This manuscript has been reproduced from the microfilm master. UMI films the
text directly from the original or copy submitted.
Thus, som e thesis and
dissertation copies are in typewriter face, while others may be from any type of
computer printer.
The quality o f this reproduction is dependent upon the quality o f the copy
submitted. Broken or indistinct print, colored or poor quality illustrations and
photographs, print bleedthrough, substandard margins, and improper alignment
can adversely affect reproduction.
In the unlikely event that the author did not send UMI a complete manuscript and
there are missing pages, these will be noted. Also, if unauthorized copyright
material had to be removed, a note will indicate the deletion.
Oversize materials (e.g., maps, drawings, charts) are reproduced by sectioning
the original, beginning at the upper left-hand comer and continuing from left to
right in equal sections with small overlaps. Each original is also photographed in
one exposure and is included in reduced form at the back of the book.
Photographs included in the original manuscript have been reproduced
xerographically in this copy. Higher quality 6” x 9” black and white photographic
prints are available for any photographs or illustrations appearing in this copy for
an additional charge. Contact UMI directly to order.
UMI
Bell & Howell Information and Learning
300 North Zeeb Road, Ann Arbor, Ml 48106-1346 USA
800-521-0600
by
K o n stan tin Zahariev
B.Sc., U niversity of Sofia, 1992
A D issertation S u b m itted in P a rtia l Fulfillment of th e Requirem ents for th e Degree of
D O C TO R O F PH ILO SO PH Y in the School of E a rth an d Ocean Sciences
We accept this d issertation as conforming to th e required standard
Dr. Chris G arrett
Supervisor (School of E arth and O cean Sciences)
D r. Ken Denman
Depart^ rental M em ber (School of E a rth and Ocean Sciences)
Dr. Andrew Weaver
D epartm ental M ember (School of E a rth and Ocean Sciences)
D r. N orm M cFarlane
O utside M ember (C anadian C entre for C lim ate Modelling and Analysis)
D r. Eric D ’Asaro
E x tern al Exam iner (University of W ashington)
© K o n stan tin Zahariev, 1998 U niversity of V ictoria
All rights reserved. This dissertation m ay not be reproduced in whole or in p a rt, by photocopying or other m eans, w ithout th e permission of th e author.
Il
A b s tr a c t
T h e oceanic surface boundary layer is of great im portance and interest as
its dynam ics provides for the exchange of energy, m om entum , heat an d m a tte r be
tw een th e atm osphere and the ocean. It is crucial to have a thorough understanding
of physical processes th a t m ight have a significant influence on its properties zmd
variability. In this stu d y 1 consider several different facets of m ixed lay er/b o u n d ary
layer dynam ics.
One aspects concerns th e consequences of th e nonlinearity of th e equation
of s ta te in m ixed layer models. T he n onlinearity of the equation of s ta te gives rise to a te rm in th e averaged surface buoyancy flux which can be com parable in
m agnitude to o th er term s. Its m agnitude is shown to be proportional to the area
enclosed by th e seasonal cycle of sea-surface te m p e ra tu re T versus th e oceanic heat
content 1-L. T h e te rm always represents a buoyancy in p u t into the ocean and is
com pensated exactly by the buoyancy loss via cabbeling (densification on m ixing)
whenever th e m ixed layer entrains w ater w ith different properties from below.
A nother problem of interest is th e role of th e coherent w ind-induced vor
tices, com m only known as Langmuir circulation, in generating th e surface m ixed
layer. A sim ple param eterization of th e m ixing due to Langm uir circulation is
exam ined in the light of an oceanic d ataset. Some evidence for th e validity of th e
p aram eterizatio n is found, thus draw ing a tte n tio n to Langm uir’s assertion th a t
Langm uir circulation is one of the key physical processes in th e oceanic boundary layer.
T h e th ird aspect of surface boundary layer dynam ics explored is th e m ean
effect on m ixed layer entrain m en t of periodic vertical m ovem ent of isopycnals in
th e therm o d in e due to non-breaking in ternal waves (referred to as heaving). Sea
sonal m odel runs incorporating idealized heaving show th a t heaving can produce
significant seasonal differences in sea-surface te m p e ra tu re com pared to a reference
case w ithout heaving. It is inferred th a t by periodically stretch in g and com pressing
th e m ixed layer, heaving causes instabilities th a t result in additional en train m en t
of colder w ater from below. A heaving num ber is proposed, and two param e-
terizations of heaving for use in m ixed layer models are suggested.
Dr. Chris G a rre tt
Supervisor (School of E a rth and O cean Sciences)
Dr. Ken D en m an
D epart^ rental M em ber (School of E a rth and Ocean Sciences)
____________________________________________________ Dr. A ndrew Weaver
D ep artm en tal M em ber (School of E a rth and Ocean Sciences)
_________________________________________ Dr. N orm M cFarlane
O utside M em ber (C anadian C entre for C lim ate M odelling and Analysis)
Dr. Eric D ’A saro
IV
T able o f C o n ten ts
List of T a b l e s ... vii
List of Figures ... viii
A cknow ledgm ents... xv
D ed ic a tio n ... xvi
C hapter 1. I n tr o d u c tio n ... 1
1.1 M otivation and o u t l i n e ... 1
1.2 T h e oceanic surface boundaxy l a y e r ... 3
1.2.1 O v erv iew ... 3 1.2.2 F o rc in g ... 4 1.2.3 P ro c e sse s... 9 1.3 O b s e rv a tio n s ... 19 1.3.1 OWS P a p a ... 19 1.3.2 LOTUS ... 20 1.3.3 M ILE ... 20 1.3.4 P A T C H E X ... 21 1.3.5 GATE, phase I I I ... 21
C hapter 2. M ixed layer m o d e l s ... 23
2.2 T h e Price-W eller-Pinkel m o d e l... 26
2.3 T he Mellor-YcLmada Level 2 m o d e l... 31
2.4 O th er m o d e l s ... 33
C h ap ter 3. Noolinear e q u atio n of state and ap p aren t buoyancy flux . . . . 39
3.1 I n tr o d u c tio n ... 39
3.2 T h e o r y ... 41
3.2.1 R ep resen tatio n of th e term involving clQ ... 41
3.2.2 G eneral properties of th e seasonal cycle of versus "K 47 3.3 Seasonal cycle m odelling - a te st c a s e ... 50
3.4 C alculation of th e NES term from OWS P a p a d a t a ... 55
3.5 Discussion a n d c o n c lu s io n s ... 57
C h ap ter 4. Langm uir circulation and m ixed layer d e e p e n i n g ... 61
4.1 I n tr o d u c tio n ... 61
4.2 The LC criterion for ML d e e p e n i n g ... 63
4.3 Verification of th e LC criterion using LO TU S d a t a ... 66
