Mesoscale-resolving simulations of summer and winter bora...

2 Mesoscale-resolving simulations of summer and winter bora

3 events in the Adriatic Sea

4 Benoit Cushman-Roisin1 and Konstantin A. Korotenko1

5 Received 31 January 2006; revised 26 June 2006; accepted 11 September 2007; published XX Month 2007.

6 [1] This paper presents simulations of the Adriatic Sea response to two distinct bora wind7 events, one in summer (11–20 August 2001), when the water is stratified, and the other in8 winter (11–20 February 2003), when it is vertically homogeneous. The simulations9 employ the DieCAST model on a 1.2-min grid (about 2-km resolution) and resolve the10 mesoscale variability because the grid size falls below the first baroclinic deformation11 radius (about 5–10 km) and the model has very low horizontal dissipation. The model is12 initialized with seasonally averaged temperature and salinity data and spun up with13 climatological winds. The summer of August 2001 event leads to the generation of a14 coastal current directed paradoxically to the left of the wind and identified with the15 summertime Istrian Coastal Countercurrent. Analysis of the physics simulated by the16 model leads to the conclusion that this current is caused by baroclinic geostrophic17 adjustment of the Istrian coastal waters following a rapid but strong wind impulse.18 According to both satellite observations and model simulations, the current persists for19 more than a week after the bora event. The winter event of February 2003 generates a20 slightly less complicated situation because the shallow northern Adriatic is then vertically21 homogeneous, but in situ observations at the time of this particular event permit a22 comparison of model results with observations and thus an evaluation of the model23 performance. In particular, the model simulates correctly a series of currents, bifurcations,24 and confluences of the wind-driven currents across the northern Adriatic basin. A lesson25 learned is that a bora event, though generally strong, especially in winter, leaves a26 legacy that does not obliterate completely what existed prior to the event. In other words,27 the state of the northern Adriatic basin is fashioned by sequential events, and one bora28 event may not be viewed in isolation but must be considered as one episode in an29 unfolding succession of events.

30 Citation: Cushman-Roisin, B., and K. A. Korotenko (2007), Mesoscale-resolving simulations of summer and winter bora events in

31 the Adriatic Sea, J. Geophys. Res., 112, XXXXXX, doi:10.1029/2006JC003516.

33 1. Introduction

34 [2] The Adriatic Sea basin is a semienclosed sea with35 varied topography (Figure 1), and its circulation is driven by36 four types of forcings [Cushman-Roisin et al., 2001]: Wind37 stress, freshwater runoff, surface buoyancy fluxes, and38 water exchange through Otranto Strait. Winds over the39 Adriatic can be classified in various types [Poulain and40 Raicich, 2001], each leaving a mark on the sea after its41 occurrence.42 [3] Among the distinct wind types over the Adriatic, bora43 is of special significance [Jurcec and Brzovic, 1995].44 Its nature as a cold air mass rushing down the Dinaric45 mountains (along the eastern coast) and spilling over the sea46 has long been the subject of meteorological studies47 [Prettner, 1866; Yoshino, 1976; Smith, 1985, 1987; Pirazzoli

48and Tomasin, 1999]. Finding its way throughwind gaps, bora49exhibits a multijet structure [Kuzmic, 1986, 1993], sometimes50called vorticity banners [Grubisic, 2004]. The lateral shear on51the flank of these jets is indeed marked by strong vorticities,52the wind-stress curl of which generates gyres in the water53[Orlic et al., 1994]. In addition to being a strongwind, bora in54winter also brings cold dry air that causes heat loss and55evaporation, leading to dense water formation on the shelf56[Vested et al., 1998; Beg Paklar et al., 2001; Vilibic et al.,572004].58[4] A number of modeling studies of the Adriatic59response to bora have been published over the last few60decades. Stravisi [1977] and Orlic et al. [1986] used61vertically integrated (2D) and overly viscous models of62the northern basin but were able to show that the curl of the63bora wind generates a cyclonic gyre in the very northern64portion of the basin. Kuzmic [1986, 1991], Orlic et al.65[1994], Bergamasco and Gacic [1996] investigated the66Adriatic response to a schematic bora wind field with67low-resolution models. These models showed to which68extent wind-driven motions could perturb the seasonal

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 112, XXXXXX, doi:10.1029/2006JC003516, 2007ClickHere

for

FullArticle

1Thayer School of Engineering, Dartmouth College, Hanover, NewHampshire, USA.

