High frequency variability of the surface layers in the Skagerrak during SKAGEX

Transcription

1 Continental Shelf Research 19 (1999) 1021}1047 High frequency variability of the surface layers in the Skagerrak during SKAGEX Bo Gustafsson* Department of Oceanography, Earth Science Center, Go( teborg University, Box 460, S Go( teborg, Sweden Received 19 September 1997; received in revised form 5 November 1998; accepted 15 December 1998 Abstract This paper presents spatial and temporal distributions of freshwater height and potential energy in the Skagerrak during the period of Skagerrak Experiment (SKAGEX). In SKAGEX extensive quasi-synoptic hydrographic measurements were taken during one month. These "elds provide a basis for an examination of the dynamics of the surface layer. Qualitatively they show a time-series of destruction and re-establishment of the cyclonic circulation. The response to changes in the magnitude and direction of the wind is very rapid. It appears to take approximately one week to develop a strong cyclonic circulation from a situation with very weak baroclinic currents. Freshwater in#ow from the Kattegat calculated from recording current meters correlates well with the changes of total freshwater volume within the Skagerrak. In#ow of freshwater-in#uenced water along the Jutland coast was during this particular period of equal importance in re-"lling the Skagerrak freshwater pool. By tracking an outbreak of freshwater from the Kattegat along the coasts of Skagerrak, it is qualitatively veri"ed that the freshwater #ow within the coastal currents is in geostrophic balance and proportional to the freshwater height squared. Another estimation using the current meter data shows that the in#ow of potential energy from the Kattegat may give a substantial contribution to the total potential energy of the Skagerrak, at least during calm weather conditions. The in#ow of potential energy along the Jutland coast was also of importance during the SKAGEX period. However, comparing the conditions of the SKAGEX period with a previous calculation of the pro"le potential energy (Gustafsson and Stigebrandt, Journal of Sea Research, 35, 39}53), it is concluded that the advection of potential energy through the open boundaries is not strong enough to explain the high pro"le potential energy content during winter-time. Thus, the in#ows seem to be of major importance for the energetics of the Skagerrak during the calm summer period, but of minor importance during the windier winter Elsevier Science Ltd. All rights reserved. Keywords: Skagerrak; SKAGEK; Circulation; Freshwater; Potential energy * Corresponding author. Tel.: ; fax: ; bogu@oce.gu.se /99/$ - see front matter 1999 Elsevier Science Ltd. All rights reserved. PII: S ( 9 9 )

2 1022 B. Gustafsson / Continental Shelf Research 19 (1999) 1021} Introduction The Skagerrak is located in between the North Sea and the Baltic Sea (see Fig. 1). The basin is rather deep, with an average depth of 210 m (e.g., Svansson, 1975) it is considerably deeper than adjacent parts of both the North Sea and the Kattegat. The largest portion of the water body is of northern North Sea or Atlantic origin, entering north of Jutland. Trapped on the steep slope of the Norwegian Trench, it makes a cyclonic turn around the basin. The strength of this circulation is about 5 10 m s (Rodhe, 1987; Rydberg et al., 1996). The upper 10}100 m is strongly in#uenced by freshwater emerging from the Baltic Sea (& m s, BergstroK m and Carlsson, 1994; Omstedt et al., 1997), local rivers (&2000 m s, Svansson, 1975) and the southern North Sea (&1500}4500 m s, Rodhe, 1987; Rydberg et al., Fig. 1. Map of the study area.

3 1996). The in#uence manifests itself in strong haline strati"cation in the upper layers of Skagerrak. The freshwater emerging from the Baltic Sea is mixed with high saline water in the Kattegat and in the straits of the Baltic sill before it enters the Skagerrak. The in#ow of freshwater in#uenced surface water to the Skagerrak from the Kattegat is regulated by a geostrophically balanced front, the Skagerrak}Kattegat front, usually located from the northern tip of Jutland across to the Swedish coast (e.g., Stigebrandt (1983), Rodhe (1996) or Jakobsen (1997)). The mean circulation of the freshwater-in#uenced layers in Skagerrak is, as the deeper circulation, cyclonic. However, the surface circulation is mainly forced by the variable local winds (Gustafsson and Stigebrandt, 1996, hereafter GS96). The winds do not only force turbulent vertical exchange (entrainment) with deeper layers, but also Ekman #ows through the open boundaries as well as downwelling at the coasts. The latter mechanism appears to be the most important forcing of the cyclonic circulation (GS96, see also Gustafsson, 1997). The mean circulation of denser waters appears to be forced by the estuarine circulation of the freshwater-in#uenced layers, while direct in#uence of the winds only contributes to variability on quite short time-scales (Rodhe, 1996). Thus, the primary forcing of both the freshwater-in#uenced layers and deeper layers of the Atlantic water appears to be local, i.e., the wind forces motion and entrainment in the surface layers and the entrainment forces the circulation of the deeper layers. The strong impact of the winds makes the appearance of the freshwater-in#uenced surface layer quite variable on short time-scales (e.g., Aure and Svtre, 1981; Gustafsson, 1997) as well as on the seasonal time-scale (Svtre et al., 1988 or GS96) The SKAGEX measurements B. Gustafsson / Continental Shelf Research 19 (1999) 1021} In the years 1990 and 1991, the international SKAGEX project supervised by ICES was launched. It consisted of four multi-ship surveys: SKAGEX 1 in May}June 1990, SKAGEX 2 in September 1990, SKAGEX 3 in January 1991 and SKAGEX 4 in May The "rst, SKAGEX 1, was by far the largest with participation of 17 research vessels working at sea for a full months time. The large hydrographic data set collected during the SKAGEX 1 experiment consists of a series of almost synoptic and densely spaced observations allowing the construction of three-dimensional pictures of the state of the Skagerrak (see Fig. 2 where the main station net is shown). The main station net was essentially visited every third day. In total, the hydrographic data set includes some 1400 vertical pro"les of salinity and temperature measurements. The hydrographic data is complemented with recording current meter data from moorings in a section across the northern Kattegat and another across the western Skagerrak (sections A and G, respectively, in Fig. 2). The current measurements in the northern Kattegat are used in the present paper. This section, between Jutland and Sweden, was covered by 15 current meters distributed on six stations (see Fig. 2). All of the current meters worked during the period May 24 to June 10, thereafter two of the current meters malfunctioned. Several specialized studies of physics, biology and chemistry were also conducted during the "eld program.

