ICES mar. Sei. Symp., 195: 32-39. 1992
Variability of convective conditions in the Greenland Sea
Jens Meincke, Steingn'mur Jönsson, and James H. Swift
M eincke, J ., Jö n s so n , S ., an d Swift, J. H. 1992. Variability o f convective conditions in
th e G r e e n la n d Sea. - IC E S m ar. Sei. S ym p., 195: 32-39.
R ecent o bservations a n d d a ta com pilations show decadal an d in teran n u al v ariations in
the d e p th o f w intertim e convection in the G r e e n la n d Sea. In a qualitative study the
fluctuations arc related to changes in wind an d t h erm o h alin e forcing. C h a n g es in both
wind-stress curl an d sea-ice cover concur with the results from h y d ro g rap h ic o b se r­
vations indicating that no renew al o f d e e p w a te r has ta k e n place d u rin g th e 1980s.
Jens M eincke: Institut f ü r M eeresku n d e der Universität, H am burg, G erm any; Steingn'­
m u r Jö n sso n : The M arine Research Institute a n d U niversity o f A ku reyri, A ku reyri,
Iceland; Ja m es H. Sw ift: Scripps Institution o f O ceanography, L a Jolla, California,
U SA .
Introduction
T h e process of convective renew al of w aters in the
G reen lan d Sea is a m ajor influence on the therm ohaline
circulation o f two systems (A ag aard et a l., 1985): on the
o n e hand it is on e o f the sources for th e form ation of
interm ed iate w aters which leave the area as overflows
across th e G re e n lan d -S co tlan d ridge an d drive the
global circulation system o f the N o rth A tlantic D eep
W ater. O n the o th e r hand it contributes significantly to
the d eep circulation internal to the A rctic O cean/N ordic
Seas system, which is confined to the area n orth of the
Bering Strait and north of the G re e n lan d -S co tlan d
ridge. T he division betw een the two systems can be
defined by the sigma - 1 level o f 32.785, which is found
aroun d 1500-m d ep th in the N ordic Seas. This is the
m axim um d epth of w inter convection th a t determ ines
w h ether convective renew al is limited to interm ediate
w ater o r affects both interm ediate and deep levels.
W ith this background a study on the variability of
convective depths in the G reen lan d Sea is im po rtant for
the role of the oceans in climate. This contribution will
illustrate the variability and discuss the mechanism
causing it.
The convective system
A much im proved database and conceptual modelling
have led to a fairly consistent picture of convective w ater
mass transform ation in th e A rctic O cean/N ordic Seas
system. T h e th e rm o haline part is d om inated by the
32
regionally varying interaction o f th ree m em bers in TSspace (A ag aard et al., 1985; R udels and Q u ad fasel,
1991), illustrated in Figure 2: cold and fresh w ater from
run off and m elting, cold and saline w ater from brine
released during sea-ice form ation , and w arm and saline
w ater con tribu ted from A tlantic inflows.
In the A rctic O cean wide shelf areas th e brine from
seasonal sea-ice form ation accum ulates in troughs and
trenches (M idttun, 1985) before spilling ov er th e shelf
edge. It sinks dow n th e slopes as dense high salinity
plum es at n e a r freezing tem p eratu res, entraining in
particularly w arm in term ediate w ater of A tla ntic origin,
and results in A rctic O cean D e e p W a te r (A O D W ) , the
w arm est and m ost saline d eep w ater c o m p o n en t in the
A rctic O cean/N ordic Seas system.
In the case o f the d eep G reen lan d Sea th e saline
plum es from sea-ice form ation enrich the relatively fresh
u p p e r A rctic layer with salt, such th a t subsequ en t
plum es p en etrate the halocline into w arm in term ediate
layers. T o preserve continuity, the w arm w ater rises to
the surface, w here it melts the ice and th e reb y stops the
convection. A fte r cooling and new ice form ation a new
haline convection cycle can start (R udels, 1990). This
results: in a stepwise increase of convective depths
(R udels et a l. , 1989), in extrem ely w eak m ean stratifi­
cation (Fig. 2), and in G reen lan d Sea D e e p W ater
(G S D W ), the coldest and freshest d eep w ater co m ­
ponent. T h e o th e r d eep w ater co m po nents, which are
TS-wise b etw een the characteristics of those from the
Arctic O cean and the G reen lan d Sea, are a p roduct of
mixing betw een the latter (A ag aard et al., 1991).