Copyright 2007 by the American Geophysical Union.0148-0227/07/2006JC003516$09.00

XXXXXX 1 of 12

http://dx.doi.org/10.1029/2006JC003516

69 circulation but were incapable of reproducing the associated70 mesoscale activity. Building on work by Rachev and Purini71 [2001], Loglisci et al. [2004] developed a coupled atmo-72 sphere-ocean model specifically devoted to the air-sea73 interactions during a bora event. Beg Paklar et al. [2001]74 and Wang [2005] used the Princeton Ocean Model to75 simulate a bora event of mid-January 2001. These last76 several studies emphasized the heat loss resulting from77 the cold bora rather than the details of the wind-driven78 circulation, which, in their respective models, was not79 allowed to develop mesoscale instabilities, either because80 of inadequate spatial resolution [Rachev and Purini, 2001;81 Loglisci et al., 2004] or a more dissipative numerical82 scheme [Beg Paklar et al., 2001; Wang, 2005]. In a similar83 vein, Pullen et al. [2003, 2007] coupled an atmospheric84 model (COAMPS, 4-km resolution) with an oceanic model85 (NCOM, 2-km resolution) to study the response of the86 northern Adriatic basin to strong forcing. The emphasis of

87this work, however, is not so much on the physics of the88Adriatic as on ways to improve the performance of the89model coupling.90[5] In 2003–2004, several cruises and drifter launches91provided data on the mesoscale variability in the northern92and middle basins of the Adriatic and captured a bora event93with unprecedented detail [Lee et al., 2005]. The data are94providing an important new source of information, and the95numerical model presented below was specifically devel-96oped to serve as a platform to investigate in particular the97mesoscale dynamics observed during these recent observa-98tional campaigns.99[6] The approach is to use a numerical model with the100least possible amount of horizontal dissipation, in order to101allow the instabilities of the flow to develop as freely as102possible. The model must also be capable of handling103abrupt topography, such as the steep channels and104escarpments off the Croatian coast. Under these constraints,

Figure 1. Geography and bathymetry of the Adriatic Sea, with depth contours in meters. The twostraight lines across the northern basin, labeled A and 1, indicate lines along which sections will bepresented later.

XXXXXX CUSHMAN-ROISIN AND KOROTENKO: ADRIATIC BORA SIMULATIONS

2 of 12

XXXXXX

105 DieCAST [Dietrich, 1997; Cushman-Roisin et al., 2007]106 stands as the model of choice, because of its fourth-107 order resolution in the horizontal (allowing large Reynolds108 numbers at the grid level) and its z-level discretization in the109 vertical (best suited for the representation of topographic110 steps). The application of this model to the Adriatic basin111 must also insure that the grid resolution is sufficiently high112 to resolve scales at the baroclinic radius of deformation, of113 about 10 km [Cushman-Roisin et al., 2001].

114 2. Model Description

115 2.1. General Model Description

116 [7] The DieCAST ocean model (http://fluid.stanford.edu/117 yhtseng/research/DieCAST/users_manual.pdf) is a z-level,118 finite-difference, three-dimensional, primitive-equations,119 hydrostatic, Boussinesq model with very low dissipation120 thanks to a fully fourth-order numerical scheme and a121 weakly filtered leap-frog time integration. For details of122 the governing equations, the reader is referred to Dietrich123 [1997] and to Appendix A of the paper by Staneva et al.124 [2001], which presents an application to the Black Sea.125 Applications to the Adriatic Sea are given by Rachev and126 Purini [2001], Loglisci et al. [2004] and Cushman-Roisin et127 al. [2007].128 [8] Here and as in the work by Cushman-Roisin et al.129 [2007], the horizontal resolution is 1/50� (1.2 nautical miles,130 about 2 km), and the dimension of the mesh is 370 � 272,131 covering the entire Adriatic basin from 12.25�E to 19.6�E132 and from 40.4�N to 45.8�N. This resolution allows for a133 faithful representation of the larger and intermediate-size134 islands and channels in the Adriatic basin, especially along135 the Croatian coast.

136 2.2. Initialization and Forcings

137 [9] Initialization of the model is performed as described138 by Cushman-Roisin et al. [2007], namely by specification139 of temperature and salinity distributions corresponding to140 the season in which the event occurs (e.g., summer for an141 August event and winter for a February event). The velocity142 field is spun up from rest.143 [10] As mentioned in section 1, there are four types of144 forcings acting on the Adriatic Sea: river runoff, surface145 winds, surface buoyancy fluxes and water exchange through146 Otranto Strait, all of which are included in the present147 model. Freshwater fluxes from the 38 largest rivers around148 the perimeter of the basin are specified from climatological149 data sets [Raicich, 1994, 1996], with daily values interpo-150 lated from perpetual annual cycles. River runoff is imple-151 mented in the model as a freshwater source in the topmost152 level of the grid cell closest to the river mouth. An153 exception is made for the Po River because of its size:154 Its discharge is divided among its four branches, and155 the discharge at each mouth is spread over the three156 closest grid cells. Also, actual daily discharge and temper-157 ature values for the periods concerned are used in our158 simulations.159 [11] For surface winds, climatological wind data of160 Hellerman and Rosenstein [1983] are used during the161 spin-up phase of the model. Once spun up, the model is162 forced by hourly winds obtained from COAMPS [Hodur et163 al., 2002; Pullen et al., 2003] during the event under study

164and a few days beyond. This wind field given on a 4-km165spatial resolution and then interpolated onto our 2-km grid166is adequate to resolve the multijet nature of bora.167[12] Buoyancy forcings used for model spin-up are taken168from Artegiani et al. [1997], while values used in the actual169event simulations are those of COAMPS [Hodur et al.,1702002; Pullen et al., 2003]. For water exchange through171Otranto Strait, values are taken from simulations of a wider172model run on a seasonal scale, as described by Cushman-173Roisin et al. [2007].

