Estuaries and Coasts (2010) 33:1164–1175 DOI 10.1007/s12237-010-9319-9 Wind Modulation of Dissolved Oxygen in Chesapeake Bay Malcolm E. Scully Received: 17 June 2009 / Revised: 21 November 2009 / Accepted: 13 June 2010 / Published online: 29 June 2010 # Coastal and Estuarine Research Federation 2010 Abstract A numerical circulation model with a simplified dissolved oxygen module is used to examine the importance of wind-driven ventilation of hypoxic waters in Chesapeake Bay. The model demonstrates that the interaction between wind-driven lateral circulation and enhanced vertical mixing over shoal regions is the dominant mechanism for providing oxygen to hypoxic sub-pycnocline waters. The effectiveness of this mechanism is strongly influenced by the direction of the wind forcing. Winds from the south are most effective at supplying oxygen to hypoxic regions, and winds from the west are shown to be least effective. Simple numerical simulations demonstrate that the volume of hypoxia in the bay is nearly 2.5 times bigger when the mean wind is from the southwest as compared to the southeast. These results provide support for a recent analysis that suggests much of the long-term variability of hypoxia in Chesapeake Bay can be explained by variations in the summertime wind direction. Keywords Hypoxia . Chesapeake Bay . Dissolved oxygen . Wind . Vertical mixing Introduction It is largely accepted that anthropogenic eutrophication has increased the extent and severity of hypoxia in coastal and estuarine waters (Diaz 2001). Yet, assessing human impacts and predicting the potential benefit of nutrient reductions M. E. Scully (*) Center for Coastal Physical Oceanography, Old Dominion University, Norfolk, VA 23508, USA e-mail: [email protected] are often confounded by the natural complexity and variability of these systems. Dissolved oxygen is modulated by both physical and biological processes that vary over time scales ranging from minutes to decades. These physical and biological processes often are tightly coupled, and distinguishing the importance of different mechanisms can be extremely difficult. A complete understanding of the physical processes that modulate hypoxia remains an obstacle to quantifying the impact of eutrophication on estuarine systems. The traditional view is that hypoxia in Chesapeake Bay develops during the summertime when oxygen utilization exceeds replenishment by vertical mixing (Officer et al. 1984). This view essentially assumes a one-dimensional oxygen balance in the vertical. However, Malone et al. (1986) suggested that the rotational response to winddriven forcing may provide an important mechanism for exchanging both nutrients and dissolved oxygen between the surface and sub-pycnocline waters in Chesapeake Bay. Sanford et al. (1990) also noted significant synoptic scale variability in near bottom oxygen measurements, which were attributed to wind-driven lateral oscillations of the pycnocline. Wind straining of the density field can exert a first-order control on the vertical density stratification and thus play an important role in modulating vertical mixing (Scully et al. 2005; Chen and Sanford 2009). Both O'Donnell et al. (2008) and Wilson et al. (2008) suggested that wind straining of the density field plays an important role in controlling hypoxia in Western Long Island Sound by modulating the vertical mixing. More recently, Scully (2010) demonstrated that the interannual variations of the summertime volume of hypoxic water in Chesapeake Bay are positively correlated with the duration of westerly winds and negatively correlated with the duration of southeasterly winds. This study found that Estuaries and Coasts (2010) 33:1164–1175 more of the inter-annual variation in the volume of hypoxic water was explained by wind direction than any other variable considered. Scully (2010) further suggested that increases in hypoxic conditions in the bay over the past several decades may have been enhanced by shifts in the summertime wind forcing caused by decadal scale climate variability. While these results highlight the importance of wind forcing to the oxygen dynamics of Chesapeake Bay, they do not provide any insight into the physical mechanisms driving this behavior. Despite the overall importance of wind to this system, observations conclusively demonstrating the importance of wind to the oxygen dynamics are lacking. While such observations are possible, the spatial resolution necessary to accurately capture the oxygen budget would be challenging, making numerical approaches an attractive alternative. In this paper, a realistic three-dimensional circulation model is used to conduct a detailed study of the physical response of dissolved oxygen in Chesapeake Bay to wind-driven forcing. The model is used to assess the importance of wind direction and provide an oxygen budget for the sub-pycnocline hypoxic zone in the bay. Methods Model Simulations The numerical experiments in this study were conducted utilizing the Regional Ocean Modeling System (ROMS; Shchepetkin and McWilliams 2005). Simulations employ the numerical grid developed for the open source Chesapeake Bay ROMS Community Model. This is a 150×100 curvilinear grid with 20 vertical terrain-following coordinates. It includes all of the main tributaries to the bay as well as the shelf region immediately adjacent to the bay mouth. The model was run in an idealized manner, using only M2 tidal forcing at the oceanic boundary and constant river discharge for all of the tributaries. The discharge for each tributary was calculated from the annual mean from the seaward-most gauging station maintained by the US Geological Survey. Turbulence closure is achieved using the k–ε model with the stability functions of Kantha and Clayson (1994). The background diffusivity for both momentum and scalars is set to 10−5 m2/s consistent with value used by Li et al. (2005). A very idealized dissolved oxygen model was implemented. The goal was to use as simple a model as possible, so that the physical processes that modulate dissolved oxygen could be isolated from biological variability. To do this, oxygen is introduced into the model as an additional tracer. A constant water column respiration rate of 0.8 gO2/ m3/day was applied everywhere inside the estuarine portion of the model domain. This rate is within the range of measured values for Chesapeake Bay (Kemp et al. 1992). A 1165 gradient flux condition was imposed at the sea surface following the formulation proposed by Marino and Howarth (1993), assuming a constant piston velocity of 3 cm/h. This is consistent with a wind speed of roughly 4 m/s, the approximate summertime average over Chesapeake Bay. Oxygen at both the oceanic and river boundaries was fixed to saturation values. Oxygen concentrations were not allowed to become negative, essentially imposing a respiration rate of zero for anoxic conditions. While this ignores the potential influence of sulfide reduction, it is in keeping with the goal of using the simplest possible model for oxygen dynamics. To account for wind forcing, a spatially uniform surface momentum flux was applied to the model domain. Values of wind stress were calculated using the drag formulation of Large and Pond (1981). The sea-level response of the Bay to wind forcing consists of both a local response to the surface wind stress as well as a remote effect driven by coastal upwelling/downwelling (Wang 1979). With the limited coastal domain, the model does not fully capture this remote barotropic process that is driven by large-scale shelf dynamics. To include this effect, an empirically derived subtidal setup/set down based on wind strength and direction was added to the tidal boundary forcing. This empirical formulation was derived using 20 years of observed water levels at the Chesapeake Bay Bridge tunnel and winds observed at Chesapeake Light. The observed water levels were de-tided, and the residual was regressed against the north and east components of the wind. This analysis gave the following relationship for the subtidal water level fluctuations at the mouth of the Bay: h ¼ 0:12 N þ 0:12 E ð1Þ where N and E are the north and east components of the wind velocity, respectively (in meters per second, positive for winds from north and from east). To examine the role of wind forcing, a series of numerical experiments (see Table 1) were conducted using the model configuration described above. For the base model run (A1), no wind forcing was used. The model was run until both the salinity field and oxygen distributions were steady at subtidal time scales. To examine the importance of individual wind events from different directions, simulations using an idealized 3-day wind event were conducted (B1–B4). For these runs, the idealized forcing consists of a wind that increased from 0 to 6 m/s and back to zero sinusoidally over a period of 3 days. This forcing was chosen to be largely representative of the typical wind events observed in summertime over Chesapeake Bay. The wind event begins on day 3 of the simulations, ends on day 6, and is followed by 4 days without wind. Model runs were conducted with wind forcing from the north (B1), east (B2), south (B3), and west (B4). 1166 Table 1 Description of model runs Estuaries and Coasts (2010) 33:1164–1175 Model ID Model description A1 B1 B2 B3 B4 C1 C2 C3 D1 D2 D3 Base mode run with no wind forcing 3-day sinusoidal wind event from north (max 6 m/s) 3-day sinusoidal wind event from east (max 6 m/s) 3-day sinusoidal wind event from south (max 6 m/s) 3-day sinusoidal wind event from west (max 6 m/s) Constant wind speed (4 m/s) with oscillatory direction (W to E with mean S) Constant wind speed (4 m/s) with oscillatory direction (SW to NE with mean SE) Constant wind speed (4 m/s) with oscillatory direction (NW to SE with mean SW) Observed winds at Thomas Point Light for May–June 1996 Observed winds at Thomas Point Light for May–June 1998 Observed winds at Thomas Point Light for May–June 1996, magnitude reduced by 10% In reality, wind direction over Chesapeake Bay is rarely constant for 3 days. More typically, wind direction varies at synoptic time scales associated with the passage of weather systems. A series of additional simulations were conducted with winds that varied in direction in an idealized way (C1– C3). For these runs, wind speed was held constant at 4 m/s while the wind direction oscillated back and forth through a 180° arc centered on the mean wind direction. Consistent with the typical summer time conditions, mean wind direction for these three runs was south (C1), southeast (C2), and southwest (C3). Wind direction oscillated through a 180° arc with a period of 6 days. Finally, a series of model runs were conducted using observed winds (D1–D3). These runs focused on the summers of 1996 and 1998—two years with very similar river discharge and estimated nitrogen loading but very different amounts of observed hypoxia (Hagy et al. 2004). For these runs, a spatially uniform surface wind stress was calculated using the observed wind speed and direction measured at the Thomas Point Light for the summer of 1996 (D1) and 1998 (D2). The Thomas Point Light is located along the central portion of the main stem of the Bay (Fig. 1) and is assumed to largely represent the wind forcing over the entire system. To highlight only the role of wind, no other forcing was changed (tides and river discharge were held constant). The model was started using the equilibrated fields from the model run with the oscillatory southerly wind (C1). Because the model results will be compared to the annual surveys of hypoxia reported by Hagy et al. (2004), which were generally conducted in early July, the model was run for a 2-month period beginning on May 1st of each year. This assumes that the wind forcing over the months of May and June is most relevant to the extent of hypoxia observed in early July. Because the mean wind speed in 1996 was slightly greater than 1998, an additional run was conducted where the magnitude of the wind in 1996 was reduced by 10% (D3). Model Analysis There are several definitions and criteria for hypoxia and anoxia in estuarine waters. For the purposes of this paper, hypoxic conditions are defined when water column concentrations of dissolved oxygen fall below 1 mg/L. Using this criteria, the total volume of hypoxic water in the model domain can be calculated at any given time. While the total volumetric extent of hypoxia is of interest, in order to synthesize and compare results from different model runs, it is useful to define a fixed volume that represents the region most susceptible to hypoxia in the bay, but that does not change between different model runs. This fixed volume is defined based on the location of the 1-mg/L