4 1024 B. Gustafsson / Continental Shelf Research 19 (1999) 1021}1047 Fig. 2. Map of the Skagerrak. The basic station net of SKAGEX is indicated. The recording current meters n the Kattegat were deployed at the stations of section A and the current meters in the Skagerrak were located along section G. The program and some results from the hydrographic observations during SKAGEX 1 are presented in two papers, Danielssen et al. (1991,1997) (hereafter D91 and D97). In D91, it was concluded from the measurements that strong northwesterly winds not only block the out#ow of the Norwegian coastal current, but also block the in#ow to the Skagerrak of Jutland coastal water. In the situation of strong northwest wind intense upwelling was found o! the Norwegian southwest coast. Former Skagerrak surface water was also found as far as west of Hanstholm (Fig. 2). Quite persistent cyclonic and anti-cyclonic eddies in the northern Kattegat were revealed from drifter tracks. In D97, the blocking of the in#ow of the Jutland coastal water is analyzed somewhat in detail. It was noted that Jutland coastal water of high nitrate concentration was found on several occasions o! Hanstholm, but high nitrate concentrations were never found o! Hirtshals during the SKAGEX 1. From recording current meter measurements it was concluded that the currents close to the Norwegian coast were almost of uniform strength at all depths until June 5, there after the #ow was more con"ned to the less saline Norwegian coastal current. In the middle of June the #ow of high saline Atlantic water along the Norwegian coast ceased completely About the present investigation In GS96 the long-term average and annual variations of the freshwater height and pro"le potential energy distributions in the Skagerrak were examined. They

5 B. Gustafsson / Continental Shelf Research 19 (1999) 1021} demonstrated that these quantities provide a powerful tool for both quali"cation and quanti"cation of the dynamics of the freshwater-in#uenced surface layers. However, the analysis in GS96 was based on historical data scattered in time and space, and the results were in the form of long-term average and standard deviation "elds. Only along the section between Hirtshals and Torungen, see Fig. 2, there were enough of measurements to examine the seasonal cycles. The unique spatial and temporal resolution of the hydrographic data set from SKAGEX 1 makes it possible to follow the variations of the "elds of freshwater height and pro"le potential energy every third day during the month of the experiment. The coverage is frequent enough to calculate time-series of spatially integrated freshwater and potential energy content in the Skagerrak. The rates of change estimated from the time-series are compared with estimates of the exchange with adjacent seas. In Section 2, the concepts of freshwater height and pro"le potential energy are introduced. An Appendix complements the Section where the relationship between these quantities and baroclinic pressure variations are derived. It is also shown that they may be interpreted as stream-functions for baroclinic geostrophic motions. In Section 3, the results of calculating the freshwater height and pro"le potential energy from the SKAGEX 1 data set are presented. Budgets of freshwater and potential energy content in the Skagerrak based on the time-series of content and estimates of exchange with adjacent seas are presented in Section 4. The paper ends with a short discussion of the "ndings of the present investigation. 2. The quantities freshwater height and pro5le potential energy 2.1. Dexnitions The buoyancy of the surface water in the Skagerrak is determined essentially by its content of freshwater. Temperature plays only a minor role. The fact that the temperature usually is rather horizontally uniform implies that its in#uence on horizontal buoyancy gradients is still less. Being proportional to the total buoyancy of the surface water (if temperature e!ects are neglected) the amount of freshwater in a vertical column is a variable of great dynamic signi"cance in the Skagerrak. The freshwater height, F, is de"ned as F" 1 S max[s!s(z), 0] dz (1) where S is a reference salinity and S(z) is the salinity at the depth z. In order to avoid negative contributions, the integrand is kept positive by setting the minimum value to zero. Preferably, S should be the salinity of a water mass una!ected by freshwater supplied regionally. In the Skagerrak a reference salinity of 35.0 psu is a reasonable choice since it is close to the salinity of the Atlantic and northern North Sea waters that form the water masses below the surface layers of Skagerrak.

