W 1.57 x 10- ( + 12 0.81 3 2 2 (8)
s (I+10. (-)-T) )r g (s)()
where r is in pm and Ws is cm/sec. In this equation we have left out the
cloud formation removal mechanism because it acts on a much longer time
scale.
C. FLUX SPECTRUM ANALYSIS
The CEWCOM-73 data have been used to evaluate the surface flux term
of Equation 7. Since the other terms of the expression were measured, one
simply calculates Fs using a rearrangement of the terms.
A period was chosen from CEWCOP.1-78 where all data were available,
the synoptic conditions were fairly stable, the wind speed was reasonably
constant and a good mixed layer was present (Table 1). There was a slowly
decreasing continental influence indicative of steady NW winds (A' in Table
I and Figure 2). Subsidence was assumed to be negligible so that We = dh/dt.
Thus, the final expression used is:
Fs = h + (h -W ) V (9)dt (d--W v
The scatter and uncertainty of the aerosol and mixed layer depth mea-
surements introduces certain trade-offs between averaging times and statisti-
cal validity. For short averaging times, the random scatter of the
individual terms is considerable leading to large variations in the time
derivative terms. By increasing the averaging time to four hours, the FIs "signal-to-noise" becomes more reasonable. A time series of the individual
termms of Equation 9 is shown in Figure 3 for two different particle sizes.
The average surface production flux spectrum is shown in Figure 4. This
flux spectrum applies to [he average surface conditions for the entire 20
hour period (average wind speed was about 3 m/sec).
5
44 - - :7D === 4 - 4 - 4 4
THE BDM CORPORATION LaI-. U. wWL aW
~ ~ - Co U~.4>.2 co co Ul N 0O r~l le, IT
C5,=4.- -4 1- Il.4~i C: . ~ .- -~- -4
=: =44I= 1-D =-
-14 W ........ lj L W L U
c:- == = : =4 = = CD= 4====44 = 4
--C - - - - - - - -
~4- 0 rUS- LiL'L C-ULL U'' .1i u
v~ C.J c:, CD a4i C=j *:
r u + .++ o 4 ++ +4.4
4o F.- - - -'- 4 --0i
.C. =4= =4
0) . r -i -n -i-
ujC UJ j Ij ;CC), Z; t 0 o0
42 t14 c:,= = c
44 4- '-f4, -
V . . ..i -4 -44 4= = -4~ -4 - -4 - -. - - -.
L) L in n jA -- Lr. 6
ITHE BDM CORPORATION
CEWCOM-78
40- ~- RADON 2
a---- Aerosol
30- 15EGC-
z0
D20- 10.
101
0~~~~ 1 II-
11~~ 12 1 4 1 8 1 02
10Y (MY,198Figure 2. Tinie seres of atospheric 22 Rn ac iiy(oi ie asn
Kasmi an Bessn,197) ndconinnta arosl oefice0
(dashd lin ),SA-
~ I / / -S7
THE 8DM CORPORATION
10- CEWCOM-78
8 r=2pim r=l5pm S-v
7D C .
6--05~
,a
C, aC)E r -
~0
U, -
ExC -2 - L
.4- i0 ) 0
RH=80% (aC)(on 4-
a a
-6-1 wi-vU-
-8- dh/dt V
-10- o h dV/dtL
16 0 8 16 0 8Time (hrs)8
ITH[ 8DM CORPORATIONI'I
10' RH=80% CEWCOM-78
8.0 rn/s
E
100
li , , I I I , I i I ,
1 10Radius (pm)
IFi gure 4. The ensemble average surface flux spectru. F (r) for the C,,- .
analysis period (average wind speed, 8 m/sec).
II9I
THE BDM CORPORATION
The relative Contributions of entrainment, graviational fallout and
sur~tic pruduction .iW nilly illustrated by defining an equivalent surface
prJu1ULtiun vertical eloity, W F s/V so that in equilibium (dVs/dt = 0)r 5 5
I,4 F ' ,. U + t (10)
For, the avergje conditions found in the analysis period, the entrain-
r.ent and Stokes terms were roughly equal at a particle radius of 5 ,m
(Fi ,re 5).
U. FLUX: A',r .... l". ....~ FEF-L ,' , ,Z: ,! .. ... ',4 1 i D ', PF ED
:-J .... I the roudsonably constant wind speeds during the 20 hour
,G,".,-7., pr i , ,-twre aule to improve the statistical certainty for the
Fs Icu1iti,: L cbining all the data from the period, assuming that the
t1,l ,qoul; be a r|ea-,onable representation for U = 8 m/sec wind speed. Thus,
.H.OW fldw ?Idv tha ,_ surface flux spectrum at a single wind speed. In order to
ucl il it. Lhu fl-ax it other wind speeds, we note that the right hand side of
L;i,ition 10 is ,uirly independent of wind speed for equilibrium conditions.
Sr~ref ore, the flux at one wind speed can be r' it,.d to the flux at other
,.-iri speeds it ttoe equilibrium volume spectra are known.
Vs ( U)
F ( ' ) : F ( ) V 1s {U 2) (21)
We have available from JASI'N a large set of ensemble averages of
aerosol volume spectra at different wind speeds (Figure 6 and Table I).