oxygen contour from the base model run (A1). This volume is largely located below the pycnocline over regions where the water column depth is greater than 10 m. In the absence of wind forcing, unrealistically large volumes of the bay are predicted to be hypoxic particularly in the lower bay, so this volume is limited to the region north of 37.5° N latitude. This fixed volume is then used for several calculations including evaluation of the terms in the subpycnocline oxygen budget. The equation for the conservation of oxygen can be presented as: @O2 @ @O2 Kz þ r O2 ¼ R @z @t @z ð2Þ where the first term on the left-hand side represents the time rate of change of oxygen and the second term represents the divergence in advective flux. The two terms on the righthand side represent the flux divergence due to vertical mixing and loss by respiration, respectively. For the purpose of this analysis, the advective fluxes are separated into a longitudinal and lateral component. It should be noted that this distinction is somewhat arbitrary for a estuarine systems with complex bathymetry as the two fluxes are strongly related via the continuity constraint. For Estuaries and Coasts (2010) 33:1164–1175 1167 Fig. 1 Modeled average dissolved oxygen concentration (run C1). a Average bottom oxygen concentration; b average oxygen concentration along the thalweg of the main stem of Chesapeake Bay. Magenta lines in a denote the locations of the upper and lower Bay cross sections. Magenta star indicates location of the Thomas Point Light NDBC station 39.5o Thomas Point Light 39o Upper Bay Cross-Section 38.5o 38o Lower Bay Cross-Section 37.5o 37o o 76.6o 77 0 1 2 3 76.2o 4 5 mg/L 75.8o 6 7 8 0 -5 -10 -15 -20 -25 -30 o 39 simplicity, the longitudinal flux is calculated as the divergence between the flux into and out of the two ends of the volume. The remaining flux is attributed to lateral processes and includes both lateral and vertical advection. Each term is averaged spatially over the entire fixed subpycnocline volume defined above. Results Distribution of Dissolved Oxygen Despite the simplified oxygen dynamics, the model reproduces an overall distribution of dissolved oxygen concentration that is consistent with typical summertime observations. The model predicts hypoxic conditions below the pycnocline for significant portions of the mid and upper o o 38.5 38 Degrees North 37.5 o bay. The hypoxic zone is generally confined to the deep central portion of the bay extending roughly from the mouth of the York River in Virginia to the region parallel to Baltimore, MD. For the period from 1950 through 2001, Hagy et al. (2004) report that the volume of water in the Bay with concentration less than 1 mg/L varied from 0.8 to 9.6 km3. In comparison, for the case with constant wind speed of 4 m/s and a mean direction from the south (C1), the model predicts 7.6 km3. While the overall distribution is largely consistent with observations, it should be noted that using this relatively simple approach does generally overpredict the southern extent of the hypoxic region. Idealized 3-Day Wind Events As described in “Methods” section, simulations of individual wind events were run for winds from four different 1168 Estuaries and Coasts (2010) 33:1164–1175 a) Modeled wind stress 0.08 Pascals 0.06 0.04 0.02 0 0 5 10 15 b) Modeled volume of hypoxic water (< 1 mg/L) hypoxic volume (km3) 14 12 10 N S E W 8 6 4 0 5 10 15 model day Fig. 2 a Surface wind stress magnitude for idealized model runs; b time series of predicted hypoxia (defined as the volume of water with oxygen concentration less than 1 mg/L north of latitude 37° N) for model runs with wind from the north (solid black line), south (solid gray line), east (dashed black line), and west (dashed gray line) directions: north, south, east, and west (model runs B1– B4). The time series of wind stress magnitude is shown in Fig. 2a. For all wind directions, there is a significant reduction in the volume of hypoxic water during the wind event, followed by a gradual return to pre-wind conditions once the wind subsides (Fig. 2b). However, there are marked differences between the different wind directions. The wind event from the south has the greatest reduction in the volume of hypoxic water followed by the slowest return b) South Wind a) North Wind 1.5 lat. adv. mixing 0.5 0 -0.5 -1 1.5 long. adv. respiration 1 gO2/m3/day gO2/m3/day 1 0.5 0 -0.5 0 5 10 model day -1 15 c) East Wind 1.5 1 1 0.5 0 5 10 model day 15 0.5 0 -0.5 -0.5 -1 0 d) West Wind 1.5 gO2/m3/day gO2/m3/day Fig. 3 Time series of terms in the sub-pycnocline oxygen budget for the modeled a north, b south, c east, and d west wind events. Solid black line denotes respiration term, solid gray line denotes longitudinal advection, dashed black line denotes vertical mixing, and dashed gray line denotes lateral advection to pre-wind conditions. In contrast, the rapid decrease in the volume of hypoxic water associated with the northerly wind is followed by the most rapid return. The smallest reduction in the volume of hypoxic water is associated with the wind event from the west. It is noteworthy that the minimum volume of hypoxic water predicted during the south wind event (5.5 km3) is almost a factor of two smaller than the minimum volume predicted during the west wind event (9.3 km3). Using the numerical model, the sub-pycnocline oxygen budget can be evaluated for the idealized 3-day wind events (Fig. 3). There are several notable results from this budget. First, for all wind directions, the magnitude of the lateral advective term is either equal to or greater than the flux due to vertical mixing (Table 2). For the wind event from the south, which has the greatest overall increase in subpycnocline oxygen content, the lateral advective flux exceeds vertical mixing by over a factor of 4. The greatest lateral advective flux is predicted in response to the southerly wind event. The west wind simulation has the smallest increase in sub-pycnocline oxygen content and the smallest flux associated with lateral advection. Despite the fact that the wind from the south has the greatest reduction in the volume of hypoxic water, it has the lowest overall flux due to vertical mixing. In all cases, longitudinal advection of oxygen into the hypoxic zone is minimal. Because the model does not allow negative concentrations of oxygen, the respiration rate must