DREDGING FOR ENVIRONMENTAL BENEFIT: MODELS OF CIRCULATION AND FLUSHING DYNAMICS IN THE PROVIDENCE RIVER ESTUARY

In estuarine dynamics, circulation in the form of mixing and exchange have a direct link to the water quality. Edgewood Shoals is a highly anthropogenically impacted region of the Providence River, bordered by three cities, and receives the outfall from six wastewater treatment facilities. Edgewood Shoals also experiences low dissolved oxygen levels during the summer months. The Edgewood Shoals is classified as a circulation-restricted zone, where hydrodynamic exchange is limited due to the steep bathymetric gradient created by an adjacent federal shipping channel. The US Army Corps of Engineers (USACE) is in the process of determining if there are options for placement of a CAD (Confined Aquatic Disposal) cell for contaminated sediment disposal in regions of the Providence River. Edgewood Shoals is under consideration for the placement one of these CAD Cells. The purpose of this project is to first model an Edgewood Shoals reference case, verify this model run against existing hydrodynamic data, and finally to use the model to alter the bathymetry of the Shoal in a way that would enhance hydrodynamic exchange. Dredging scenarios created in this study aim to cover two objectives. The first is to increase the amount of exchange between Edgewood Shoals and the adjacent deep channel of the Providence River, improving the flushing dynamics on Edgewood Shoals. The second is to achieve this goal while remaining practical for use by USACE. The Regional Ocean Modeling System is applied to investigate these changes to circulation using simulated drifters and numerical dyes to characterize local residence times and exchange. It is evident from this study that the model is describing flushing times that are unrealistically fast. Therefore, results are presented as a percent-change from the reference case. Results indicate that an east-west oriented channel dredged in the northern section of the Shoal decreases the flushing time by 60%, and filling in the Port Edgewood Turning Basin decreases the flushing time by 30%.

Integrated into their plan is a CAD Cell in a shallow, 1km-wide area of the Providence River known as Edgewood Shoals (Figure 1a, 1b). Edgewood Shoals is known for intermittent hypoxia due to weak hydrodynamic exchange with the rest of Narragansett Bay. The circulation that severely restricts the lateral exchange of water between the Shoal and the Ship Channel has been studied as a result of 15+ years of observational data collection. The purpose of this project is to use these data to augment the results of a 3-dimensional hydrodynamic model in order to analyze the tidal and non-tidal patterns of flow on Edgewood Shoals, and identify the specific pathways through which exchange of chemical constituents with the main estuary occurs. Using these results, I propose a series of bathymetric modifications that

Circulation Dynamics and Dissolved Oxygen:
It has been well-established that the physical processes of an estuary or subestuary have a direct link to the water quality (Stram et al., 2005  Hypoxia, or the decrease in dissolved oxygen in the water column to a level of less than 3 mg/l, has significant impacts on the ecology of the estuary. Increases in biological productivity leads to amplified respiration rates, resulting in a dissolved oxygen minimum zone that occurs beneath the bloom in subpycnoclinal waters. Causes behind these isolated blooms are a combination of nitrogen loading from anthropogenic sources (Saarman et al. 2008), and weak lateral movement of bottom water within an estuary or sub-estuary with the main body of the estuary (Abdelrhman, 2005).
Weak lateral movement of a water column is an important factor in dissolved oxygen distribution in an estuary (Saarman et al. 2008, Abdelrhman, 2005, Deacutis et al, 2006, Deacutis, 2008Brietburg, 2002. Water has two ways of being re-oxygenated following periods of blooms and hypoxia-causing biological activity. This water can be re-aerated at the surface, or can be advected in from another place where it was recently interacting with surface water. In this region of Narragansett Bay, reoxygenated water is sourced from either the interaction with local surface water, or the lower Bay, where water is more likely to have been mixed. The sub-regions of the Providence River are susceptible to reductions in the lateral transport of water from the adjacent Ship Channel, which is the main conduit for lower-Bay water. Brietburg (2002) identified local residence times as being one of the most important factors influencing dissolved oxygen concentrations. Residence times are highly attributed to local hydrographic and geographic features, and can ultimately determine the water quality of a sub-region. Circulation patterns are the controlling mechanisms behind the residence times of a subregion of an estuary (Fujjuwara et al., 2002). Without regular flushing, water that sits in an area of the bay has the ability to become hypoxic, and a single event can advect this low-oxygen water to a different part of the bay. Deacutis et al. (2006) found that dissolved oxygen minima in highly stratified water columns in the Providence River rivaled those of the less-stratified water column of Greenwich Bay, RI, where residence times were determined to be longer.
A highly density-stratified water column inhibits mid-water column to low-water column parcels from interacting with surface water to experience oxygen replenishment. Near bottom, or subpycnoclinal hypoxia almost always occurs during periods of highly stratified conditions (Brietburg, 2002).