4.4 Im plications for u p p er ocean r e s t r a t i f i c a t i o n ... 78
C h ap ter 5. Isopycnal heaving and its effect on en train m en t ... SO 5.1 I n tr o d u c tio n ... 80
5.2 Models an d sensitivities to th e background diffusivity . . . 81
5.2.1 A relationship betw een background diffusion and en tra in m en t 83
VI
5.3 Effects of isopycnai heaving on e n tra inm e n t ... 91
5.3.1 A model w ith isopycnal h e a v i n g ... 91
5.3.2 Seasonal runs w ith different heaving param eters . . . . 98 5.3.3 A detailed look a t a three-day period in th e su m m er . . 104
5.3.4 Heaving n u m b e r ... 110
5.4 P aram eterizing heaving for use in m ixed layer m o d e l s ... 115
5.4.1 Heaving as an equivalent 115
5.4.2 Heaving as a correction to th e dynam ic sta b ility criterion 119
5.5 S um m ary and conclusions ... 124
C h ap ter 6. O u t l o o k ... 126
VIII
L ist o f F igu res
1.1 Some of the m echanisms affecting the turbulence in the oceanic su r
face boundary layer: Langmuir circulation, breaking surface waves,
shear-induced turbulence, convection, breaking internal waves (re
drawn from T horpe (1985))... 10
1.2 A schem atic of th e nonlinear dependence of density of seaw ater on
tem perature. W arm w ater expands more th a n cool w ater contracts. If water w ith tem p eratu re T l, density Z)l, mixes with w ater w ith
tem perature T2 > T l , density D'2 < D l, in equal volumes, th e m ixture, while of tem p eratu re T m = { T l + T 2 ) / 2 ^ will have density
D { T m ) larger th an th e m ean density D m = ( D l + D 2 ) / 2 ... 12
1.3 A schematic representation of Langmuir circulation (redraw n from
Pollard (1977))... 14 1.4 Ten-day tim e series of isotherm s versus dep th an d tim e, taken from
th e LOTUS d ataset (Briscoe and Weller 1984; Bowers et al. 1986). The high-frequency variability in the isotherm s’ position in an d
below the seasonal therm ocline at 20 m etres d ep th is indicative of
3.1 A schem atic of m ixed layer deepening due to m echanical stirrin g an d associated entrainm ent. A well m ixed surface layer of d ep th
h, te m p e ra tu re T , and a te m p e ra tu re ju m p A T across its base
deepens, in tim e to a depth h + 6h, lowering its te m p e ra tu re by
6 T to T - S T ... 44
3.2 T h e annual cycle of versus 'H a t OW S Echo (35°N, 48°W ),
m odified from Gill an d Turner (1976)... 48
3.3 Idealized forcing w ith zero m ean an n u al heat flux. Top: insolation
Qg, to ta l heat loss (tbe sum of sensible, la te n t, an d longwave),
an d net h eat flux Q. Bottom : eastw ard wind stress ... 51 3.4 T h e annual cycle of T^ versus Ti from m odel o u tp u t, if th e NES is
excluded (upper solid fines), an d including it (dashed fines). T h e
pairs of fines for each case delineate th e diurnal ranges. T h e lower curve shows th e loop, with th e NES b u t when th e diu rn al cycle is
not resolved. T h e m inim um annual h e a t content is a free p a ra m e ter which does not en ter the calculations of th e NES term , an d is set
to zero. Each circle corresponds to th e m iddle of a m o n th ... 53
3.5 (top) T he daily range (i.e. daily m inim um and daily m ax im u m
values) of m ixed-layer depth h if th e NES is excluded (solid fines) an d including it (dashed fines); h is slightly larger on average in th e
NES case, (b o tto m ) T he daily range of sea-surface te m p e ra tu re T^
if th e NES is excluded. Its inclusion results in slightly lower average
3.6 T h e annual cycle of versus H for th e year 1972 from OWS P a p a
d ata. T he m inim um annual h e at co n ten t is set to zero. Each circle
corresponds to th e middle of a m o n th ... 58
4.1 A schem atic diagram showing th e interactio n betw een Langm uir circulation and pre-existing stratifica tio n . I: the general problem , II: startin g from a linear stratifica tio n , and III: sta rtin g from a two-layer stratification. Redrawn from LG97... 64
4.2 Time-series of th e difference betw een th e tem p eratu re at a d e p th of 0.6 m (considered a sea-surface tem p eratu re), and the te m p e r a tu re a t a d e p th of 5 m etres. T h e surface tem p eratu re is consis ten tly lower by about 0.06°C even a t night when the u pper ocean appeared well m ixed... 68
4.3 Top (right) panel: an exam ple of tem p e ra tu re contours an d cor responding estim ates of h (u p p er thick line) and (lower thick line); B ottom (left) panel: Forcing (wind speed and insolation) for th e same period... 69
4.4 Com parison of th e heat content estim a te s for the upper 51 m etres of ocean as derived from tem p eratu re profiles (shaded) and as im plied from the local surface heat fluxes (th ick black)... 71
4.5 T h e caption is on page 7 0 ... 74
4.6 T h e caption is on page 7 0 ... 75
4.7 T h e caption is on page 7 0 ... 76
5.1 T em p eratu re profiles after 50-day runs with, th e MY-2 m odel for different background diffusivity values, and for forcing w ith zero
m ean to ta l heat flux, constant eastw ard w ind stress of 0.1 P a. From
right to left a t th e surface: initial profile; 50th day final profiles for
ATyg = 0; 1 X 10 ^;5 X 10 ^... 87
5.2 Idealized forcing w ith zero m ean annual heat flux. Top: insolation
Q^, to ta l heat loss (th e sum of sensible, la te n t, and longwave),
and to ta l heat flux Çt- B ottom : eastw ard wind stress ... 89
5.3 Seasonal runs w ith MY-2 m odel for two different background diffu sivity values: lxlO ~^m ^s~^ (m iddle), and 5 X 10~“m^s~^ (b ottom ).
Com pare th e m axim um seasonal SST at th e end of sum m er with
th e reference (/f^g == 0 (top) case)... 90 5.4 Ten-day tim e series isopycnal plot of dep th versus tim e, from the
LOTUS d ataset. Note th e coherent heaving movement of isopyc
nals regardless of depth. Also shown is the estim ated m ixed layer
d e p th ... 92
5.5 Seasonal runs w ith the P W P m odel for three different background
diffusivity values: 0.7 x 10~^m^s~^ (1), 1 x 10 (2), and 1.5 X 10~^m^s~^ (3). Also shown for com parison is the reference
(ATbg = 0 (top) case)... 95 5.6 O ne-day (year-day 202) modified P W P m odel o u tp u t showing heav
X II
5.7 Seasonal runs w ith a modified P W P model for two different am plitudes of isopycnal heaving: = 2.0m (1), an d = 5.0m (2),
w ith a fixed wave period of 1.6 h. Those are co m p ared to the
reference (top) no-heaving case... 100
5.8 Mixed layer d ep th from a modified P W P m odel fo r two differ
ent am plitudes of isopycnal heaving: A^ = 2.0m (to p panel), and A^ = 5.0m (b o tto m panel), ivith a fixed wave p erio d of 1.6 h.