approach zero for anoxic conditions. As a result, the volume averaged respiration rate goes down when a large percentage of the volume goes anoxic, and differences in the calculated respiration rate simply reflect differences in the overall extent of anoxic 0 5 10 model day 15 -1 0 5 10 model day 15 Estuaries and Coasts (2010) 33:1164–1175 1169 ∂O2/∂t ∂/∂z Kz∂O2/∂z v∂O2/∂y+w∂O2/∂z u∂O2/∂x Respiration North South East West 0.397 0.470 0.452 0.019 −0.543 0.447 0.181 0.800 −0.032 −0.502 0.285 0.244 0.505 −0.031 −0.433 0.154 0.283 0.253 0.022 −0.403 All values are reported as gO2/m3 /day. Values are averaged over a fixed volume defined by the 1-mg/L contour from base model run (A1) without wind forcing. Values are averaged over the duration of the 3-day wind event conditions between model runs. It is important to point out that the vertical mixing term only accounts for a direct flux of oxygen into the fixed volume. Mixing that occurs elsewhere and is advected in appears as an advective flux in this analysis. As will be discussed below, this is an important mechanism for providing oxygen to the subpycnocline waters. Model Runs with Constant Wind Speed and Variable Direction The results from model runs B1-B4 demonstrate that the oxygen dynamics in Chesapeake Bay are sensitive to the direction of the wind forcing. However, the response to wind forcing presented above suggests that neither the oxygen nor salinity dynamics are in steady state at synoptic time scales. This is particularly apparent for the southerly wind case where neither the oxygen nor the stratification has returned to their original values by the end of the model run. While significant insight can be gained from the simple numerical simulations presented above, the model runs with constant wind speed and variable wind direction allow a more realistic examination of this non-steady-state behavior. Figure 4 shows the predicted volume of hypoxic water for simulations C1, C2, and C3. The model starts from the equilibrated run without wind forcing (A1), which predicts large volumes of water with oxygen concentration below 1 mg/L. With the addition of wind forcing, the volume of hypoxic water is reduced rapidly. While there are oscillations associated with the shifting wind direction, the oxygen field becomes approximately steady over longer time scales. The total volume of hypoxic water predicted by the model is strongly dependent upon the overall mean wind direction. The greatest extent of hypoxia is predicted for the case with a mean wind direction from the southwest (C3) while the least hypoxia is predicted for the run with a mean wind from the southeast (C2). In fact, the model predicts that a shift in mean wind direction from southeast to southwest results in a nearly 2.5-fold increase in the volume of hypoxic water in the bay. While this non-steady response at time scales less than several days makes evaluating the importance of the instantaneous wind direction more difficult, the overall dependence on the mean direction is noteworthy. Model Runs with Observed Wind Forcing (1996 and 1998) The sensitivity to wind direction is highlighted further by comparing the two years with very similar river discharge, nutrient loading, and wind speed but with different wind direction. The average Susquehanna River discharge between January and May was 2,355 and 2,363 m3/s for 1996 and 1998, respectively. Estimated nitrogen loading also was very similar with average spring loading rates of 3.84×105 and 3.80×105 kg/day for 1996 and 1998, respectively (Hagy et al. 2004). Despite these similarities, Hagy et al. (2004) report that the estimated volume of water with oxygen concentration less than 1 mg/L was over twice as large in 1998 (9.61 km3) as in 1996 (4.54 km3). The early summer wind magnitudes observed in these two years were very similar. In contrast, there were significant differences in the observed wind direction (Fig. 5). Based on the Thomas Point Light data, the wind speeds averaged over May and June are 4.9 and 4.8 m/s, for 1996 and 1998, respectively. During 1996, the wind was directed primarily from the south/southeast—the direction found to result in the greatest wind-driven ventilation. In contrast, the histogram from 1998 shows two distinct peaks, with a significant duration of summer winds from the west/ northwest—the direction least conducive to bottom water ventilation. The time series of the predicted volume of hypoxic water for the two summers is shown in Fig. 6. While there is considerable variability in the time series for each year, the 18 hypoxic volume (km3) Table 2 Oxygen budget for idealized wind runs SW S SE 16 14 12 10 8 6 4 2 0 5 10 15 20 25 model day 30 35 40 Fig. 4 Time series of predicted hypoxic volume for model runs with constant wind speed and oscillatory wind direction (runs C1, C2, and C3). Wind direction oscillates through a 180° arc with a period of 6 days, centered on a mean direction from the southwest (dashed gray line), south (solid black line), and southeast (dashed black line) 1170 Estuaries and Coasts (2010) 33:1164–1175 Fig. 5 Frequency histograms for wind observed at the Thomas Point Light NDBC station (38°53′54″ N, 76°26′12″ W) for the period of May through June. a Wind speed 1996; b wind speed 1998; c wind direction 1996; d wind direction 1998 a) Wind Speed 1996 0.10 b) Wind Speed 1998 0.10 mean = 4.9 m/s 0.08 0.08 0.06 0.06 0.04 0.04 0.02 0.02 0 5 10 15 mean = 4.8 m/s 0 5 m/s c) Wind Direction 1996 0.15 0.10 0.10 0.05 0.05 E model predicts a greater average volume of hypoxic water for the summer of 1998. Averaged over the month of June, the predicted volumes of hypoxic water (<1 mg/L) were 5.2 and 9.4 km3 for 1996 and 1998, respectively. These values are consistent with the observations reported by Hagy et al. (2004) for these two summers and demonstrate that relatively subtle changes in wind direction can have a significant impact on the extent of hypoxia in Chesapeake Bay. To confirm that these results are not actually the result of slight differences in the wind magnitude, the summer of 1996 was rerun with a 10% reduction in wind speed (D3). This reduced the mean wind speed for the summer of 1996 from 4.9 to 4.4 m/s, a reduction to roughly 2 standard deviations below the long-term (1986–2007) May–June mean at this location. Despite this reduction, the average S direction 15 d) Wind Direction 1998 0.15 N 10 