6 1026 B. Gustafsson / Continental Shelf Research 19 (1999) 1021}1047 The potential energy of a water column or water mass is a measure of its ability to force motion. Potential energy may increase not only due to horizontal convergence in the upper layers but also due to vertical mixing and is thus a highly dynamic variable. The pro"le potential energy, P, is the excess potential energy built up in a strati"ed water column as compared to one with uniform density. P is de"ned in the following way, P" g ρ max[ρ!ρ(z), 0]z dz (2) Here g is the acceleration of gravity. As the reference density, ρ, we take the density of water of salinity 35 psu and temperature 83C (Atlantic Water). The density, ρ(z), is calculated by taking into account both salinity and temperature e!ects. The temperature e!ect is included mainly because of completeness. For practical reasons, the reference density should be assigned a value so low that reference water is present also in relatively shallow areas. On the other hand, it is important that the reference density is taken su$ciently high to ensure that waters participating in the surface circulation are included. The local value of the pro"le potential energy is rather sensitive to the choice of reference density. However, the horizontal structures and variations are not strongly dependent on the reference density. For clarity, it is desirable that the pro"le potential energy and the freshwater height are determined by integrating the same water. The reference temperature is taken 23C higher than in GS96 since it was found that most of the waters having salinity above 35 psu were warmer than 63C during the SKAGEX. An increase of the reference temperature from 6to83C reduces the average potential energy with some 40%, which should be kept in mind when comparing the potential energy estimates in the present paper with those in GS96. In general, a high freshwater height without a corresponding high pro"le potential energy indicates thin layers with low density (salinity), while the converse, low freshwater height and high pro"le potential energy, indicates relatively high density (salinity) in thick layers Dynamic implications An important strength of freshwater height and pro"le potential energy as diagnostic tools is the direct coupling to geostrophic dynamics. In a basin where density is solely dependent on salinity and where the deep water with salinity higher than S is at rest, i.e., the reduced gravity approximation, the freshwater height is directly proportional to the stream-function of the geostrophic surface currents and the pro"le potential energy is proportional to the geostrophic #ow stream-function. The derivations of these relations are left to the Appendix. The expression of the stream-function for a geostrophic surface current is, " gβs f F (3)

7 B. Gustafsson / Continental Shelf Research 19 (1999) 1021} and the transport stream-function is, ψ" 1 P. (4) f Using g"9.8ms, β"8 10 (see Eq. (A.6)), S "35 psu and f" s the constants of proportionality becomes 2290 m s and 8330 s, respectively. In a frontal region where the freshwater height gradient is strong, we expect strong surface currents. Correspondingly, a front with a strong gradient in pro"le potential energy is an indication of strong vertically integrated currents. A comparison of surface current speed and the vertically integrated current reveals the vertical structure of the baroclinic currents. For example, strong surface currents accompanied by relative low integrated currents indicate that the #ow is con"ned to a surface near layer. 3. Freshwater height and pro5le potential energy during the SKAGEX1 The freshwater height and pro"le potential energy were calculated for each T}S pro"le and horizontal "elds were drawn for each of the ten days when synoptic measurements were taken throughout Skagerrak. However, the pictures are not strictly synoptic but are composed of measurements from the day before and the day after the nominal day. The drawing of these "elds was done by hand in order to enhance clarity and to avoid smoothing which would have been unavoidable if standard numerical interpolation had been used. The distributions of freshwater height and pro"le potential energy are shown in Figs. 5}9. The wind conditions during the period were quite variable including events with wind speeds exceeding 10 m s. Wind conditions are shown in Fig. 3 and Table 1. Fig. 3. The measured winds at Ma seskak r during May}June 1990.

8 1028 B. Gustafsson / Continental Shelf Research 19 (1999) 1021}1047 Table 1 Wind conditions at station Ma seskak r during SKAGEX Dates Wind speed (daily average in m s ) Wind direction Wind energy input (mw m ) May 21}24 2}10 West 21.1 May 25}29 4}9 Northwest 12.9 May 29}June 4 4}8 South}southeast 4.5 June 5}6 3}4 Northwest 1.2 June 7}11 4}8 East 5.2 June 11}14 4}6 West 2.4 June 16}18 3}6 Southwest 6.5 June 18} 3}6 Southeast 2.9 The average wind energy input during each period is calculated according to Eq. (5) directly from the wind measurements, which are given in 6 h intervals. The distribution of freshwater height on May 24 (Fig. 5a) indicates upwelling at the Norwegian coast o! Torungen and downwelling at the Danish coast. This should be due to the westerly wind on May 24. The distribution of pro"le potential energy also re#ects these up- and downwelling areas (Fig. 5b). It appears that there is no out#ow to the North Sea along the Norwegian coast. There exists a large slick with somewhat elevated freshwater height in the center of Skagerrak (F'3 m) without any increase of pro"le potential energy, indicating a shallow layer with low salinity, while there are thick layers with rather high salinity in the area outside Hirtshals. A well-developed front between Kattegat and Skagerrak is very clearly visible in the distribution of freshwater height and also indicated, but not as clearly, the distribution of pro"le potential energy. Thus, it was a quite strong surface current along the Skagerrak} Kattegat front, but the pro"le potential energy "eld indicates that the transport was not very large. The latter is consistent with the current measurements which do not indicate an out#ow from the Kattegat (Fig. 4). By the wind shift toward northwest on May 25 the blocking of the Norwegian coastal current is eliminated and a baroclinic propagation of freshwater takes place along the Norwegian coast (Fig. 5c and d). However, the head of the current appears to separate from the coast. At the head, the gradient of the freshwater height is quite strong with ca. 4 m change in a distance of less than 20 km, indicating an average geostrophic surface velocity of some 45 cm s as estimated from Eq. (A.8). The pro"le potential energy is not very high at the coast, wherefore it is obviously a rapidly propagating shallow layer of low salinity, presumably from the northeastern corner of the Skagerrak where large amounts of freshwater usually are accumulated (GS96). During the same period, there is intense upwelling o! the Norwegian southwest edge to which the freshwater propagating along the coast has not yet reached. The northerly wind component is strong enough to force the freshwater-in#uenced layer away from the Swedish coast, resulting in upwelling. The current meters in the northern Kattegat show a southward #ow in the surface layer during the period May 25}27 (Fig. 4). Thus, the intensity of the upwelling at the Swedish Skagerrak coast is