1753. Response to Summer Bora (11–20 August 2001)

176[13] For the simulation of a bora event in summer, the177period of 11–20 August 2001 was selected because it178includes a well defined bora event lasting two days (11–17912 August) and is accompanied by a valuable set of satellite180images, especially from Sea-viewing Wide Field-of-view181Sensor (SeaWiFS), because of clear skies during the bora182event proper and some days afterward.183[14] On 9 August, prior to the bora event, wind blew from184WSWand then turned into a slight SE sirocco by 10 August.185This sequence drove the Po River plume eastward across the186northern basin and then against the Istrian peninsula. The Po187River discharge had been normal for this time of year, about188900 to 1000 m3/s. When bora began to blow on 11 August189(Figure 2), the wind jet over Rijeka caused upwelling in the190Bay of Rijeka, juxtaposing cold and saline waters on the191eastern flank of Istria (in the bay) next to warm and fresh192water of Po River origin on the western flank (open sea). As193the surface chlorophyll distribution of 12 August shows194(Figure 3), the water west of Istria is loaded with chloro-195phyll from the Po, while the chlorophyll-free water inside196the bay seems to have originated from further south along197the Croatian coast. Because of their respective origins, the198chlorophyll-rich water offshore is warmer and fresher than199the clear water in the bay.200[15] As the bora wind ceases and some of the dense water201has been dragged seaward out of the bay, the situation is202away from equilibrium and an adjustment becomes203necessary. The numerical model (Figure 4) shows that a204southward current develops along the southwestern shore of205Istria within a day or so, i.e., on the inertial timescale as we206expect for geostrophic adjustment. The current reaches a207maximum speed of 25 cm/s about 10 km from the coast.208Evidence of its existence is seen in satellite pictures (some209of them included in Figure 3, others shown in Figure 5) in210which a tongue of high-chlorophyll water proceeds south-211eastward down the coast of Istria and, beyond the southern212tip, veers to the right toward of the middle of the basin. This213southward current along Istria can perhaps be identified214with the occasional Istrian Coastal Countercurrent described215by Supic et al. [2000]. We shall return to this point in the216discussion (section 5).217[16] The fate of the southward current is eventually to roll218up in an anticyclonic vortex, as shown in the comparative219plot of Figure 5, with computed sea surface density (left220column) compared to SeaWiFS satellite images (right221column). The anticyclonic vorticity was presumably that222imparted to the sea on the right flank of the bora jet223(clockwise wind-stress curl) in the region of the current’s224origin.


3 of 12

XXXXXX

225 [17] The northern Adriatic response to a summer bora226 event includes not only the aforementioned upwelling in227 Rijeka Bay, the upwind flow against Istria, the Istrian228 Coastal Countercurrent and its decay into a vortex, but a229 host of other mesoscale instabilities, most of them presum-230 ably due to the mixed barotropic-baroclinic nature of the231 vorticity-laden flow in the presence of summer stratifica-232 tion. There is no room to discuss and analyze all those here,233 and we shall only mention a swirl occurring on the flank of234 the upwind flow toward Istria, which like most other235 mesoscale structures in the sea at the time is reproduced236 by the model.237 [18] In the satellite image of 14 August (Figure 3, right)238 this swirl is seen as the thumb on the southern flank of the239 Po plume, located slightly south of the midpoint between240 the Po delta and the southern tip of Istria. It is extruded from241 the Po plume by an anticyclonic vortex, seen in the242 simulated currents (Figure 4, bottom right). The feature243 appears to be an instability of the eastward flow. In this244 region, just north of the bora jet blowing from Rijeka Bay,245 the wind-stress curl is clockwise, and this has imparted to246 the water in the area some anticyclonic vorticity. As seen in247 the vertical sections presented in Figures 6a and 6b, this248 feature is accompanied by simultaneous upwelling249 with northward flow (just west of the 46.9 km mark in250 Figures 6a and 6b) and downwelling with southward flow251 (just east of the 46.9 km mark in the figures). Such pattern is252 symptomatic of a barotropic instability that feeds kinetic

253energy from the vorticity-laden flow into potential energy of254tilted density surfaces.