m/s W N N E S direction W N volume of hypoxic water predicted by the model (7.4 km3) was still over 40% greater than predicted for 1998, when averaged over the month of June. The terms in the sub-pycnocline oxygen budget highlight the differences between these two years (Table 3). In both 1996 and 1998, both lateral advection and vertical mixing balance the consumption of oxygen through respiration. Lateral advection exceeds vertical mixing by a factor of 2 and 1.6 for 1996 and 1998, respectively. Thus, the differences in hypoxia between the two years largely reflect differences in the magnitude of the lateral advection term and not differences in vertical mixing. In contrast to the slight differences in vertical mixing between the two years (0.204 vs. 0.228 gO2/m3/day), there is significantly greater ventilation by lateral advection in 1996 (0.463 gO2/ 14 hypoxic volume (km3) 12 1996 1998 Table 3 Oxygen budget for runs with observed wind forcing 10 1996 8 1998 6 ∂O2/∂t −0.025 0.001 4 ∂/∂z Kz∂O2/∂z v∂O2/∂y+w∂O2/∂z u∂O2/∂x Respiration 0.228 0.463 −0.005 −0.711 0.204 0.323 0.001 −0.527 2 0 May Jun model day Jul Fig. 6 Time series of predicted hypoxic volume for model runs forced with observed winds from Thomas Point Light for the summer of 1996 (dashed black line) and 1998 (solid gray line) All values are reported as gO2/m3 /day. Values are average over a fixed volume defined by the 1-mg/L contour from base model run (A1) without wind forcing. Values are average over the entire month of June Estuaries and Coasts (2010) 33:1164–1175 1171 m3/day) as compared to 1998 (0.323 gO2/m3/day). Given the consistency in the other forcing for these two runs, the difference in lateral advection is attributed to differences in wind direction. Discussion Longitudinal Straining It is instructive to use the results from the numerical simulations to examine the mechanisms that result in the asymmetric response of the oxygen dynamics to changes in wind direction. The traditional view of hypoxia in estuarine waters is that vertical mixing across the pycnocline controls the supply of oxygen to sub-pycnocline waters. It is often assumed that the strength of vertical mixing is directly related to the overall level of density stratification. Since the interaction between the axial density gradient and residual estuarine circulation is the mechanism by which stratification is created, it follows that wind-driven modulation of the along-channel residual flow may play a key role in modulating the extent of hypoxia. The model results for the idealized 3-day wind events (B1–B4) show that the wind has a clear impact on the strength of the subtidal residual estuarine velocity (Fig. 7a). Winds from both the north and west tend to increase the strength of the residual velocity, while winds from the south and east reduce or even reverse the near-bed residual flow. This result demonstrates that the simple along-channel straining mechanism proposed by Scully et al. (2005) is modified by the earth’s rotation in Chesapeake Bay. For estuarine systems wider than the external Rossby radius of deformation, the integrated wind-driven surface transport is expected to be 90° to the right of the wind (in the northern hemisphere). The width of the Chesapeake Bay is smaller than the external Rossby radius of deformation but of the same order as the internal Rossby radius. a) along-channel velocity Given the importance of lateral advective processes to the oxygen budget presented above, it is instructive to examine what controls the strength of lateral exchange. The strongest lateral flows are predicted for winds from the south, where the surface transport is to the right of c) across-channel velocity 2 2.5 6 1 2 4 cm/s cm/s Lateral Circulation b) top-to-bottom salinity diff. 8 1.5 2 N S E W 0 -2 -4 Therefore, the integrated wind-driven Ekman transport has a component in both the along-wind and orthogonal direction. As a result, both north and west winds induce transport with a component in the down-estuary direction near the surface. This down-estuary surface transport is balanced with an increased up-estuary flow near the bed. The opposite is observed for both the south and east wind cases (Fig. 7a). The straining of the density field caused by the windinduced changes in the along-channel residual circulation plays a key role in the overall changes in stratification. Winds from the north, south, and east all reduce the overall degree of salinity stratification during the wind events (Fig. 7b). In contrast, stratification is observed to increase during the west wind event before eventually decreasing slightly. These patterns of stratification are largely consistent with the rotationally modified wind-driven along-channel circulation. However, the differences between the north and west wind events, both of which increase the along-channel residual circulation, also demonstrate the importance of lateral circulation in modulating stratification. Winds from the north induce a strong lateral circulation, while winds from the west have a much weaker lateral response (Fig. 7c). The stronger lateral flow in response to winds from the north significantly tilts the isopycnals toward a more vertical position, thus reducing the overall degree of stratification. The lateral tilting of the isopycnals is significantly reduced in response to winds from the west, and the overall degree of stratification is more strongly controlled by the alongchannel straining. 