Amendment_0001_Package_Shiner_W912P421B0008_U00001_01June2021.pdf

W912P421B0008

SECTION SF 30 BLOCK 14 CONTINUATION PAGE

SUMMARY OF CHANGES

SECTION 00 21 13 - INSTRUCTIONS TO BIDDERS

The following have been modified:

52.252-3 ALTERATIONS IN SOLICITATION (APR 1984)

Portions of this solicitation are altered as follows:

0001 Amendment 0001:

1. Amendment 0001 has modified Clause 52.252-3 with additional information added to the solicitation.

2. Added: Phone number for Bid opening:

Call in number 1-877-336-4440 Access number 733187 Time: 3:00 P.M. Eastern Standard Time

3. Added: Question and Answer document from pre-bid meeting

4. Added: University at Buffalo (U.B.) Final Project Report.

5. Added: Dive Inspection of Seawall at Broderick Park.

6. The bid opening date has been changed from 31 May 2021 to 14 June 2021.

(End of Summary of Changes)

Emerald Shiner Project

Pre-Bid meeting 18 May 2021 from 01:00 till 02:00 Eastern Stand Time.

1. Contractor:

What is the makeup of the one-foot overhangs on the Niagara River Side of the project? Is there more information/drawings on it?

ANSWER: The existing precast concrete block overhangs the timber crib by approximately one.

This was determined by underwater dive inspection performed by Corps of Engineers divers. A copy of the diver inspection report is attached. There are no available drawings of the existing concrete and timber crib structure. The cross sections of the existing structure shown on the contract drawings are based on site inspections and drawings of similar structures in the Buffalo Harbor area.

2. Contractor:

What is to be used to fill in voids on Niagara River side of the wall? Grout?

ANSWER: This is either concrete or grout as shown and/on noted on contract drawings C-305 through C-308.

3. Contractor:

Has this concept been done else were?

ANSWER: The design is unique for this project and is based on a study performed by the University of Buffalo for the Corps of Engineers. A copy of the University of Buffalo study report is attached.

Questions received via email

4. Contractor:

Can USACE provide clarification on the foam material used to fill the baffles? Manufacturer or material id?

ANSWER: Requirements for foam are specified in Paragraph 2.2 of contract specification Section 01 35 13.

5. Contractor:

Bid date is Memorial Day. Is this correct, will there be an extension?

ANSWER: Solicitation response date has been updated.

6. Contractor: Please clarify what is to be included in each of the bid items, provide Measurement and Payment Specification.

ANSWER: Contract specification Section 01 22 00.00 10 PRICE AND PAYMENT PROCEDURES contains this information for the bid items.

7. Contractor: Please clarify what is to be included in each of the bid items, provide Measurement and Payment Specification.

ANSWER: Contract specification Section 01 22 00.00 10 PRICE AND PAYMENT PROCEDURES contains this information for the bid items.

8. Contractor: If able, please provide a copy of the dive inspection report.

ANSWER: The dive inspection report has been posted.

9. Contractor: Is the tremie concrete 3000 psi or 5000 psi?

ANSWER: 5000 psi per Paragraphs 1.3.3 and 2.2 of contract specification Section 03 30 53.

10. Contractor: Please confirm crib grout to be 3000 psi.

ANSWER: Grout is 3000 psi per Paragraph 2.1.6 of contract specification Section 03 30 53.

11. Given the current volatility of the steel market, can USACE benchmark steel prices at time of bid?

ANSWER: No

12. Contractor: Please clarify the extent of work for Vertical Concrete Surface Repair. Is this for base bid area only or does it include repairs on Option E?

ANSWER: Vertical Concrete Surface Repair is a base bid item per Paragraph 1.2.4.1 of contract specification Section 01 22 00.00 10 that involves work over the height of the vertical concrete surface of the existing concrete wall to remove loose existing concrete, scarify the existing concrete surface and place new concrete between the existing concrete wall and new steel plate. This does not include Option E work.

