Ecosystem Modeling in Cobscook Bay, Maine: A Boreal, Macrotidal Estuary 2004 Northeastern Naturalist 11 (Special Issue 2):23–50 Modeling Tidal Circulation and Exchange in Cobscook Bay, Maine DAVID A. BROOKS1 Abstract - Cobscook and Passamaquoddy Bays and their connecting passages lie at the entrance to the Bay of Fundy, on the eastern boundary between the United States and Canada where the mean tidal range is about 6 m. Vigorous tidal currents maintain cold temperatures and efficient exchange with offshore waters year-round. Over the last several decades, a net-pen salmon aquaculture industry has developed in both bays. Recent outbreaks of fish diseases have led to heightened concerns about tidal coupling between net-pen sites and potential pathways for disease transmission. This paper summarizes a Cobscook Bay circulation study by Brooks et al. (1999) and presents some new results to improve understanding of the tidal circulation and potential exchange pathways linking the bays. A dipole pair of back-eddies forms in the central part of Cobscook Bay on each flood and plays an important role in the dispersion and retention of particles. Direct and indirect observations support the existence of the eddy pair and some associated flow details. Model flushing times are a day or two in the Outer Bay where most aquaculture lease sites are located but a week or longer in the inner arms of the Bay. The inner part of South Bay appears to be a repository for particulate matter. Model experiments including both bays and their connecting passages suggest that most water entering Cobscook Bay comes from Head Harbor Passage adjacent to Campobello Island, illustrating the importance of well-coordinated international plans for ecosystem management. Introduction and Background From earliest reported experience, Cobscook Bay and the adjoining Passamaquoddy Bay (Fig. 1) have been associated with an unusually rich array of natural resources. When western Europeans first arrived in this region, they entered waters already well known and exploited for great quantities of alewives, herring, pollock, cod, haddock, and other fish species. The French explorer Sieur du Monts sailed into the bays in 1604 and established a colony on an island in the Saint Croix River, ultimately determining the eastern boundary between the United States and Canada (Brooks 1984). The colonists suffered a disastrous winter, losing more than half their number to disease and conflicts with a well-established native population. The survivors retreated to Port Royal, (now Annapolis Royal, NS) the Department of Oceanography, Texas A&M University, College Station, TX 77843; dbrooks@ocean.tamu.edu. 24 Northeastern Naturalist Vol. 11, Special Issue 2 following spring, but not before recording that the native-American name Passamaquoddy referred to great quantities of pollock taken from the Bay (Kilby 1888). Figure 1. The Cobscook–Passamaquoddy Bay archipelago at the entrance to the Bay of Fundy, showing the mainland, passages, and islands defining the bays. The international border is shown by the dash-dotted line. The inset box identifies the domain of the circulation model for Cobscook Bay. 2004 D.A. Brooks 25 Cobscook Bay, ME, is located at the mouth of the Bay of Fundy, where the mean tidal range is 5.7 m (Figs. 1 and 2). The consequent vigorous exchange with the offshore waters is largely responsible for the unusually productive and diverse ecosystem of the Bay environment (Larsen 2004). In the first half of the twentieth century, abundant herring supported a prominent sardine industry that collapsed in the late 1950s, for reasons not well understood. Likewise, stocks of cod and haddock have significantly declined in recent decades. Currently, Cobscook Bay supports a wide variety of benthic organisms (Trott 2004), including scallops, clams, mussels, urchins, and macroalgae, all of which are commercially important. In the early 1990s, commercial net-pen salmon aquaculture was introduced in Cobscook and Passamaquoddy Bays. Since then, aquaculture has become an important contributor to the economy on both sides of the border. The aquaculture industry in Maine generates revenues of about $75 million annually, of which about 90% is from salmon farms, Figure 2. Map showing the detailed geographic features and place-names defining Cobscook Bay, ME. Note the location of the Dennys and Pennamaquan Rivers. For reference, the Bay is divided into Inner, Central, and Outer sub-regions. 26 Northeastern Naturalist Vol. 11, Special Issue 2 placing salmon aquaculture second only to lobsters in importance to the total fishery economy (Sowles and Churchill 2004). The increasing density of salmon pens leads to questions about sustainable levels for aquaculture in the bays. Concerns focus on nutrient loading and the potential for eutrophication associated with concentrated sources of fish wastes and unconsumed fish feed near pen sites. The central questions relate to the significance of aquaculture- related nutrient sources compared to natural ones, and the capacity of the tidal circulation to flush the Bay of pollutants and waste products. Recent outbreaks of fish diseases, particularly infectious salmon anemia (ISA), have lead to fallowing of fish farms and destruction of fish on both sides of the border, severely impacting the local economy. There is also growing concern about potential impacts of farmed-fish diseases on the wild salmon population (National Research Council 2003). The recent ISA occurrences have highlighted the need for international coordination and management of aquaculture operations based on sound scientific knowledge and cooperation between the various stakeholders. To address such questions, it is necessary to understand the capacity of the tidal circulation to flush pollutants from the bays. It is also important to assess the effectiveness of tidal water movements and mixing that may couple the bays and provide cross-boundary pathways for transmission of pathogens. As a step in this direction, a hydrodynamic circulation model recently was used to study the tidal circulation, dispersion, and flushing times in Cobscook Bay (Brooks et al. 1999). The modeling was part of a larger ecosystem study that modeled energy flow through the Bay and considered nutrient and oxygen distributions in the Bay, as well as seasonal measurements of the temperature and salinity, bottom sediments, and the distribution and growth rates of micro- and macroalgae (Campbell 2004). The present article briefly reviews results from the model