Sea Surface Temperature Patterns and Air-Sea

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 88, NO. C14, PAGES 9871-9882, NOVEMBER 20, 1983

Sea Surface Temperature Patterns and Air-Sea Fluxes During MARSEN 1979, Phase 1

In

the German Bight

KRISTINA B. KATSAROS Department of Atmospheric Sciences, U niversity of Washington

ARMANDO FIÚZA AND FÁTIMA SOUSA Grupo de Oceanografia, Departamento de Física, Universidade de Lisboa

VOLKER AMANN Institutfür Nachrichtentechnik, Deutsche Forschungs-und Versuchsanstalt für Luft-und Raumfahrt

An analysis was made of remote\y sensed sea surface temperatures (SST) obtained from aircraft and satellites and of data from hydrologica\ surveys conducted in the German Bight during the Marine Remote Sensing Experiment (MARSEN) phase I, in the North Sea August 15 to September 15, 1979. The slgnatureof a thermal double frontal structure associated with a coastal front resulting from freshwater runo/f and extending along the 30-m bottom contour at the northeastern edge of the sub­ manne glaCial valley of the Elbe river was the most prominent feature in the SST field. Air-sea ftuxes of heat and momentum were computed for the same period fram field observations by using recently developed parametenzatlOn schemes. lt was possible to group the SST patterns according to the intensity of the. wmd stress and of the net heat gain or loss by the sea. It was found that the thermal signal of the front IS more eVldent at the surface when the wind stress is greater than about 0.5 N m- 2 . In the summer of 1979 these occasions were also associated with weak heating or with a net cooling of the sea. During penods of weaker wmd stress and strong solar heating, a shallow thermocline develops which tends to Isolate the frontal cold water fram the interface and the SST pattern becomes less organized. Frontal eddles related to the barochmclty and to the current shear at the frontal zone were visible in the surface distributions of density and on the satelJite infrared imagery.

INTRODUCTION The Marine Remote Sensing (MARSEN) experiment 1979 had as one of its objectives to demonstrate the use of remotely sensed data in scientific investigations. The proper interpreta­ tion of the remotely sensed data was ensured by simultaneous conventional observations. The study reported here concerns sea surface temperature (SST) patterns observed in the German Bight during phase 1 of the MARSEN experiment, August 15 to September 15, 1979. This period of time fell within the "Year of the German Bight," when a detailed oceanographic measurement program was carried out in this area. We, therefore, have much sup­ porting information on the hydrographic structure which sub­ stantially aids interpretation of the SST patterns. There exist in the literature some examples where remotely sensed SST's have been successfully merged with hydrographic data. Good examples are the Grand Banks Experiment [La­ Via/ette, 1981J and a study of the California current [Bern­

stein et ai., 1977]. The stimulus for the detailed study of the German Bight in late summer was the hint of a thermal front in the area during that season. A double thermal frontal structure was in fact seen to be present during 1979 with varying sharpness strongly dependent on the meteorological conditions. In interpreting the SST patterns, we started out with many hypotheses as to what causes the observed features and their changes, including advection, upwelling, tidal mixing, and Copyright 1983 by the American Geophysical Union. Paper number 3CI 181. 0148-0227 j 83jOO3C-1181$05.00

forcing by wind and surface fluxes. The dynamics of how the front is formed is, however, not the central part of this study. The water masses that constitute its structure are described and discussed in Becker et ai. [this issueJ (hereinafter referred to as BFJ). What we are demonstrating with this contribution is the importance of the meteorological forcing for the thermal signature of the front at the sea surface. The meteorological forcing includes wind mixing and heating or cooling by radi­ ation and turbulent surface ftuxes. SEA SURFACE TEMPERATURE DATA