2554. Response toWinterBora (11–20February2003)

256[19] For the simulation of a winter bora, the period of 11–25720 February 2003 was selected because it corresponds with258the intensive in situ data collection campaigns reported by259Lee et al. [2005], which includes a significant bora event260(for a snapshot of the wind field, see Figure 2, right).261Fortunately, clear skies provide good satellite coverage262during the same period (P.-M. Poulain et al., The circulation263and temperature-salinity-pigment fields in the northern264Adriatic Sea in winter 2003, submitted to Journal of265Geophysical Research, 2007) (hereinafter referred to as266Poulain et al., submitted manuscript, 2007). This bora event267and the attending observations have been described [Lee et268al., 2005; Poulain et al., submitted manuscript, 2007],269and there is no need to give here another description of270the events and data. We shall instead focus on the simu-271lations by the present model and compare them with field272observations.273[20] Figure 7 contrasts the sea surface temperature before274(12 February) and after (19 February) the bora event. Two275major differences show the effect of bora winds. First, the276meridional gradient South of Istria has greatly intensified, to277the point of becoming a sharp front. Second, the previously

Figure 2. Wind stress (white arrows) superimposed on wind-stress curl (color) over the northernAdriatic during two bora events, on 12 August 2001 and 19 February 2003. Note the banded structure ofbora, with a jet blowing from Trieste in the northeastern corner, a second jet blowing from Rijeka, Southof Istria, and a third one at 44.4�N. (left) The summer bora has slightly broader jets than (right) the winterbora, especially in the northernmost part of the basin, circa 45�N. (Data are from COAMPS, courtesy ofJames D. Doyle).


4 of 12

XXXXXX

278 broad temperature gradient in the northern basin, between279 the Po River delta and Istria, has, too, become more280 accentuated. It has also rotated to take a southwest–north-281 east orientation and developed a few meanders, including282 one south of the Po River delta. It is not a priori clear how283 the relatively zonal wind jets (Figure 2, right) could create284 such an oblique feature.

285[21] These temperature changes and the resulting286changes in density are naturally accompanied by currents.287Figure 8 shows the surface currents before and after bora,288for the same dates as shown in Figure 7. A first current289generated by the wind event is a westward coastal jet290hugging the northern coast of the Adriatic basin, from291Trieste to Venice and down to the Po River Delta. This

Figure 3. Sea-viewing Wide Field-of-view Sensor (SeaWiFS) and infrared satellite images of 12 and14 August 2001 showing the surface concentration of chlorophyll a and sea surface temperature,respectively. Note how the warm, chlorophyll-rich waters from the Po River have crossed the northernbasin and reached the western shore of Istria, while the eastern (interior) side of Istria is flanked by colderand chlorophyll-free waters that have been recently upwelled by the bora wind of the previous day. By14 August a bulge protrudes on the southern flank of the Po plume in midbasin. This is the manifestationof a postwind instability discussed later in the text.


5 of 12

XXXXXX

292 current is the response to the bora jet at the level of Trieste293 and is very much expected. Drifter data [Lee et al., 2005;294 Poulain et al., submitted manuscript, 2007] confirm its295 existence. Simulated and observed velocities agree not only296 in location and direction but also in magnitude (Table 1).297 [22] Next, we note the strengthening of the westward298 current South of the Istrian Peninsula, obviously driven by299 the bora jet at the level of Pula and in geostrophic (thermal300 wind) balance with the developing zonal front. A feature301 noted in the post-bora current chart (Figure 8, right) is the302 surprising bifurcation of this current at the southern tip of303 Istria, with a minor branch turning sharply northward and304 following the coastline up to 45�N (same latitude as the Po305 River Delta), where it veers offshore. The other and more306 important branch of the current begins to diverge upon

307approaching the Italian coast, with about two thirds of the308flow turning left to join the Western Adriatic Current309(WAC) down the Italian coastline and the remaining third310turning right to augment a current flowing diagonally across311the basin from the Po River Delta to the Gulf of Trieste,312which we shall hereafter call the oblique current. The313smaller branch, that which detaches at the southern tip of314Istria and veers westward at 45�N, joins the oblique current315and reinforces it.316[23] The physical reason for the bifurcation at the south-317ern tip of Istria, the northward branch flowing along the318coast and its abrupt veering eastward toward the open sea at31945�N may seem at first puzzling. Indeed, one would have320expected a counterwind (eastward) flow at that latitude of321lull in the wind, acting as the flow compensating for the two

Figure 4. Sea surface currents (daily averaged) during the August 2001 bora event and the followingfew days. Note the development and persistence of a southward current along the western shore of Istria.This is identified with the Istrian Coastal Countercurrent (ICCC). Note also the anticyclonic (clockwise)eddy at (13�E, 44.2�N) becoming closed on 12 August and persisting for the following 2 days, and themeandering of the coastal current along the Italian coast, the meanders of which do not travel.


