Modeling the Circulation in Penobscot Bay, Maine Huijie Xue 1, Yu Xu 1, David Brooks 2, Neal Pettigrew 1, John Wallinga 1 1. School of Marine Sciences, University of Maine, Orono, ME 4469-5741. 2. Department of Oceanography, Texas A& M University, College Station, TX 77843-3148. Abstract Penobscot Bay, with approximate dimensions 5 x 1 km, is the largest estuarine embayment along the Maine coast. It can be characterized by two deep channels on its eastern and western sides, which are separated by several islands and a shoal region in the middle of the Bay. Subtidal circulation in Penobscot Bay is influenced by winds, fresh water discharge from the Penobscot River, and the southwestward Maine Coastal Current flowing pass the mouth of the Bay. The Princeton Ocean Model was adapted to Penobscot Bay to simulate the circulation for the spring and summer of 1998. Observed winds at nearby Matinicus Rock and realistic river discharge rates were used to force the model. Open boundary conditions were specified using the results from a Gulf of Maine climatological model. Simulations were somewhat sensitive to the mixing coefficient in the model. When a background viscosity of 5x1-6 m 2 /s was used, the model reproduced the observed three-layer structure in the outer western bay with outflows near the surface and the bottom and inflows in the middle of the water column. In contrast, a two-layer estuarine like circulation was found in the outer eastern bay with outflows in the upper water column and inflows in the lower water column. 1. Introduction Penobscot Bay (Figure 1), where the Penobscot River watershed drains into the sea, is the largest estuarine embayment in the Gulf of Maine, the second largest on the U.S. east coast. There are two deep channels on the eastern and western side of the bay, separated by several islands and a shoal region in the 1
middle. Penobscot Bay has been historically and remains a very important fishery ground. Approximately 7 species of fin and shellfish are harvested in the bay and on the surrounding shelf, and the bay itself accounts for roughly 5% of the lobster landings for the entire state of Maine (Conkling, 1996). Penobscot Bay Gulf of Maine EMCC Orono Belfast Islesboro Castine Mount Desert Island North Deer Isle Haven Swans Rockland Vinalhaven Isle au Haut m Figure 1. A location map of Penobscot Bay and the bathymetry of the region. Knowledge of the circulation is fundamental to the understanding of Penobscot Bay as an ecological system. Based on the data of short-term moorings 2
deployed during the summer of 1969, 197, and 1974 (Normandeau Associates, 1975), Humphreys and Pearce (1981) found a counterclockwise circulation around Islesboro in the case of southerly winds. On the contrary, under similar southerly wind conditions Burgund (1995) found a clockwise circulation around Islesboro in the summer of 1994 using visually tracked surface drifters. The southwestward flowing Eastern Maine Coastal Current (EMCC) passes the mouth of Penobscot Bay. Sea surface temperature (SST) patterns as seen from Landsat images sometimes indicate the cold EMCC water entering the lower bay that may have profound effects on the circulation (pettigrew, 1998). On the other hand, it is often suggested that the outflow from Penobscot Bay causes the EMCC to be deflected offshore (Brooks, 1994; Bisagni et al, 1996). Pettigrew et al (1998) also observed that the outflow overrides the coastal current. Previous efforts to model the circulation in Penobscot Bay include the work of Fidler (1978), Humpherys and Pearce (1981), and Burgund (1995). These models lacked the density driven flow, which was the most serious shortcoming of applying simple hydrodynamic models to Penobscot Bay that has apparent estuarine nature. In addition these models had relatively coarse resolution and poor representation of the bay geometry, As a part of an integrated effort to create a Penobscot Bay Geophysical Infomation System describing the ecological characteristics of Penoscot Bay, we have developed a Penobscot Bay circulation model that simulates time-dependent, three-dimensional velocity, temperature and salinity in the bay and on the adjacent shelf. The principal objective of the circulation model is to identify and characterize features such as fronts and eddies and their associated temporal variability. It is hoped that the model will provide valuable information for understanding marine ecological conditions, and for predicting the effects of marine resource use and management activities. 