study and focuses on the impact of its central result: a pair of back-set eddies that develops in the Central Bay on each flooding tide. The paper also introduces some newer results from a circulation model that encompasses both bays and illustrates some aspects of the tidal coupling between them. The tides The unusually large tidal range in Cobscook and Passamaquoddy Bays results from their immediate proximity to the Bay of Fundy, well known for extreme tides. The large range is due to a near-resonance of the semidiurnal tide of the North Atlantic Ocean with the Gulf of Maine-Bay of Fundy system (Garrett 1972). The principal lunar semidiurnal tidal constituent (M-2), by far the most important in the Quoddy region, has a period of 12.42 hours, or 12 hours and about 25 2004 D.A. Brooks 27 minutes. Generally, two highs and two lows occur per solar day, but the times of high and low water are delayed each day by about 50 minutes because of the moon’s prograde orbital motion. For discussion purposes, it is convenient to reckon the M-2 tidal period as exactly 12 lunar hours, noting that one lunar hour equals 1.035 solar hours. Previous studies Because of its large tidal range and nearly enclosed nature, the Cobscook-Passamaquoddy (“Quoddy”) region has obvious potential as a site for tidal-power generation (Brooks 1992, Trites 1961). Most of the available oceanographic data from the region have been collected as parts of studies related to various tidal power projects (Larsen and Webb 1997). Historical physical data from Cobscook Bay consist of temperature and salinity measurements at a few locations in different seasons (Loucks et al. 1974, Trites and Garrett 1983) and some residual current hints from a few drift bottles that escaped from Passamaquoddy Bay (Chevrier and Trites 1960). In a literature review of the Quoddy region, McGrail (1973) noted that, aside from tidal height and tidal current predictions, data from Cobscook Bay were sparse. To provide the data necessary to initialize a computer model, six survey cruises were conducted between early spring and late fall 1995 as part of the Cobscook Bay ecosystem study (Phinney et al. 2004). Brooks and Churchill (1991) and Brooks (1992) modeled the circulation in Cobscook Bay and tidal-power schemes in the Passamaquoddy region, respectively. Panchang et al. (1997) modeled waste transport at several aquaculture sites in Cobscook Bay. The Circulation Model The numerical model used for this study is a derivative of the Princeton Ocean Model known as ECOM-si (Blumberg and Mellor 1987, Casulli and Cheng 1992). Bottom topography is represented with a stretched vertical coordinate to preserve full vertical resolution in variable water depth. Vertical mixing is determined by the turbulence closure scheme of Mellor and Yamada (1982), and horizontal mixing follows the Smagorinsky (1963) method, which depends on the velocity gradients and the grid spacing. A free surface boundary allows application of tidal forcing and propagation of surface gravity waves. The coastal outline and geographic place names of land and water areas for Cobscook Bay are shown in Figure 2. The digitized outline of Cobscook Bay and several smoothed depth contours are shown in Figure 3. The horizontal resolution of the model grid is 255 m, and the vertical structure is represented by 10 cells, or 11 levels, including the surface and the bottom. The topography-following coordinate 28 Northeastern Naturalist Vol. 11, Special Issue 2 preserves the full vertical resolution of the model at all horizontal locations, leading to a grid with about 21,000 cells representing the waters of the Bay. Fresh water from the Dennys and Pennamaquan Rivers is included at the locations shown in Figure 2. The model exchanges water and other information across the open boundary between Eastport and Lubec. The water depth relative to mean low water (MLW) in each model grid cell was determined by manual interpolation from National Oceanic and Atmospheric Administration (NOAA) chart 13328, which has sufficient soundings such that at least one depth value typically was available in each cell. The mean was used in cells with more than one chart sounding; no additional smoothing was applied in the model calculations. The shaded region in Figure 3, with depths greater than 20 m below MLW, defines the principal tidal channel. The model used does not allow for wetting and drying of grid cells, so the intertidal zone is never exposed in the model and is not fully represented. To avoid numerical difficulties, the model intertidal region was “dredged” to maintain a small non-zero depth (0.45 m) at low water. Figure 3. Map showing the representation of bottom depth of Cobscook Bay, ME, below mean low water in the model. The contours are smoothed and show depths of 10, 15, and 20 m. The shaded region is the main tidal channel with depths greater than 20 m. Depths were interpolated from NOAA Chart 13328. 2004 D.A. Brooks 29 Cobscook Bay has approximately 37 km2 of intertidal flats, compared to a total Bay surface area of about 74 km2 at low water (Larsen et al. 2004), so this inadequacy is a significant shortcoming. Figure 4. From the 24 July 1995 cruise, (A) temperature (°C) and (B) salinity (ppt) on the vertical section between Birch and Gove Points that defines the boundary between the Central and Outer Bays of Cobscook Bay. The view is outward toward the open boundary, with Birch Point on the left. The section was occupied near the time of a low tide. B A 30 Northeastern Naturalist Vol. 11, Special Issue 2 Initial data The survey cruises sampled important sections at the same phase of the tide during successive springs and neaps in the spring, summer, and fall seasons (a winter cruise was not scheduled). The cruise tracks were chosen to provide vertical sections of temperature and salinity and other properties along lines separating the eastern, central, and western regions of the Bay. Phinney et al. 2004 give additional details about the data sets and station locations. The model experiments discussed here were initialized with data collected 3 May 1995. Figures 4A and 4B show the temperature and salinity on the vertical section between Birch and Gove Points on 24 July 1995, which can be compared with the May data on the same section shown by Brooks et al. 1999; both sections were completed near a neap low tide. The ranges of temperature and salinity on the section are small, and there is only weak vertical stratification in the presence of the strong tidal mixing, but there are obvious horizontal