Those are com pared to th e m ixed layer dep th (daily range) for the
reference case... 102 5.9 Seasonal runs w ith a modified P W P model for th re e different pe
riods of isopycnal heaving: = 2.4 h (1), — 1.2 h (2), and = 0.4 h (3), w ith a fixed wave am plitude of A^. o f 2.0m . Those
are com pared to th e reference (top) no-heaving case... 103
5.10 Mixed layer d e p th from a modified PW P model for two different periods of isopycnal heaving: = 1.2 h (top p an el), an d =
0.4 h (b o tto m panel), w ith a fixed wave am p litu d e of 2.0 m . Those are com pared to the m ixed layer dep th (daily range) for the
reference case... 105
5.11 Contour plots of d ep th versus tim e, for days 200-202. Top panel:
m ean (over a heaving period) tem perature contours for th e refer ence case (no heaving). B ottom panel: m ean te m p e ra tu re contours
for a heaving case w ith an am plitude of 2 m etres a n d a period of
5.12 C ontour plot of depth versus tim e, for days 200-202, of th e m ean te m p e ra tu re difference (over a heaving period) betw een th e refer
ence case (no heaving) and a heaving case w ith an a m p litu d e of
2 m etres and a period of 1.6 hours. T h e plot shows th e u p p e r 20 m etres of m odel ocean. T he m ixed layer depth for th e reference
case is shown in w hite... 108
5.13 Seasonal runs w ith a modified P W P m odel for a heaving w ith =
2.0 m and — 1.6 A: original ru n with continuous heaving (1)
a n d a run w ith heaving sw itched off from 8:00am to 4:00pm each
day (2). T h e no-heaving case is also shown for reference... I l l
5.14 Values of for pairs of d ata-p o in ts with the sam e heaving n um ber R jj b u t different nondim ensional frequency Tq/ T ^ . P airs
of points are denoted by the sam e symbol. Pairs: “o” , = 1.64:
R ^ = 2.22; “A ” , R ^ = 2.82; R ^ = 4.47; R ^ = 7.69. 114
5.15 D ependence of th e scaled m axim um seasonal SST difference
th e heaving num ber R ^ (top panel), an d on th e scaled background diffusivity R ^ (bottom panel), w ith corresponding in terp o lated val
ues (dashed lines; in the top panel th e dashed line indicates a crude
piece-wise linear interpolation)... 116 5.16 Correspondence between th e scaled background diffusivity R ^ an d
th e heaving num ber Ry^ (top p an el), and equivalent background diffusivity (in m^s~^) as a function of R ^ (b o tto m p an el),
w ith corresponding interpolated values (dashed lines, in d icatin g a
XIV
5.17 D ependence of th e scaled m axim um seasonal SST difference on
th e critical Richardson num ber R b ^ (top panel), an d a correspon dence betw een th e stability ad ju stm en t factor a n d th e heaving
num ber (b o tto m panel), w ith corresponding in te rp o la te d val ues (dashed lines; in th e bo tto m panel the dashed line indicates a
crude piece-wise linear in terp o latio n )... 121 5.18 A com parison betw een seasonal H — curves for a heaving ru n
(black line) w ith = 3.0 and a diffusivity run w ith an equivalent
background diffusivity of = 0.8 x 10~^m^s~^ (top panel). T h e
b o tto m panel shows a com parison betw een th e sam e heaving run (black line) an d an adjusted critical R ichardson n u m b er ru n w ith
A ck n o w led g m en ts
I would like to thank my supervisor Dr. Chris G a rre tt for his guidance an d
generous su p p o rt. I also th ank m y external exam iner, D r. Eric D’Asaro, an d th e
m em bers of m y com m ittee, Dr. Ken D enm an, Dr. Andrew Weaver, and Dr. N orm M cFarlane, for th e ir support and for valuable com m ents and suggestions. I th a n k
m y fam ily for th e ir love and su pport. I would also like to th a n k the U niversity of
X V I
D e d ic a tio n
C h a p t e r 1
I n tr o d u c tio n
1 .1 M o t iv a t io n a n d o u t lin e
O ceanic an d atm ospheric variability fascinates w ith its com plexity on any tem p o ral and sp atial scale. T he oceanic surface boundary layer^ is of p artic u la r
im p o rtan ce an d in terest, as its dynam ics provides for the exchange of energy, m o m en tu m , h eat and m a tte r betw een th e atm osphere and th e ocean. Its seasonal
variability controls th e exchange of heat w ith the therm ocline. T h e properties an d dynam ics of th e m ixed layer influence chem ical and biological processes in th e
ocean: alm ost all biological a ctiv ity takes place in the u p p er layer of th e ocean.
M ixed layer dynam ics governs th e sea-surface tem p eratu re a n d th u s th e fluxes of
h eat, m oisture, an d gases to th e atm osphere; it is one of th e m o st im p o rta n t factors d eterm in in g clim ate.
Consequently, it is crucial to have a thorough u n d ersta n d in g of physical
processes th a t m ight have a signiflcant influence on the pro p erties an d variability
of th e surface boun d ary layer. In th is stu d y I consider several different facets of m ixed la y e r/b o u n d a ry layer dynam ics.
^The term s “oceanic surface boundary layer” , “surface boundary layer” , “surface m ixed layer” , and “m ixed layer” m ay be used interchangeably in tliis study. T h ey denote the su rface layer o f water o f quasi uniform tem perature, salinity, and buoyan cy/d en sity, but m ay also include th e underlying therm ocline d epending on co n tex t. In the la tter case, the term “boundary layer” is usually preferred.
1. In troduction
O ne of th e aspects I look at pertains to th e consequences of th e nonlinearity
of th e eq u atio n of s ta te in m ixed layer models. As noted by G arrett et al. (1993) for th e M editerraneaji Sea, th e nonlinearity of th e eq u atio n of state gives rise to an
additional te rm in the averaged surface buoyancy flux which can be com parable in m agnitude to th e o th er term s. T he term always represents a buoyancy input into
the ocean even when th e an n u al mean heat flux is zero, so a com pensating buoyancy
loss over th e seasonal cycle m ust occur, via cabbeling (densification on m ixing) whenever th e m ixed layer entrains water w ith different properties from below. A
sim ple representation of th e m agnitude of th is te rm is derived, relating it to th e the seasonal cycle of sea-surface tem perature versus th e oceanic heat content
and general properties of th e seasonal cycle of versus H are discussed^.
A n o th er problem in this area of interest is th e role of the coherent wind-
induced vortices, com m only known as Langm uir circulation, in establishing and
m aintaining th e surface m ixed layer. Following up work by Li and G arrett (1997) on a sim ple param eterizatio n of th e mixing due to L angm uir circulation from which
a Langm uir circulation criterion was derived for use in bulk m ixed layer models, 1 collaborated w ith the authors to verify th eir form ulation by directly exam ining an oceanic d a ta set, looking for m ixed layer deepening events th a t did not appear to be
caused by shear instability a t th e base of th e m ixed layer b u t were predicted w ith the new criterion. Im plications for seasonal an d d a y tim e ocean restratification are
also discussed".
^This reseajTch was published as Zahariev and Garrett (1997). ^This research was published as Li, Zahariev, and G arrett (1 9 9 5 ).
T h e th ird aspect of surface boundary layer dynam ics I explore is th e mean
effect on m ixed layer entrainm ent of periodic isopycnal up-down m ovem ent in the
therm ocline due to non-breaking internal waves (henceforth referred to as “heav ing” ). M odel o u tp u t, such as sea-surface tem perature, from one-dim ensional mixed
layer m odels shows th a t models are sensitive to th e vaiue of background diffusiv
ity iifyg used below th e mixed layer, in the therm ocline. It is thus im p o rtan t to estabhsh w h eth er represents m ixing processes in th e therm ocline or is a proxy
for processes o m itted from the models. One such process is isopycnal heaving. I
exam ine th e sensitivity of the m ean entrainm ent ra te to th e heaving a t th e base of the m ixed layer by considering a modification of th e P rice et al. (1986) m ixed layer model th a t allows for periodic isopycnal displacem ent superim posed on th e other
variabihty. Results from seasonal runs w ith this m odel and idealized forcing are
presented, along w ith estim ates of the equivalent background diffusivity. A heav
ing num ber to characterize heaving is proposed, and two p aram eterizations of
heaving for use in m ixed layer models are suggested.
1.2 T h e o c e a n ic su r fa c e b o u n d a r y la y er
1 .2 .1 O v e r v ie w
T h e ocean typically m anifests a surface boundary layer in which turbulent mixing driven from above by surface ffuxes of buoyancy and m o m en tu m and from below by shear a t the base of the layer creates a quasi-uniform stratification within
it. Thus it is conventionally referred to as a “m ixed layer” , although observations frequently reveal th e existence of velocity shear w ithin th e layer (e.g. Davis et
1. Introduction
al. (1981a), W eller and P lu eddem an n (1996)) indicating im perfect m ixing a n d /o r
the presence of organized stru c tu re s (e.g. Sm ith et al. (1987)).