2. The Model The Princeton Ocean Model (POM) (Blumberg and Mellor, 1987) is used in this study. The POM is a three-dimensional, fully nonlinear, primitive equation, ocean circulation model that includes complete thermal dynamics. It contains the second moment turbulence closure scheme of Mellor and Yamada (1982) to provide vertical mixing coefficients. The model can be forced with surface wind, surface heat and fresh water fluxes, river discharges, and boundary forcing from the open ocean (including sea level and tidal and subtidal flows) and solve for the time-dependent, three-dimensional velocity, temperature, salinity, vertical mixing coefficients, and the sea surface elevation. An orthogonal curvilinear grid with 151x121 discrete points covers Penobscot Bay and the adjoining Blue Hill Bay, a domain size roughly 1 km by 8 km (Figure 2). The grid size varies from less than 4 m inside the Bay to about one kilometer offshore. There are 15 sigma levels in the vertical. The model s 3
offshore boundary is placed on the shelf, about 3km seaward from the mouth of Penobscot Bay so that the coastal current may enter the study area and flow along the shelf portion of the model domain. N * PR WPB EPB * MR 1 km Figure 2. Orthogonal curvilinear grid of the Penobscot Bay circulation model. Asterisks mark the C-MAN station at Matinicus Rock (MR) and the Penobscot River inflow (PR). Dots show mooring locations in the western Penobscot Bay (WPB) and the eastern Penobscot Bay (EPB), respectively. In addition, observations were made along the sections across the outer western and eastern bay (heavy curves) on 29 April 1998. This paper presents model simulations of the period between April and September 1998. This period was chosen since lobster larvae usually settle in Penobscot Bay during summer. Hourly winds from the C-MAN meteorological station at Matinicus Rock and the daily Penobscot River discharge rates were used to force the model (Figure 3). The spring freshet of 1998 was somewhat unusual 4
with two peaks in April separated by a period of about two weeks. The river discharge rose again in June and July reflecting a relatively wet summer. Winds were strong but variable in the month of April. Southerly components became dominant starting in mid-may and continued throughout the summer except for two northeasters in the end of June and August. Two iepisodes are highlighted in Figure 3 (vertical lines). The first corresponds with the April hydrographic survey during which winds were from southwest, and the second is a rather strong southeasterly wind event when Penobscot Bay responded quite differently. Penobscot River Discharge Rate (m 3 /s) Wind Stress (dyn/cm 2 ) April Hydro Survey SE Figure 3. Penobscot River discharge rate and wind stress at Matinicus Rock from April to September 1998. Open boundary conditions, including both the tidal (M 2 only) and the subtidal forcing, were obtained from a larger scale climatological Gulf of Maine model (, 2). A gravity wave radiation condition was applied to the velocity component perpendicular to open boundaries. An upwind-advection scheme was applied to temperature, salinity, and the velocity component parallel to the boundary so that, in case of inflow, the boundary conditions derived from the climatological Gulf of Maine model were imported by inward velocities. 3. Comparisons of the Model Results with Observations 3.1. Time series Model simulations were performed for the period from April to September 1998. Results were averaged over the M 2 tidal cycle and saved to illustrate the subtidal circulation. During the same time period, two moorings with downward 5
looking Acoustic Doppler Current Profilers (ADCPs) were deployed in outer Penobscot Bay (see Figure 2). The ADCP in the western bay (WPB) was configured for 8m depth bins, giving a total of 11 bins between 1 and 9m. Four out of the eleven bins are shown in Figure 4a. The ADCP in the eastern bay (EPB) was configured for 2m depth bins. Due to instrument malfunction, only the top three bins of data centered at 8, 1, and 12 m were determined useful, and two of which are shown in Figure 4b. (a) (b) Figure 4. The west (a) and the east (b) Penobscot Bay ADCP tidal residual velocity time series. The northward current component is plotted along the ordinate as a positive, and the eastward component is plotted from its time origin in the positive direction along the abscissa. The depths of the bins are indicated. 6