T he therm oh aline part o f the convective system
Figure 1. T o p o g ra p h y o f th e N ordic Seas. FS = F ra m S trait, B = B o reas Basin, G = G r e e n la n d B asin, B + G = G r e e n la n d Sea
B asin, L = L o fo ten Basin, N = N orw egian B asin, I = Iceland S ea Basin.
ATLANTIC
INFLOW
RIVER RUN-OFF
BRINE
24
34
Figure 2. S chem atic T S-relationships for the A rctic O cea n/
N o rd ic Seas system. A = G r e e n la n d S ea, B = A rctic O cea n ,
C = N o rw egian Sea. T h e e n d m e m b e rs are na m e d .
closely interacts w ith the w ind-driven p art. N ext in
im p ortance to the large-scale wind forcing, which gov­
erns the flux of A tlan tic w aters and P olar w aters
thro u g h o u t the system, the basin-scale wind forcing is
particularly im po rtan t. L egutke (1990) has show n th a t it
is the matching betw een the along-isobath circulation
and the co n to u r integral of the wind stress along f/Hcontours a ro u n d the basis that determ ines the effective­
ness of wind forcing. This has two consequences: with
varying scales o f wind forcing, different-sized cyclonic
gyres are excited in accordance with the different sizes of
topographic basins (see Fig. 1). F or the G re e n la n d Sea
this ranges from a one-gyre situation covering both the
G reen lan d and B oreas Basin to a two-gyre situation with
separate cyclonic circulations for each o f the two basins.
T h e second consequence is related to the varying in ten­
sity of wind forcing. It results in a varying deform ation of
the interface betw een the d eep and interm ed iate waters,
i.e. a varying intensity of the “d om in g ” of the d eep w ater
in the gyres.
34.900
34.896
-
1.20
GSP
34.888
DWP
c l - 1.25
195 0
1960
1970
1980
1990
Figure 3. T im e series o f a verage p o tential t e m p e r a tu re below
2000 m in th e G r e e n la n d Basins. D a ta from sta tions with
b o tto m d e p th exceeding 3000 m. T h e tw o single dots m ark
c o rre sp o n d in g salinities o b ta in e d d uring th e IC E S D e e p W a te r
Project ( D W P ) (C lark e et a!., 1990) an d d uring the G r e e n la n d
Sea Project (G S P ) (G S P G r o u p , 1991 ). 0 - an d S-scales were
choscn such th a t in th e event o f constant d e e p -w a te r density 0 an d S -variations would show as parallel curves.
Observed variations in convective
activity
T he reasons for the scarcity o f time-series inform ation
on the convective renew al of w ater masses in the A rctic
O cean /N ordic Seas system are obvious: extrem e diffi­
culty o f access to the A rctic O cean and signal am plitudes
close to the m e asu rem en t resolution for the traditional
ocean ographic p ara m e te rs are tw o m ajor reasons. It is
only since th e mid-1970s tha t salinities have been ro u ­
tinely m e asu red to 0.002 psu and tha t an throp ogenic
tracers have beco m e available as ad eq u ate param eters.
C onsequ en tly , a tim e series on convective renew al over
a decadal tim e scale can only be inferred from te m p e ra ­
ture m easu rem en ts in th e G re e n la n d Basin (Fig. 3).
T hey show tw o w arm (late 1950s an d late 1980s) an d one
cold (early 1970s) p eriod for the d eep w ater, suggesting
a time scale in the o rd e r o f 30 years. Since the w ater
masses o f the gyre of the G re e n la n d Basin are to p o ­
graphically tra p p e d and have horizontal te m p e ra tu re
gradients m uch sm aller th a n the observed variations in
tim e, the observ ed cooling can be in terp reted as being
do m in ate d by renew al of G S D W , w hereas the w arm ing
periods are d om in ate d by mixing with w aters a ro u n d the
gyre.