9 B. Gustafsson / Continental Shelf Research 19 (1999) 1021} Fig. 4. The surface near currents in northern Kattegat. Positive values show current velocity in the northern direction. The moorings are located along section A (indicated in Fig. 2) with A1 being closest to the Swedish coast. probably enhanced by the lack of out#ow of buoyant surface water from the Kattegat. The front between Kattegat and Skagerrak surface waters is destroyed or perhaps advected South of LaK sok where measurements are too sparse to resolve a front. Between the 24th and the 27th the potential energy decreases o! Hirtshals while the freshwater height remains fairly constant which implies that surface layer salinity decreases. Thus, there was presumably an advection from the Kattegat frontal area or from the coastal currents along Swedish and Norwegian coasts to the area o! Hirtshals. The freshwater also extends further to the south along the Danish West Coast on the 27th as compared to the 24th. In D91 it is argued that, as for the freshwater o! Hirtshals, this water was transported across the Skagerrak during this event. However, this does not appear to be obvious from the structures of the "elds studied in the present paper. These rather indicate a combination of the wind piling up the freshwater located near the northern Danish coast already on May 24 and a reversal of the Jutland coastal current close to Hanstholm. The latter is also shown to happen in current meter data presented in D91, where a weak westward current exists near the coast. On May 30, there is almost no visible sign of the cyclonic circulation in the distributions of freshwater height and pro"le potential energy (Fig. 6a and b). There is upwelling o! the Norwegian southern coast, downwelling at the northern Danish coast, but in general rather low freshwater height and very low energy levels. This indicates that the winds of the previous days, mostly northwesterly until May 29, have reduced the pro"le potential energy as well as the horizontal gradients of the remaining freshwater by spreading out the buoyant surface layer in the Skagerrak. Thus, if there are any freshwater transports at all these should be determined by the barotropic current "eld and by wind driven shallow Ekman currents. The vanishing of

10 1030 B. Gustafsson / Continental Shelf Research 19 (1999) 1021}1047 Fig. 5. Freshwater height (m) and pro"le potential energy (m s ) the periods May 23}25 (upper left and right panels) and May 26}28 (lower left and right panels). the Norwegian coastal current, seen on the 30th, may be the result of direct wind forcing driving the low-saline surface waters o!shore or a combination of o!shore wind transport together with a lack of re"ll by downstream freshwater. That is, if there is no supply of freshwater to the Norwegian coastal current by downwelling at the

11 B. Gustafsson / Continental Shelf Research 19 (1999) 1021} Fig. 6. Freshwater height (m) and pro"le potential energy (m s ) the periods May 29}31 (upper left and right panels) and June 1}3 (lower left and right panels). coast of low saline water from central Skagerrak or by a coastal current along the Swedish Skagerrak coast, the Norwegian coastal current will eventually empty itself into the North Sea. It is also evident that a re"ll of freshwater from the Kattegat has started by the establishment of the Baltic current along the Swedish coast. This is

12 1032 B. Gustafsson / Continental Shelf Research 19 (1999) 1021}1047 explained by the wind shifting to southerly on May 29. However, the freshwater height is still comparatively low in the northern Kattegat. From May 29 to June 4, the prevailing wind was from the south to southeast with highest speeds during the second part (around 10 m s ). The wind shift had a quite large impact on the state of Skagerrak, as seen on June 2 when the cyclonic circulation pattern with increased potential energy along the coasts is clearly identi"able (Fig. 6c and d). The coastal current system is now fed with low-saline water from the Kattegat. The current measurements in northern Kattegat indicate surface #ow toward the Skagerrak along the Swedish coast and between LaK sok and Jutland, which is also evident from the gradients of freshwater height and pro"le potential energy. The period June 5}6 is characterized by fairly light northwesterly winds (2}6ms ). On June 5, the freshwater height has increased both along the Norwegian and the Danish coasts (Fig. 7a). Rather high pro"le potential energy is found in a band somewhat o!shore from Hanstholm. The amount of freshwater north and northwest of Denmark appears to have increased from June 2 to June 5. The Norwegian coastal current appears to have reached Kristiansand (Fig. 7a and b), with energy levels of some 15 m s. The signature of the Norwegian coastal current is quite di!erent from the former re-establishment on May 27. At this time the cross-shore gradients of freshwater height are not very sharp and water with increasing pro"le potential energy is nicely piled up against the coast. The current is composed of thicker layers of higher salinities than on May 27, which makes the shear of the current smaller and thus less sensitive to disturbances. O!shore the Swedish coast we "nd very low values of freshwater height indicating upwelling. However, since the three innermost stations of the section are missing it is not certain that upwelling really occurs. It also appears unlikely that the rather low wind speeds are able to force a substantial upwelling. Both in the northeast and at the Kattegat front rather high values of potential energy are found ('30 m s ) and these maxima are associated with very sharp energy fronts. The quite high strength of the Kattegat front may manifest a blocking of the out#ow from Kattegat, if so upwelling at the Swedish coast, if there is any, may be explained by the lack of supply of low saline surface water combined with northwest winds. If the #ow of the Baltic current is blocked there must be an ageostrophic or barotropic geostrophic return #ow close to the Swedish coast returning to Kattegat the volume transport along the front that is forced by the gradients across the front. A persistent eddy in the area south of the front was observed throughout the SKAGEX period by drifter tracks (D91). There are no signs of extremes in either freshwater height or pro"le potential energy that would indicate a closed baroclinic geostrophic circulation. However, the distance between the hydrographic sections may be too large for such a feature to be resolved. From June 8 until the end of the experiment measurements are lacking from the Hanstholm}Kristiansand section, at least in the data provided to the author, which should be kept in mind when analyzing the maps from the remaining days. The more out-stretched appearance of the Kattegat front on June 8 compared with that on the 5th (Fig. 7a and c) is maybe an artifact due to lack of observations at the border between Kattegat and Skagerrak. Another explanation is that the blocking of