Aircraft and Satellite Data Remotely sensed SST patterns in the German Bight consist­ ed of the results obtained with an infrared radiometer (a Barnes PRT-5 operating in the wavelength band 9.5-11.5 11m), ftown at altitudes of 300- 450 m on a Dornier 28 aircraft of the Deutsche Forschungs-und Versuchsanstalt für Luft-und Raumfahrt e.V. (DFVLR) during 11 missions. The procedures for evaluating the surface temperatures are summarized in Figure 1. The airborne radiometer readings were corrected for e/fects of sky reftection (Lv) and of atmo­ spheric interference (óa) , respectively, by using simultaneous readings from an auxiliary zenith looking radiometer mounted on the top of the aircraft and by ftying repeatedly over the same portion of the sea at dilferent altitudes, generally over the forschungs-plattform Nordsee (FPN) and/ or over the re­ search vessel Tabasis, from where synchronous SST measure­ ments were conducted. The details of this procedure follow Kraan's [1977J review. In-flight calibrations of the PRT-5 were conducted with a portable blackbody during aircraft turns between the legs of

9871

9872

KATIAROS ET AL.:

I

SST

TpRT , MEASURED UNCORRECTED RADIATION TEMPERATURE

LlC-'N FLlGHT CALlBRATION CORRECTION T R {hl, RAD. TEMP. AT ALTITUDE h

~-r~

PRT,

,

,

.1

L

J

-1

l0 / /

---r

1-

T R {DI, RAD. TEMP. AT SEA LEVEL

Llr -

LJi

SK Y REFLE CTI NG CO RRECTlON EFFE CT T S , SKIN TEMP.

• INTERFACE EFFECT T B. BULK TEMP.

TB = TpRT

+

LlC + LlO + Llr +Lli

with selr-recording current meters were maintained in the area. The Friedrich Heincke covered the area with stations made with a conductivity-temperature-depth (CTD) system from August 14 to 20. An almost simultaneous cruise was conduc­ ted with the Gauss (August 16-20) along a much narrower sampling grid. The grid points are indicated in Plate I and 4. The Gauss again covered part or the study area August 21 - 23, and John Murray made a CTD cruise immediately afterward (August 23-26). The Tabasis also conducted some hydrologic observations during August 24- 26, while moving slowly through a large sampling grid. The John Murray made an­ other cruise between August 31 and September 3, rollowing a lmost the same pattern as in its previous cruise. The Tabasis operated in the area between September 4 and 9. The ship data were used to produce surface maps of temper­ ature, salinity, and density. The Tabasis charts show very little detail and will not be used here. The other distributions are displayed and discussed below. Vertical sections or hydrological parameters corresponding to the same data base are found in the companion paper (BFJ) where the water masses in the German Bight are analyzed. Mitt e/staedt and Soetje [1982] described the residual currents meas ured during MARSEN phase I with the moored current meters. AIR SEA FLUXES

OBSERVED MAGNITUDES

CORRECTION

PATTERNS IN GERMAN BIGHT

Ll C

PORTABLE BLACKBOOY SOURCE

lIc,. 0 .1' lIT omb

LlO

VERTICAL PROFILES

lIo· 0. 1,._,o.5°C

Llr

ZENITH VIEWING PRT2

~r

Ll i

SURFACE OBSERVATIONS

lIi '" O.zo C

• 0 .1, ...,0.6 °C

Fig. 1. Scheme of sea surface tempera ture evaluation by airborne radlOmetry on the Dornier 28 aircraft ; tJ.r is the correction for the sky radIance reflected by the sea in ta the radiometer, Lla is the correction fo r atmospheric absorption and emissio n, LlC is the correction for chopper emission, Lli is the correction for the interfacial boundary layer temperature gradient, and Ll T.mb i~ the change in ambient tem­ perature of the radiometer.

Ca/cu/ations of Momenlum, Heal, and Vapor Fluxes Through the Surface Layer

Direct meas urements of turbulent surface layer fluxes were not made during phase I of MARSEN. However, the formulae below (which incJude the bulk aerodynamic formuJae) express the fluxes of momentum, hea t, and vapor: 2

(Ia)

r = PIOCD(Ü10 - üs )2

(lb)

r=PIOu.