gradients. The freshest and warmest water is found south of the center of the channel, on the Gove Point side. In summer (July), the midchannel temperatures are slightly lower than at the channel sides because relatively cooler offshore waters are advected inward and mixed upward in the channel. In spring (May), the effect is reversed and the mid channel temperatures are slightly higher because relatively warmer offshore waters are advected into the Bay. Forcing and boundary conditions Tidal forcing was applied by oscillating the sea level height at the open boundary with the M-2 period and with amplitude 2.85 m, corresponding to the mean semidiurnal tidal range of 5.7 m. Only the M-2 tidal constituent is considered, so there is no spring-neap variation, diurnal inequality, or lunar ellipticity factor in the model tides. River run-off was specified from monthly mean fluxes determined from gauged flow in the Dennys River. For the Pennamaquan River, which had no flow data available, the flux was estimated to be 45% of the Dennys River flux, based on the relative area of the drainage basins of the two rivers. The maximum flow of 8 m3s-1 in the Dennys River occurred in April, with a secondary maximum of 5.5 m3s-1 in January. The temperature and salinity in the grid cells nearest the river mouths were set at 5 °C and 15 ppt. The Model Tidal Regime The model was started from May initial conditions and run for ten days to achieve a steady-state. For the subsequent M-2 cycle, the dependent variables at all levels in all grid cells were saved for analysis. 2004 D.A. Brooks 31 Because of the large volume of model data, only a few highlights can be shown in printed figures. Animation sequences, showing successive model output frames like a movie, provide an effective way to present the tidal circulation. Examples can be viewed on the Cobscook web site (http:// cobscook.tamu.edu). Brooks et al. (1999) give a detailed description of the model tidal cycle, as well as more information about the model and boundary conditions. The present discussion is limited to the central result and its implications. The eddy dipole The flood and ebb currents generally follow the deep channel from the open boundary into the Central Bay, with maximum current speeds reaching about 2 m s-1. The rising flood is partially blocked by the narrow opening leading into the Inner Bay, and as a result a pair of counter rotating “back-set” eddies forms in the Central Bay. The eddy doublet or dipole begins to form at about hour 9 in the model cycle, near the time of maximum flood. The dipole reaches its greatest development near hour 11 and remains clearly evident in the surface currents at hour 12, when the next ebb is just beginning off Gove Point and sea level is high at Eastport (Figs. 5A,B; times are lunar hours after high water at Eastport). The northern eddy supplies water to the rising tide in Pennamaquan and East Bays, and the southern eddy directs part of the inflow into South Bay. The development of the dipole largely determines how the flood spreads throughout the Central Bay, and the eddies, particularly the southern one, play an important role in the horizontal and vertical dispersion of particles and tracers. The southern eddy of the dipole extends nearly to the sea floor, as shown by an east-west vertical section extending from the tip of Denbow Neck to Seward Neck (Fig. 6). The contours show the northsouth component of the water velocity averaged over one tidal cycle, defining the tidal-residual circulation crossing the section. The dashed contours indicate southward flow into South Bay and conversely for the solid contours. The tidal-residual flow in the plane of the section, shown by the arrows, is essentially horizontal and eastward (toward Seward Neck) everywhere, indicating that the section is located slightly south of the eddy center. Averaged over multiple tidal cycles, there can be no net inflow or outflow in South Bay, and this is qualitatively apparent from Fig. 6. The maximum residual speed is about 12 cm s-1 on the Denbow Neck side, which can be compared to the maximum instantaneous surface speed of about 50 cm s-1 across the section at hour 11 in the model tidal cycle (Fig. 5B). 32 Northeastern Naturalist Vol. 11, Special Issue 2 Figure 5A. Model surface currents in Cobscook Bay at 11 and 12 lunar hours after high water at Eastport. 2004 D.A. Brooks 33 Figure 5B. E x p a n d e d view of model surface currents at 11 and 12 lunar hours in the Central Bay region, showing development of the eddy dipole. Currents after 11 hours Currents after 12 hours 34 Northeastern Naturalist Vol. 11, Special Issue 2 Tidal volumes McGrail (1973) lists low water and intertidal volumes in Cobscook Bay of 0.56 km3 and 0.49 km3, respectively. The corresponding model values are 0.95 km3 and 0.54 km3. Thus, on average about half a cubic kilometer of seawater enters the Bay on each flood and leaves on each ebb — roughly comparable to the mean outflow of the Mississippi River during the same interval (6.2 solar hours). By comparison, rivers bring an insignificant amount of fresh water into the Bay, less than one percent of the intertidal volume, but the freshening influence is noticeable on the Birch-Gove section (Fig. 4). The model low water volume is significantly larger than that reported by McGrail. The difference is not explained by the model treatment of the intertidal zone, which is never allowed to dry completely. It is not clear where the outer boundary of Cobscook Bay was defined by McGrail, yet this choice obviously affects the low-water and high-water volumes. The model high-water area, determined as the grid cell area (0.065 km2) times the number of “wet” cells (1392), is 90.5 km2, which can be Figure 6. Vertical section of tidal-residual currents crossing the southern eddy in the Central Bay (Fig. 5B). The transect runs from west on the left to east on the right. Solid contours indicate northward currents, and dashed contours (negative values) indicate southward currents crossing the section. Contours are labeled in cm/s. Arrows show flow in the plane of the figure. From Baca (1998). 