In the western bay, velocities were consistently northward (indicating inflows) between 1 and 7 meters except around 14 June when the strong southeasterly wind event caused southward flows throughout the water column. Velocities near the bottom were predominantly southward (indicating outflows). Although velocities were not measured in the top 1 m, from hydrographic data Pettigrew (1998) showed a surface layer (from the surface down to about 5m) of southward Penobscot River outflow, which led to his postulate of a three-layer structure in the outer western bay. In contrast, the upper layer currents in the eastern bay were steadily southward except for the mid-june event (Figure 4b). The 1997 ADCP record showed that the outflows extended from the surface to about 3 m depth, whereas below 3 meters the currents reversed and flowed northward into the bay (Pettigrew, 1998). (a) 7
(b) Figure 5. (a) and (b) are model predicted velocity time series at the WPB and the EPB mooring site, respectively, from the experiment with ν equal to 5x1-5 m 2 /s. Two model experiments were carried out with different values of background viscosity, ν. In the POM, the background viscosity is added to the viscosity determined by the Mellor-Yamada closure scheme (Mellor and Yamada, 1982), and it becomes predominant when the local Richardson number is large. Figures 5 and 6 show model predicted velocity time series at the mooring sites in the experiment with ν equal to 5x1-5 m 2 /s and 5x1-6 m 2 /s, respectively. Differences between the two experiments are larger on the western side (Figures 5a and 6a). Southward flows in the upper water column tend to be weaker. On the other hand, flows near the bottom change from predominantly northward to predominantly southward by reducing the background viscosity from 5x1-5 m 2 /s to 5x1-6 m 2 /s. The eastern bay (Figures 5b and 6b) is less sensitive to the change of ν except near the bottom where the northward flows become more intermittent. When the smaller background viscosity was used, the model reproduced the three-layer structure in the western bay and the two-layer structure in the eastern bay postulated by Pettigrew (1998) (Figure 6). However, the surface southward flows reached more than 2 m on the western side but less than 2 m 8
on the eastern side. The former was too deep, and the latter too shallow when compared with the observed velocity time series shown in Figure 4. We are continuing the sensitivity study with different model parameterizations and open boundary conditions and hope to achieve better agreement of layer thickness. Despite the differences in layer thickness, the model captured the temporal variability on the synoptic scale, especially the abnormal response to the mid-june southeasterly event. Southward flows were found throughout the water column on the western side, whereas the surface flows on the eastern side reversed, changing from southward to northward. Although not observed, model results suggested that bottom flows on the eastern side reversed direction as well, changing from northward to southward during this event. (a) Figure 6. (a) and (b) are modeled velocity time series at the WPB and the EPB mooring site, respectively, from the experiment with ν equal to 5x1-6 m 2 /s. 9
(b) 3.2. Spatial variability It is clear from both the observed and the modeled velocity time series that the subtidal circulation in Penobscot Bay can be highly variable, especially in the western bay. The circulation also exhibits rich spatial variability. Two instances were selected to demonstrate the strong contrasts between the circulation patterns. The first corresponded to the hydrographic survey on 29 April. Model results were compared with the tidally-averaged, ADCP and CTD cross-sections made in the outer western and eastern bay (see Figure 2 for locations of the sections). The second was sampled during the strong southeasterly wind event from 13 to15 June. During the hydrographic survey, wind was from the southwest and light (Figure 3). Cross-sectional distributions of velocity and density in the outer bay are shown in Figure 7. In the western bay, isopycnals were almost flat near the surface and the surface density was rather low. During this time of the year, temperature doesn t vary much and density gradients arise from salinity effects associated with the river outflow. Isopycnals in the mid-water column tilted downward toward the east. The corresponding along-channel velocity (Figure 7a) shows that outflows occurred on the eastern side. On the western side of the channel, flows were northward on the top and southward near the bottom. The 1