H = PlOCpCH(ÜIO -

the flight pattern. These were later used for correcting the radiometer readings for the effect of the ambient temperature change; this occurs beca use the reflection coefficient of the PRT-5 golden chopper is only about 90% [Lorenz, 1973], and, as it is located outside the temperature controlled refer­ ence chamber of the radiometer, it offsets the calibration bya ractor (t. c) of 0.1 times the ambient temperature change (t. T.mb) [Kraan, 1977]. The interfacial effect (skin temperature being different from bulk temperature) was removed from the airborne records by using the FPN or Tabasis measurements contemporary with the overfiights; this effect (t.i) amounted to about 0.2°C. The corrected records were then divided in 0.25°C ranges, starting from the lowest observed temperature in each flight, and hand-contoured SST charts were drawn from them. Thermal infrared images (A VHRR) obtained at the Univer­ sity of Dundee ground receiving station, both with the NOAA 6 and TIROS N satellites, were also available for a few oc­ casions when cloud-free conditions prevailed over the study area. Ship Data

During MARSEN phase I, several research vessels conduc­ ted hydrological surveys in the German Bight and moorings

(2a)

H = PlOcpu.8*

üsXes -

elO)

(2b)

E=PlOU.q*

(3a)

E = PlOCe(ü1 0 - ü,)[qsCT.) - ql o(TlO )]

(3b)

where r is the momentum flux (or s urface stress), H is the heat flux, E is the vapor flux , PIO is the air density 10 m above the sea surface, u'" is lhe friction or scaling velocity, 8* is the scaling potential temperature, q* is lhe scaling specific humidi­ ty, ÜlO is the wind speed at 10 m, Üs is lhe sea surface current speed (in the direction or the wind speed), cp is the specific heal of moist air at constant pressure, s is the potential tem per­ ature of air at the sea surface, li 10 is the potential temperature of air at lO m, q,(7;.) is the specific humidity of air at the sea surface at temperature f" and qlO(Tlo ) is the specific humidity of air at 10m at temperature TIO' Overbars denote hourly means. C D is the drag coefficienl or bulk momentum transfer coefficient, and C H and C e are the bulk heat and vapor trans­ fer coefficients, respectively. Hourly observations of the bulk sea temperature at 7 m depth and of wind speed, air temperature, reI ative humidity, and air pressure at 46 m were recorded at the Nordsee re­ search platform (FPN) at 54°42.5'N, 7° 1O.3' E. To adapt these observations to the requirements of the above formulae (namely, that the mean quantities apply at 10 m height in a

e

KATSAROS ET AL.:

SST

neutrally stratified atmosphere), the following assumptions are made: 1. The logarithmic, sta bility adjusted , surface layer ftux­ profile relations [Businger et ai., 1971] hold for mean wind, mean potential temperature, and mean specific humidity as follows:

(4)

(5) ifz - ifs = _1_ [ln(z/ zQ) -1/I2(z/ LuJ]

Ekk

q*

(6)

where the subscript z refers to height Z above the sea surface ; subscript s refers to the sea surface; k is the von Ká rmán constant; Zo is the roughness length, or the intercept of the wind profile; ZT is the intercept of the potential temperature profile; zQ is the intercept of the specific humidity profile; Ek is the ratio of eddy diffusivity of momentum to eddy diffusivity of heat, KM/K H ; 1/11 and 1/12 are the Businger-Dyer stability functions discussed by Businger et ai. [1971]; and Lv is the virtual Monin-Obukhov length. The Monin-Obukhov length is a height scaling parameter which depends on the balance between buoyancy a nd shear-produced turbulence. It changes sign from positive for stable to negative for unstable con­ ditions. 2. Here, Zo can be parameterized by Charnock's relation, Zo = IXU* 2/ g, where g is the acceleration due to gravity and IX-Charnock's constant-is chosen as follows : Under neutral conditions (so that 1/1 I = O), combine Charnock's relation and (la), (Ib), and (4) to get IX = C (D

10g _

ul O

~

Us

)2 exp(-k/ YC D

(7)