Dissolved Oxygen Dynamics on Edgewood Shoals, Providence River
In a Bay-wide low dissolved oxygen event in 2001, the oxygen minimum zone of the Providence River occurred beneath the shallow pycnocline on Edgewood Shoals (Deacutis et al. 2006). Bottom water in this case was severely hypoxic (<0.8 mg/l) to borderline anoxic (<0.1 mg/l), and has lower dissolved oxygen levels than bottom water in the adjacent Ship Channel. During this time period in which the low-dissolved oxygen event occurred, the conditions of the water column were highly stratified.  2006,2008,2009,2010,2012 and 2013 during the months of July and August. The circulation pattern leading to a highly stratified water column on Edgewood Shoals is to blame for the higher probability of low-oxygen events. It is believed that there is a direct link between the hydrodynamic disconnect between Edgewood Shoals and the adjacent Ship Channel, and high probability of low-oxygen events that occur there. It is believed that the system can be altered in such a way to allow for a higher rate of water exchange with the Ship Channel, and a higher rate of vertical mixing. If these factors can be improved, the health of the ecosystem on Edgewood Shoals will improve in a significant way. The section of the bay that comprises Edgewood Shoals is subject to significant anthropogenic pressures, including bacterial contamination, pollution from heavy metals and excessive nutrient loading. This nutrient loading is sourced from a combination of land-surface runoff, wastewater treatment facility discharge, and the discharge contribution of local tributaries (Deacutis, 2008

Circulation on Edgewood Shoals: Previous Studies
Circulation in the Providence River Estuary has been studied previously by , Rogers (2008), LaSota (2009) and Balt (2012 (Rogers, 2008;Kincaid, 2012) characterize hydrodynamic patterns in this region for both instantaneous (tidal) and residual (tidal cycle frequencies removed) flows.
Residual flow patterns observed in both numerical and laboratory models include a strong net southward flow in the surface water of the Providence River Ship Channel, a northward deep return flow in the bottom and eastern edge of the Ship Channel, and the formation of a persistent clockwise gyre on Edgewood Shoals, as is discussed below.

Impacts of Dredging on Estuarine Circulation
Bathymetric modifications by dredging can be grouped into four categories.
Channel deepening, channel widening, channel creation, and fill. The impacts of dredging on estuarine circulation have been heavily studied in Tampa Bay, FL in reference to maintenance dredging projects for their shipping lanes (Zhu et al, 2014, Goodwin, 1987. These studies have exemplified that widening and deepening the main shipping channel in Tampa Bay, FL will increase the tidal range, and decrease the tidal phase from the mouth to the head of the bay. More importantly, it was discovered that widening and deepening channels will cause a positive shift in nontidal, or residual circulation (Goodwin, 1987). Goodwin (1987) finds that with deepening and widening of channels in shallow areas, increasingly rapid transfer of dissolved chemical constituents is observed. Additionally, increased salinity in upper reaches of Tampa Bay have been used as a metric for increased tidal flushing.
Circulation restriction zones are a side-effect of maintenance dredging in a shallow estuary. Circulation restrictions are described as shallow zones on the edges of deep, maintained zones, and are therefore heavily affected by irregular bottom topography in shallow, partially to well-mixed estuaries. These zones experience a decrease in lateral exchange with the adjacent deep channel, increasing local residence times for dissolved chemical constituents within the water column (Abdelrhman, 2005). The creation of deeper zones, caused by maintenance dredging of commercial ship channels, has the potential to increase the number of circulation restriction zones in a particular estuary (Goodwin, 1987).