0 5 10 model day 1 15 0.5 0 -1 -2 0 5 10 model day Fig. 7 Predicted a subtidal along-channel velocity, b top-to-bottom salinity difference, and c subtidal lateral velocity for the four modeled wind directions: north (solid black line), south (solid gray line), east (dashed black line), and west (dashed gray line). Values were 15 -3 0 5 10 model day 15 averaged over the entire hypoxic layer in the main stem of Chesapeake Bay and low-pass filtered. Salinity stratification is reported as the difference between the hypoxic layer and the overlying water 1172 0 0.50 0.25 0.10 0.05 0.01 0 -5 depth (m) the wind (toward the east), with a compensatory transport in the lower hypoxic layer toward the west (negative in Fig. 7c). The lateral response is weaker and in the opposite direction (bottom flow to the east) for winds from the north. For both east and west winds, the lateral flow in the lower layer is smaller. The strength and direction of lateral flow is coupled to the along-channel dynamics. Both north and west winds increase the along-channel residual shear. For an inviscid flow, this increase in along-channel shear will act to increase the slope of the pycnocline via the thermal wind balance (upwelling to the east and downwelling to the west). The tilting of the pycnocline is accomplished by a lateral flow that is consistent with Ekman dynamics. However, because the estuary is of finite width, some component of the surface transport is in the same direction as the wind stress. For the case of the west wind, the lateral surface transport induces downwelling on the east side and upwelling on the west, largely canceling the thermal wind response. A similar cancelation results from winds from the east. The result is much weaker lateral flows for both east and west winds. The greater response to north and south winds occurs because the transport orthogonal to the wind stress induces a pattern of upwelling and downwelling that reinforces the thermal wind response. The degree of salinity stratification also impacts the strength of lateral circulation. The lateral flow tilts isopycnal surfaces, setting up an adverse pressure gradient that tends to suppress lateral flow (Seim and Gregg 1997; Chant and Wilson 1997; Lerczak and Geyer 2004). As a result, stronger density stratification tends to result in weaker lateral circulation. Because of this interaction, winds from the south not only induce a lateral flow in the bottom water that is opposite to the flow observed during north winds but it is also stronger due to the overall lower levels of stratification. This suggests the somewhat surprising result that, as long as some level of density stratification is maintained, the strength of stratification plays a greater role in limiting the lateral advective exchange of oxygen than in limiting the direct turbulent flux of oxygen. This interpretation is supported by calculations of the gradient Richardson number (Ri). It is generally accepted that turbulent mixing by shear instability cannot develop when Ri >0.25 (Miles 1961; Howard 1961). Once the level of stratification is sufficient for Ri to exceed this threshold, it is expected that direct turbulent flux through the pycnocline is effectively reduced to zero. As a result, further increases in the strength of stratification cannot decrease the turbulent flux through the pycnocline any further. The model results suggest that, for typical summer forcing conditions, significant portions of the estuarine cross section maintain a value of Ri in the pycnocline that is never reduced below its critical value (Fig. 8). In contrast, Estuaries and Coasts (2010) 33:1164–1175 -10 -15 Percentange to time where Ri<0.25 2 4 6 8 10 width (km) Fig. 8 Contours of the total percentage of time that the gradient Richardson number (Ri) is below 0.25 for the model run C1 regions over the shallow shoal regions frequently experience conditions of Ri <0.25. Thus, oxygen is readily mixing downward over the shoals, providing a source for advective transport into the deeper hypoxic regions. The Importance of Bathymetry The lateral differences in the frequency of subcritical value of Ri highlight the importance of lateral bathymetry to the oxygen dynamics. The bathymetry in most coastal plain estuaries is not symmetric about the channel. Typically, the deepest portion of the channel is located on the right (looking up-estuary) side of the estuary, with a broad shoal region to the left. The mid and upper regions of Chesapeake Bay are generally consistent with this morphology. In contrast, in parts of the lower Bay, the deepest part of the channel is located on the left (west)-hand side of the channel with a more pronounced shoal region to the right (east). The interactions between wind-driven lateral circulations and this bathymetry play a key role in the observed asymmetric response of the oxygen dynamics to wind direction. This can be demonstrated most clearly by looking at the response of the oxygen field to winds from the north and south at two locations, one in the upper Bay and one in the lower Bay. Figure 9 shows contours of dissolved oxygen at these two locations averaged over both the north and south wind events (runs B1 and B3, respectively). Also shown are the differences in oxygen concentration for each simulation, at both locations. The three upper panels (Fig. 9a–c) are from a cross section in the upper bay near the mouth of the Severn River (see Fig. 1). At this location, there is a broad gently sloping shoal on the western side of the channel with Estuaries and Coasts (2010) 33:1164–1175 b) North Wind c) South minus North 0 0 -5 -5 -5 -10 depth (m) 0 depth (m) depth (m) a) South Wind -10 -10 mg/L -15 -15 Upper Bay 2 4 6 8 0 10 2 4 mg/L 6 -15 8 4 2 6 8 -5 -5 Lower Bay 0 5 10 width (km) 15 a comparatively narrow and steep shoal region on the eastern side. During the wind event from the south, the surface water is deflected toward the eastern side of the Bay due to the Coriolis force. There is a compensatory flow in the lower layer that advects hypoxic bottom water laterally up onto the broad western shoal. On the eastern side of the channel, oxygenated surface water downwells and is advected west replacing the hypoxic water below the pycnocline in the channel. Because of the steep lateral bathymetry on the eastern side, the oxygenated surface water only has to travel a relatively short distance to reach the region in the deep channel where persistent hypoxia is observed. In contrast, during winds from the north, the surface water is deflected toward the west and the bottom waters are advected to the east. The oxygenated surface water that downwells along the western shore has a considerably larger distance to travel to reach the hypoxic zone in the channel. The difference between the oxygen concentrations for the north and south wind events is clearly dominated by the lateral rotational response of the oxygen field. Because of the bathymetry at the upper Bay location, there is a much greater lateral flux of oxygen into the hypoxic region in response to the wind from the south. This is in contrast to the contours shown for the cross section in the lower bay north of the mouth of the York River (Fig. 9d, e). While the general response to the wind forcing is the same at this location, the bathymetry here