13. Contractor: What does the river bottom consist of along the work area?

ANSWER: The provided dive inspection report states: “The bottom consists of gravel and small cobbles with various timber, stone and other debris.” There is more debris in the recess area than along the remaining length of wall.

14. Contractor: What is the difference between Bid Item 9 and 13? Both are for Grouting of Ferry Landing 80 CY.

ANSWER: Bid Item 13 is an option that would only be exercised if additional grout is needed beyond that provided by Base Bid Item 9 for the Grouting of Ferry Landing.

This supporting document is an informational reference. The Project

Specifications govern actual requirements.

Final Project Report:

Research and Design for Passage of the Emerald Shiner

Contract no. W912P4-14-C-0019

Submitted by

J.F. Atkinson, S. Bennett, J. Qiao, and K. Vorenkamp Department of Civil, Structural and Environmental Engineering and Department of Geography

University at Buffalo, State University of New York

Submitted to

Bryan Hinterberger Project Manager Buffalo District, US Army Corps of Engineers

July 6, 2018 i

Executive Summary

This project aims to develop a design for a fish passage structure to allow emerald shiner minnows (Notropis atherinoides) to move upstream from the upper Niagara River into Lake Erie. The focus of the study is along the Broderick Park seawall on Unity Island, which is the proposed site for the fish passage structure, and also is scheduled to be repaired, after having weathered many years in the Niagara River. The study began in September 2014 and comprises four main parts: (1) model and field evaluation of hydraulic conditions in the upper Niagara River and mapping of hydraulic barriers and possible passageways between about the Peace Bridge upstream and the northern tip of Unity Island downstream; (2) more detailed field study of flow and turbulence characteristics along the seawall, evaluating conditions that might affect shiner ability to swim towards Lake Erie; (3) laboratory study of fish swimming ability, focusing both on single fish and also on tests with small schools (up to 20 fish); and (4) application of fine scale hydrodynamic modeling derived as an extension of Part 1 to evaluate, along with input from representatives of the US Army Corps of Engineers (USACE), different possible designs that would accommodate fish swimming ability determined from Part 3, and combine that model with a neural network model to simulate virtual fish moving in a turbulent environment.

Environments to be tested include the natural channel and also one amended by passage structures, and the virtual tests are proposed as a means of pre-evaluation of the structure design.

Part 4 also includes additional laboratory study of fish behavior with and without a passage structure in a flume. Each of these parts of the project are reported in each of the chapters of this report, and in many cases the chapters are partially edited versions of papers that have either already been published, are in review for publication, or are in preparation to be submitted.

Discussion, conclusions, and suggestions for additional work, as applicable, are included in each chapter.

Parts 1 and 2 each resulted in a Master of Science thesis in the Department of Civil, Structural and Environmental Engineering (CSEE) at the University at Buffalo (UB), and material from each thesis was extracted to produce peer reviewed publications. Parts 1 and 2 of this report are reproductions of those publications (Allen et al. 2017 and Delavan et al. 2017, respectively), with some reformatting to fit the presentation of this report. Results from Parts 3 and 4 will form the basis for PhD dissertations in CSEE. Because of the nature of the components that make up each main chapter, there is some unavoidable duplication of introductory material. On the other hand, each chapter is more or less self-contained.

The main results of Part 1 are a characterization of different flow regimes in the upper river resulting from different water levels in Lake Erie, and development of a hydrodynamic model to investigate distributions of velocity in the river. A frequency analysis of Lake Erie water levels is conducted to define high (exceeded 10% of the time), median (exceeded 50% of the time), and low (exceeded 90% of the time) water levels. The river is modeled with a software hydrodynamic code, Flow3D, using a two-dimensional, depth-averaged approach, with each of the three water levels as upstream boundary conditions. Using an assumed critical velocity of 40 cm/s, which was determined as the maximum sustainable swimming speed for the shiners, it is determined that there are few passage pathways available, and then only for the lower water levels (lower water level translates to slower river velocities). A least cost path analysis also is ii developed as a means of identifying possible passage pathways in the river. This method was investigated for possible application to evaluate different passage designs.