13 B. Gustafsson / Continental Shelf Research 19 (1999) 1021} Fig. 7. Freshwater height (m) and pro"le potential energy (m s ) the periods June 4}6 (upper left and right panels) and June 7}9 (lower left and right panels). the Kattegat out#ow is somehow eliminated and therefore the Baltic current #ows freely into the Skagerrak. The latter explanation is consistent with the wind conditions, because on June 7 the wind shifted to east and that direction lasted until June 11 with peak speeds at some 10 m s on June 8. High values of pro"le potential energy

14 1034 B. Gustafsson / Continental Shelf Research 19 (1999) 1021}1047 are found along both the Swedish and Norwegian coasts (Fig. 7d), especially in northeastern corner ('35 m s ). Thus, the cyclonic coastal current system is developed, probably enhanced by both the sudden out#ow of Kattegat surface water of high freshwater content between the June 5 and June 8 and by wind forced downwelling at the Norwegian coast. The cross-shore gradients of both freshwater height and pro"le potential energy are quite strong along the Norwegian coast, especially at the out#ow from the northeastern corner. Here, the distance between the hydrographic stations is ca. 8 km. Thus we "nd a freshwater gradient of the order of 3 m per 8 km which corresponds to average geostrophic surface current speed of 0.9 m s. The pro"le potential energy gradient is some 20 m s per ca. 12 km which gives an average transport of 14 m s. Integrated across the current, the geostrophic #ow is at least m s as estimated from Eq. (4). The strength of the coastal currents is reduced on June 11 as indicated by the distribution of pro"le potential energy (Fig. 8b). This is most probably a consequence of the easterly wind decreasing in strength from 10 m s on June 8 to some 2}5ms on June 9 and 10, and eventually shifting direction to west on June 11. The west wind with strength of 4}7ms dominates until June 14. A local minimum of the freshwater height is found in conjunction with the front in the northeastern corner (Fig. 8a), separating an area with higher freshwater height from the coastal downwelling area. This feature does not appear that clearly in the pro"le potential energy "eld. However, geostrophic clock-wise circulation within the patch of less saline water is probable, although the pro"le potential energy is somewhat reduced by the rising the pycnocline. Starting on the 11th, the pro"le potential energy increases north of Jutland, most probably due to downwelling by the westerly wind. The increase appears to be coherent with an increase in strength of the Kattegat front (compare Fig. 8b and d) due to increasing freshwater height and pro"le potential energy in northern Kattegat. On the 14th, the potential energy is again quite high along the coasts, indicating strong geostrophic coastal #ow. The freshwater height has generally increased in central Skagerrak (Fig. 8c), and the earlier visible patch with enhanced freshwater height is no longer visible. However the local minimum is still found in the northeast, also in pro"le potential energy. The anti-cyclonic eddy that was present on June 11 has disappeared while the cyclonic eddy has become stronger. The wind has shifted to southwest on June 16 and it continued to blow from southwest until June 18 when it slowly shifted to southeast. On June 16 the wind was rather strong with a speed of about 9}12 m s, but it rather rapidly decreased to some 5 m s. On June 17, the pro"le potential energy along the Danish coast is fairly high all the way from Tybor+n (Fig. 9b), with a quite high maximum o! Hirtshals ('20 m s ), indicating a large #ow in the Jutland coastal current. However, the pro"le potential energy has now decreased somewhat along the Norwegian coast from the high values of the 14th of June. The freshwater height in the northern Kattegat (Fig. 9a) has decreased considerably, especially along the Swedish coast. The current meter observations show a southward #ow of surface water along the Swedish Kattegat coast between June 11 and June 15, while there is a northward #ow in the channel between LaK sok and Jutland during the same period (Fig. 4). However, the

15 B. Gustafsson / Continental Shelf Research 19 (1999) 1021} Fig. 8. Freshwater height (m) and pro"le potential energy (m s ) the periods June 10}12 (upper left and right panels) and June 13}15 (lower left and right panels). decrease in freshwater height along the Swedish Kattegat coast is not visible before June 17, when the current along the Swedish coast already has shifted to the north again. The most probable explanation is that the wind shift from west to southwest enables a geostrophic #ow from the frontal region and large amounts of

16 1036 B. Gustafsson / Continental Shelf Research 19 (1999) 1021}1047 Fig. 9. Freshwater height (m) and pro"le potential energy (m s ) the periods June 16}18 (upper left and right panels) and June 19}21 (lower left and right panels). freshwater-in#uenced water #owed thereby northward into the Skagerrak. Thus, the decrease in freshwater height should be a consequence of divergence along the Swedish coast due to southward or near zero #ow in the Kattegat and northward #ow by the Baltic current further to the north.