6 of 12

XXXXXX

322 westward jets at the latitudes of Trieste and Pula. However,323 one should not forget about Ekman dynamics, which can324 explain veering to the right of Rijeka Bay waters once they325 reach the southern tip of Istria. Another and not exclusive326 reason is friction: The flow hugging the coast inside Rijeka327 Bay is subjected to a boundary shear stress, which creates328 anticyclonic vorticity; once in the open sea, water with329 anticyclonic vorticity has a tendency to turn to the right.330 [24] The strength of the oblique current on 20 February331 2003, by the end of the bora event, is naturally explained by332 geostrophy. Computer simulations as well as observations333 reveal that the current is accompanied by a relatively strong334 thermal gradient (see Figure 1 of Lee et al. [2005]). Thus it335 is a geostrophic current in thermal wind balance with a336 cross-current density gradient. What is more difficult to337 explain is its oblique orientation, at a definite angle with338 respect to the main wind direction. To explore this question,

339we ran our model with the bora wind but no accompanying340cooling of the sea surface (simulations not shown), and the341result showed a much weaker and more zonal flow, nearly342aligned with the wind axis (although obviously in the lull343between the Trieste and Rijeka jets), leading us to conclude344that the surface heat loss is responsible for the obliquity of345the current. The actual heat loss is greater in the Po plume346region, and this not only creates a stronger density contrast,347and hence a stronger current, but also anchors the current348further south at its western origin. The need to return waters349to the Gulf of Trieste, where bora creates a sea surface350depression, sets the terminal point of the current. The351current chooses the straight path between both ends, hence352its oblique angle.353[25] Figure 9 (left) shows the velocity distribution at 15 m354depth on 20 February 2003. By comparing with the surface355velocity distribution on the same day (Figure 8, right), we

Figure 5. Ultimate decay of the southward current along Istria, according to both (left) model(computed sea surface density) and (right) observations (SeaWiFS images). By 17 August, the currentclosest to the Istrian coast has reversed, now flowing northward (black arrow in top left plot), while thesouthward countercurrent is found offshore (yellow arrow in top left plot) then rolls up in an anticycloniceddy by 23 August (circled area in bottom left plot). The anticyclonic vorticity was that imparted to thesea by the sheared bora jet in the region of the current’s origin.


7 of 12

XXXXXX

Figure 6a. Density and currents along cross-basin Section 1 (indicated in Figure 1) according to themodel simulation of 11 to 14 August 2001. The summertime vertical stratification is disrupted by thelighter Po water along the Italian coast (left end of plots), the upwelling-downwelling pattern to its east(straddling the 46.9-km mark), and weak upwelling against Istria (right end of plots). The verticalvelocity component is exaggerated to reveal it.

Figure 6b. Currents across Section 1 (indicated in Figure 1) according to the model simulation of 11 to14 August 2001. Positive values indicate northward currents. Note the Western Adriatic Current (WAC)along the Italian coast (left end of plots), an anticyclonic formation to its east (around the 46.9-km mark),associated with the upwellling noted in Figure 6a, and the ICCC along Istria (far right of plots).


8 of 12

XXXXXX

Figure 7. Simulated sea surface temperature on 12 and 20 February 2003, before and after the 10-daybora event. Note the formation of two fronts: a zonal front extending westward from the southern tip ofIstria (color variation from orange to yellow) and a wavy front running from the Po River delta to thenorthern corner of Istria (color variation from yellow to green).

Figure 8. Simulated surface currents on 12 and 20 February 2003. Note the double bifurcation of thewind-driven current rounding the southern tip of Istria, with a first, weaker northward branch hugging thecoastline and a second, stronger northward branch emerging in midbasin. The remaining stem ofthe current cuts across the basin. The zonal part of the current is in geostrophic equilibrium with the frontat the same latitude. Note also the reverse current originating from an eddy near the Po River Delta andcutting obliquely across the northern basin to the Gulf of Trieste.


9 of 12

XXXXXX

356 note that currents are similarly aligned (barotropic) about357 everywhere, except at one location. Along the western shore358 of Istria, there is almost no flow at the surface while there is359 a noticeable southward current at 15 m. Such flow actually360 occupy the depth range of 9 to 23 m and is about 3 cm/s361 strong. Three days later, after the bora wind has died away,362 this subsurface current has reached the surface, forming a363 well developed countercurrent about 10 cm/s strong. Such364 countercurrent at the conclusion of a wintertime bora365 typically occurs off the southern segment of the Istrian366 coast and is explained by the negative (anticyclonic) vor-367 ticity imparted by the bora wind jet and the sea-level rise in368 the area [Orlic et al., 1994; Kuzmic et al., 2006].369 [26] Before concluding this discussion of the simulation370 of the mid-February 2003 bora event, we present vertical371 sections to complement the observations which were for the372 most part on the surface. Figure 10 shows the salinity and373 velocity along longitudinal section A indicated in Figure 1.374 The obvious features are the homogeneity of the Adriatic375 winter waters to a depth of about 20 m and the strong zonal376 front at the level of Pula (at about position 40 km along the377 section), together with its associated geostrophic current (up378 to speed of 22 cm/s at the surface and decreasing with379 depth). What the section reveals that was not known from

380the surface plots is the downwelling-upwelling pattern381associated with this current. The current is accompanied382by downwelling on both flanks and narrow but intense383upwelling at its center.