The Providence River and Harbor Dredged Material Management Plan
Dredging has occurred in the Providence River since 1853.  , 2005). The deepest of these cells was 28 meters below the river floor. 1.5 million cubic meters of additional material was dredged in order to create these CAD cells. This material, below a certain depth, was determine suitable for offshore disposal. Unsuitable material was placed into the CAD Cells (USACE, 2005).
The current dredging project is scheduled to begin in late 2019 to early 2020. It is estimated that a similar amount of material will need to be removed from the Providence River and Harbor area. For the upcoming PRHDMMP, Edgewood Shoals is being considered for the placement of a CAD cell, due to the Shoal's proximity to the Ship Channel, and a lack of viable space in the footprint of the Ship Channel for additional CAD cells. This project requires the building of additional channels to be used for access to and from the Cell. If USACE is planning a dredging project on Edgewood Shoals for the placement of a CAD Cell, our hypothesis is that there is a specific bathymetric modification that can be applied to the Shoal which may induce flushing and exchange. Working with USACE to design a series of scenarios, I propose a series of bathymetric alternatives that may prove to be environmentally beneficial to Edgewood Shoals.

This project involves the integration of the 2005 ADCP data and the 2010 Tilt
Current Meter data to identify key circulation patterns on Edgewood Shoals. Then, the Regional Ocean Modeling System (ROMS), a finite-difference numerical model, is run using environmental forcing from Summer 2010 and is used as a reference test to ensure that these key circulation patterns are accurately represented. Finally, using the validated numerical model, I can then alter the bathymetry on Edgewood Shoals to test a suite of dredging scenarios, and analyze the key changes to the circulation pattern based on the bathymetric alterations.
ROMS (Shchepetkin andMcWilliams 2003, 2005) is a 3-dimensional, terrainfollowing, free-surface numerical model that solves the Reynolds-averaged Navier-Stokes equations (RANS), as well as the equations for the conservation of energy and scalars using simplifying assumptions (Haidvogel et al. 2008, Shchepetkin andMcWilliams, 2003 Providence WWTF's. A correction factor is applied to account for groundwater discharge rates throughout the basin (Rogers, 2008). Figure    which allows water to leave the domain as well as enter at the boundary based on hydrographic data collected at the mouth of Narragansett Bay. Radiation with nudging has been proven effective in active/passive radiation conditions (Haidvogel et al, 2008

Initial conditions:
Initial conditions are obtained from a ROMS re-start file containing existing gridded conditions for Summer 2010 with an additional 3-day spin-up period to obtain realistic density stratification. The model was spun-up from decimal day 180 to decimal day 183, and re-started on decimal day 180, with a barotropic time-step of 20 seconds and a baroclinic time-step of 10 seconds. The boundary file was checked to ensure that water levels at the east passage boundary follow closely to those of NOAA PORTS in Newport, RI. Experimental runs were completed for 40 modeled days, starting on decimal-day 180 and ending on decimal day 220.

Grid Generation:
The alterations made to the NB-ROMS existing model for this project are as follows: alteration of bathymetry files to create nine bathymetric alternatives, or dredging scenarios, the addition of environmental forcing files using real-time data from Summer, 2010, and the addition of station files to receive data output from a series of locations in Edgewood Shoals.
The NB-ROMS grid is a 175 (East-West) by 350 (North-South) node curvilinear grid with 15 terrain-following sigma layers in the vertical. The grid includes of all of Narragansett Bay with the boundary set at the mouths of the East Passage, the West Passage and the Sakonnet River in a roughly east-west orientation.
This boundary was determined to be far enough South in Narragansett Bay as to not affect the study area. The new bathymetric grid files were created using a MATLAB script that allowed the user to make changes to the already-existing grid file by creating and loading a series of depth planes to interpolate onto the existing grid.

Analysis:
Numerical tracers in the form of Lagrangian drifters and numerical dyes were used in this study to "tag" parcels of water and monitor their movement over time.
Numerical tracer concentration was analyzed in a designated box on Edgewood Shoal

RESULTS:
There are two accepted ways to use numerical models to gain insight into the flushing dynamics of a sub-estuary (Abdelrhman, 2005). The first is to monitor the movement of a numerical conservative tracer set at an initial concentration in the water column and tracked throughout the model run. The second is to simulate flow and transport to provide insight into natural flushing patterns on the both the tidal and subtidal scale. Both of these methods were used in this study to first explore the flushing pattern of Edgewood Shoal, and secondly to engineer a dredging scenario to enhance the natural effect.