2004 D.A. Brooks 35 compared to the high-water area of 111 km2 determined by analysis of satellite images (Larsen et al. 2004). Thus 20.5 km2 of nearshore intertidal is excluded from the model domain because of grid limitations (cf., Figs. 2 and 9). The model low-water area exclusive of the intertidal, represented by 1136 grid cells, is 73.8 km2, in almost exact agreement with Larsen et al. 2004. An independent estimate of the intertidal volume can be calculated from the mean tidal range (2.85 m) and the satellite-derived intertidal area (37 km2). Assuming a linear bottom slope in the intertidal region, the total volume that must be exchanged with each tide is 0.53 km3, in close agreement with both the model result and the McGrail value. The model tidal prism, or relative intertidal fraction, defined as (HW-LW)/HW volume, is 0.38, using values adjusted to account for the “dredged” intertidal in the model. Thus, about one-third of the highwater volume moves in and out of the Bay on each tide. Bulk flushing time estimates If the intertidal volume simply moved in and out of the Bay as a surface layer with no vertical mixing, there would be no exchange with deeper waters, leading to an infinite residence time for the deep waters. However, there clearly is strong tidal mixing in the Bay and considerable exchange between the intertidal and deeper waters during each tidal cycle. If the vertical mixing were complete on each flooding tide and none of the ebbing water re-entered on the next flood, then the bay-averaged flushing time would be about 3 tidal cycles or 1.5 days, since about onethird of the volume would be exchanged with each tide. This turns out to be a reasonable overall value for the main tidal channel in the Outer Bay, but there are wide variations over the entire Bay, as we shall see. The eddy dipole: is it real? The dipole pair of eddies that forms on each flood apparently plays a central role in the dispersion of suspended and dissolved materials in Cobscook Bay, so it is important to look for corroborating support for the model result. Direct support comes from a brief record of current measurements collected at a mooring adjacent to an aquaculture site near Sheep Cove (Fig. 2). The mooring was located just west and inshore of Red Island, about 1 km west of Birch Point (Brooks and Churchill 1991). Figure 7A shows three-day records of the eastward (solid line) and northward (dashed line) currents measured at that mooring and smoothed with a 2-hr low-pass filter. One of the interesting features is the reversal from westward to eastward flow that occurs a few hours after each flood begins. Brooks and Churchill (1991) suspected that the reversal was caused by eddying in the lee of Red Island during the flooding tide. However, the present model results show 36 Northeastern Naturalist Vol. 11, Special Issue 2 Time (hours) Current speed (m/sec) Modeled Sheep Cove Currents Figure 7B. Modeled surface currents at Sheep Cove from the grid cell nearest the location of the measured currents shown in Figure 7A. Note similar pattern of flow reversal during each flood cycle. Figure 7A. Measured near-surface currents at Sheep Cove site (see Fig. 2 for location). Smoothed with 2-hr low-pass filter. Solid line shows east-west component. Note flow reversal on each flood, e.g., at about 7, 19, 31 ... hours. Sheep Cove 2hrlp: east (—), north (--) Time (hours) Current speed (cm/sec) 2004 D.A. Brooks 37 that the reversal happens as the northern member of the eddy dipole expands eastward after the flood begins, causing the current at the mooring location to set toward the southeast or east (Figs. 5A,B). The model currents at the grid cell nearest the mooring location (Fig. 7B) reveal a similar reversal on each flood cycle, with similar times and duration of reversals. The comparison extends to interesting details such as the multiple-staged ebbs evident in both the measured and modeled currents. The diurnal inequality apparent in the measured current is not reproduced by the model, because only the lunar semidiurnal forcing component was used. Recently, the eddy dipole has been directly observed by tracing the surface flow with simple drifters (Fig. 8). Several replications of this experiment were conducted by students at Shead Memorial High School in Eastport, with the guidance of technical arts instructor Scott Fraser. The students designed and built the neutral surface drifters from ballasted PVC pipe without drogues. For the case shown (taken from the students’ website, www.cobscook.org, where results from additional experiments can be viewed), the drifters were released along the Birch- Gove transect just as the flood was beginning. The students tracked the drifters from small boats using portable Global Positioning System receivers. The drifter tracks in Figure 8 reveal the existence of a counter-rotating eddy pair in the Central Bay. Landsat images of the Bay showing color-coded distributions of surface turbidity provide indirect evidence of the eddy dipole (Fig. 9; Larsen et al. 2004). The “class 3” green color marks relatively clear offshore water entering the Bay near the time of maximum tidal flood. The pattern reveals the northward and southward splitting of the flood as it enters the constricted passage between Leighton and Denbow necks. Low-altitude aircraft photographs also reveal part of the southern eddy in surface sediments stirred up by scallop draggers (photos by Laurice Churchill shown in Brooks et al. 1997). Accumulations of muddy sediment on the bottom in the Central Bay appears to be associated with the two eddies (Kelley and Kelley 2004), indicating persistence of the eddy process over long times. The association suggests that the eddy recirculation may concentrate sediments derived from net-pens, which raises questions about potentially elevated levels of contaminants. It is also known that the bacterium Aeromonas salmonicida, responsible for the virulent fish disease furunculosis, can be accumulated in sediments from fish farms (Husevag and Lunestad 1995, Stewart 1998). Finally, local fisherman have reported that the tide sets northward along the western side of Gove Point during most of the tidal cycle (pers. comm. to D.A. Brooks). Taken together, the direct and indirect observations as well as local experience point to a persistent eddy dipole pattern that forms 38 Northeastern Naturalist Vol. 11, Special Issue 2 in the Central Bay during the flooding tide. Thus, with some confidence we can address the consequences of the recurring pattern in the tidal currents. Mixing and Residence Times The paths followed by neutrally-buoyant particles illustrate the chaotic nature of the tidal stirring in Cobscook Bay. Figure 10 shows the tracks of two particles released at the surface in adjacent model grid cells just as the flood begins at Eastport. The two particles initially were separated by only 255 m (the grid spacing), but they soon diverged and followed very different