17 B. Gustafsson / Continental Shelf Research 19 (1999) 1021} During the last days of the experiment the wind was primarily from the southeast with speeds peaking at 8 m s on June 19. This appears to contract the Kattegat front toward the Swedish coast and reduce the pro"le potential energy in the Jutland coastal current (Fig. 9d). At the outer stations along the western part of the Norwegian coast both the freshwater height and the pro"le potential energy are quite high (Fig. 9c and d). It is however hard to tell whether this is an e!ect of a wave-like pulse propagating along the coast or if it is due to local downwelling. 4. Budget estimates In GS96, a long-term average potential energy budget for Skagerrak was presented. It was found that diapycnal mixing and advection of energy from the Kattegat and the southern North Sea all were only minor contributors to the forcing of the baroclinic cyclonic circulation in the freshwater-in#uenced surface layers of Skagerrak. The coverage of the SKAGEX 1 hydrographic data set is good enough for an integration of almost instantaneous freshwater and pro"le potential energy content. The estimations are probably the best one is able to do, not only at present but also in the foreseeable future since it would be very di$cult to gather so many ships without the contribution from the former eastern countries. These countries countributed with about half of the ships during SKAGEX 1. Thus, it is possible to directly correlate the changes in freshwater content and total pro"le potential energy with estimated sources and sinks Variations in freshwater content and average proxle potential energy Time-series of the freshwater and pro"le potential energy content inside the Skagerrak have been calculated from the horizontal "elds described above. The "elds have been interpolated numerically on a one by one nautical mile ("1852 m) grid thereafter the integration of freshwater and potential energy content is performed. The numerical interpolation results in smoother "elds than those previously presented graphically, but the total amounts should be reasonably conserved. The integration area is limited by the outermost section at the North Sea border (section H) and an eastward line from the northernmost tip of Jutland across to Sweden (indicated by lines in Fig. 2). The integration was limited by the coastline of a high resolution (grid-size is 1852 m) digital bathymetry provided by the Danish Hydraulic Institute. The total surface area of integration is km and thus larger than the area used in GS96 where the area northwest of Jutland was excluded. The time-series of average freshwater height and average pro"le potential energy is shown in Fig. 10a. From these time-series the three-day averaged rates of change of the quantities are also calculated (Fig. 10b). The time evolutions of spatially averaged freshwater height and pro"le potential energy do roughly follow each other. Both freshwater height and pro"le potential energy are minimal on May 30, as is expected from the previously displayed distributions (see Fig. 6a and b). The minimum in freshwater height can be explained by a net

18 1038 B. Gustafsson / Continental Shelf Research 19 (1999) 1021}1047 Fig. 10. The time evolution of horizontally averaged freshwater height and pro"le potential energy (a) and the rate of change of freshwater volume and potential energy content (b). out#ow of up to m s the preceding days. However, no measurements were taken in the northeastern corner between May 27 and June 2, wherefore the decrease of freshwater volume may be less. The area not covered by observations is some 10}15% of the whole area of integration. Therefore an uncertainty of the estimate of the freshwater height in the northeastern corner of as much as 1 m will alter the total average with only some 0.1}0.15 m. Observations in the northeastern corner are also missing on June 20 (Fig. 9c and d). During the period from May 30 to June 5 the average freshwater height increases with approximately 0.5 m and the average pro"le potential energy with some 5 m s. The increase in freshwater height implies a net in#ow of the order of m s between May 30 and June 2. The increase of pro"le potential energy implies a net average energy supply of 7}8mWm (ca. 250}290 MW) between May 30 and June 5. These increases in freshwater height and pro"le potential energy are coherent with the wind shifting from northwest to south

19 late on May 29 (see Fig. 3 and Table 1) and the establishment of the cyclonic circulation in the Skagerrak. Between June 5 and June 14 both quantities remain fairly constant, implying an approximate balance between supply and loss. At the end of the SKAGEX period, both quantities decrease again. The contribution to the pro"le potential energy due to wind driven diapycnal mixing was rather small during the whole experiment as the wind speeds were rather small. The energy supply from the wind is calculated according to = "ρu H /κ B. Gustafsson / Continental Shelf Research 19 (1999) 1021} where the friction velocity u H is calculated from observed wind speeds at Ma seskak r according to Smith (1980). The constant κ is equal to The variations of the energy supply from the wind are presented in Table 1. During the period of large increase of the pro"le potential energy the average energy supply by wind stress was 4.5 mw m. However, only a fraction of the turbulent energy supply is available for mixing. The #ux Richardson number being the ratio between work against the buoyancy forces and turbulent energy supply has a value of about 5% (see, e.g., Stigebrandt, 1981). Thus, the contribution to the increase of pro"le potential energy by wind forced diapycnal mixing should be 0.23 mw m which is negligible in comparision with the observed rate of change Exchange with the Kattegat and the North Sea The water and freshwater #ows from the Kattegat to the Skagerrak have been calculated from the current meter data obtained in a section across northern Kattegat. In Fig. 11, the net water #ow and the freshwater #ow from May 24 to June 10 are (5) Fig. 11. The water and freshwater #ow from the Kattegat to the Skagerrak as calculated from current measurements in northern Kattegat.