The Reference Case
All differences between Dredging Scenario modeled cases will be described as      Channel to rise. Figure 16 shows an average of six surface elevation differences during a spring (red line) and neap (black line) tidal cycle. Over the 12-hour cycle, it is evident that a pattern emerges in the surface elevation difference between the Channel and the Shoal. As the Providence River "fills" during the tidal flood, the surface elevation difference between the Shoal and the Ship Channel decreases. During the slack before tidal flood, in the neap cycle, the Ship Channel sits roughly 0.5 cm higher than the Shoal.    Basin to the ambient depth of the Shoal (2 meters). These filling scenarios are environmentally and economically conscious, and would provide a beneficial opportunity for clean material disposal that would otherwise need to be transported to Rhode Island Sound for offshore disposal.

Results of Model Runs of Dredged Scenarios
Several lines of evidence, including repeat DO surveys, moored and spatial circulation data and lab circulation models, suggest Edgewood Shoals suffers from chronically poor water quality related to restricted flushing due to distinct hydrographic regimes created by dredging a deep channel adjacent to a shallow Shoal.
I am using an existing hydrodynamic model combined with existing circulation data to further define specific details of circulation on and flushing from Edgewood Shoals.
Modeling cases are also used to explore a number of strategic dredging scenarios for Edgewood Shoals, defined in consultation with USACE, and to characterize how these might lead to improved flushing and water quality. Eight experimental dredging scenarios will be compared in this section (Runs 1a,1b,1c,1d,2a,2b,2c,2d). The majority of the results for Scenario 3a will be presented in Appendix C, as this was a test for buoyancy-driven flows and did not characterize a realistic dredging scenario.   E-folding times (Monsen et al., 2002), an accepted way to describe efficiency in river or estuarine flushing (the time it takes for each scenario to reach 37% of its original concentration (100%)) are featured in When scenarios are grouped with their respective Shoal modifications, it is evident that out of the "Dredging Northern Access Channel", Runs 1a, 1b and 1c, have the greatest effect in lessening the amount of dye constituent that remains on the Shoal. In the scenarios that place a CAD cell in the southwestern section of the Shoal (1c, 2b, 2c) Run 1c has the lowest retention time. Run 1c, which also fills in the turning basin, has the lowest retention time for water parcels out of the scenarios that fill in the Turning Basin.  Drifters were released at an incoming tide during the neap cycle. Figure 27: Modeled surface elevation (zeta) difference between the Ship Channel and the Turning Basin (TB) on Edgewood Shoal over 6 averaged neap and spring cycles, for each dredging scenario (Run 1a in red, 1b in blue 1c in pink and 1d in green). A negative value indicates that the Shoal surface elevation is higher than the Ship Channel, whereas a positive value indicates that the Ship Channel has a higher surface elevation than the Shoal. Figure 28: Modeled surface elevation (zeta) difference between the Ship Channel and the Turning Basin (TB) on Edgewood Shoal over 6 averaged neap and spring cycles, for each dredging scenario (Run 2a in red, 2b in blue 2c in pink and 2d in green). A negative value indicates that the Shoal surface elevation is higher than the Ship Channel, whereas a positive value indicates that the Ship Channel has a higher surface elevation than the Shoal.
Modeled differences in sea-surface height were calculated between the Turning Basin and the Ship Channel. From dredging design 1 (Runs 1a, 1b, 1c and 1d in figure   27) there is a roughly 0.1 cm increase in the difference between the reference case seasurface height gradient, and the dredged cases during the neap tide. During the spring tide, the Shoal in run 1c sits higher, earlier in the tidal cycle than the reference case.
This implies that a sea-surface height gradient is created between the Shoal and the Ship Channel that is greater than the reference case. This sea-surface height gradient leads to east-west flows that would move a constituent from the turning basin, off of the Shoal, which is consistent with both the drifter as well as numerical dye results.
The same pattern occurs in the spring cycle in model runs 2a and 2c (figure 28), where the Turning Basin (Shoal) sits higher than the adjacent ship channel, forming a seasurface height gradient that produces off-Shoal flows.  There is an understanding of estuarine flushing and water parcel residence times as being a function of the freshwater input rate from rivers, balanced by the outgoing flows in the lower estuary as it interacts with shelf-water (Asselin and Spaulding, 1993). Since this method does not take into account bathymetric structures that retain water, this formulation leads to relatively low residence times, and high estimated flushing rates. This is especially true in sub-regions of Narragansett Bay and the Providence River Estuary. In the case of Edgewood Shoals, applying this type of residence time calculation for the average Providence River water parcel is unrealistic, as it does not take into account the key circulation and flushing patterns that occur on the Shoal.