Smith [1980] gives a regression formula for C D on ulO for neutral conditions of C D = (0.61 + 0.63 tlI O) x 10- 3. Substi­ tuting this regression into the expression for IX gives IX as a weak function of Ul O and us ' For a representative selection of values for Ul O a nd Us (see assumption 8 also), a mean value for IX can be computed and used in Charnock's relation. The mean IX used was consistent with Smith's expression for neutral C D . One must also choose a value for k; 0.4 was used in these calculations. 3. Either ZT is set equal to Zo, and C H is given by (2)-(6) or C H is specified from experiments, 8. is found from (2b), and ZT is found from (5). For example, Smilh [1980] found C H = 0.00083 under stable condiiions and C T = 0.0011 under unsta­ ble conditions. These values were used under assumption (3b) for the MARSEN ftux calculations. 4. Here, zQ is equal to ZT' 5. The air temperature at the sea surface is assumed equal to the bulk sea temperature at 7 m depth. 6. The air at the sea surface is saturated. 7. The air above the sea surface is not saturated (so that rising and falling air parcels change temperature dry adia­ batically). 8. The sea surface current speed (us) in the direction of the wind is 3% of the wind speed at 10 m. (This assumption is an approximation in view of the relatively strong tidal ftows in the study area but is adequate in comparison to the overall accuracy of the present bulk parameterization scheme [see further Geernaert, 1983].)

9873

PATTERNS IN GERMAN BIGHT

9. The hourly observations at FPN represent hourly means. Though the bulk sea temperature is measured to ± 0.05°C, assumption 5 may be reasonable because the air temperature is measured only to ±0.5°e. Assumption 8 is made on the basis of wind tunnel studies. With FPN observations, Charnock's relation, expressions for the virtual Monin-Obukhov length and the Businger-Dyer stability functions, and (2b) for {)*, (4)-(6) represent three si­ multaneous nonlinear equations in u.' 010 , and q •. These equations can be solved iteratively by repeated substitution given first guesses for u.' Blo, and q •. Except under stable conditions with low wind speeds, the iterations converge nicely almost regardless of the first guesses. Given the iterated solutions for u.' 6*, and q* , the mome~tum, heat, and vapor ftuxes can be computed ftom (la)-(3a) . Details of the algo­ rithm can be found in Kat saros and Dempsey [1983]. Calculation of lhe Radiative Heat Fluxes

The radiative heat ftuxes were calculated from hourly obser­ vations at Helgol a nd Island (54°25'N, 7°50' E) and synoptic upper levei data as follows: For insolation the scheme of Lumb [1964] was used. This empirical formulation was devel­ oped with measurements and cloud observations from the weather ship Juliell e at 52°30'N, 20 0 W and tested with data from Alfa (62°N , 33°W). One might expect these two stations to have a somewhat simílar radiation climate as the Helgol­ and station. Direct measurements of short-wave irradiance made on the J ohn Murray , August 20 to September3 were about 30-40 W m- 2 lower in the daily average. However, on some days (August 30, September 3) when the ship operated in the vicinity of Helgoland, the agreement is within 1 W m - 2 . The high incidence of cloudiness in the German Bight would make insolation very variable over the area. Shortwave albedo was considered by applying an analytical expression fitted to Payne's [1972] data [see Lind el ai., 1983]. The longwave exitance was simply calculated by using the Stefan-Boltzmann equation with infrared emittance of water taken as 0.97. The longwave ifradiance scheme is that of Lind and Katsaros [1982]. This is a physical model essentially with­ out empirical constants although a somewhat subjective choice of the value of cloud emittance is made. This scheme employs the World Meteorological Organization (WMO) hourly observations at surface statÍons. Upper levei synoptic charts are used to develop vertical temperature profiles, which are used in estimating cloud base temperatures, and the irradi­ ance from the moist atmospheric boundary layer. The infrared irradiance parameterization was tested against measurements obtained in the Joint Air Sea Interaction (JASIN) experiment off Scotland in the sa me season as MARSEN, phase 1, and is therefore expected to provide a reasona ble estimate for the conditions existing in the German Bight. The net radiative balance E ne1 is obtained from E nc1 = Esl

-

IXE sl -

Mil

+ aE 11

(8)

Esl is solar irradiance, IX is albedo, Mq is longwave exitance and equals 8(JT,4, where T, is the interface temperature and (J is the Stefan-Boltzmann constant; 8 is emittance of the sea surface taken as 0.985, and a is the sea surface absorptance of longwave irradiance from the sky EI taken as 0.964 (from spectral measurements by Mikhay lov and ZolOlarev [1970]). Here, e and a are slighly different from each other because of the different wave1ength regions in which the sea surface and sky radiate due to different absolute temperatures.