20 1040 B. Gustafsson / Continental Shelf Research 19 (1999) 1021}1047 shown. The variations of the net #ow are consistent with the volume changes in the Kattegat and the Baltic estimated from sea level variations (not shown), wherefore we may be con"dent that the daily averaged #ow is rather accurately determined. The #ow is southward during the period May 24 to May 29 and northward for the remaining measurement period. The freshwater #ow is also directed to the south for the initial period. The maximum #ows are almost m s of water and m s of freshwater. During the period May 30 to June 10 a total of 45 km of water and 20 km of freshwater leaves the Kattegat. The increase of the average freshwater height in Skagerrak due to the out#ow from Kattegat is ca m if the same area is used for the integration as used above. The "elds of freshwater height and pro"le potential energy do not indicate any signi"cant strength of the Norwegian coastal current before June 8, possibly because the out#ow of freshwater along the Norwegian coast probably was quite small until this day. The estimated increase in average freshwater height due to #ow from Kattegat "ts very well with the estimated increase of freshwater content in the Skagerrak during the same period. However, it appears that the freshwater volume in the Skagerrak changes most rapidly between the May 30 and June 2 while the current measurements indicate that on June 2 only half of the total freshwater out#ow has occurred. This could be due to an overestimation of the freshwater volume on June 2 because the northeastern Skagerrak and northern Kattegat are quite poorly resolved, but it could also be a consequence of in#ow of freshwater along the Jutland coast. It is possible to estimate the freshwater #ow along the coast of Jutland from the distribution of freshwater height presented in this paper and the current meter records presented in D91. The horizontal resolution of the hydrographic measurements o! the West Coast of Jutland is quite variable throughout the SKAGEX. However, it is evident from the freshwater distributions that freshwater was exported from the Skagerrak to the North Sea during the period May 27}30 which is also supported from the current measurements which show westward mean currents (ca. 10 cm s ) near the coast o! Hanstholm between May 26 to May 28. During May 29}31, there are only very weak currents, but during June 1}3 the average currents are quite strong ('40 cm s ) and are directed into the Skagerrak. The estimated out#ow of freshwater along the Jutland coast during May 26}28 is not very large. If a freshwater height of 2 m and a current width of 20 km (which probably is an overestimation) are assumed, the freshwater #ow becomes only 4000 m s. The #ow may be quite underestimated if the strong northwest wind generates very shallow currents above the uppermost current meters, which were located at 10 m depth or more. The in#ow of freshwater from the North Sea during June 1}3 was considerably larger. The width of the current was of the order of 40 km and the freshwater height was close to 2 m resulting in a freshwater #ow of some m s. Thus, the in#ow of freshwater to Skagerrak along the Jutland coast may have contributed substantially to the increasing freshwater content. An estimate of the in#ow of potential energy from the Kattegat may also be obtained from the current meter data. The in#ow of potential energy may be expressed as ε Q where ε is the energy density and Q the volume #ow (see GS96). In a two-layered system where density is solely dependent on salinity, the energy

21 B. Gustafsson / Continental Shelf Research 19 (1999) 1021} Fig. 12. The #ow of potential energy from the Kattegat as calculated from current measurements in comparison with the rate of change of the potential energy content within the Skagerrak. density may be expressed in terms of the freshwater height; ε" 2P h "gβs F where h is the pycnocline depth and β is the expansion coe$cient of seawater with respect to salinity, cf. Eq. (A.6) in the Appendix. Because the vertical resolution of the current measurements is quite poor, an assumption of the pycnocline depth has to be done. There are at least two possible ways to calculate the energy #ow. The most straightforward is to use the freshwater height and volume #ow (Fig. 11), ε Q + gβs FM Q (6) where FM is a typical freshwater height that may be found from the previously shown horizontal distributions of freshwater height and Q the volume #ow as measured by the current meters. The second possibility is to use the freshwater #ow (Fig. 11) and assume a typical pycnocline depth, hm. In this case the energy #ow is given by ε Q + gβs hm Q. (7) The second approximation may be argued to be the most consistent because only #ow containing freshwater is accounted for, however, the pycnocline depth is utterly uncertain. The in#ow of potential energy expressed as spatial average which is calculated by using Eqs. (6) and (7), and the potential energy change of the Skagerrak are shown in Fig. 12. The in#ow appears to be coherent with the actual change, however with lower magnitude. In the calculation a typical freshwater height of 7 m and a typical pycnocline depth of 14 m were used. The two approximations di!er by ca. 50%. It would perhaps be possible to argue for an increase of the pycnocline depth

22 1042 B. Gustafsson / Continental Shelf Research 19 (1999) 1021}1047 to some 20 m and thereby one can get a better agreement between in#ow and change of energy content. However, using the "rst approximation it would be necessary to increase the freshwater height by a factor of two, which is not consistent with observations. However, the estimated in#ow of potential energy may account for a substantial part of the total energy input during the SKAGEX. The energy #ow along the Jutland coast June 1}3, associated with the abovementioned in#ow of freshwater could be of the order of 6 mw m (218 MW), if Eq. (7) is used with a freshwater #ow of m s and a pycnocline depth of 25 m. The #ow along the Jutland coast may thus have been of the same magnitude as the in#ow from the Kattegat. The Norwegian coastal current was not very prominent during the "rst half of the experiment as previously discussed in connection with the freshwater and pro"le potential energy distributions (Section 3). The current meter recordings presented in D91, where the cyclonic circulation did not show any signi"cant depth dependence before June 5, con"rm this. Thus, the export of freshwater and pro"le potential energy by the Norwegian coastal current was most probably insigni"cant until that date Geostrophic freshwater yow Stigebrandt (1987) presented a simple relation between the freshwater height and geostrophic freshwater #ow in a system where the density is determined by salinity alone. The relation states that Q " gβs F 2f (8) where f is the Coriolis parameter and F is the freshwater height at the coast. Stigebrandt (1987) used the relation to calculate the #ow of saline water from the Arkona Basin to the Baltic proper. In GS96 Eq. (8) was applied to the average freshwater height o! the southwestern edge of Norway whereby an average geostrophic freshwater transport of m s was found which is quite close to the total freshwater supply to the Skagerrak. BjoK rk (1989) uses a similar geostrophic relation with good results to calculate the out#ow of freshwater-in#uenced water from the Arctic Ocean through the Fram Strait and the Canadian archipelago. Thus, Eq. (8) has proved to be a useful estimator of the freshwater #ow in many practical cases. However, Lundberg and Walin (1990) showed that the shape of the isohalines within the coastal current in#uences the freshwater #ow in a way that two coastal currents with identical pro"les at the coast may have di!erent freshwater #ow although the volume #ow is the same. The geostrophic freshwater #ow has been estimated using Eq. (8) for three stations along the Swedish and Norwegian coasts. The estimations are shown in Fig. 13 together with those of freshwater #ow from the Kattegat calculated by current meters. The increase of freshwater #ow appears to be lagged with some 2}3 d between the stations, which is consistent with a propagation speed of 70}80 cm s. However, the peak value at the hydrographic stations occurs at approximately the same time, which