T hroughout th e course of a year, the mixed layer executes a ch aracteristic seaaonal cycle, varying its thickness in response to th e forcing. In th e w inter, it
deepens because of h eat loss a t th e surface; in the spring progressively shallower
m ixed layers are form ed u n d er conditions of net ocean h eatin g u n til a new seasonal
therm ocline is form ed for th e year. O n top of this variability, each d iurnal period a shallow m ixed layer m ay be form ed in th e daytim e because o f stro n g p e n e tra tin g
solar heating, deepening at night back to its seasonal d ep th . T h e d e p th of m ixing can also vary w ith geographical location. Especially in th e w in ter, it could be as
little as 50 m etres in th e subtropics and as much as 500 m etres or m ore in th e northern N orth A tlan tic ocean.
This stu d y focuses on th e m id -latitu d e mixed layer, aw ay from eq u ato rial
and polar regions. M oreover, it considers open ocean conditions away from late ra l boundaries and horizontal variability, so th a t a one-dim ensional (1-D) m odel in
the vertical of local forcing and response can be adequately applied. In addition, salinity/freshw ater variability is not considered.
1 .2 .2 F o r c in g
The oceanic surface bo u n d ary layer is forced by fluxes o f h eat, m om entum ,
and freshwater.
The u p p er ocean gains h eat by absorption of solar rad ia tio n in th e daytim e,
and (norm ally) loses h e a t from th e surface due to tu rb u le n t tra n sfe r via sensible and latent heat h e a t fluxes, and due to long-wave radiation.
T h e sensible heat flux
Q h ~ P a P p a ^ ^ ’ (1-1)
where are air density a n d specific heat, respectively, results fro m a tu rb u
lent tran sfer of heat to /fro m th e atm osphere, depending on th e sign o f th e surface te m p e ra tu re difference betw een th e atm osphere and the ocean. Since th e air tu rb u
lence n ear th e surface is largely caused by shear, th e rate of sensible h e a t transfer
depends also on wind speed. T hus, on dim ensional grounds, is p aram eterized
by a “bu lk ” (i.e. involving averaged quantities) form ula (S m ith 1980; Large and Pond 1982; Sm ith 1988)
0 ^ = / > . c , . c „ i 7 „ ( r . - r j , (1.2)
w here is a dimensionless coefficient, the wind speed a t 10 m e tre s height, th e sea-surface tem p eratu re, th e 10-meter air tem p eratu re. U n d er m oderate
w ind conditions with wind speeds of 5 — 25 ms ~^ , which often prevail over the ocean, th e atm ospheric b o undary layer is nearly neutrally stratified . Different
values/ form ulae for the n eu tral have been suggested; e.g. S m ith et al. (1980)
suggests
lO^Cjj = <
0.83 for stab le conditions (T^ < Z)^),
(1.3)
1. Introduction.
Typical m ean values of Qf^ are in th e range 10 — 20 W m ~ ^ (N urser 1996), and could co n stitu te h e a t loss or heat gain by the ocean depending on th e sign of th e
air-sea te m p e ra tu re difference. T he la te n t h eat flux
Q^ = L y w q , (1.4)
where L y is th e la te n t heat of vaporization and q is hum idity, results from evapo
ratio n of w ater, ta k in g heat away from th e ocean. A gain on dim ensional grounds
th e evaporation r a te is param eterized as
E = P ^ c^U ^o {q^ -qJ , (1.5)
where Cg is a dim ensionless coefflcient, q^ is the satu ra tio n value of th e spe
cific hum idity a t te m p e ratu re T^, q^ is th e specific h u m id ity of air at 10 m etres.
S m ith (1980) suggests for th e n eu tral Cg
lO^Cg = 1.5 (1.6)
b u t later revises it to
(Smith. 1988). T he laten t heat of vaporization is
L y = 2.5008 X 10® - 2.3 x 10®T [Jkg (1.8)
where T is the w ater te m p e ra tu re (G ill 1982). Thus th e la te n t heat flux is
Q e ~ L y E = LyP^C^U^Q^q^ — q ^ ). (1-9)
A typical mean value of is 100 W m ^ or more (N urser 1996), and co n stitu tes heat loss by the ocean.
T he long-wave rad iatio n h eat flux is the result of em ission of infrared rad ia
tion by th e ocean. It is p aram eterized as a flux em itted by a black body, corrected for th e departure of th e ocean from a black body behaviour, for atm osphere mois
ture, an d for cloudiness. An exam ple is
= eo -r/(0 .3 9 - 0 .0 5 e / / ') ( l - 0 . 6 n /) , (1.10)
where e = 0.985 is th e e stim ated ocean emissivity, <j is S tefan ’s co n stan t, is th e
10-metre vapour pressure, is th e fraction of sky covered by clouds (G ill 1982). T ypical m ean values of in th e range 30 — 70 W m ~ ^ , and always represent
heat loss by the ocean.
Solar radiation is a source of heat for the ocean. T h e incom ing solar radia tion can be estim ated by em pirical form ulae involving th e solar co n stan t an d cor
1. In tro duction
Dobson an d S m ith 1988). F urtherm ore, th e insolation is absorbed w ith d e p th de
pen d in g on w ater clarity, w ith th e red sp ectral com ponents absorbed w ith in th e
u p p er 1 m etre of ocean, and the rest p e n e tra tin g deeper. P aram eterizations of th is ab so rp tio n profile involve an exponential d e p th dependence. This can be a single
ex ponential fit (Ivanoff 1977), th e m ore com m only used 2-exponent fit (P au lso n
an d Sim pson 1977), 3-exponent fit (W oods et a/.1984) or a fit w ith even m ore
ex p onents (Sim pson and Dickey 1981; W oods 1980). Using m ore exponents w ould typ ically im prove the accuracy of th e fit n ear th e surface, in the top 5 m e tre s of
ocean.
T h e ocean surface gains m o m en tu m by tu rb u len t transfer from th e a tm o sphere v ia th e w ind stress
r = —p j u u , (1.11)
v irtu a lly all of which goes into driving th e m ean current (R ichm an and G a rre tt 1977), im p a rtin g a “friction velocity”
“ ■
=O
w here is th e density of w ater. S ta rtin g w ith Taylor (1916), the drag of th e
where is a dim ensionless drag coefficient dependent on wind speed a n d on
atm ospheric stability. Different form ulae for the n eutral c^ have been suggested, e.g. S m ith (1980) suggests
10 c^ = 0.61 +0.063C/^jj for 6m s < < 22ms (1.14)
which is very close to a m ore recent form ula by Yell and and Taylor (1996). A typical w ind stress value is 0.1 N m ^ for of 8 — 9 m s
T h e u p p er ocean is also forced by a freshw ater flux, w ith rain an d ice m elting
being a freshw ater source, an d evaporation and freezing being a freshw ater sink for th e ocean.
1 .2 .3 P r o c e s s e s
T h e u pper ocean is host to a variety of processes, some of which are schem at
ically depicted in Fig. 1.1. Here I will describe briefly only th e processes considered in subsequent chapters.