T h ere are two salinity values of sufficient quality
ad d ed to Figure 3. T h e salinity scale was chosen such
th a t in th e case of constant density th e te m p e ra tu re and
salinity curves w ould be parallel. This is not the case,
indicating th a t non-isopycnic mixing occurs (A a g a a rd et
al., 1991).
T he in terp retatio n o f th e w arm ing period for the
1980s in the d eep w ater can be substan tiate d from
34
increased observational activity, resulting in m ore
detailed estim ates of the annu al d epth o f convective
overturning. T he D e e p W a te r Project (C larke et al.,
1990) surveyed the G re e n la n d Sea in w inter and early
su m m er 1982 and found the convection d ep th at 500 m.
F o r the p eriod 1986-1989 (G S P -G ro u p , 1990) convec­
tive depths w ere d e term in ed to be 200,1300 a nd 1600 m
for the G re e n la n d Sea gyre. In co m b ination with
analyses of an th rop ogenic tracers covering the period
1972 to 1989, which show a cessation o f d eep w ater
form ation from 1980 o nw ards (Schlosser et al., 1991), it
can be concluded th a t the 1980s have to be characterized
by no detectable renew al o f w aters in th e G re e n la n d Sea
below 1600 m. O b servation s by N agurny and P opov
(1985) suggest convection dow n to the b o tto m in w inter
1984. This is reg ard ed as q u estio nable, since salinity
tim e series presen ted for the G S D W by P opov (pers.
co m m .) show u n reason ab le fluctuations.
T h ere are two pieces o f observational evidence fo r the
cooling p erio d in the early 1970s. M alm b erg (1983)
re p o rts having observ ed convective conditions of the
central G re e n la n d Basin dow n to 3500 m in w inter 1971.
T h e o th e r evidence com es from tritium d eterm inations
(Schlosser et al., 1991), which suggest intensive d eep
w ater form ation having occurred betw een th e mid-1960s
and the early 1970s.
Causes of convective variability
T h e description o f th e convective variability o f the
G reen lan d Sea has revealed decadal and interan nual
tim e scales. T he literature offers only a few discussions
on possible causes. A a g aard (1968) correlated long-term
changes and in teran nual changes of in term ed iate and
d eep w ater te m p e ra tu re s w ith a w inter cooling index
derived from air te m p e ra tu re observations. T he results
suppo rt N a n se n ’s (1906) concep t o f the o v erturning of
cooled surface w aters, neglecting any haline co m p o n en t
in the system.
In discussing the wind causing variability in the co n ­
vective intensity we tak e up Jô n ss o n ’s (1991a) sugges­
tion: an analysis of wind-stress curl o ver the N ordic Seas
by m eans of em pirical o rthog onal functions. Jönsson
(1991b) shows significant decadal and interan n u al vari­
ations o f intensity and spatial scales. T he low er m odes,
which com pare in scale to the G re e n la n d Sea, show
in teran n u al variability w ith high energy in th e 1960s and
early 1970s and low energy in the late 1970s an d 1980s.
Figure 4 is a relevant p resen tatio n of this scale. T h e
higher m odes com p are in scale with th e individual basins
(Fig. 1), have little energy in the 1960s an d 1970s, and
reach high levels in the 1980s.