3845. Discussion and Conclusions

385[27] The DieCAST model employed here successfully386simulated two distinct bora events over the Adriatic Sea,387one in summer when the water is vertically stratified and the388other in winter when density differences exist only in the389horizontal direction. The summer case (11–20 August3902001) is complex, with numerous mesoscale features391arising, moving and decaying. Paradoxically, the current392along the Istrian coast is found to be directed to the left of393the wind, in opposition to Ekman dynamics. Far from being394an artifact of the model, this current is corroborated by395observations and explained by the tip-shaped form of the396Istrian Peninsula and the jet nature of the bora wind.397[28] This southward current along Istria may be identified398with the occasional Istrian Coastal Countercurrent described399by Supic et al. [2000]. According to these authors,400the normally northward current along Istria, which is a leg401of the overall cyclonic circulation in the Adriatic basin,

t1.1 Table 1. Comparison Between Simulated and Observed Velocities in the Northern Adriatic Basin on 19–

20 February 2003a

Location of Current Latitude Longitude Simulated Value Drifter Value ADCP Valuet1.2

Southern tip of Istria 44.6�N 13.8�E 30–35 (no data) 31–34t1.3Zonal current 44.5�N 13.5�E 25–30 (no data) 20–30t1.4Oblique current 45.0�N 13.1�E 20–25 22–26 10–25t1.5Northern shore 44.3�N 12.7�E 18–24 14–16 12–28t1.6

aValues are in cm/s.t1.7

Figure 9. Comparison of (left) currents at 15 m depth on 20 February 2003 and (right) surface currentson 23 February 2003. This reveals how the southward current along Istria is a submerged current on20 February (seen at 15 m in left plot of Figure 9 but with no surface expression according to the rightplot of Figure 8) and emerges at the surface 3 days later (right plot).


10 of 12

XXXXXX

402 occasionally reverses in summer. In situ data collected403 during the years 1968 to 1997 have revealed a southward404 current in the month of August of some but not all years.405 While Supic et al. [2000] suggest that the appearance of this406 reversed current, dubbed the Istrian Coastal Countercurrent407 (ICCC), is somewhat correlated with air-sea fluxes and the408 Po River discharge over a month-long period or longer prior409 to the current reversal, this does not exclude that a brief bora410 event may be the necessary trigger. Indeed, our numerical411 simulations suggest that two ingredients are necessary for412 the establishment of the ICCC: first, a preconditioning413 phase that ends with different water characteristics on the414 eastern and western shores of Istria, and, second, a bora415 event that brings the latter water against the former. The416 ICCC then arises as the geostrophic adjustment to the417 imbalance caused by the bora-induced water displacement.418 This mechanism is presented here merely as a conjecture. A419 further study, across multiple years, should be conducted to420 assess whether an August occurrence of an extended Po421 River plume followed by a bora event is a systematic cause422 of the ICCC. Note also that this conjecture is not exclusive,423 for there could be different reasons for the occurrence of the424 ICCC, such as the creation of an anticyclonic circulation425 pattern in the northern Adriatic under persistent wind and426 surface flux patterns.427 [29] While the ICCC appears to be a spill of less dense428 water over denser wind-upwelled water leading to a persis-429 tent geostrophically adjusted flow, the net effect is quite430 paradoxical. Normally, when a strong wind blows, Ekman431 dynamics cause a flow in the water to the right (in the432 Northern Hemisphere) of the wind. In the case of a NE bora,433 the current would be northwestward, i.e., up the Istrian434 coast, not the opposite as observed and modeled. The435 explanation of a leftward current is twofold. First, Ekman436 drift lasts only as long as the wind blows and thus ceases437 quickly when the event is brief like the bora episode of 10–438 11 August 2001. Second, the jet-like nature of bora confines439 the upwelling (and its temporary Ekman drift) to the Gulf of440 Trieste and Rijeka Bay, while the waters off Istria are not441 upwelled. On the contrary, the counterwind return flow