C a b b e li n g
T h e equation of s ta te for seaw ater is nonlinear, th a t is to say th e density p is a nonlinear function of te m p e ra tu re T , salinity s, and pressure p (M illero
and Poisson 1981). For th e u p p er ocean a t tem p erate latitu d es th e te m p e ra tu re dependence is th e m ost im p o rta n t one. T h e tem p eratu re expansion coefficient
1. Introduction 10 breaking w av es Langm ulr circulation CONVECTION shear-induced turbulence breaking internal w aves
Fig. 1.1. Some of th e m echanism s affecting the tu rbulence in the oceanic sur face boundary layer: Langm uir circulation, breaking surface waves, shear-induced turbulence, convection, breaking internal waves (redraw n from T horpe (1985)).
nonlirieaxity m eans th a t if equal volumes of two w ater m asses of different proper ties are m ixed, the resulting m ixture, while having th e ir m ean tem p eratu re, will
have a d en sity larger th a n th e ir m ean density (Fig. 1.2). In th e special case when th e two w ater masses have different salinities and tem p eratu res b u t th e sam e or
nearly th e sam e density, th e resulting m ix tu re will be heavier th a n b o th and will
ten d to sink. A pparently W itte (1902) was th e first to recognize th e possible im portance of this phenom enon (Foster 1972), visualizing a frontal situ atio n when
th e two w ater masses are ju x tap o sed and la te ra l m ixing occurs, causing sinking in
a narrow region between th e m . T he sam e setup was considered also by G arrett and H om e (1978). However, cabbeling need not occur only in horizontal fronts.
Indeed, it occurs whenever th ere is m ixing of w ater of different properties, e.g. whenever th e m ixed layer entrains colder w ater from below. C abbeling m ight be
im p o rtan t for th e form ation of th e densest ocean w ater, e.g. in th e form ation of
th e A tlantic B o tto m W ater in th e Weddell Sea (Foster 1972). T h e salinity con trib u tio n to cabbeling opposes th a t of tem p eratu re, b u t is typically 2 orders of
m agnitude sm aller (K ihnatov and Kuzm in 1991).
L a n g m u ir c ir c u la tio n
In 1927, while crossing th e A tlantic Ocean, the chem ist Irw ing Langm uir observed floating seaweed arranged in streaks roughly parallel to th e wind di
rection, an d w ith a spacing of 100 to 200 m etres betw een th em . In subsequent
years he m ad e a series of experim ents in a lake, and showed (L angm uir 1938) th a t these streaks axe m anifestations of a subsurface organized stru c tu re — pairs of
I. Introduction 12 D1 I W c (D Q 02 T1 Tm T2 Temperature ■
Fig. 1.2. A schem atic of th e nonlinear dependence of density of seaw ater on tem p eratu re. W arm w ater expands m ore th a n cool w ater co n tracts. If w ater w ith te m p e ra tu re T l , d en sity D l , m ixes w ith w ater with te m p eratu re T 2 > T*l, density
D2 < D l , in equal volum es, th e m ixtu re, while of tem p eratu re T m = { T l + T 2 ) /2 ,
downwelling regions betw een the rolls (hence surface convergences an d streaks of flotsam above th em ). Langm uir concluded from his experim ents th a t these vor
tices, later n am ed after him , “ap p arently co n stitute the essential m echanism by
which th e [m ixed layer] is produced” . Furtherm ore, envisioning m ixed layer en- train m en t due to this Langm uir circulation (LC), he sta ted th a t th e “currents
thus set up a t th e b o tto m of th e [mixed layer] m ay sweep off th e u p p er p a rt of th e therm ocline, m ak in g it th in an d of increased gradient” .
A fter m a n y years of observations (in lakes and in the ocean, m ore recently in 1982 an d 1990 off th e coast of California (S m ith 1992; P lu ed d em an n et a/.1996)),
th e conceptual p ictu re for these subsurface vortices has rem ained largely as de scribed by Langm uir. Fig. 1.3 presents a schem atic diagram of LC by Pol
lard (1977). T h e spacing between streaklines, or “windrows” , can range from 2 to hundreds of m eters (Leibovich 1983). Observations (e.g. W eller an d P rice (1988)) and m odelling (e.g. Skyllingstad and Denbo (1995)) suggest th a t, as tim e pro
gresses, sm aller-scale cells are swept into th e larger scale, while a t th e sam e tim e
new sm all-scale cells form . T h e horizontal scale for the spacing, a t least of th e largest cells, is generally assum ed to correlate w ith the d ep th of th e therm ocline,
i.e. the cells have an aspect ratio of 0 (1) and penetrate to th e base of th e m ixed layer (ML). Som e observations (e.g. Sm ith (1992)) have confirm ed this, while o th
ers (e.g. W eller a n d Price (1988)) have found th a t LC is confined to th e u p p er half of th e ML. T h e localized downwelling velocities below the streaks are from 0.03 to
0.20 m s ^ (W eller and P rice 1988), sometim es m ore (Pollard an d T hom as 1989).
T h e upwelling is significantly slower an d broader. There is a c h aracteristic down wind surface horizontal je t. LC is always nearly aligned w ith th e w ind, an d if th e
1. Introduction 14
wind
c < 15w - 1-2 cm/s
-w - 10 cm/s
Fig. 1.3. A schem atic representation of Langm uir circulation (redraw n from Pol lard (1977)).
wind suddenly changes direction, LC gets reoriented in th e new direction w ithin a
few m inutes to a b o u t half an hour, w ith sm aller scale LC reorienting m o re quickly
th a n larger scale LC. These tim es are also indicative of the form ation tim es for LC (Leibovich 1983).
M any theories for the form ation of LC have been proposed, b u t th e com
m only accepted theory is th a t of C raik (1977) and Leibovich (1977a). It rehes on
an in teractio n between a sheared horizontal cu rren t and the surface waves, and involves an in stab ih ty of an infinitesim al dow nw ind je t w ith its induced vertical
v o rticity of altern atin g sign on b o th sides. T h e Stokes drift, which decays ver tically w ith d ep th , tilts these vorticities an d th u s creates the pairs of roUs w ith
downwelling in between. This p a tte rn gets reinforced, as w ater m oving tow ards th e surface convergence line gets accelerated by th e w ind stress which in tu r n leads
to m ore downwelhng, intensification of th e circulation, and stronger ho rizo n tal sur face je t.
I n t e r n a l w a v e s
G rav ity waves in the interior of th e ocean are com m onplace. In analogy to
surface grav ity waves, one can th in k of a d istu rb an ce of an interface betw een two layers of slightly different densities which will feel th e restoring force of gravity
(although m uch reduced because of th e sim ilar densities) and reverse direction,
overshoot its equilibrium position, and so on, resu ltin g in wave propagation. How ever, in tern a l waves can be supported in any continuously varying s ta b le s tra ti
fication. Because th e restoring force is m uch w eaker th a n in th e case of surface waves, in tern a l waves have spatial scales an d tim e scales which are ty p ically m uch
I. Introduction 16
larger. Away from the u p p e r ocean, typical internal waves have vertical displace m ents of th e order of tens of m eters and periods from tens of m in u tes to m any
hours (G a rre tt and M unk 1979). In the u p p er ocean, intern al waves typically have
sm aller am plitudes and sh o rte r periods (Pinkel 1975; Kase and C larke 1978; Dillon an d Caldwell 1980; Levine et a/.1983), b u t are still large com pared w ith surface
waves (Fig. 1.4). In tern al waves are com m only m anifested in single-sensor observa
tions as sh o rt period fluctuations in th e tem p e ra tu re (M unk 1981). N ansen (1902) m ight have been the first to rep o rt them . A ttrib u tin g these te m p e ra tu re fluctua
tions to in tern al waves, an d thus estim ating th eir instantaneous vertical velocity
by w = —{ d T f d t ) J { d T / d z ) assum es no significant horizontal advection a n d /o r diffusion (Briscoe 1975).