2 24.1 model reproduced these patterns in the density and velocity (Figure 8a and 8b) although the outflows on the eastern side of the channel were much too strong. Surface density in the model was slightly lower than observed and the bottom density in the model was higher. depth (m) depth (m) 2 4 6 8 1 2 4 6 8 1 6 8 2 23.6 24 24.2-2 4 1-2 23.2-2 8 2 12 2 4 6 8 24.1 24.2 24 12 2 4 6 8 2 6 44 2 24.25 24.2-2 -4 (a) 24.1 (b) 2 4 23.5 23.723.9 24.3-5 -15-1 5 1 1 5 5 1 2 4 6 24.1 24.5 24.3 24.5 2 4 6 cross-channel distance (km) cross-channel distance (km) Figure 7. Observed along-channel velocity in cm/s (upper panels) and density in kg/m 3 (lower panels) in the outer western bay (a and b) and in the outer eastern bay (c and d) on 29 April 1998. Positive (negative) velocities indicate northward (southward) flows, i.e., inflows (outflows). 1 (c) (d) 5 24.5 11
In the eastern bay, the observed along-channel velocity (Figure 7c) suggests that outflows occurred above 3m depth and inflows near the bottom and to the eastern side of the channel. Isopycnals tilted upward toward the east with the lowest density in the upper western side of the channel (Figure 7d). Again the model appeared to reproduce the basic features of the velocity and density (Figure 8c and 8d). However, the tilt of the isopycnals was not as steep and outflows didn t reach sufficient depth. These findings are consistent with the time series results discussed in the previous section. (a) (c) (b) (d) Figure 8. Model simulated along-channel velocity in cm/s (upper panels) and density in kg/m 3 (lower panels) in the outer western bay (left panels) and in the outer eastern bay (right panels) on 29 April 1998. 12
Figure 9 shows the modeled surface velocity and salinity on 29 April 1998. Due to the westerly component of the wind, surface flows on the shelf were eastward. Below the surface (e.g., at 1m) the EMCC still moved southwestward on the shelf at about 15 cm/s (not shown). Inside Penobscot Bay, southwesterly winds drove the northward flows west of Islesboro. The Penobscot River outflow was mostly confined to the east of Islesboro (as seen from the surface salinity) which appeared to form a semi-enclosed clockwise circulation around Islesboro, except for the area between Islesboro and North Haven where the surface currents were northeastward, i.e., downwind. In the outer western bay, the surface currents moved northeastward on the western side and southeastward on the eastern side. The surface inflows on the western side came from the coastal water west of Penobscot Bay. Driven by the wind and the river outflow, surface currents in the outer eastern bay moved southeastward. wind: 7.9 m/s Figure 9. Modeled surface velocity and salinity on 29 April 1998. During the strong southeasterly wind event between 13 and 15 June, circulation patterns in the bay were quite different. Figure 1 shows the alongchannel velocity and density distributions at the same western and eastern outer bay cross-sections. In the outer western bay, southward flows, although weak at the surface, extended throughout the water column on the western side of the channel. Northward flows, driven by the strong wind, filled the eastern and shallower part of the channel and were quite strong. Similarly, in the outer eastern bay, outflows were on the western side of the channel whereas inflows on the eastern side. Overall, the water column was less stratified during this time of the year compared with that at the end of April during the spring freshet. Although 13