9874

KATSAROS ET AL.:

SST PATTERNS IN GERMAN BIGHT

AUGUST 1979

SEPTEM8ER

~ tWE14ST15 . 16FPN· 17 18 19 20 21 22 23 2425 26 27 28 293031 I 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 . . . . . . . . . . . . . . . . . . . . . . . . . . . . ..

~

i ?fr1~,,\\~À~~'-' ~\\\\l'r>~"~T~U~a Fig. 2a.

Wind speed and direction at the No rdsee Platform in the middle of the study area displayed as a "stíck" diagram with the meteorological convention (i.e, the sticks extend toward where the wind comes from).

Since the radiative fluxes are calculated by using parame­ terization schemes with mean parameters from one station only, the results will not be representa tive of the fluxes oc­ curring from hour to hour at any particular location in the area. However, by taking daily averages, discrepancies are ex­ pected to be reduced. In this study it is the relative magnitudes and the sign of the net heat flux which is of interest. METEOROLOGICAL CONDITIONS

During the first phase of MARSEN weak atmospheric pres­ sure gradients, typical of summer conditions, prevailed Over the German Bight, the area being on the average inside the northernmost extension of the subtropical high. This pattern was broken by the passage of an intense low pressure center, August 26-29, with WNW winds as high as 20 m S-1 (see Figure 2a). From September 10 the lcelandic low was the prominent feature controlling the atmospheric circulation over the North Sea, bringing a sequence of frontal systems with strong winds over the area. During the period of August 15 to September 10 there Were several intervals of relative calm and strong net heat fiux into the ocean alternating with windier periods when the heating was much less or the net heat fiux resulted in some cooling of the sea. The calculated wind streSS and net heat flux, Sea to air, 300

200 W

;;;-z

NET HEAT FLUX

100

-100

-200

.3

WINO STRESS

N mZ

.2

.1

15

AUGUST

30 I

SEPTEMBER

15

Fig. 2b. CaJculated daily averages of wind stress and net heat ftux across the air-sea interface for the period August 15 to September 15, 1979, In the German Bight central region. Positive net hea t ftux indi­ cates cooli ng of the sea.

averaged OVer the day are plotted in Figure 2b. The following discussion of the SST patterns will focus on a few periods of time when the meteorological conditions remained approxi­ mately constant. These periods are marked ori Figure 2b. August 16-20 was characterized by weak wind stresS and strong solar heating. The period August 21-26 had a moder­ ate wind stress and much less net heating, < 100 Wj m 2 • A third period is the stormy period during August 27 and 28 when strong wind streSS neady 0.3 Wjm 2 , and over 100 W j m 2 of net cooling occurred. The fourth period is August 29 to September 2 which has extremely light winds with strong solar heating. During the fifth period, September 3-12, variable conditions prevailed. ATMOSPHERIC FORCING OF SEA SURFACE TEMPERATURE PAT­ TERNS AND ASSOCIATED HYDROGRAPHIC STRUCTURE

The thermal surface features observed in the German Bight during MARSEN phase 1 showed essentially two different patterns which were related to air-sea fluxes. During periods of Iight winds (wind stress under 0.05 N m - 2, say), which were generally related to c1earer skies and to a net heat gain by the Sea (heat fluxes into the Sea exceeding about 100 Wm - 2), a patchy structure tended to prevail in the SST field, with patch dimensions of 20-50 km and with some tendency for zonal patterns to dominate Plates 1 and 3b and 5b). On the other hand, under moderate to strong winds (wind streSSeS of 0.1 N m - 2 or greater), corresponding to weak net heat gains by the sea or to heat losses, a more organized SST pattern was ob­ served; this consisted of a NNW-SSE oriented band of low temperatures a few tens of kilometers wide, extending from the northern Iimit of the area under study, up to the vicinity of Helgoland Island (Plates 2a, 2b, and 3a). On both sides of this feature relatively strong thermal gradients of the order of 1°Cj 5 km characterize this feature as a double thermal front. This front was very nearly aligned with the 30-m bottom con­ tour line defining the northern limit of the glacial Elbe river estuary, thus indicating topographic control. (The topography is mapped in Figure 4.) From the vertical cross sections of hydrologic parameters presented by BFJ one can conclude that the SST front is a surfacc manifestation of a frontal structure which separates coastal fresher and lighter water, with a strong influence from the discharge of the Elbe river, from saltier and heavier ofT­ shore North Sea water. The water of the frontal zone is itself colder and has an intermediate salinity relative to inshore and offshore waters and, as was shown by BFJ, constitutes a separ­ ate water mass which is formed by mixing of Elbe estuary water with North Sea deep water. In the period August 16-20 when there was a net heat gain by the sea (Figure 2b), the thermal signal of the front at the surface weakens and almost disappears (Plates 1 and 3b) due to the formation of a shallow thermocline only about 10 m deep. During subsequent events of modera te to strong net heat loss periods (Figure 2b), the thermal stratification eroded and the front extended to the surface in full strength (Plates 2 and 3b).