is quite different. The broad shoal region is on the eastern side of the channel, with the steeper bathymetry to the west. In contrast to the upper Bay, Fig. 9f now shows that there is a much greater lateral flux of oxygen into the hypoxic region in response to the wind from the north. depth (m) -5 depth (m) 0 -20 -10 -15 mg/L -20 0 0 5 2 4 4 6 8 10 f) South minus North 0 -15 0 2 0 -10 -2 10 e) North Wind d) South Wind depth (m) Fig. 9 Contours of dissolved oxygen concentration averaged over 3-day wind event: a south wind upper bay, b north wind upper bay, c difference between north and south winds in upper bay (positive values indicate greater oxygen concentration resulting from south winds), d south wind lower bay, e north wind lower bay, and f difference between north and south winds in lower bay (positive values indicate greater oxygen concentration resulting from south winds). Solid black line denotes the 1-mg/L oxygen contour that serves as the vertical boundary of the volume used for budget calculations 1173 2 10 width (km) 4 -10 -15 mg/L -20 6 15 -2 8 0 5 0 10 width (km) 2 4 15 It is important to note that, in the absence of turbulent mixing, the lateral advective fluxes shown in Fig. 9 would not result in a net flux of oxygen into the hypoxic region. Without mixing, this hypoxic water would simply relax laterally back into the channel as the wind subsides. However, significant turbulent oxygen flux occurs when the hypoxic water is laterally advected onto the shoals (Fig. 10). Oxygen enters the water column through the air– water interface. This surface flux of oxygen is directly proportional to the concentration gradient between the surface water and the atmosphere. When water with low oxygen concentration is advected from the deeper channel up onto the shoals, it brings water with low oxygen concentration to the surface, enhancing the surface flux of oxygen into the water column. When this water relaxes back into the channel at the end of the wind event, the enhanced surface flux results in a net influx of oxygen to the hypoxic layer. The lateral bathymetry in the mid to upper bay enhances the vertical flux of oxygen in response to winds from the south. In this region, southerly winds drive hypoxic water from the channel onto the broader western shoal, bringing water with lower oxygen concentration to the surface. This enhances the surface oxygen flux over the western shoal region. This broad shoal region also typically has large areas where Ri <0.25, allowing significant vertical oxygen flux to occur through turbulent mixing within the water column (Fig. 10). This newly oxygenated water is then advected back into the channel as the winds relaxes. Under northerly winds, the lateral flow advects hypoxic water from the channel onto the narrower eastern shoal. Here, both the surface area of water with low oxygen concentration and total area where Ri <0.25 are reduced, resulting in 1174 b) North Wind--Upper Bay a) South Wind--Upper Bay 0 0 -5 depth (m) 5 1 -10 -15 -5 1 -10 -15 2 4 6 0 8 2 10 4 6 8 10 g O2 m-2 day-1 13 2 4 6 12 14 16 0 9 depth (m) depth (m) -5 1 -10 -15 -20 Model Sensitivity One of the conclusions from the analysis presented above is that lateral advective processes play a dominant role in the ventilation of hypoxic regions in Chesapeake Bay. However, the relative importance of lateral advection to direct vertical mixing is dependent upon the choice of background diffusivity. Li et al. (2005) showed that model predictions of the strength of the residual circulation, stratification, and 9 5 1 -10 -15 -20 5 less vertical oxygen flux. The net result is significantly greater oxygen flux in response to south winds in the upper and mid bay because of the interactions between the lateral flow and the bathymetry. The opposite appears to be true in the lower bay, where the larger shoal is located on the eastern side of the channel. Here, winds from the north drive water with low concentrations of oxygen onto the broader eastern shoal, where significant oxygen flux can occur. The effectiveness of winds from the south is limited in the lower bay because of the relatively smaller amount of shoal region located to the west. The total extent of hypoxia in the bay appears to be reduced the most by southerly winds because hypoxia is generally more severe in the mid to upper portions of Chesapeake Bay, where the large shoal areas are found to the west of the channel. However, these results suggest that southerly winds are less effective at ventilating regions of lower Chesapeake Bay. 10 13 5 -5 8 d) North Wind--Lower Bay c) South Wind--Lower Bay 0 13 9 5 9 depth (m) Fig. 10 Cross-sectionally averaged contours of vertical turbulent oxygen flux for a south winds upper bay, b north winds upper bay, c south winds lower bay, and d north winds lower bay. Contour interval is 2 gO2m−2 day−1. Solid black line denotes the 1-mg/L oxygen contour that serves as the vertical boundary of the volume used in budget calculations. Locations for cross sections are shown in Fig. 1 Estuaries and Coasts (2010) 33:1164–1175 10 width (km) 15 5 10 width (km) 15 salt content of Chesapeake Bay were all sensitive to the choice of background diffusivity. In their work, a value of 10−5 m2/s was found to match observations best. This is the value used in this study as well, but it is important to note that both the total extent of hypoxia predicted and the relative importance of vertical mixing are sensitive to the value that is used. To test the sensitivity of the model to the value of background diffusivity, the model was run with the value set to 10−4 and 10−6 m2/s. Increasing the background diffusivity decreased the predicted extent of hypoxia, while reducing this value increased hypoxia. Even though the overall extent of hypoxia was sensitive to the value of background diffusivity, the asymmetric response to wind direction was robust. Regardless of the value of background diffusivity, southerly winds always resulted in the smallest amounts of hypoxia and westerly winds always resulted in the most. Further, the interactions between the lateral flow, bathymetry, and vertical mixing described above remained an important mechanism for supplying oxygen to the hypoxic region. Similarly, varying the value for the respiration rate and piston velocity changed the extent of hypoxia predicted by the model. However, it is important to stress