paths. One (dashed track) remained close to Seward Neck during the flood, leaving the particle near the center of the channel off Birch Point, where the ensuing ebb swept it out of the Bay. In contrast, the other particle (solid track) was carried farther inward on the flood, where it interacted with the northern eddy and became trapped for the remainder of the 10-day experiment. The sensitivity of the particle tracks to small changes in their initial position illustrates the non-linear nature of the tidal flow, called “Lagrangian chaos” by Zimmerman (1986). In an identical experiment, except that the particles were released near the bottom instead of at the surface, both particles were eventually flushed from the Bay, but only after many tidal cycles (Brooks et al. Figure 8. Map of Cobscook Bay, showing tracks of six neutral surface drifters over one tidal cycle starting at the beginning of the flood tide.. The drifters were made and the experiment was conducted by Shead Memorial High School students. 2004 D.A. Brooks 39 Figure 9. Landsat satellite Thematic Mapper image of Cobscook Bay taken near the time of maximum flood. The color scheme indicates relative levels of surface turbidity. The class 3 green color marks relatively clear offshore water entering the Bay. Image provided by Cynthia Erickson, Bigelow Laboratory for Ocean Sciences. 1999). The second experiment indicates a different fate for the nearbottom particles, which were sequestered in South Bay for most of the 10-day model run. The difference between surface and bottom particle trajectories shows the importance of the three-dimensional structure of the velocity field. Tidally-averaged characteristics A useful map of flushing time for the Bay can be developed by seeding model runs with neutrally-buoyant surface particles released in every grid cell at many different stages of the tidal cycle. For each run one particle was released at each grid cell at the surface and tracked for up to eight days after a two-day spin up period. Particles that were not flushed from the Bay during each run were assigned a residence time of eight days. The tidal-mean residence time was obtained by averaging over 12 separate model runs with surface releases at each (lunar) hour of the tidal cycle (Fig. 11A). In the main channel of the Outer Bay, most particles escaped regardless of the phase of the tide at the time of release, although there are a few regions where particles did not escape when released into the flood Classes 40 Northeastern Naturalist Vol. 11, Special Issue 2 (e.g., Johnson Bay) and a few places where none escaped regardless of when they were released (e.g., Bar Harbor). Residence times less than two days are almost entirely confined to the Outer Bay with the lowest values on the eastern side, where the deep channel lies close to Moose Island. In the extremities of the inner arms of the Bay, especially South and Whiting Bays, none of the particles escaped, regardless of the phase of the tide when released. Some influence of the Central-Bay eddy dipole on residence time is apparent. For example, residence times of three days or less occur in a larger region of the northern eddy than in the southern one, and the particles released into the northern eddy escaped the Bay during a greater range of tidal phases than in the southern eddy. This result is consistent with the tendency, noted earlier, for materials to be preferentially directed toward South Bay and to be sequestered there. The map of residence time shown in Figure 11A was determined from advection of neutral particles released at the surface at all locations Figure 10. Tracks of two neutral surface particles released in adjacent grid cells (open circles) as the tidal flood begins. Stars on the tracks indicate time intervals of 6 lunar hours, i.e., one-half tidal cycle. The particle indicated by the dashed line was ejected from the bay in one tidal cycle; the other escaped the ebb and became trapped for the duration of the model run. 2004 D.A. Brooks 41 uniformly over one tidal cycle. The result will be different for local releases at specific sites, or with intermittent or non-uniform release rates, or at subsurface locations. Sinking, dissolution, or aggregation of particles, all of which may be important to the fate of unconsumed fish feed pellets or waste products, for instance, have not been considered here. It is also important to recall that the influence of the wind has not been included in these model experiments. The importance of the vertical structure of the flow on the bulk flushing characteristics of the Bay is illustrated by a model experiment in which the three-dimensional distribution of a passive tracer (“dye”) is tracked by the model. In this experiment, every grid cell in the model domain is initialized with a concentration of 100 units, and then the concentration at all grid cells is determined as a function of time from the same advection-diffusion equation that governs evolution of the model temperature and salinity fields. The vertical and horizontal turbulent eddy diffusion parameters were determined as explained earlier. Figure 11B shows the time required for the surface-to-bottom average of the dye concentration to decay by one e-folding scale, i.e., decrease to 37% (e-1) of the initial concentration. The vigorous tidal currents and proximity to the offshore waters results in a vertically-averaged e-folding flushing time of about one day (two tidal cycles) in most parts of the Outer Bay, and also in the main tidal channel of the Central Bay. Exceptions are Johnson Bay near Lubec, where the required time is closer to two days, and the protected Bar Harbor and Carryingplace Coves, where the time is 3–4 days. The distal arms of the Central and Inner Bays have flushing times greater than a week, consistent with the surface flushing times derived from the particle-tracking model experiments (Fig. 11A). The influence of the dipole eddy is evident in the vertically-averaged flushing times in the region where the constriction leading to the Inner Bay causes the northward and southward diversion of water in the eddy pattern. Because of the counter-rotating dipole pattern, the vertically-averaged flushing times in the Central Bay are one to two days greater adjacent to Birch and Gove Points than off Leighton Point and Denbow Neck. Next we consider an experiment in which a fixed source concentration of 100 units is maintained at a single surface grid cell while the model tracks the concentration evolution at all grid cells. As before, the