442from Italy toward Istria between the bora jets of Rijeka443and Trieste (Figure 2) brings more freshwater against Istria444(Figure 3), just the opposite of what upwelling would do!445This was already noted by Orlic et al. [1986]. The resulting446along-basin density gradient between upwelled waters in447Rijeka Bay and the less dense water along the western shore448of Istria is what triggers the southward current. In sum, the449paradoxical direction of the surface current is due to the450geography of the coastline and the narrowness of the wind451jet.452[30] It is worth noting that Zavatarelli and Pinardi453[2003], using a model forced with climatological wind454fields, did reproduce what appears to be the ICCC for the455summer season and particularly for the month of September,456at least with their less dissipative, 1.5-km grid resolution457[Zavatarelli and Pinardi, 2003, Figures 15d and 16b]. Their458current strength (about 15 cm/s) is higher than the mean459value observed by Supic et al. [2000] (less than 8 cm/s) and460our event-specific value (about 12 cm/s). Given the low461temporal (climatological) and spatial (only 1.125 degrees)462resolution of the ECMWF reanalysis winds used by463Zavatarelli and Pinardi [2003], it is not possible to464associate their countercurrent along Istria with short and465strong wind events such as bora, unless there is a cumula-466tive effect in the mean. Zavatarelli and Pinardi [2003] offer467no dynamical explanation of their result.468[31] The simulation of the February 2003 bora event469reproduces the intricate pattern of current bifurcations and470confluences noted in the field observations. The model also471reproduces the so-called oblique current that flowed472diagonally from the Po River Delta to the Gulf of Trieste.473An earlier version of the simulation, one with climatological474heat flux instead of the actual heat flux, was unable to475produce this current, and from this, we conclude that this476bora event did not only drive waters but also created new477water densities, which in turn were driven by the wind, in478contrast with summer bora.479[32] Upon closer examination it further appears that it is480not so much the cooling caused by bora as its gradient481(stronger cooling to the north than to the south [see Kuzmic

Figure 10. Salinity and velocity along longitudinal Section A on 20 February 2003. (top) Salinity(colors) and velocity vectors in the plane of the section (arrows); the vertical velocity is exaggeratedto reveal the structure of its distribution. (bottom) Magnitude of the velocity component normal tothe section; positive values indicate eastward flow (as indicated by arrows accompanying Section A inFigure 1).


11 of 12

XXXXXX

482 et al., 2006, Figure 2]) that is the root cause of this current.483 Indeed, the resulting temperature gradient (increasingly484 lower temperatures to the north) causes a density gradient485 that is largely responsible for the current. Note that the486 lower salinity of the Po River plume helps strengthen the487 current by lowering the density on the other side.

488 [33] Acknowledgments. Support for this study was provided by the489 US Office of Naval Research under grant N00014-02-1-0065 to Dartmouth490 College. The authors also wish to thank David E. Dietrich for technical491 support with the model, James D. Doyle for providing COAMPS wind data492 sets, Elena Mauri for supplying satellite pictures, and Pierre-Marie Poulain493 and Vlado Malaeic for helpful comments on an earlier version of this paper.

494 References495 Artegiani, A., D. Bregant, E. Paschini, N. Pinardi, F. Raicich, and A. Russo496 (1997), The Adriatic Sea general circulation. Part I: Air-sea interactions497 and water mass structure, J. Phys. Oceanogr., 27, 1492–1514.498 Beg Paklar, G., V. Isakov, D. Koracin, V. Kourafalou, and M. Orlic (2001),499 A case study of bora-driven flow and density changes on the Adriatic500 Shelf (January 1987), Cont. Shelf Res., 21, 1751–1783.501 Bergamasco, A., and M. Gacic (1996), Baroclinic response of the Adriatic502 Sea to an episode of bora wind, J. Phys. Oceanogr., 26, 1354–1369.503 Cushman-Roisin, B., M. Gacic, P.-M. Poulain, and A. Artegiani (2001),504 Physical Oceanography of the Adriatic Sea: Past, Present and Future,505 304 pp., Kluwer Acad., Norwell, Mass.506 Cushman-Roisin, B., K. A. Korotenko, C. E. Galos, and D. E. Dietrich507 (2007), Simulation and characterization of the Adriatic Sea mesoscale508 variability, J. Geophys. Res., 112, C03S14, doi:10.1029/2006JC003515.509 Dietrich, D. E. (1997), Application of a modified ‘‘A’’ grid ocean model510 having reduced numerical dispersion to the Gulf of Mexico circulation,511 Dyn. Atmos. Oceans, 27, 201–217.512 Grubisic, V. (2004), Bora-driven potential vorticity banners over the513 Adriatic, Q. J. R. Meteorol. Soc., 130, 2571–2603.514 Hellerman, S., and M. Rosenstein (1983), Normal monthly wind stress over515 the world ocean with error estimates, J. Phys. Oceanogr., 13, 1093–516 1104.517 Hodur, R. M., X. Hong, J. D. Doyle, J. Pullen, J. Cummings, P. Martin, and518 M. A. Rennick (2002), The Coupled Ocean/Atmosphere Mesoscale519 Prediction System (COAMPS), Oceanography, 15(1), 88–98.520 Jurcec, V., and N. Brzovic (1995), The Adriatic bora: Special case studies,521 Geofizika, 12, 15–32.522 Kuzmic, M. (1986), Wind-curl vs. variable viscosity: A northern Adriatic523 related modelling study, Geofizika, 3, 64–74.524 Kuzmic, M. (1991), Exploring the effects of bura over the northern525 Adriatic: CZCS imagery and a mathematical prediction, Int. J. Remote526 Sens., 12, 207–214.527 Kuzmic, M. (1993), A satellite observation of the Adriatic Sea response to a528 spatially heterogeneous wind, Geofizika, 10, 1–18.529 Kuzmic, M., I. Janekovic, J. W. Book, P. J. Martin, and J. D. Doyle (2006),530 Modeling the northern Adriatic double-gyre response to intense bora531 wind: A revisit, J. Geophys. Res., 111, C03S13, doi:10.1029/532 2005JC003377.533 Lee, C. M., et al. (2005), Northern Adriatic response to a wintertime bora534 wind event, Eos Trans. AGU, 86, 157–165.