KATSAROS ET AL.:

SST

PATTERNS IN GERMAN BIGHT

(o) FR. HEINCKE­ ' 14- 20 AUG. 1979

55~~~~~+-~~H~-~

°N

17

./. .

Jielgo ond , •

.~~ .. :-~ .:/

54~---+--~---+-­

(C) JO N MURRAY 31 UG. -3 SEPT. f'JPIP'_~(7r?~

Plale L

Sea surface lemperalure, salinilY, and densily (in sigma t units) dislributians abtained in lhe German Bight during lhe summer af 1979 under canditians af weak wind slress a nd net healing af lhe sea.

9875

541

541

°N

551

f"A - " )U'

, I

I ~

SOE

q

T-oC

I

"I

"I

p:..-~j

7°E

7°E

_ q

SOE

SOE

_ e:r=- 0:t7' ..",.

7°E

7°E

q

SOE

q

SOE

__ e:r=- 0:t7' ..",.

Pia te 2, Sea surrace tempera ture, salini ty, a nd d ensity (in sigma t units) distributi ons o btained in the G erma n Bight during the summer or 1979 und er co nditio ns or m odera te to stro ng wind stress a nd net hea t loss o r small hea t gain by the sea,

7°E

(b) JOHN MURRAY 23-26_AUG . 1979 _ ou­

17

. r .,

7°E

SOE

21- 23 AUG, 1979

(o) GAUSS

I

O'T

\O

00

:t ...,

ã

tI:!

Z

>

m )ó 3:

C)

z '"Z

)ó

~

>

'1:1

-I

{/} {/}

r

~ >

Sl

)ó

~

Ã

>

a-­

......

KATSAROS ET AL.:

9877

SST PATTERNS IN GERMAN BIGHT

55r-------~~~

°N

(a) Helgolond ..

54 f - - - - - - - f - - - - - - - - - - j 22.AUGUST,

24 AUGUST

-­

26 AUGUST

7°E

-­

28 AUGUST . 1979 '"

55r-------~~~

°N

(b)

16.2 ...:

_I

29 .AUGUST . 1979

31 AlJGUST 1979

SEPTEMBER 1979

54r-----~~~

3 SEPTEMBER 1979

4 Sf::PTEMB~R 1979

5 S~P.TEMB~R 1979

Plate 3. Airborne sea surface temperatures obtained on eleven missions on the dates indicated (a) under moderate to strong wind stress and net heat loss or small heat gain by the sea, (b) under weak wind stress and with net heating of the sea, (e) under intermediate conditions.

9878

KATSAROS ET AL.:

SST PATTERNS IN GERMAN BIGHT

(a)

TIROS- N 9 SEPT ' 79 14 : 11 GMT Fig. 3.

Infrared images ablained an NOAA 6 and TIROS N salelliles. (a) Examples af lhe signalure af lhe Gerrnan

Bighl franl rnissing. (b) Examples af lhe Gerrnan Bighl franl being recagnizable.