T h e equations for a Boussinesq fluid (Boussinesq 1903; Spiegel an d Veronis
1960) of varying stratificatio n w ith buoyancy frequency N [ z ) lead to th e dispersion relation (Phillips 1977)
ui = N cos 6 + f sin 0, (1.15)
where w is th e internal wave frequency, / is th e local inertial frequency, 6 is th e
angle to th e horizontal of th e vector w avenum ber. Varying 9 from 0 to tt/2 , thus
intern al waves can exist only in th e range of frequencies
Depth [m] Depth [m]
■drÆilLK
^ 9 ^
Fig. 1.4. Ten-day tim e series of isotherm s versus d ep th and tim e, tak en from the LOTUS d a tase t (Briscoe and W eller 1984; Bowers et al. 1986). T he high-firequency variability in th e isotherm s’ position in an d below th e seasonal therm ocline at 20
I. Introduction 18
G arrett and Munk (1975), using m any different observations o f internal wave activ ity in th e ocean away from th e upper layer, compiled a representation
of th e d istribution of internal wave energy in wavenumber and frequency space, subsequently called the GM spectrum . A fter tests and refinements it was con cluded th a t the observed characteristics of deep-ocean internal waves a re nearly
universal in space and tim e, thus the GM spectrum can be regarded as a universal equilibrium spectrum , the result of a dynam ic balance between wave generation,
in teractio n , and dissipation. In th e u p p er ocean, where such an equilibrium is
not to be expected due to th e forcing an d variability, the internal wave field of higher frequencies deviates from th e GM spectrum , and exhibits high coherence
w ith d ep th (Fig. 1.4).
Internal waves can be generated by a variety of mechanisms. Surface gen e ratio n by the atm osphere could involve travelling pressure fields/ fronts (Keller
an d M unk 1970; Polyanskaya 1969; Leonov and Miropolsky 1973), a travelling
buoyancy flux (M agaard 1973), and a travelling wind stress field (Tom czak 1966;
Tom czak 1967; Krauss 1972a; Krauss 1972b), the la tte r being th e m ost plausi ble source (Thorpe 1975). Surface generation by interaction of pairs of surface waves w ith sim ilar wavenumbers and frequencies is also possible (T h o rp e 1966;
N esterov 1972; Brekhovskikh et a/.1972). T h e steady flow of th e stratified ocean
over b o tto m topography is another source (H uppert and Miles 1969; B ell 1975),
generating standing internal waves analogous to lee waves in the atm osphere. T idal (barotropic) currents over topography can also generate internal waves of tid a l fre quency (Cox and Sandstrom 1962; Baines 1973; BeU 1975; Baines 1982). In the
ocean interior, in tern a l waves can be g en erated a n d /o r modified by wave-wave in teractions (O lbers 1976; McComas and B re th e rto n 1977; Pom phrey et a/.19S0).
D issipation of in tern al waves could occur th ro u g h sh ear in stab ih ty (T horpe 1973)
an d wave b reaking (O rlanski and Bryan 1969; O rlanski and Ross 1973; G argett and Holloway 1984), and also through viscous a tte n u a tio n (LeBlond 1966).
1 .3 O b se r v a tio n s
I will briefly describe th e sets of observations of th e upper ocean which are
referred to in subsequent chapters.
1 .3 .1 O W S P a p a
Since 1952 (T ab ata 1965), extensive oceanographic observations of the up
p e r ocean were m ade by Canadian w eatherships a t O cean W eather S tatio n (OW S) “P ” (50°N, 145°W ) in th e northeast Pacific O cean, u n til the w eatherships were
retired in 1981. T h e d a ta consist of m eteorological observations and b a th y th e r
m ograph (B T ) te m p e ra tu re records over m any years a t 3-hour intervals (th e BT records are a t 3-hour intervals since 1965). T h e m eteorological observations consist
of w ind speed, w ind direction, sea-surface te m p e ra tu re , air tem p eratu re, w et-bulb
a n d dew -point tem p eratu res, cloud cover, an d surface air pressure. T h e B T tem p e ra tu re d a ta are generally a t 5-metre intervals from th e surface to 295 m.
I. Introduction 20
1 .3 .2 L O T U S
T h e Long-Term Upper-Ocean S tudy (LOTUS) experim ent was conducted from M ay 1982 to M ay 1984 a t 34°N, 70°W in the deep w ater over th e H at
te r as abyssal p lan e in the Sargasso Sea in th e western su b tro p ical A tlantic Ocean.
T h e d a ta consist of m eteorological observations, tem p e ra tu re and current m eter records (Briscoe an d Weller 1984; S tram m a et a/. 1986). T hese are divided into
four periods of several m onths each, w ith gaps in between w hen th e moorings were
replaced an d w ith o th e r gaps caused by d a ta losses. T h e m eteorological observa
tions consist of w ind speed, wind direction, sea-surface te m p e ra tu re (at a d ep th of
0.6 m ), air te m p e ra tu re , barom etric air pressure, and directly m easured insolation. T he hum idity m easurem ents were unreliable. Mooring d a ta consist of te m p e ra tu re
and east an d n o rth currents generally a t depths of 5, 10, 15, 20, 25, 35, 50, 65, 75,
100 m etres and som e in deeper w ater. D ata are available a t 15-m inute intervals.
1 .3 .3 M IL E
T h e M ixed Layer E xperim ent (M ILE) was conducted from A ugust 19 to
Septem ber 6, 1977, a t 49 37 N, 145°6^W, in the vicinity of OWS Papa. T he d a ta
consist of m eteorological observations, tem p eratu re and cu rren t m eter records (Davis
et al.l981a), a n d also two series of casts through the u p p er 200 m etres of ocean
w ith a m icro stru c tu re profiler (Dillon and Caldwell 1980). T h e meteorological observations consist of wind speed, wind direction, sea-surface te m p eratu re, air
tem p eratu re, w et-bulb and dew-point tem peratures, surface air pressure, cloud cover, m ade every 3 hours, and solar radiation m easurem ents m ade every 1 hour.
T h e MILE-1 m ooring has com plete 19-day records of tem p eratu re and east an d n o rth currents a t depths of 11, 23, 26, 29, 32, 35, 41, 44, 47, 50, 75, 92, 125 an d
175 m etres and 16-day records at 20 an d 38 m etres. T he MILE-2 mooring has
com plete records of tem perature and east and n o rth currents a t depths of 5, 10, 15, 20, 42 and 54 m etres. T he d ata are available a t 112.5-second intervals.
1 .3 .4 P A T C H E X
T he PATCHes E x p e rim en t (PA TCH EX ) was conducted in October 1986
aro u n d 34°N, 127°W , 800 km west of P oint Conception, off th e coast of California.
T h e d a ta include meteorological observations, te m p e ra tu re and horizontal velocity
records (B rainerd and Gregg 1993). T h e meteorological observations consist of wind speed, w ind direction, sea-surface tem p eratu re, air tem p eratu re, wet-bulb
and dew -point tem peratures, cloud cover, surface air pressure, and insolation. An A dvanced M icrostructure Profiler recorded 11 days of tem p eratu re, conductivity,
te m p e ra tu re gradient, and velocity fluctuations profiles down to 300 m, at an
average ra te of 2.4 profiles per hour. Relative current velocities were m easured using an acoustic current m eter, a t a ra te of 4 drops per day for th e last 7 days.
1 .3 .5 G A T E , p h a se I II
T h e G A R P A tlantic Tropical E xperim ent (G A TE), phase III, was con
d u cted from A ugust 30 to Septem ber 19, 1974, a t 9°N, 23°W in the A tlantic O cean, to stu d y in ternal waves in th e u p p er ocean (Kase an d Clarke 1978). T h e
d a ta consist of a triangular horizontal array of moorings, recording tem p eratu re and currents, an d GTD profiles from a drifting ship. T h e CTD profiles reached
1. Intro ductio n 22
down to 500 m , at 3-hour intervals, and th e re were also som e sh o rt-term CTD
series from 20 to 70 m, a t 108-second intervals. The d a ta a t th e m oorings are at 225-second intervals.