not shown, the surface had been warmed by several degrees, but the surface salinity was higher. (a) (c) (b) (d) Figure 1. As in figure 8 except for 14 June 1998. Figure 11 shows the corresponding surface circulation and salinity during the 14 June event. Driven by the strong southeasterly wind, the surface water moved northward into the bay. However, the northward flows were confined mostly to the eastern side of the channels. The circulation pattern around Islesboro was rather complicated. Most of the Penobscot River outflow moved along the channel west of Islesboro. The discharge rate on 14 June was not much 14
lower than that on 29 April, but the surface salinity in the bay was much higher because 29 April was at the end of the spring freshet (see Figure 3). wind: 16.9 m/s Figure 11. Modeled surface velocity and salinity on 14 June 1998 4. Summary The Princeton Ocean Model was applied to Penobscot Bay, Maine to simulate the subtidal circulation in the spring and summer of 1998. Preliminary results indicate that the simulation was sensitive to the value of the background viscosity. When a background viscosity of 5x1-6 m 2 /s was used, the model reproduced the observed three-layer and two-layer structure in the outer western and eastern bay, respectively. However, the southward surface flow reached a greater depth than the observed in the outer western bay, whereas the southward surface flow was weaker and shallower than the observed in the outer eastern bay. More sensitivity experiments with different model parameterizations and open boundary conditions that set up the density contrast between the water inside the bay and the coastal water are needed to obtain a better agreement between the observed and the model predicted layer thickness. The model responded to the synoptic scale forcing and produced temporal and spatial variability resembling the observations, in particular for the 29 April and the 14 June instances. In addition, the model also showed a semi-enclosed clockwise circulation around Islesboro in the case of southwesterly wind, consistent with the Burgund s (1995) finding. Additional observations are needed to verify the circulation pattern and the associated variability in the inner bay. In 15
the future, a multi-year model simulation and a Lagrangian trajectory study will be carried out, with the goal of providing insights into the interannual variability of lobster settlement in the Penobscot Bay region. 5. Acknowledgements This study was supported by the NOAA Sea Grant project R-CE233 and NOAA NESDIS Penobscot Bay Marine Resource Collaborative to the University of Maine and Texas A & M University. 6. References Bisagni, J. J., D. J. Gifford and C. M. Ruhsam, 1996. The spatial and temporal distribution of the Maine Coastal Current during 1982. Cont. Shelf Res., 16, 1-24. Blumberg, A. F., and G. L. Mellor, 1987. A description of a three dimensional coastal ocean circulation model. In Three-dimensional Coastal Ocean Models, Vol. 4, N. Heaps ed., Amer. Geophys. Union, 1-16. Brooks, D. A., 1994. A model study of the buoyancy-driven circulation in the Gulf of Maine. J. Phys. Oceanogr., 24, 2387-2412. Burgund, H. R., 1995. The currents of Penobscot Bay, Maine: Observations and a numerical model. Department of Geology and Geophysics, Yale University. 71 pp. Conkling, P. W., 1996. The marine ecology of Penobscot Bay. In Penobscot the Forest, River and Bay. D. D. Platt edit, Island Institute, Rockland, ME, 32-44. Fidler, R. B., 1979. An approach for hydrodynamic modelling of Maine s estuaries. MS thesis, University of Maine. Humphreys, A. C. and B. R. Pearce, 1981. A hydrodynamic investigation of the Penobscot estuary. Technical Report, Dept. Civil Engineering, University of Maine, 81 pp. Mellor, G. L., and T. Yamada, 1982. Development of a turbulence closure model for geophysical fluid problems. Rev. Geophys. Space Phys., 2, 851-875. Normandeau Associates, Inc., 1975. Environmental survey of upper Penobscot Bay, Maine. Unpublished Report. Pettigrew, N. R., D. W. Townsend, H. Xue, J. P. Wallinga, P. J. Brickley, and R. D. Hetland, 1997. Observation of the Eastern Maine Coastal Current and its offshore extensions. J. Geophys. Res., 13, 3,623-3,64. Pettigrew, N. R., 1998. Vernal circulation patterns and processes in Penobscot Bay: Preliminary Interpretation of Data. A final report for year 1 of the Penobscot Bay Experiment. University of Maine, 23 pp. Xue, H., F. Chai, and N. R. Pettigrew, 2. A model study of the seasonal circulation in the Gulf of Maine. J. Phys. Oceanogr, 3, 1111-1135. 16