535Loglisci, N., M. W. Qian, N. Rachev, C. Cassardo, A. Longhetto, R. Purini,536P. Trivero, S. Ferrarese, and C. Giraud (2004), Development of an537atmosphere-ocean coupled model and its application over the Adriatic538Sea during a severe weather event of Bora wind, J. Geophys. Res., 109,539D01102, doi:10.1029/2003JD003956.540Orlic, M., M. Kuzmic, and Z. Vucak (1986), Wind-curl currents in the541Northern Adriatic and formulation of bottom friction, Oceanol. Acta, 9,542425–431.543Orlic, M., M. Kuzmic, and Z. Pasaric (1994), Response of the Adriatic Sea544to the bora and sirocco forcing, Cont. Shelf Res., 14, 91–116.545Pirazzoli, P. A., and A. Tomasin (1999), Recent abatement of easterly winds546in the northern Adriatic, Int. J. Climatol., 19, 1205–1219.547Poulain, P.-M., and F. Raicich (2001), Forcings, in PhysicalOceanography548of the Adriatic Sea: Past, Present and Future, edited byB.Cushman-Roisin549et al., pp. 45–65, Kluwer Acad., Norwell, Mass.550Prettner, J. (1866), Die Bora und der Tauernwind, Z. Osterr. Gesellsch.551Meteorol., 1, 210–214, 225–230.552Pullen, J., J. D. Doyle, R. Hodur, A. Ogston, J. W. Book, H. Perkins, and553R. Signell (2003), Coupled ocean-atmosphere nested modeling of the554Adriatic Sea during winter and spring 2001, J. Geophys. Res.,555108(C10), 3320, doi:10.1029/2003JC001780.556Pullen, J., J. D. Doyle, T. Haack, C. Dorman, R. P. Signell, and C. M. Lee557(2007), Bora event variability and the role of air-sea feedback, J. Geo-558phys. Res., 112, C03S18, doi:10.1029/2006JC003726.559Rachev, N., and R. Purini (2001), The Adriatic response to the bora forcing:560A numerical study, Nuovo Cimento Soc. Ital. Fis. C, 24, 303–311.561Raicich, F. (1994), Note on the flow rates of the Adriatic rivers, Tech. Rep.562RF 02/94, 8 pp., CNR Ist. Sper. Tallassogr., Trieste, Italy.563Raicich, F. (1996), On the fresh water balance of the Adriatic Sea, J. Mar.564Syst., 9, 305–319.565Smith, R. B. (1985), On severe downslope winds, J. Atmos. Sci., 42, 2595–5662603.567Smith, R. B. (1987), Aerial observations of the Yugoslavian bora, J. Atmos.568Sci., 44, 269–297.569Staneva, J. V., D. E. Dietrich, E. V. Stanev, and M. J. Bowman (2001), Rim570current and coastal eddy mechanisms in an eddy-resolving Black Sea571general circulation model, J. Mar. Syst., 31, 137–157.572Stravisi, F. (1977), Bora driven circulation in Northern Adriatic, Boll. Geo-573fis. Teor. Appl., 19, 95–102.574Supic, N., M. Orlic, and D. Degobbis (2000), Istrian Coastal Countercurrent575and its year-to-year variability, Estuarine Coastal Shelf Sci., 51, 385–576397.577Vested, H. J., P. Berg, and T. Uhrenholdt (1998), Dense water formation in578the Northern Adriatic, J. Mar. Syst., 18, 135–160.579Vilibic, I., B. Grbec, and N. Supic (2004), Dense water generation in the580north Adriatic in 1999 and its recirculation along the Jabuka Pit, Deep581Sea Res., Part I, 51, 1457–1474.582Wang, X. H. (2005), Circulation and heat budget of the northern Adriatic583Sea (Italy) due to a Bora event in January 2001: A numerical model584study, Ocean Modell., 10, 253–271.585Yoshino, M. M., (Ed.) (1976), Local Wind Bora, 289 p., Univ. of Tokyo586Press, Tokyo.587Zavatarelli, M., and N. Pinardi (2003), The Adriatic Sea modelling system:588A nested approach, Ann. Geophys., 21, 345–364.

��590B. Cushman-Roisin and K. A. Korotenko, Thayer School of Engineering,591Dartmouth College, Hanover, NH 03755-8000, USA. ([email protected])


12 of 12

XXXXXX

Mesoscale-resolving simulations of summer and winter bora...

Documents

Transcript of Mesoscale-resolving simulations of summer and winter bora...