results are best viewed as animation sequences available at the cobscook.tamu.edu website. Figure 12 shows a single frame “snapshot” from an experiment with a continuous surface source located in the channel between Birch and Gove Points. The colors show percent concentration at the surface 192 hours (eight days) after the source became active. The influence of the southern eddy in the 42 Northeastern Naturalist Vol. 11, Special Issue 2 Figure 11B. Vertically-averaged time (in days) for a conservative tracer to exponentially decay by one e-folding scale, i.e., to 37% of the original 100% concentration initialized in every grid cell at every level. From Baca (1998). Figure 11A. C o l o r - c o d e d map of tidal-average residence time (in days) for particles released in each grid cell at the surface. Particles not flushed from the bay after 8 days were assigned a residence time of 8 days. Central Bay is obvious, with elevated tracer concentration apparent in the counterclockwise motion. As noted earlier, the surface flood is directed preferentially toward South Bay by the Birch Point promontory, so the dye is less evident in the northern eddy and in the northern arms of Central Bay. The pattern resembles the surface distribution of sediments stirred up in South Bay by scallop-dragging vessels (cf., photo by Laurice Churchill shown in Brooks et al. 1997). After eight days, almost no tracer remains at the surface in the Outer Bay, where the flushing is most effective. Surface concentrations are less than 20% in the Inner Bay, indicating that most of the material 2004 D.A. Brooks 43 injected at the source point is flushed from the Bay before it can pass through the Falls Island passages. It is interesting to examine the evolution of tracer concentration near the bottom, where the accumulation of particulate matter or pollutants may be important to the benthic community. In this experiment, the surface source of 100 units in the same location between Birch and Gove Points is active only for the first 24 hours. Thereafter the source is Figure 12. Surface tracer (“dye”) concentration eight days after a source has been activated and maintained with a steady value of 100% at a location in the Central Bay (red dot). The influence of the southern member of the eddy dipole is apparent. From Brooks et al. (1997). Figure 13. Concentration of tracer at the bottom seven days after a surface source was deactivated. Values are listed in percentage of source concentration in the Central Bay at the same location as in Figure 12. Note expansion of concentration scale. From Brooks et al. (1997). 44 Northeastern Naturalist Vol. 11, Special Issue 2 turned off and the model tracks the spatial and temporal evolution and decay of the tracer distribution at all levels. Figure 13 shows a single frame of the tracer concentration at the bottom 192 hours after the experiment begins, or seven days after the source becomes inactive. Since the surface source is neutrally buoyant, the concentration near the bottom would remain zero except for vertical advection and turbulent mixing, which bring the tracer into contact with the subsurface circulation and eventually the bottom. The website movie shows that the tidal flushing action rapidly clears most of the Bay of the tracer. The frame in Figure 13 shows that seven days after the source became inactive bottom tracer concentrations were reduced to less than 1% everywhere except in the inner part of South Bay, where the highest remaining concentration was about 4% in a small region at the southern tip (note the expanded color scale in Fig. 13 vs. Fig. 12). The movie confirms that the persistence of detectable tracer levels in South Bay occurs because the material emitted at the source location is preferentially steered toward the south as each flood forms off Birch Point and then is captured by the southern half of the eddy dipole. Although the absolute concentrations are low, the particulate matter represented by the tracer may include phytoplankton, benthic diatoms, and macroalgal detritus, important food sources for filterfeeding organisms such as scallops. Implications for Aquaculture Figure 14 shows the locations of authorized aquaculture leases in Cobscook Bay (Maine Department of Marine Resources 1996) and smoothed contours of surface flushing times derived from Figure 11A. Many of these lease sites are currently inactive (Sowles and Churchill 2004), and most are located in the Outer Bay where flushing times are less than a few days, according to the model. The inner part of South Bay and Dennys and Whiting Bays have much longer flushing times and do not appear to be suitable for lease sites. Disease vectors Passamaquoddy and Cobscook Bays recently have been impacted by outbreaks of infectious salmon anemia (ISA). In 1997, several sites were infected near Deer Island (Fig. 1), and in February 2001, the virus was first officially documented in Maine near Treat Island (Fig. 2). Cobscook Bay was completely fallowed from February to May 2002, with a loss of about 1.6 million fish (French 2002). Isolated cases of ISA were discovered near Eastport in July 2003. The ISA pathogen is known to survive outside a living host for at least several days in water with temperatures below 10 °C (Joint Government/ 2004 D.A. Brooks 45 Industry Working Group 2000). In a Norwegian study, Jarp (1997) concluded that “ISA is mainly transmitted from infected salmonid sources to clean sites through sea water,” noting that the risk of infection increased eightfold when sites were situated closer than 5 km, compared to cases with greater separations. A Scottish study (Joint Government/ Industry Working Group 2000) noted that a 5 km separation may be inadequate in areas with strong tides, and that sites should be separated by greater than two tidal excursions to inhibit the spread of infection. In Cobscook Bay, where tidal excursions (horizontal displacements during a flood-ebb cycle) may be ≈10 km in the main channel (Fig. 10), such separation distances are obviously impractical (Fig. 14). Currently, the smallest separation between an active US lease site (Treat Island) and a Canadian site (Campobello Island) is about 1.2 km. Several transborder lease sites in Western Passage are separated by less than 2 km. To address questions about possible tidal coupling between proximate pen sites, a circulation model was applied to the full Figure 14. The location of permitted aquaculture lease sites in 1995 is shown by the stars. Overlaid are contours of surface flushing time in