Results show that NB-ROMS is capable
Numerical Dye results indicate that Runs 1a, 1b and 1c (east-west channel design) are the most efficient in removing dye from the Shoal than the reference case.
Since exchange is classified in this area as the east-west flow of water onto and off of the Shoal, a significant increase in the eastward or westward velocities in the northwest section of the Shoal indicate that water is either moving eastward off of the Shoal into the Ship Channel, where it is transported down-bay, or that water is jumping the Shoal break and is making its way onto the Shoal in a westward motion.
It is assumed that a northward-southward increase in velocities will not have this same effect due to the orientation of the Shoal in the vicinity of the Ship Channel.

CONCLUSIONS:
The DMMP for the next maintenance dredging cycle of the Providence River and Harbor FNP includes the plan to construct a Confined Aquatic Disposal (CAD) Cell in a shallow area of the Providence River known as Edgewood Shoals. Environmental benefit for each dredging scenario is determined through analysis of flow structure on the Shoal, as well as the comparison of the retention of lagrangian drifters and numerical passive tracers (dyes) in the highly impacted section of the Shoal known as the Port Edgewood Turning Basin.
There are two major ways to decrease the residence times of water parcels on Edgewood Shoals. This is accomplished by enhancing the natural pathways that exist currently in order to get water on-to and off-of the Shoal. The first is to dredge an east-    There appears to be no significant difference between the observed residual flows in the parameter sensitivity runs where bottom roughness length is set to 1 cm, and 5 mm, respectively. However, when the bottom roughness length is increased by a factor of 10, from 1 cm to 10 cm, there appears to be a reduction in the amplitude of the residual flow pattern in both the northward and eastward directions. During the spring cycle at the Port Edgewood Channel output station (PEC, Station 80), the direction of residual flow reverses from net northward when the bottom roughness length is set to 1 cm and 5 mm, respectively, to net southward when the bottom roughness is increased by a factor of 10, from 1 cm to 10 cm.
While it is an interesting result, increasing the bottom roughness length to 10 cm is unrealistic for this application of the ROMS model. According to literature for ROMS applications in the Chesapeake Bay area a similar tidal-straining, partially-towell-mixed estuary, the bottom roughness length is set to a value of less than 1 cm.
Due to the small differences in flow observed in the sensitivity run between the set NB-ROMS value of 1 cm, and the Li and Zhong (2009)

Results
The results will be presented as follows: All TCM data with compasses will be analyzed for residual and instantaneous means, absolute maxima, standard deviation and variance. TCM's with compasses will then be analyzed individually by location based on significant environmental events: one high discharge event from the Pawtuxet River, one spring tide cycle, one neap tide cycle, one strong southerly (northward) wind event, and one strong northerly (southward) wind event. TCM's without compasses will be presented as raw, unprocessed data. The TCM station that experiences the smallest north/south and east/west variance (less than 3 cm/s) in both northward and eastward velocity is ES02, in the Turning Basin ( figure B5). The largest north/south variance is observed adjacent to the Ship Channel (23 cm/s). The TCM adjacent to the Port Edgewood Channel also has a strong north/south variance, at 18 cm/s), but a relatively small east-west variance at 3.5 cm/s. ES01, at the Save the Bay dock, shows a 10 cm/s variance in the east-west direction and a 6 cm/s variance in the north-south direction.    Figure C3: Modeled surface elevation (zeta) difference between the Ship Channel and the Turning Basin (TB) on Edgewood Shoal over 6 averaged neap and spring cycles, for each dredging scenario (Run 3a spring tide in red, 3a neap tide in blue reference neap tide in solid black, and reference spring tide in dotted black). A negative value indicates that the Shoal surface elevation is higher than the Ship Channel, whereas a positive value indicates that the Ship Channel has a higher surface elevation than the Shoal.