In Part 2, the flow and turbulence characteristics along the seawall are examined, using a combination of modeling and field measurements, to determine conditions that would provide a satisfactory environment in which fish would be able to swim and successfully move upstream.

However, because of uncertainties in our ability to design for turbulence, the design described in Part 4 is developed primarily on the basis of mean velocities and draws heavily on the results of Part 3.

Part 3 is the main experimental portion of the study and consists of fish swimming tests conducted in tubes placed in a laboratory flume. An initial series of tests considered fish swimming individually. Fish are collected from the Niagara River in various locations near the seawall and, once acclimated to the laboratory and flume conditions, are subjected to flows of varying velocity. The time to fatigue is recorded, or if the fish is able to swim at the given speed for the duration of the test (both 60 minute and 200 minute tests are reported), it is classified as swimming at sustained speed. A characterization of fish ability and dependence on a number of factors such as body size (or age), water temperature and other water quality parameters, or time of year when the fish was caught, is unknown and one goal of this part of the project was to determine these dependencies. Unfortunately, no clear dependencies were evident, and the design process (Part 4) considers all fish tested as a single population. The design condition that is focused in in Part 4 is based on a determination of the percentage of fish that can swim at a given water velocity, preferably in a sustained mode.

Because of inconsistencies in some of the data for the 60 minute swimming tests, and also to follow more well-established procedures for fish testing, attention is focused primarily results from the 200 minute tests. For these tests, it is shown that approximately half the fish tested were able to swim sustainably at a velocity of 40 cm/s. Smaller percentages are associated with higher velocities. Although this measure is used in testing the performance of different passage designs, it should be noted that this percentage does not include fish that may be able to swim occasionally at burst or prolonged speed, so the percentage of fish that may be able to swim in a passage designed for 40 cm/s may be somewhat greater than 50%.

Anecdotal evidence, particularly field observations in the study area, suggest schooling may be a strategy used by the fish to enable them to manage higher stream velocities (among other motivations such as protection from predators). Preliminary tests with small schools of fish (up to 20) showed there may be some hydrodynamic benefit of swimming in schools. Present results, based both on a critical swimming speed as well as endurance tests similar to the individual fish tests, support the idea that fish are able to swim sustainably at higher speeds when they are in a school. It is difficult to extract quantifiable results due to the wide range of behaviors observed, and to experimental uncertainties, but it appears that the sustainable swimming speed may be as high as 60 cm/s when fish are in a school, a 50% increase relative to the individual test results.

For Part 4, the hydrodynamic model of Part 1, which has a resolution of 2 m, was refined to a resolution of 8 cm in the region near the seawall and the modeling was conducted using a nested approach, in which the larger scale river model provides boundary conditions for the finer scale iii model in the region of interest. This smaller dimension is approximately the size of a large emerald shiner. The resolution of the Part 1 model is incapable of examining flow in the passage structure in any detail. In addition, improvements in the Part 1 model are made to better fit observed flow rates. These improvements are mainly due to adjustments in drag coefficients for the bottom and the seawall, and to incorporation of improved bathymetry data. All three water levels determined in Part 1 are tested, but again to be conservative, focus is placed on the higher water level.

Several different structure designs are examined, and along with input from USACE personnel a passage structure is suggested that takes into account practical considerations of the site as well as the swimming ability of the fish. The chosen passage structure is based on the installation of modular trapezoidal blocks along the seawall, and several dozen configurations of blocks were tested. The design goal is to reduce velocities near the seawall to about 40 cm/s, as determined in Part 3 as the velocity at which half the shiners should be able to swim upstream. The choice of enabling 50% of the fish to make this journey is a compromise based on the need to significantly improve (from zero or nearly zero percent) the migration ability of this species, while not over-designing the desired structure, and working within constraints imposed as part of the overall seawall restoration. Hydrodynamic modeling suggests the recommended structure will reduce flow velocities over most of the region close to the seawall to 40 cm/s for the highest water surface elevation condition. Compared to the natural condition, where river velocities can be on the order of 1 – 2 m/s, the proposed design is expected to produce a significant increase in the number of fish that can navigate upstream to Lake Erie. Importantly, the proposed design is predicted to produce no noticeable water lever change in the river.