days, smoothed from Figure 11A. 46 Northeastern Naturalist Vol. 11, Special Issue 2 Passamaquoddy–Cobscook region (Fig. 1), including the principal islands and passages that connect the bays with each other and with the offshore waters. The model (Hess 1989, 2000) is similar to the Cobscook-only version, except for a simplified vertical turbulence scheme (see Brooks 1992 for details). In the absence of adequate field data, initial temperature and salinity values were specified at open boundaries and at river mouths and interpolated at interior points. Annual- mean fresh water inflows were included for five principal rivers, and M-2 tidal sea level variations were imposed at open boundaries; wind influences were not included. Here we consider only a single drifter experiment in the passages linking the bays (Fig. 15). Figure 15. Model surface drifters tracked for 8 tidal cycles in the passages connecting Passamaquoddy and Cobscook Bays. Circles mark the release points. Symbols mark time intervals in lunar hours along the track lines. The neutrallybuoyant drifters were released sim u l t a n e o u s l y near the time of maximum flood in Head Harbor Passage. 2004 D.A. Brooks 47 Drifter tracks were determined by integrating the model velocity field starting from the initial positions of the particles (open circles). A small “random walk” based on the local horizontal turbulent mixing coefficient was added at each time step to approximate effects of tidal diffusion (Hess 1989). The drifters were released simultaneously near the time of maximum flood in Head Harbor Passage and tracked for eight tidal cycles. The track lines in Figure 15, augmented by others not shown, suggest that much of the surface water reaching Cobscook Bay flows close to the inner side of Campobello Island. For example, one of the Head Harbor drifters (no. 1) moved along the western shore of Campobello, passed near Treat Island, and thereafter became trapped in Johnson Bay. The drifter released closer to Indian Island (no. 2) escaped the inflow toward Cobscook Bay and moved toward Western Passage, where it was drawn into the narrow channel between Indian and Deer Islands; thereafter it continued northeastward along the eastern shore of Deer Island. The tendency for the flood in Head Harbor Passage to divide, with the branch adjacent to Deer Island turning northward into Western Passage, is also evident in a statistical study of model particle tracks in the region by Thompson et al. (2002). The drifter released near the west side of Western Passage (no. 3, track marked by stars), crossed Friar Roads off Eastport and then moved southward along the west side of Campobello Island, following nearly the same path as drifter 1 (marked by crosses). Drifter 4, initially only about 300 m from drifter 3, continued northward along the west side of Deer Island. Management issues The apparent coupling of the bays points to the importance of an integrated international management plan based on knowledge of the circulation patterns. Although there are uncertainties associated with, for example, inadequate resolution of the intense tidal eddying in Head Harbor and Western Passages, the preliminary model experiment suggests that sites in outer Cobscook Bay may be most susceptible to influences from sites on the western side of Campobello Island. The same Campobello sites appear to be downstream of sites on the US side of Western Passage, which suggests that an effective fallowing strategy should include all of those areas simultaneously. Additional modeling and field studies are needed for verification. Because of the vigorous tides, a 5 km separation distance between pen sites is inadequate to insure isolation, especially in and near the passages connecting the two bays. The same conclusion was reached by Stewart (1998), who noted that a Single Management Area (SMA) strategy is the only sensible approach for the Passamaquoddy region 48 Northeastern Naturalist Vol. 11, Special Issue 2 (SMAs are sub-regions thought to be isolated enough that separate fallowing cycles could be maintained in each without risking crossinfections). Stewart identified three SMAs: 1) the northern part of Passamaquoddy Bay, 2) Deer Island/Campobello Island and Letete Passage, and 3) Grand Manan Island. The preliminary results noted here, subject to field verification, suggest that SMA No. 2 should be expanded to include all of outer Cobscook Bay, plus the lower reaches of Western Passage. Acknowledgments The Cobscook Bay Marine Ecosystem Study was supported by a grant from the Andrew W. Mellon Foundation to The Nature Conservancy (TNC), with matching support provided by the participating institutions. TNC’s Barbara Vickery provided capable (and patient) program management. The circulation modeling component of the project was funded at Texas A&M University under Contract No. MEFO-12-07-94b from the Maine Chapter of TNC. The hydrographic data used to establish initial and boundary conditions for the Cobscook model were obtained from cruises in 1995 conducted aboard the R/V Otto Miller, Jr. of the Washington County Technical College’s Marine Technology Center in Eastport, ME. The fine services of Captain Tom Duym and Chris Bartlett are especially appreciated. David Phinney of the Bigelow Laboratory for Ocean Sciences served as Chief Scientist for most of the cruises, ably assisted by Doug Phinney. The Cobscook model calculations were carried out under the author’s guidance as part of the master’s thesis of Michael Baca (Baca 1998; reported in Brooks et al. 1999); several unpublished figures from his thesis are shown here (noted in captions). Rahilla Shatto and Amy Warren helped with preparation of some of the original figures; Annette deCharon of the Bigelow Laboratory helped me re-format most of the figures for this paper. I am grateful for the helpful comments and clarifying suggestions of several reviewers, particularly Thomas Trott in the preliminary stages, and later David Greenberg and Barbara Vickery. I also appreciate the thorough reading and detailed comments of an anonymous reviewer. Literature Cited Baca, M.W. 1998. A numerical study of circulation and mixing in a macrotidal estuary: Cobscook Bay, Maine. M.S. Thesis, Texas A&M University, College Station, TX. 82 pp. Blumberg, A.F., and G.L. Mellor. 1987. A description of a three-dimensional coastal ocean circulation model. Pp 1–16, In N.S. Heaps (Ed.). Three-dimensional Coastal Ocean Models. American Geophysical Union, Washington, DC. Brooks, D.A. 1984. The eastern maritime boundary between the United States and Canada. Eos 65:1209–1210. Brooks, D.A. 1992. Tides and tidal power in Passamaquoddy Bay: A numerical simulation. Continental Shelf Research 12:675–716. 