C h a p t e r 2
M ix ed layer m o d e ls
I will present the approxim ate and sim plified governing equations th a t are ad eq u ate for use in 1-D m odels of th e u p p er ocean, and describe in m ore detail
m odels referred to in subsequent chapters.
2 .1 E q u a ti o n s
T h e full equations governing th e evolution of the ocean are (G ill 1982):
T h e mass conservation equation
P ^ + V - u = 0, (2.1)
w here D / D t = + u • V is the m ateria l derivative, u is th e fluid’s (vector) velocity, an d p th e density of th e fluid.
T h e m om entum equation
^ + / ( k X u) = p V p - g + f/V^u, (2.2)
where / is th e Coriolis p a ra m e te r/in ertia l frequency, k the vertical co ordinate u nit
vector, p th e pressure, g th e gravitational acceleration, u the kinem atic viscosity
2. M ixed layer m odels 24
T h e heat equation
~ W ^ + S h . (2-3)
where T is th e te m p eratu re, th e specific h eat a t constant pressure, a = —I j p l d p j d T ) th e tem p eratu re expansion coefficient a t constant safinity and pressure, % the th er
m al conductivity, F th e rad iativ e flux, th e heating due to a change of phaae, chemical reaction, and viscous dissipation.
T he salinity equation
Ds
p — = V • (Kj^Vs), (2.4)
where s is th e salinity, an d is th e salinity diffusivity.
The equation of s ta te
p = p { p , s , T ) . (2.5)
We can sim plify these equations by m aking a num ber of approxim ations
a n d /o r assum ptions based on th e context in which they will be used, th a t is, for
upper ocean modelling. Because b o th particle and phase speeds of disturbances are much sm aller th a n th e speed of sound, an d also because th e vertical scale
of the m otion is sm all com pared to the scale height of th e ocean, we assume incom pressibility (B atchelor 1967), so D p / D t = 0 and therefore (2.1) reduces to
th e continuity equation V • u = 0; we also ignore pressure variations in the heat equation, and assum e = const. Because th e reference h y d ro statically balanced
density varies little around its m ean, and also because the m otion-induced density variations are m uch sm aller th an the reference state density, we can m ake the
Boussinesq approxim ation (Boussinesq 1903: Spiegel and Veronis 1960) and replace
density in th e equations w ith a constant reference density except when it gives
rise to buoyancy (6 = —gp^ ^(p~~Pq)} forces in the gravitational acceleration term .
In addition, we make th e boundary layer approxim ation th a t vertical gradients are
m uch g reater th an horizontal gradients, and in fact com pletely drop horizontal
gradients, assuming a local balance in the vertical as 1-D ML m odels do (Niiler an d K raus 1977). We are interested in the evolution of m ean properties, and how th ey are affected in a m ean sense by turbulence, so we perform Reynolds
decom position of all variables into mean and fluctuation parts. M olecular term s
are also neglected as sm all com pared with th e other term s. Finally, th e averaged equations for m ean qu an tities are
W = _ / ( k X Û ) - | ( ; 7 7 ) (2.8)
P = p{s,T ), (2.9)
where overlines denote averaging and mean quantities, and prim es denote turb u len t fluctuations. I[z) is th e vertical absorption profile for th e insolation. In addition
to these equations, th e equation for th e averaged turbulent kinetic energy (TK E) is also often used for calculating th e am ount of mixing a n d /o r th e entrainm ent
2. M ix ed layer m odels 26
ra te for a m ixed layer. It is, u n d er th e sam e assum ptions and approxim ations as above (K raus 1972),
1 d(q^) ' ' d U ~!~r d >, 2 , T , .
2 ~ d T = + ^ 6 - - ^ w {q ( 2 + p / p j - e, (2.10)
where = u ^ + to ^ is tw ice th e T K E per unit mass. T h e first te rm on th e
rig h t-h an d side of (2.10) is th e ra te of work by the Reynolds stress w u o n th e m ean shearing flow j d z \ th e second te rm is the rate of work of buoyancy forces;
th e th ird term is the convergence of th e tu rbulent vertical flux; the fo u rth te rm is th e dissipation rate of T K E . H enceforth overlines denoting m ean q u an tities will be dropped, but implied.
2 ,2 T h e P r ic e - W e lle r - P in k e l m o d e l
T h e model of Price, W eller, an d Pinkel (1986), henceforth th e P W P m odel,
is a 1-D bulk m ixed layer m odel, i.e. it assumes the existence of a w ell-m ixed surface layer. T he m odel solves equations (2.6 — 2.9) on a fine uniform v ertical
grid (e.g. w ith A z = 1 m ), and employs stab ility criteria to establish, m ain tain ,
and evolve a mixed layer of uniform tem p eratu re, salinity, m om entum , an d density. In accord w ith the scope of this study, let us consider the salinity to be co n sta n t. T h e surface boundary conditions axe
w 'r'(o ) = 0L (2.11)
1(0) = (2.12)
IÜ u (0) = r (2.13)
w here = Q[, + + Q is th e n et heat flux (loss) from long-wave radiation, la te n t, eind sensible fluxes, respectively; Q^ is th e incoming solar radiation; t is th e
vector wind stress. T h e tu rb u le n t fluxes at d epth are not e x p h citly calculated, b u t
are im plicitly accounted for as a result of th e adjustm ents by s ta b ility criteria. T h e
absorption profile I{ z) for th e p en etratio n of solar radiation a t d ep th is assum ed to have a double exponent dependence (Paulson and Sim pson 1977), i.e.
I { z ) = 1(0) [ r e x p ( z / ^ J - b ( l - r) exp(z/^^)] , (2.14)
so th e rad iatio n is divided into a long-wave and a short-wave sp ectral com ponent w ith a tte n u a tio n depths of and respectively. The a tte n u a tio n depths change
w ith w ater tu rb id ity . Jerlov (1976) designated several w ater ty p es based on w ater
chlorophyll concentrations. For a fairly clear, mid-ocean w ater of ty p e lA th e values are r = 0.62; = 0.6m; = 20m .
For each tim e-step , th e P W P m odel proceeds as follows.
(i) Insolation is absorbed in the water colum n, an d net heat loss
occurs from th e top grid-point, or layer, thus changing th e te m p e ra tu re profile. D ensity is calculated (from a linear or nonlinear equation of s ta te , depending on
im plem entation).
(ii) A sta tic stab ility criterion
2. M ix e d layer m odels 28
is evaluated, an d com plete m ixing of tem p eratu re, m om entum , and density occurs from th e surface downwards, iteratively, until a sta tic a lly stable stratification w ith a m ixed layer on top is achieved. Mixing at this stage is equivalent to th a t caused
by free, n o n -penetrative convection.
(iii) T h e wind stress is applied to th e to p grid-point, and m om entum
is d istrib u te d uniform ly through the m ixed layer. H orizontal currents are ro ta te d inertially.
(iv) Overall m ixed layer shear in sta b ih ty versus buoyancy sta b ih ty
is ev alu ated a t th e base of th e mixed layer by a bulk Richardson num ber criterion
Rk = > 0.65 (2.16)
where h is th e m ixed layer d ep th , 0.65 is the critical Rb value, A p and | A u | are th e
density an d velocity differences, respectively, betw een the ML and ju st below it. If Rb < 0.65, th e m ixed layer deepens by one grid-point and properties are m ixed
com pletely w ithin th e new ML. The new Rb n u m b er is then evaluated, and en
tra in m e n t and m ixing proceed iteratively until th e criterion is satisfied. This ty p e
of d y nam ic sta b ility criterion, implying th a t the ML is always either stable or in a sta te of m arg in al stabihty, has also been im plem ented by others (Pollard et a/.1973; Price et aL1978; K ronenburg 1985). Although no sound theoretical evidence th a t a
tru e critica l R ichardson num ber exists for a tu rb u le n t m ixed layer (Philhps 1977),
lab o rato ry experim ents by EUison and Turner (1959) show th a t entrainm ent by a surface half-jet decreases by a fuU order of m ag n itu d e as Rb increases from 0.4 to 0.8, suggesting a n approxim ate critical bulk R ichardson num ber in th a t range.