W ith respect to w ind-driven circulation and convec­
tion a one-gyre situation is favoured for the G reen lan d
Sea because o f the dom inance o f the low er m od es in the
"c3
CL
/
/
//_
///
///
///
///
////
////
////
/// /
////
////
////
////
////
// //
// //
// //
////
PH
n
p—1/
/ /
/ /
/ /
/ /
/ /
/ /
/ /
/ /
/ /
/ /
/ /
/ /
/ /
/
/
/
/
/
/
/
/
/
/
/
/
/
/
/
/
V
/
/
/
/
/
/
/
/
/
/
/
/
/
R/ //
/
/
/
V\ R // V\
/
/ / /
/ / / / /
/
/
/
/
/
/
/
/
/
/
/
/
/
/
/
/
/
/
/
/
/
/
/
/
/
/
/
/
/
/
/
/
/
/
/
/
/
/
/
/
/
/
/
/
/
/
/
/
/
/
/
/
/
/
/
/
/
/
/
/
/
/
/
/
/
/
/
/ FH
/
/ /
/
/ /
/
/ /
/
/ /
/
/ /
/
/
/
/
/ /
/ / v\ /
/ / / /
/ / / /
/ / / /
/ / / /
/
/
/
/
/
/
/
/
/
/
/
/
/
Y—
FH /
/
/ /
/
/ PI / /
/ / /
/
/ / /
/
/ / /
/
/ / /
/
/ / /
/
/ / /
/
/ / /
/
/
/ 0 / /
/
/
/
/
FH /
/ /
/
7]/
/ / /
/ / /
/ /
56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87
Figure 4. Interannua l variations o f wind stress curl o v er the G r e e n la n d Sea. T h e values are averages o v er an area o f 2 0 0 0 0 0 k m 2 in
the central p a rt o f th e G r e e n la n d Sea an d o v e r a o n e-y ear perio d fro m s u m m e r o f y e a r - 1 to th e following su m m er. N o te the
decrease in th e late 1970s an d 1980s.
1960s and early 1970s, with strong dom ing, u p p er layer
divergence due to E k m a n pum ping and well developed
Arctic and Polar fronts separating the convective regime
from the Polar and A tlantic w aters. This situation pres­
ents favourable conditions for d eep reaching convec­
tion. D uring the 1980s, the low er m od e forcing was less
intense and the potential fo r the generatio n of separate
gyres in the G reen lan d and B oreas Basin was much
larger, thus repeatedly interrupting the prevailing, but
w eak en ed , one-gyre situation for certain periods. Q u ad fasel and M eincke (1987) indeed observed a two-gyre
situation in 1986. This causes advective “leakages" of
the frontal zones in the area of the gyre separation,
bringing fresh P olar w aters into the surface layer and
w arm A tlantic w ater into the in term ediate layers o f the
central G reen lan d Sea. B oth effects are adverse to
convection. A ag aard and C arm ack (1989) have shown
this for excess fresh w ater input into the convective gyre
and Figure 5 shows this for an intrusion of w arm A tlantic
w ater. It occurred betw een A ugust 1986 and M arch
1987, restricting the convective d ep th in the G reen lan d
Basin to 200 m because to o much h eat had to be
rem oved by the convective cycles during one winter.
This m akes decadal and in terannual changes in the
wind field one of the likely causes of corresponding
changes in convective d epths in the G reen lan d Sea quantitative pro o f has still to be evaluated.
T u rn ing to th e rm oh alin e causes of convective vari­
ability, the convective processes in th e A rctic O cean and
G reen lan d Sea, as described in section 2 above, imply
net fluxes o f heat and salt as follows: in the A rctic O cean
the dow nslope p e n e tratio n of high salinity shelf w ater at
freezing te m p eratu res and the e n train m en t o f w arm
A tlantic w aters p roduce a net d ow nw ard flux o f h e a t and
salt for the A rctic Basins. In the G reen lan d Sea the
e n train m en t of fresh w ater in haline plum es at freezing
te m p e ra tu re s and th e p enetrative ch aracter o f the
plum es in the interm ed iate w arm levels result in an
upw ard flux o f h e a t and salt. T o co m p en sate for gains
and losses in the interior o f the A rctic O cean/N ordic
Seas system the two reservoirs have to couple by
exchanges th rough the F ram Strait. Figure 6 is a schem a­
tic diagram of this concept. Since th e size o f the reser­
voirs is vastly different, th e forcing o f the fluxes differs
and since non-linearities of the T S-relationships are
involved (see Fig. 3) the coupled system will react with
oscillations. N o tim e-series d a ta are available, but the
description of the “ G re a t Salinity A n o m a ly ” by D ick­
son et al. (1988) hints th a t such coupling exists: the
au tho rs have described an ano m alou s high freshw ater
accum ulation in the area n orth of Iceland in the late
1960s. It was believed to have originated from an
increased outflow o f fresh w ater from the A rctic O cean.