2004 D.A. Brooks 49 Brooks, D.A., and L.U. Churchill. 1991. Experiments with a terrain-following hydrodynamic model for Cobscook Bay in the Gulf of Maine. Pp. 215–216, In Spaulding, Bedford, Blumberg, and Cheng (Eds.). Estuarine and Coastal Modeling. Proceedings of the 2nd International Conference, American Society of Civil Engineering, Reston, VA. 786 pp. Brooks, D.A., M.W. Baca, and Y.-T. Lo. 1997. Tidal mixing, dispersion, and flushing in Cobscook Bay, Maine: A hydrodynamic model study. Texas A&M University, College Station, TX, Technical Report 97-2-T (December) 77 pp. Brooks, D.A., M.W. Baca, and Y.-T. Lo. 1999. Tidal circulation and residence time in a macrotidal estuary: Cobscook Bay, Maine. Estuarine, Coastal, and Shelf Science 49:647–665. Campbell, D.E. 2004. Evaluation and Emergy Analysis of the Cobscook Bay ecosystem. Northeastern Naturalist 11(Special Issue 2):355–424. Casulli, V., and R. Cheng. 1992. Semi-implicit finite difference methods for three-dimensional shallow water flow. International Journal of Numerical Methods in Fluids 15:15629–15648. Chevrier, J.R., and R.W. Trites. 1960. Drift-bottle experiments in the Quoddy region, Bay of Fundy. Journal of the Fisheries Research Board of Canada 17:743–762. French, E. 2002. Hopes anchored to Cobscook recovery. Quoddy Tides 11 January 2002. Garrett, C. 1972. Tidal resonance in the Bay of Fundy and Gulf of Maine. Nature 238:441–443. Hess, K.W. 1989. MECCA Program Documentation. NOAA Technical Report NESDIS 46, Washington, DC. 258 pp. (addendum, September 1989, 29 pp.) Hess, K.W. 2000. MECCA2 Program Documentation. NOAA Technical Report NOS CS 5, Silver Spring, MD. 49 pp. Husevag, B., and B.T. Lundestad. 1995. Presence of the fish pathogen Aeromonas salmonicida and bacteria resistant to antimicrobial agents in sediments from Norwegian fish farms. Bulletin European Association of Fish Pathologists 15:17–19. Jarp, J. 1997. Infectious Salmon Anemia (ISA) risk factors in sea-cultured Atlantic salmon Salmo salar. Diseases of Aquatic Organisms 28:79–86. Joint Government /Industry Working Group. 2000. Final Report of the Joint Government/Industry Working Group on Infectious Salmon Anemia (ISA) in Scotland. Published by the Scottish Executive. Available from: FRS Marine Laboratory Aberdeen, PO Box 101, Victoria Road, Aberdeen AB11 9DB. Also online at: www.marlab.ac.uk/PDF/JGIWGReport.pdf Kelley, J.T., and A.R. Kelley. 2004. Controls on surficial materials distribution in a rock-framed, glaciated, tidally-dominated estuary: Cobscook Bay, Maine. Northeastern Naturalist 11(Special Issue 2):51–74. Kilby, W.H. 1888. Eastport and Passamaquoddy, Shead, Eastport, ME. Reprinted in 1982 by Border Historical Society, Eastport, ME. 533 pp. with preface. Larsen, P.F. 2004. Notes on the environmental setting and biodiversity of Cobscook Bay, Maine: A boreal, macrotidal estuary. Northeastern Naturalist 11(Special Issue 2):13-22. Larsen, P.F., and R.V. Webb. 1997. An environmental bibliography of Cobscook Bay and the Quoddy region. The Nature Conservancy, Brunswick, ME. 150 pp. 50 Northeastern Naturalist Vol. 11, Special Issue 2 Larsen, P.F., S. Barker, J. Wright, and C.B. Erickson. 2004. Use of cost effective remote sensing to map and measure marine intertidal habitats in support of ecosystem modeling efforts: Cobscook Bay, Maine. Northeastern Naturalist 11(Special Issue 2):225–242. Loucks, R.H., R.W. Trites, K.F. Drinkwater, and D.J. Lawrence 1974. Physical oceanographic characteristics. Pp. 1–60, In Summary of Physical, Biological, Socio-economic, and other factors relevant to potential oil spills in the Passamaquoddy Region of the Bay of Fundy. Journal of the Fisheries Research Board of Canada, Ottawa, ON, Canada. Technical Report No. 428. 215 pp. Maine Department of Marine Resources. 1996. Aquaculture Lease Inventory. L. Churchill and K. Honey (Eds.). Maine Department of Marine Resources, West Boothbay Harbor, ME. 80 pp. McGrail, D. 1973. Physical oceanography and meteorology. Pp. 20–46, In E. Shenton and D. Horton, (Eds.). Literature Review of the Marine Environmental Data for Eastport, Maine. The Research Institute of the Gulf of Maine (TRIGOM).Portland, ME. 130 pp. + appendices. Mellor, G.L., and T. Yamada. 1982. Development of a turbulence closure model for geophysical fluid problems. Reviews of Geophysics and Space Physics 20:851–875. National Research Council. 2003. Atlantic Salmon in Maine. Report of the National Research Council of The National Academies (pre-publication version, December 2003). Washington, DC. 240 pp. Panchang, V.J., G Cheng, and C. Newell. 1997. Modeling hydrodynamics and aquaculture waste transport in coastal Maine. Estuaries 20:14–41. Phinney, D.A., C.S. Yentsch, and D.I. Phinney 2004. Primary productivity of phytoplankton and subtidal microphytobenthos in Cobscook Bay, Maine. Northeastern Naturalist 11(Special Issue 2):101–122. Smagorinsky, J. 1963. General circulation experiments with the primitive equations. Monthly Weather Review 91:99–164. Sowles, J.W., and L. Churchill. 2004. Predicted nutrient enrichment by salmon aquaculture and potential for effects in Cobscook Bay, Maine. Northeastern Naturalist 11(Special Issue 2):87–100. Stewart, J.E. 1998. Sharing the waters: An evaluation of site fallowing, year class separation, and distances between sites for fish health purposes on Atlantic salmon farms. Canadian Technical Report Fisheries and Aquatic Sciences, 2218. 56 pp. Thompson, K.R., M. Dowd, Y. Shen, and D. Greenberg. 2002. Probabilistic characterization of tidal mixing in a coastal embayment: A Markov Chain approach. Continental Shelf Research 22:1603–1614. Trites, R.W. 1961. Probable effects of proposed Passamaquoddy power project on oceanographic conditions. Journal of the Fisheries Research Board of Canada 18:163–201. Trites, R.W., and C.J.R. Garrett. 1983. Physical oceanography of the Quoddy region. Pp 9–34, In M.L.H. Thomas (Ed.). Marine and Coastal Systems of the Quoddy Region, New Brunswick. Canadian Special Publication of Fisheries and Aquatic Sciences 64. Department of Fisheries and Oceans, Ottawa, ON, Canada. 306 pp. Trott, T.J. 2004. Late 20th century qualitative faunal changes in Cobscook Bay, Maine. Northeastern Naturalist 11(Special Issue 2):325–354. Zimmerman, J.T.F. 1986. The tidal whirlpool: A review of horizontal dispersion by tidal and residual currents. Netherlands Journal of Sea Research 20:133–154.