Part 4 also includes a somewhat experimental task to develop a neural network model to create a “virtual fish” that reacts to environmental stimuli and “swims” in a way that mimics real fish behavior. In the present study the only environmental stimuli being considered are hydrodynamic conditions, primarily flow velocity and, to a lesser extent, turbulence characteristics.

Experimental results are shown for tests in which blocks are placed in a flume to demonstrate the desired reduction in flow velocity along the side of the flume, and also to show that the fish recognize the velocity advantage provided by the blocks and are willing to swim in the amended area. Detailed procedures are developed to obtain both flow data and fish swimming pathways at the same time, and this information is used in development of the virtual fish model. The experiments clearly support the proposed design, though the virtual fish model results are still somewhat preliminary. Ultimately, this type of approach is expected to add an important tool for pre-evaluation of different fish passage design options.

iv

Table of Contents

Executive Summary .......................................................................................................................... i

Part 1: Potential barriers to upstream fish passage caused by anthropogenic river modifications:

A computer modeling study of emerald shiners (Notropis atherinoides) in the upper Niagara River

1.1 Abstract

1.2 Introduction

1.3. Methods

1.3.1 Computational fluid dynamics modeling

1.3.1.1 Modeling scenarios

1.3.1.2 Model geometry

1.3.1.3 Computational mesh and boundary conditions

1.3.1.4 Runtime and initial conditions

1.3.1.5 Model validation

1.3.2 Least-cost path analysis and barrier identification

1.3.3 Barrier verification with field measurements

1.4 Results and discussion

1.4.1 Compu1tational fluid dynamics model validation

1.4.2 Horizontal velocity profiles

1.4.3. Least-cost path analysis

1.4.4 Verification of potential barrier prediction

1.5 Conclusions

References

Part 2: Anthropogenic turbulence and velocity barriers for upstream swimming fish: A field study on emerald shiners (Notropis atherinoides) in the Upper Niagara River

2.1 Abstract

2.2 Introduction

2.3 Methodology

2.3.1 Vertical profiles of velocity

2.3.2 Velocity time record collection

2.3.3 Velocity time record analysis

2.4 Results

2.4.1 Vertical profiles of water velocity

v

2.4.2 Near shore point water velocities

2.4.4 Reynolds shear stress

2.4.5 Turbulence intensity

2.4.6 Size of main energy containing eddies

2.4.7 Energy within the problematic range of eddy sizes

2.5 Discussion

References

Part 3: Swimming capability experiments for emerald shiner minnows in the upper Niagara River

3.1 Introduction

3.2 Swimming endurance tests – single fish

3.2.1 Preliminary tests – 60 minute duration, circular tube

3.2.1.1 Experimental runs

3.2.1.2 Results

3.2.1.3 Critical assessment

3.2.2 Longer duration (200 minute) swimming endurance tests

3.2.2.1 Experimental setup

3.2.2.2 Acclimation

3.2.2.3 Three swim tunnels vs. one

3.2.2.4 Results

3.2.3 Schooling experiments

3.2.3.1 Preliminary observations

3.2.3.2 Critical velocity tests

3.2.3.3 Schooling swimming endurance tests

3.2.3.4 Results

3.3 Discussion and recommendation

3.3.1 Individual swimming

3.3.2 Schooling tests

3.4 Conclusions and future directions

References

Part 4: Fish passage design along Broderick Park seawall, Buffalo, New York

4.1 Simultaneous measurement of turbulent velocities and fish movement in complex flows using large-scale particle tracking velocimetry and convolutional neural network vi