This outflow, which A a g aard and C arm ack (1989) esti35
O*
10° E
i
< -1 5
DBAR
-
500-
500-
1000-
1000-
-
I
0.5
0.5
1.1
2000-
2 00 0 -
0 (*C )
AUG. 1986
MAR. 1987
- 1.2
RV VALDIVIA
RV VALDIVIA
3000-
3000-
Figure 5. T e m p e r a tu r e distribution for s u m m e r 1986 (left) and w in ter 1987 (right) along 74°45'N startin g o ff at B e a r Island an d
running d u e west across the central G re e n la n d Basin. N o te th e w arm intrusion in th e interval 200 to 500 m o b se rv ed in w inter.
A R C T IC O C E A N
Q,S
G R E E N LA N D SEA
Q,S
oc
AO
- 1.2
GS
- 1.3
1980
TIME
FRAM
STR A IT
Figure 6. Schem atic r e p re se n ta tio n o f th e Arctic O cea n /N o rd ic Seas system as a two-basin system with different th erm o h alin e
fluxes due to different processes for d e e p w a te r fo rm ation . F luctuations arc k n o w n only fo r the d e e p G r e e n la n d Sea.
36
STAND. DEV.
DEGR. LONG
7881-90
r- 76-
>
H
-I
C
D
m 74-
72-
71-80
70-12
LONGITUDE
Figure 7. E x te n t o f sea ice in the G r e e n la n d Sea. Solid lines: W in tertim e (Jan to M ar) location o f ice edge for th e p erio d s 1971—
1980 an d 1981—1990. D a ta sources: 1971-1980 Vinje (1984), 1981-1990 Jo in t Ice C e n te r. D as h e d line: S ta n d a rd dev iatio n as
o b tain ed for the perio d 1981-1990. N o com patible estim ates were possible for th e 1970s. T h e cen tre for convection in the
G r e e n la n d Sea is m a r k e d at 75°N, 04°W.
m ated to rep resen t a 30% increase in th e m e an fresh­
w ater flux from the A rctic O cean , obviously did not
im m ediately affect the favourable convective conditions
in the G reen lan d Sea while passing en ro u te to the
Iceland Sea. H ow ev er, while prop ag ating from the Ice­
land Sea a ro u n d the su bp olar gyre the freshw ater excess
red u ced convection in the L a b ra d o r Sea in 1972 and,
after returning to the G reen lan d Sea in 1981, convective
conditions becam e less favourable. A similar event
occurred at th e beginning of this century. W e believe
th a t this is some indication o f the validity o f the m odel on
the generations of variability in d eep w ater properties by
th erm ohaline coupling of A rctic O cean/N o rd ic Seas
reservoirs.
T he third reason for the variability is m ore straight­
forw ard and relates to local th e rm ohaline forcing by seaice form ation. A s described in section 2, cycles of ice
form ation and melting are the surface expression of
convective activity in the G re e n la n d Sea. A ccording to
Rudels (1990) these cycles occur on time scales of the
o rd e r of several days and on spatial scales of the o rd e r of
kilom etres. T h ere are no rou tin e d ata allowing direct
observations of such cycles. Indirect indication can be
ob tained from th e routine ice charts, which have a tim e
resolution of the sam e o rd e r as can be expected for the
ice cycles and a spatial resolution which is m uch larger.
T h e m ost consistent p a ra m e te r in these charts o ver the
p eriod 1970 to 1990 is the ice edge location. In Figures 7
and 8 the ice edge an d its variance are plotted for decadal
and interann ual tim e scales.