that the conclusions of this paper are generally not impacted when the background diffusivity is changed by 2 orders of magnitude, or values for the respiration rate and piston velocity are varied within the range of values reported in the literature. Estuaries and Coasts (2010) 33:1164–1175 Biological Feedbacks In order to isolate the physical processes that modulate hypoxia, an extremely simplified oxygen model was used in this study. This model is clearly a gross simplification of the biological processes that impact oxygen concentration in the bay. Biological processes in Chesapeake Bay may be strongly influenced by the estuarine dynamics described in this study. For instance, Malone et al. (1986) suggest that wind-driven lateral circulations may play an important role in supplying nutrients from the sub-pycnocline waters to the surface layer. As a result, strong lateral exchange may enhance surface productivity by phytoplankton. Clearly, these biological feedbacks could significantly impact the oxygen dynamics described above. Capturing such complex interactions is extremely challenging in either an observational or modeling study. Such interactions were not captured by this study and are clearly an area that requires further study. Conclusions A series of numerical experiments were conducted to examine the dependence of the oxygen dynamics in Chesapeake Bay on wind direction. Model results demonstrate that the greatest ventilation of hypoxic bottom waters occurs in response to winds from the south with the least ventilation in response to winds from the west. Even slight changes in the mean wind direction appear to have significant consequences for oxygen dynamics. A simple numerical experiment demonstrates that a shift in mean wind direction from the southeast to southwest can result in a nearly 2.5-fold increase in the volume of hypoxic water in the bay. The sensitivity to wind direction is largely driven by the interactions between lateral wind-driven circulation, bathymetry, and vertical mixing. Large parts of the upper and mid Chesapeake Bay have an asymmetric distribution of lateral bathymetry with broad shoal regions to the west. During wind events from the south, bottom hypoxic water is laterally advected onto the broad western shoal where significant vertical oxygen flux occurs. This oxygenated water relaxes back into the channel as the wind subsides resulting in a net flux of oxygen. Winds from the west have limited lateral exchange, which prevents significant ventilation. References Chant, R.J., and R.E. Wilson. 1997. Secondary circulation in a highly stratified estuary. Journal of Geophysical Research 102: 23207– 23216. Chen, S.N., and L.P. Sanford. 2009. Axial wind effects on salinity structure and longitudinal salt transport in idealized, partially- 1175 mixed estuaries. Journal of Physical Oceanography 39: 1905– 1920. Diaz, R.J. 2001. Overview of hypoxia around the world. Journal of Environmental Quality 30: 275–281. Hagy, J.D., W.R. Boyton, C.W. Keefe, and K.V. Wood. 2004. Hypoxia in Chesapeake Bay, 1950–2001: long-term changes in relation to nutrient loading and river flow. Estuaries 27: 634–658. Howard, L.N. 1961. Note on a paper by J.W. Miles. Journal of Fluid Mechanics 10: 509–512. Kantha, L.H., and C.A. Clayson. 1994. An improved mixed layer model for geophysical application. Journal of Geophysical Research 99: 25,235–25,266. Kemp, W.M., P.A. Sampou, J. Garber, J. Tuttle, and W.R. Boynton. 1992. Seasonal depletion of oxygen from bottom waters of Chesapeake Bay: role of benthic and planktonic respiration and physical exchange processes. Marine Ecology Progress Series 85: 137–152. Large, W.G., and S. Pond. 1981. Open ocean momentum flux measurements in moderate to strong winds. Journal of Physical Oceanography 11: 324–336. Lerczak, J.A., and W.R. Geyer. 2004. Modeling the lateral circulation in straight, stratified estuaries. Journal of Physical Oceanography 34: 1410–1428. Li, M., L. Zhong, and W.C. Boicourt. 2005. Simulations of Chesapeake Bay estuary: Sensitivity to turbulence mixing parameterizations and comparison with observation. Journal of Geophysical Research 110: C12004. doi:10.1029/2004JC002585. Malone, T.C., W.M. Kemp, H.W. Ducklow, W.R. Boynton, and J.H. Tuttle. 1986. Lateral variation in the production and fate of phytoplankton in a partially stratified estuary. Marine Ecology Progress Series 32: 149–160. Marino, R., and R.W. Howarth. 1993. Atmospheric oxygen exchange in the Hudson River: Dome measurements and comparison with other natural waters. Estuaries 16: 433–445. Miles, J.W. 1961. On the stability of heterogeneous shear flows. Journal of Fluid Mechanics 10: 496–508. O'Donnell, J., H.G. Dam, W.F. Bohlen, W. Fitzgerald, P.S. Gay, A.E. Houk, D.C. Cohen, and M.M. Howard-Strobel. 2008. Intermittent ventilation in the hypoxic zone of western Long Island Sound during the summer of 2004. Journal of Geophysical Research 113: C09025. doi:10.1029/2007JC004716. Officer, C.B., R.B. Biggs, J.L. Taft, L.E. Cronin, M.A. Tyler, and W. R. Boynton. 1984. Chesapeake Bay Anoxia: Origin, development, and significance. Science 223: 22–27. Sanford, L.P., K.G. Sellner, and D.L. Breitburg. 1990. Covariability of dissolved oxygen with physical processes in the summertime Chesapeake Bay. Journal of Marine Research 48: 567–590. Scully, M.E. 2010. The importance of climate variability to winddriven modulation of hypoxia in Chesapeake Bay. Journal of Physical Oceanography 40: 1435–1440. Scully, M.E., C.T. Friedrichs, and J.M. Brubaker. 2005. Control of estuarine stratification and mixing by wind-induced straining of the estuarine density field. Estuaries 28: 321–326. Seim, H.E., and M.C. Gregg. 1997. The importance of aspiration and channel curvature in producing strong vertical mixing over a sill. Journal of Geophysical Research 102: 3451–3472. Shchepetkin, A.F., and J.C. McWilliams. 2005. The regional oceanic modeling system (ROMS): A split-explicit, free-surface, topography-following-coordinate oceanic model. Ocean Modelling 9: 347–404. Wang, D.P. 1979. Wind-driven circulation in the Chesapeake Bay, Winter, 1975. Journal of Physical Oceanography 9: 564–572. Wilson, R.E., R.L. Swanson, and H.A. Crowley. 2008. Perspectives on long-term variations in hypoxic conditions in western Long Island Sound. Journal of Geophysical Research 113: C12011. doi:10.1029/2007JC004693.
© Copyright 2026 Paperzz