Owing lo lhe strong salinily contrast between the main water masses in presence (Elbe estuary water and North Sea water), the field of mass in the German Bight is essentially determined by the salinity distribution. The density field was thus almost insensitive to the thermal evolution of the upper layers of this coastal sea during the study period. Indeed, and as Plates 1 and 2 illustrate clearly, the surface salinity and sigma t distributions did not present any major variations for the duration of the experiment; their main feature consisted in a strong, one-sided front coinciding with the thermal front described above. From the vertical sections presented by BFJ, it is obvious that the frontal zone separates inshore and offshore waters both with little or no stratification. However, the frontal zone corresponds itself to strong vertical and horizontal gradients in the analyzed fields of temperature, salinity, and density. The band of stronger density gradients, defined by the layer con­ tained between the isopycnals of 23.0 and 23.5 sigma t units (bright blue in Plates 1 and 2) coincides with the inshore half of the core of colder temperatures of the thermal double fron­ tal structure. This peculiar thermal structure is due to the "frontal water" mass being colder but of intermediate salinity relative to the neighboring water masses as defined by BFJ, who also discuss the stratification in the frontal zone. The individual SST distributions show some interesting fea­ tures. As mentioned above, the calm wind, net heat gain by the sea situation was characterized by mesoscale patchiness in the SST fields; this designation should perhaps be clarified by some examples. Comparing the SST charts in Plate I (low wind stress) with those in Plate 2 (moderate to high wind stress), one could imagine that a uniform heating of 1°-2°C of the structure shown in the latter would lead to distributions not toa dissimilar from those in Plate I. Indeed, bands of lower surface temperatures still can be identified both in Plates la and lb in the area of the front, and in Plate 3c the cold patch near Helgoland corresponds to a local maximum of frontal water, as can be seen in the percentage distributions

of water masses presented by BFJ. Lower temperatures were also observed at that same location during situations of strong wind stress and net heat loss from the sea (Plates 2b and 3b). The tendency for zonal patterns noted in the ship's SST charts under calm wind conditions may result to some extent from the east-west orientation of the lines of stations occupied during ali of the cruises. An interesting pattern, observed in the surface density fields shown in Plates 1 and 2 (particulary in those obtained with the high resolution grid of the Gauss) is apparently due to the presence of frontal eddies with individual sizes of the order of 10 km and with a separation scale of 20-40 km. One might argue that these features could result from tidal fluctuations of the frontal structure between consecutive sampling lines; how­ ever, the infrared satellite pictures of Figure 3b also show frontal eddies with the same spatial scales. These suggest that the frontal structure observed in the German Bight during the summer is affected by baroclinic instabilities. The frontal eddies may also contribute to the observed thermal patchi­ ness. The airborne SST maps shown in Plate 3 illustrate in some detail the day-to-day variability that took place within periods with different meteorological forcing in spite of the reduced area covered in each flight and of their somewhat irregular coverage. The sequence in Plate 3b shows the progressive warming of the sea surface lInder persistently calm winds right after the isolated storm of AlIgust 27 and 28. A considerable amount of small-scale strllcture is observed in the aircraft­ produced charts, which is apparently related to the almost continuous sampling along the flight Iines; these strllctures are very similar to those obtained with the fine resollltion of the Gauss data (Plates Ib and 2a) and thus do not seem to result from a possible residual inflllence of differential atmospheric interference or scattered clolld reflection in the radiometric records. The sequence of SST charts in Plate 5a shows the evollltion of the intensity of the dOllble thermal front north­ west of Helgoland under moderate to strong wind conditions.

KATSAROS ET AL.:

SST

S\ ..,..-oUll ine of double thermol (b)

9879

PATreRNS IN GERMAN BIGHT

\

Fig. 3.

(continued)