By assuming the centre o f convective activity to be at
75°N, 04°W (G S P -G ro u p , 1990), by fu rth er assuming
that a significant portion of the ice observed in the
central G re e n la n d Sea during w inter has b een locally
form ed and by implying the observed variance in the
central G reen lan d Sea to be significantly related to ice
form ation/ice melting cycles, the figures can be dis­
cussed as follows: o n the decadal tim e scale (Fig. 7) there
was m o re ice form ed during th e 1970s than during the
1980s. T aking the variance of the 1980s to hold also for
the 1970s, then the centre o f convection was occasionally
ice-free during the w inter periods of the 1970s, but
ra th e r frequently during the 1980s. This result is consist­
ent with the observed convective activity described in
section 3, if m ore ice and m ore occasional o p en w ater
m ean m o re salt brine release into the system.
O n the interan n u al time scale (Fig. 8) we find the
central w inter m onths in 1988 and 1989 just as has been
discussed above for the 1980s. In 1987, how ever, th ere
was so m uch ice th a t the cen tre o f convection was
p erm anently u n d e r an ice cover. This correspo nds well
with Figure 5, w here th e h eat co nten t of a w arm in tru ­
sion of A tlantic w ater at interm ediate d epth could n ot be
released to the a tm osphere and consequently convective
depths w ere limited to 200 m.
Conclusions
T he foregoing discussion has indicated relationships
betw een decadal and interannual variations in the depth
37
E G -IC E
BOUNDARY
80'
MEAN
M A X -M IN
FE B 1987
MAR 1988
FEB 1989
702 0 'W
10 '
O'
E 10' 0
200
400
600 km
800
Figure 8. M onthly m ean ice edge and its longitudinal variation (m ax im u m ex ten t m inus m in im u m e x ten t) for F e b r u a ry 1987,
M arch 1988. an d F eb ru ary 1989. D a ta source: Jo in t Ice C e n te r. T h e c en tre fo r convection in th e G r e e n la n d Sea is m a r k e d at 75°N,
04° W.
o f the w intertim e convection and changes in forcing by
wind and th e rm o haline processes. O n e cann ot at pres­
ent get m uch fu rth e r along this route: the limitations on
d a ta availability and d a ta consistency are one reason.
W hereas limitations o n d a ta availability are obvious for
the ocean o bservations, the d ata consistency problem
concerns the longer term w ind and ice d ata for the
spatial scale o f th e circulation in the G reen lan d Sea.
T h ere have been significant changes in o b servation tech­
niques for sea ice and in m o d e l-suppo rted interpolation
schem es for gridded m eteorological d ata within the last
two decades. A second reason is the fact that wind and
th e rm o haline forcing are not in d e p en d en t of each oth e r.
C o nsequ en tly , progress in fu rth er u n d erstanding vari­
ability in G re e n la n d Sea convection is expected from
mesoscale circulation modelling on the o ne h and and
from co ntinuation of relevant observations in the
G re e n la n d Sea , and hopefully in the F ram Strait and the
A rctic O cean, on the oth er.
A cknow ledgm ents
This p a p e r is based on d ata exchange and discussions
am ong m any colleagues w orking on the A rctic O cean/
N ordic Seas system, notably K. A a g a a rd , E . C arm ack ,
E. F ah rbach, D. Q uadfasel, and B. R udels.
38
References
A a g a a rd , K. 1968. T e m p e ra tu r e variations in th e G r e e n la n d
Sea d e e p w ater. D e e p -S e a R e s ., 15: 281-296.
A a g a a rd , K ., an d C a rm a c k , E. C. 1989. T h e role o f sea ice and
f resh w ater in the A rctic circulation. J. G eo p h y s. R e s.. 94:
14485-14498.
A a g a a rd , K ., F a h rb ach , E ., M ein ck e, J . , a n d Swift, J. H. 1991.
Saline outflow s from the Arctic O cea n : its co n trib u tio n to the
d e e p w aters o f th e G r e e n la n d an d N o rw eg ian Seas. J. G e o ­
phys. R e s., 9 6 ( C 1 1), 20.433-20.441.