It is a simple matter to apply this data to Equation 11, using the
CEW'CUM-74 aerosol flux and equil ibrium spectrum to generate the surface
volur,e flux spectra as a function of wind speed (Figure 7 and Table III).
10
THE BDM CORPORATION
101C EWCOM -78
X-- WF
EU
~i 0
UX
01
Radius (pm)
Figure 5. Enseii;ble average contrib~tion of entrainment (W ),Stokes fallout(WS) and surface flux W F F s /V s assuming a stste of dynamiccq i Ii br ium.
THE BDM CORPORATION
102RH=87% JASIN
12 hr uave
E
x 9
101 1 10
Radius (pm)Figure G. Ensemble average equilibrium sea salt volume spectra from JASIN.
The number to the right of the spectrum is the wind speed categoryin meters/second.
12
ITHE BDM CORPORATION
TABLE 11. Equilibrium sea-salt aerosol-spectrain Pm2/cm3 as a function of radius (Gm)and wind speed (misec).
r- 0.8 2 5 10 15
5 2.8 3.4 3 0.3 0.05
7 10 10 5 1 0.35
9 18 25 13 6.4 4.2
11 20 30 20 15 15
13 24 33 22 22 22
16, 28 35 28 28 28
8* 9 11 2 0.2 0.4
*CEWCOM-78 Analysis Period
11
THE BDM CORPORATION
RH=80% 16
713102/1
9
E 10
.-
(A V
x~cn
*,~ C) ,.
14J
C,- .-
Ci
1000Radius w Im
]4)
L1
THE BDM CORPORATION
1
TABLE III. Surface sea-salt aerosol flux,F (r), (pm2/cm n/sec) as a function ofparticle radius (pm) and wind speedI (m/sec)
0.8 2 5 10 15
5 1.3 1.1 2.5 1.0 0.33
7 4.5 3.1 4.2 3.3 2.3
9 8.2 7.7 11 21 27
11 9.1 9.2 17 49 48
13 11 10 19 72 140
16 17 11 24 92 180
3* 4.1 3.4 1.7 2.3 2.6
* CEWCOM-78 Analysis Period
1II
II 15
' Ii
THE BDM CORPORATION
E. NON-EQUILIBRIUM
For time periods of a few hours, the aerosol spectrum may not be in astate of dynamic equilibrium. If we rewrite Equation 7 in the following
form:
dVs + I - (12)dt T h
where the time constant, Tp, is
Tp = h/(We + Ws) (13)
then we see the analogy of aerosols and a capacitor charged by an applied"voltage," Fs, through a "resistance," h. In this analogy, the "capacitance,"
is (We + Ws) - .The response time of the aerosol density is a strong function of
particle radius because Ws - r2 . Values of Tp for S = 0.8 and h = 400 m
are given in Table IV.
TABLE IV. The equilibrium time constant for aerosolsat S = 0.8, h = 400 in and We = 0.4 cm/sec.
0.5 1 5 10 15
Tp, hours 23 22 11 3.5 0.5
The boundary layer mixing time, TIi, which represents the time required
for changes in aerosol density to be evenly distributed throughout the
marine layer is:
T= h/W (14)
16
I THE BDM CORPORATION
For the analysis period (h = 400 m and W * 0.6 m/sec) we find Tm = 0.16
hours. Therefore, short term variations in mixing volume (dh/dt) will
is much slower than the mixing time. Thus, for time periods on the order
of one hour, changes in the aerosol density (h dVs/dt) will be highly cor-
related with the mixing volume term (Vs dh/dt).
This effect, which is particularly noticeable for smaller particles
(see Figure 3 and Table III) is shown in Figure 8.
17
Ii
THE BDM CORPORATION
1.5- CEWCOM-78o r =.8 pm
1.0 x r=214m 0
00 5 x- 0.5
E
-10-0.5-0
-1 .0 -x x
-1.5I II I
-1.5 -1.0 -0.5 0 0.5 1.0 1.5
-dh/dt (cm/sec)
Figure 3. Changes in aerosol spectral density (dV /dt) vers us changes inmixing volume (-dh/dts). This graph ilustrates the dominanceof these terms in Equation 9 for short time periods.
18
THE BDM CORPORATION
INITIAL DISTRIBUTION LIST
DODAAD Code No. of Copies
1. Defense Documentation Center S47031 12Cameron Station, Building 5Alexandria, Virginia 22314
2. Library, Code 0212 2Naval Postgraduate SchoolMonterey, California 93940
3. Dean of Research, Code 012 N62271Naval Postgraduate SchoolMonterey, California 93940
4. C. W. Fairall, The BDM Corporation 5Naval Postgraduate SchoolMonterey, California 93940
5. Assoc. Professor K. L. DavidsonCode 63DsNaval Postgraduate SchoolMonterey, California 93940
6. Professor G. SchacherCode 61sqNaval Postgraduate SchoolMonterey, California 93940
7. Director, Naval Research Laboratory N00173 6Attn: Code 2627Washington, D.C. 20375
8. Office of Naval Research N62879Branch Office1030 East Green StreetPasadena, California 91106
9. Mr. Robin Simpson, ONR RepresentativeStanford University, Room 165Durond Aeronautics BuildingStanford, California 94305
I
II 19