(v) Local shear flow stability is evaluated below th e m ixed layer w ith a gradient Richardson num ber criterion
where 0.25 is th e critical Rg value; gradients are evaluated by taking first differences
over neighbouring grid-points. If R g is found to be less th an 0.25 a t a given pair
of grid-points, p a rtia l m ixing of properties occurs betw een them . If densities prior to the partial m ixing are and the density exchanged is e.g.
(2-18)
w ith a com plete m ixing for Rg = 0; a constant (0.3) slightly larger th a n 0.25 is used to hasten convergence, the larger value having no appreciable effect in the
solutions (Price et a/.19S6). Then the Rg array is recalculated, reevaluated, ajid
new adjustm ents are m ade, until Rg > 0.25 everywhere.
The m o tivation for adding gradient Richardson num ber dependent m ixing
below th e m ixed layer was to sm ooth out sharp te m p era tu re ju m p s at th e ML base, m aking m odel profiles more like the profiles from observations, and also to
relieve local shear instabilities in th e stratified fluid below the ML. However, the im plem entation of this criterion in the model leads to “unm ixing” of th e m ixed layer established w ith the bulk Richardson num ber criterion, as p artial m ixing
adjusts the te m p e ra tu re at the bo tto m ML grid-point as well. As a result, the final ML dep th diagnosed by the m odel is not the ML dep th established by overall
2. M ix ed layer m odels 30
m ixed layer stability, as evident e.g. from Fig. 2 of Schudlicb and P ric e (1992) and th e accom panying figure caption. Thus, the Rg procedure seems to deliver
m ore th a n it was designed to do; this does not seem to be com m only recognized, as th e m odel has been described or im plied by others as setting its ML d e p th w ith
a bulk criterion, w ith R g m ixing only resolving instabilities below such a m ixed
layer (A rcher et a/.1993; K a n th a an d Clayson 1994; Large et a/.1994; L arge an d Craw ford 1995).
M odel diagnostics (not shown) revealed th a t to satisfy R g > 0.25 every
where, up to tens of thousands of ad ju stm en ts at each tim e-step m ight b e needed for a region w ith a few tens of grid-points. In addition, while looking o n occasion
a t spikes in th e P W P m odel o u tp u t such as ML dep th and sea-surface te m p e r
atu re, I was able to trace th e m back to th e Rg criterion procedure. It appears th a t, afte r m any sm all ad ju stm en ts, th e final ML d ep th m ight in co rp o rate a m ore
or less a rb itra ry fluctuation. A pparently, th e problem of m axim izing Rg , by over
shooting discrete-step ad ju stm en ts, to achieve Rg > 0.25 everywhere for a n arra y of grid-points, does not have a unique solution. So th e final ML d e p th m ay not
be consistent from one tim e-step to th e next as different “solutions” are realized. T he spikes disappear, however, if th e ML dep th achieved from overall m ix ed layer
sta b ih ty w ith th e bulk Richardson num ber criterion is enforced, while s till allow
ing for ad ju stm en ts in th e therm ocline below it^. Therefore, th e m ixed lay er d e p th
^This can be done as follows. For each iterative adjustm ent due to the gradient R ichardson num ber criterion, i f partial m ixing aifected the b o tto m grid-point o f the m ixed layer established b y th e R h
criterion, a com p lete m ixing is performed w ithin this layer to restore its hom ogeneity before th e n ext iteration is a ttem p ted .
based on overall stab ility is consistent in tim e, and in fact it seems difficult to ju stify m odifications of th is d ep th via th e gradient Richardson nu m b er criterion.
2 .3 T h e M e llo r - Y a m a d a L e v e l 2 m o d e l
A hierarchy of tu rb u le n t closure m odels, based on an order-of-m agnitude
analysis of sm all deviations of th e Reynolds stresses and heat fluxes from local isotropy, was described by M ellor and Y am ada (1974), refined by th em la te r (MeUor
a n d Y am ada 1982), and reexam ined by G alperin et al. (1988). In the lim iting case of neax isotropy, algebraic relations axe o b tain ed for all tu rb u le n t q u an tities. The
resulting m odel, classified as Level 2 closure (MY-2), is a ttra c tiv e because of its
sim plicity and robustness (G alperin et a/. 1988), and has been applied in various situ atio n s (M artin 1985; M a rtin 1986; K an th a an d Clayson 1994; Large et a/.1994). T h e equations, in term s of buoyancy instead of te m p e ra tu re and salinity separately,
are d U d d t = - / ( k X U ) - — (2.20) - r - T d V — V — = —w u • -5---- \-w b (2.21) A.. d z
w here (2.21) is the averaged T K E equation (2.10), simplified by neglecting the
convergence of the tu rb u le n t vertical flux, and assum ing a balance betw een shear p ro d u ctio n , buoyancy forcing, and dissipation. T h e dissipation is assum ed to be
2. M ix e d layer m odels 32
“m a ste r length scale” of turbulence, to which all o th e r relevant length scales are
p ro portional. T his m aster scale is calculated from B lackadar’s (1962) interpolation fo rm u la
w here z is distance from a boundary, k % 0.4 is von K arm an ’s constant, is a
m easu re of th e extent of th e turbulent field. T h e form ula interpolates between
I — > H.\z\ as |z|/Zg — V 0, and I — y as — j- oo; is calculated from
w here q is rm s tu rb u len t velocity, and 0.2 is the chosen value of th e em pirical
co n sta n t in agreem ent w ith results from higher level MY models (Mofjeld and
Lavelle 1984; M artin 1985).
T h e tu rb u len t fluxes in (2.19 — 2.21) are param eterized by eddy diffusivity a n d eddy viscosity
w b = - K ^ { z ) ^ = - I q S ^ ^ (2.24)
w here th e vertically-varying eddy coefficients are p roducts of the m aster length
scale Z(z), rm s tu rb u len t velocity g(z), and a gradient Richardson num ber Rg de p e n d e n t sta b ility functions involving more p roportionality constants. A fter
su b stitu tin g the values of th e proportionality constants tak en from Mellor and Ya-
m ad a (1982), wherever needed, th e final equations are
8
-
i
(«.
8
) -
0.
5 _ ________0.4 0- 3.0 8 0^ _________
M (1 - 34.680H)(1 - 6.130^) ^
where = —{l/q)^db/dz, and I and are calculated from (2.22) and (2.23) above.
It can be shown th a t 6"^ > 0, > 0, and th a t they are zero for Rg > 0.19.
2 .4 O th er m o d e ls
One-dimensional m odels of the m ixed layer generally fall into two classes. Turbulence closure ML m odels (also called K-theory m odels) can be organized
and classified as parts of a single hierarchy (Mellor and Y am ad a 1974; Mellor and
Y am ada 1982; Galperin et a/.1988) based on an order of m ag n itu d e analysis of sm all deviations of the second m om ents of turbulence from th e s ta te of local isotropy.
T hese models employ vertically-varying eddy diffusivity an d ed d y viscosity. T h e
M ellor-Yamada Level 2 m odel described in the previous section is an exam ple of a K -theory model. Details on other K -theory models will n o t be given here; th e
reader is referred to th e th ree papers cited above. Bulk ML m odels assume th e