A a g a a rd , K ., Swift, J. H ., an d C a rm a c k , E. C. 1985. T h e r m o ­
haline circulation in th e A rctic M e d ite rra n e a n Seas. J. G e o ­
phys. R e s., 90: 4833-^846.
C la rk e , A . , Swift, J. H ., R e id , J. C ., an d K o lte rm a n n , K. P.
1990. T h e fo rm atio n o f th e G r e e n la n d S ea D e e p W ater:
d o u b le diffusion o r d e e p convection. D e e p -S e a R e s., 37:
1385-1424.
D ickson, R. R ., M ein ck e, J . , M a lm b e rg , S v .-A ., a n d L e e , A . J.
1988. T h e " G r e a t Salinity A n o m a ly ” in th e N o r th e rn N o rth
A tlan tic 1968-1982, Progr. O c e a n o g r., 20: 103-151.
G S P G r o u p 1990. G r e e n la n d S ea P ro ject - a v e n tu re to w ard
im p ro v e d u n d erstan d in g o f the o cean 's role in clim ate. E O S ,
7 1 :7 5 0 - 7 5 1 ,7 5 4 -7 5 5 .
Jö n s so n , S. 1991a. T h e sources o f fresh w ater in th e Iceland Sea
an d the m cchanism s go v ern in g its in teran n u al variability.
IC E S Sym p. o n V ariability, N o. 5, 12 pp. M a rie h a m n 1991.
Jö n s so n , S. 1991b. S easonal an d in teran n u al variability o f wind
stress curl o v er th e N o rd ic Seas. J. G eo p h y s. R e s., 96: C2,
2649-2659.
L e g u tk e , S. 1990. A n um erical investigation o f th e circulation
in th e G r e e n la n d an d N o rw eg ian Seas. J. Phys. O c e a n o g r.,
2 1 :1 1 8 -1 4 8 .
M alm b erg , S v.-A . 1983. H y d ro g rap h ic investigations in the
Iceland and G r e e n la n d Seas in late w inter 1971 - "D e e p
W a te r P ro je ct” , Jö k u ll, 33: 133-140.
M id ttun. L. 1985. F o rm a tio n o f dense b ottom w a te r in the
B arents Sea. D eep -S ea R e s., 23: 1233-1241.
N ag u rn y, A . P ., an d P opov, A . V . 1985. Intensive u p w e llin g o f
d eep an d b o tto m w aters an d th eir f orm ation at th e surface in
th e region o f the G r e e n la n d Hollow. M eteorologiya i G idrologya, 7: 71-75.
N an sen , F. 1906. N o rth e rn w aters: C aptain R oald A d m u n d se n ’s o ceanographic o bservations in the A rctic Seas 1901.
V id enskabs Selskabets Skrifter, I., M a t h .-N atur. Klasse.
Christiania.
Q u ad fase l, D ., an d M ein cke, J. 1987. N o te on the therm al
stru cture o f th e G r e e n la n d Sea G yres. D eep-S ea R e s., 34:
1883-1888.
Rudels, B. 1990. H alin e convection in th e G r e e n la n d Sea.
D eep -S ea R e s., 37: 1491-1511.
R u d els, B.. an d Q u ad fase l, D. 1991. C on v ectio n an d d e e p
w a te r fo rm atio n in the A rctic O cea n - G r e e n la n d S ea sys­
tem . J. M ar. Sys., 2(3/4), 435-450.
R u d els, B ., Q u ad fase l, D ., Fried rich , H .. a n d Houssais, M .-N .
1989. G r e e n la n d Sea convection in th e w in ter 1987-1988, J.
G eo p h y s. R e s., 94: 3223-3227.
Schlosser, P ., Bönisch, G . , R h e in , M ., an d B ay er, R. 1991.
R e d u ctio n o f d e e p w a te r fo rm atio n in the G r e e n la n d Sea
d uring th e 1980s: evidence from tracer d a ta . Science, 251:
1054-1056.
39