f ro"1

9880

KATSA ROS

ET

AL.: SST PATTERNS IN GERMAN BIGHT

The potential of airborne infrared radiometery is quite well demonstrated here, particula rly when comparing the detail and the strength of the fronta l structure observed in the ai r­ borne SST chart of August 24 with the coarse resolution of the corresponding J ohn Murray SST map shown on Plate 2b. The last three ftights provided the sequence in Plate 3c and illus­ trate a transitional situation during a 3-day cycle of intensi­ fication and decay in wind stress and of a slight increase of the heat gain by the sea (Figure 2b). Because of the general trend observed in the three SST charts, we conclude that vertical mixing induced by the wind stress prevailed during this partic­ ular period. In Figure 3 the TIROS N and NOAA 6 infrared pictures ava ila ble for the MARSEN phase 1 period are shown. The two typicaJ situations characterized above are clearly illus­ trated with the satellite imagery. The pictures obtained on August 18 a nd on September 9, both corresponding to very similar situations of Iight winds and with net heat fluxes into the sea, show no trace of the thermaJ front. However, the images o btained on August 29, right after the summer storm, and on September 6, under a pulse of moderate wind stress, show c1early the thermal front as a meridional band of cooler water about 50 km wide, with its sides distorted by the smaJl scale eddies already described. The morning images (NOAA 6, around 0830 GMT) show the front somewhat betler than those obtained in the early afternoon (TIROS N, around 1430 GMT); this may be due to instrumental diJferences or to skin effects related to the solar warming during those relatively cloud-free days. RESIDUAL CIRCULATION

The dynamical interpretation of the frontal structure cannot be sought in models based on tidally induced mixing on the shallow water side of stratified waters [e.g., Simpson and Hunler, 1974; Pingree el ai., 1978], as in the area under study the water column was essentially unstratified on both sides of the frontal zone. Instead, the gravitational circulation induced by the salinity-density gradients deriving from land runoff, coupled with the topographic control of the residual (i.e., non­ tidal) circulation, apparently determine the dynamics of the front. Other mechanisms due to upwelling, mesoscale eddies, long waves, internai tides, and waves are ruled out beca use of the vertical homogeneity away from the front and the re­ strictive configuration of the coastline. The diret action of the wind stress or the resulting effects of wind setup a lso do not appear to determine the strength of the front , which rem ained essentially unchanged for the duration of the MARSEN phase 1 experiment (as exemplified by the sigma I fields in Figures 3 and 4) under fairly large variations of wind direction and in­ tensity. A description of the residual circulation in the German Bight as observed during the ex periment is given below in an attempt to contribute to the understanding of the dynamics of the front. Ali the charts of residua l currents in the North Sea, based on observa tions or on numerical models, show a northerly coastal f10w in the Germa n Bight a rea [see, for exa mple, the review by Backhaus, 1980]. The current measurements con­ ducted during MARSEN phase 1, with two moorings loca ted near the coast in depths of 10 a nd 20 m, provided the residuais illustrated for positions WT and ST in Figure 4, which show some variability. This may be due to strong dependence of the circulation in such shallow a reas on the wind. The results obtained by Backhaus [1980J with a wind-driven multilayer

model including tidal and density induced f10ws indicate tha t northerly winds of 10 m S - 1 are capable of reversing the whole residual circulation in the German Bight. Although winds of such an intensity were only observed for a short time during the study period (Figure 2a) , the shallow areas a re certainly more sensitive to the mechanical forcing by th e wind and the variability observed there is not surprising. Figure 4 shows that the residual currents near the surface an d close to the bottom were consistently northward a nd had almost no vertical shear at the moorings installed in total depths of 40 m or more to the west of the front (positions PU, WB, A). How­ ever, the current meters located in the central part of the area, between depths of 30 and 40 m in the glacial Elbe ri ver valley to the west and northwest of Helgoland (positions FDB, ZE, and BG), displayed strong shea rs between near surface a nd bottom residuais, the FDB position even providing a near­ bottom southward f1ow, while near the s urface the current was predominantly to the north. These observations generally agree with the tide and density-induced residual currents o b­ tained by Backhaus [1980J with his multilayer numerical model. He shows a strong shear between a general cyclonic near surface ftow and a dee p southward circuJa tion, stronger near the northern margin of the submerged valley a nd weak­ ening oJfshore. The currents meas ured at locations B and FPN (Figure 4) cor responding to the position of the core of the frontal zone, showed vertical speed shear (at B) or strong variability in direction (at FPN). These confirm the baroclinic character of the front as opposed to the pronounced baro­ tropy of the offs hore northwa rd residual f10w (as observed at positio ns PU, WB, A). The mechanisms mai ntaining this front appear to have some similarity with those producing a coastal front off South

,'.