23 B. Gustafsson / Continental Shelf Research 19 (1999) 1021} Fig. 13. Estimation of the geostrophic freshwater #ow from a few hydrographic stations along the coasts of Skagerrak. The #ow from Kattegat is estimated from current meters (Fig. 11). is probably an e!ect of the wind changing direction from east to west. It should be recalled that the residence time for freshwater in the Skagerrak is about 40 days (GS96) while a baroclinic signal may travel around the coasts in less than one week. From this it follows that there should be large local e!ects along the coasts of Skagerrak which make simple explanations hard to "nd. Thus, although the geostrophic freshwater #ow appears to be consistent with a propagation of a signal from the Kattegat, the e!ect-of &local' wind forced downwelling of freshwater is not excluded as a major factor. Until June 4, the prevailing wind was from south to southeast, which contracts the coastal current along the Swedish coast. The southerly wind was followed by two days of light wind from the northwest, which was not able to produce any downwelling. When the wind eventually on June 7 turns to the east, which forces downwelling along the Norwegian coast, the geostrophic freshwater #ow at Torungen already has increased to approximately m s and does not increase very much more. However, even though downwelling directly do not produce the signal visible o! Torungen the signal may be a result of export from the storage of freshwater in the northeastern corner of Skagerrak. 5. Concluding remarks In the present paper the rapid response of the freshwater-in#uenced surface layers to variations in the magnitude and direction of the wind stress and to variations in upstream conditions was illustrated by quasi-synoptic distributions of freshwater height and pro"le potential energy. The development of the circulation from May 30, when the cyclonic circulation was completely

24 1044 B. Gustafsson / Continental Shelf Research 19 (1999) 1021}1047 gone, to June 8 when it was fully developed, indicates that the spin-up time is some 9 days. The budget estimates of total freshwater and pro"le potential energy reveal that the variations of the in#ow of low saline surface water from the Kattegat and along the Jutland coast may be of considerable importance for the circulation in the Skagerrak even on relatively short time-scales. This may at "rst appear to contradict the conclusion that the local wind-forced downwelling is the major forcing for the surface circulation (GS96). However, the annual variation of the strength of the circulation is quite large (GS96) with a maximum in winter and minimum in summer. Also when comparing the pro"le potential energy levels found in the present analysis (maximum of ca. 30 m s in the Norwegian coastal current) with the average winter values found in GS96 outside Torungen (ca. 110 m s ), although the pro"le potential energy is ca. 40% higher in GS96 due to a di!erent choice of reference density, one has to conclude that the advective sources for potential energy cannot explain a factor of two more energy. Thus, the probable conclusion is that the supply of low saline surface water from the Kattegat and along the Jutland coast does give a contribution to the forcing of the circulation, but during the winter the forcing due to wind-forced downwelling is about three times larger. This is in accordance with previous observations of the response of the circulation to variable winds (Aure and Svtre, 1981). This can also be expected from the fact that the residence time for freshwater in Skagerrak is some 40 days (GS96) while the transit time of freshwater in the coastal currents from source to sink is only about a week. In the present paper the contribution of freshwater and potential energy to the Skagerrak by in#ow of water along the Jutland coast are to a large extent part of a re-circulation of Skagerrak surface water (see also D97). Rydberg et al. (1996) concluded that a re-circulation of freshwater-in#uenced surface water through a section o! Hanstholm is really a prominent feature, with a long-term average freshwater #ow of ca m s. The budget calculations show that the recirculation is of importance for the maintenance of the circulation by providing up to 220 MW (or 6 mw m ) during parts of the SKAGEX 1 experiment. The annual cycles of freshwater height and pro"le potential energy o! Hirtshals covaries with a maximum in August}September (F+3 m and P+40 m s ) and a minimum during early spring (F+1 m and P+5m s ), see GS96. Thus, the late summer average conditions are similar to the conditions in the middle of the SKAGEX period if the higher reference density is taken into account. This leads us to the conclusion that the relative importance of the contribution of potential energy by the Jutland coastal current is large during late summer when the pro"le potential energy is generally low in Skagerrak, but insigni"cant during wintertime. Acknowledgements The author is in great debt to the numerous scientists who gathered the SKAGEX data set at sea and ICES for providing the quality checked data, Swedish Meteoro-