4.1.1 Abstract

4.1.2 Introduction

4.1.3 Methodology

4.1.3.1 Fish Collection

4.1.3.2 Experimental setup

4.1.4 Results

4.1.4.1 Data Analysis Scheme

4.1.4.2 Image Preprocessing

4.1.4.3 Fish Trajectory

4.1.4.4 Fish Detection Model Training and Inference

4.1.4.6 Moving Object Detection

4.1.5 Application of the data analysis scheme

4.1.5.1 Image Preprocessing

4.1.5.2 PTV Analysis and Fish Trajectory Analysis

4.1.6 Discussion

4.1.6.2 Instantaneous Hydrodynamic Environment

4.1.6.3 Uncertainty of Measurements

4.1.7 Conclusion

References:

4.2 Unconfined fish passage design and assessment for emerald shiners in upper Niagara River

4.2.1 Abstract

4.2.2 Introduction

4.2.3 Methodology

4.2.3.1 Study Site and Design Constraints

4.2.3.2 Numerical modeling

4.2.4 Fish passage design (numerical model application)

4.2.5 Physical experiments

4.2.5.1 Fish collection

4.2.5.2 Experimental setup

4.2.5.3 Experimental results

4.2.6 Discussion

4.2.6.1 Coupling CFD and physical modeling for fish passage design vii

4.2.6.2 Unconfined modular fish passage

4.2.6.3 Fish passage connectivity

4.2.6.4 Broader implication of unconfined fish passage on riverbank stability

4.2.7. Conclusions

References

4.3 Virtual fish development from fish migration dataset in lab environment

4.3.1 Abstract

4.3.2 Introduction

4.3.2.1 Fish Perception

4.3.2.2 Machine learning, neural networks and behavior modeling

4.3.2.3 Contribution

4.3.3. Methodology

4.3.3.1 Model development

4.3.4 Model architecture

4.3.4.1 Evaluation Metrics

4.3.5 Data Preparation

4.3.5.1 Data Collection

4.3.5.2 Data Selection and Preprocessing

4.3.6. Results

4.3.7. Discussion

4.3.7.1 Factors affecting fish path-choosing decisions

4.3.8 Future work

4.3.8.1 Refine and expand training inputs

4.3.8.2 Couple virtual fish with CFD simulation of fish passage

4.3.9 Conclusion

References

Part 1: Potential barriers to upstream fish passage caused by anthropogenic river modifications: A computer modeling study of emerald shiners (Notropis atherinoides) in the upper Niagara River

I.W. Allena, S.K. Delavana,1, A.R. Hannesb, and A. Pérez-Fuentetajac aCivil, Structural, and Environmental Engineering, University at Buffalo, The State University of New York, United States bUnited States Army Corps of Engineers, Buffalo District, United States cBiology Department and Great Lakes Center, Buffalo State, The State University of New York, United States 1Current address: United States Army Corps of Engineers, Walla Walla District, United States

(published in Ecological Engineering 103, 76-85 (2017))

1.1 Abstract

Anthropogenic modifications at the head of the Niagara River attributed to land reclamation and the installation of vertical, bulkhead seawalls have altered the river channel, leading to increased water velocities and reduced availability of aquatic habitat. As a result of the increased velocities in the channel, concerns have been raised that the habitat connectivity of the emerald shiner (Notropis atherinoides), a key forage species for sportfish and threatened avian species, has been reduced between Lake Erie and the upper Niagara River. In this study, the influence of water velocity stressors on emerald shiner movement was assessed by identifying potential hydrodynamic barriers to upstream fish movement using a combination of depth-averaged velocities simulated in a 2-D hydrodynamic model and field measurements collected with a current meter. Results indicate that high velocities (in excess of 2.0 m/s) immediately downstream of the Peace Bridge, an international road crossing between Buffalo, NY, USA and Fort Erie, Ontario, Canada, restrict potential emerald shiner movement to within close proximity of either shoreline. Along both shorelines, several additional threats to successful upstream passage exist, the largest of which are located along vertical bulkhead seawalls. The extent of the identified potential barriers fluctuates slightly with water level, but the locations remain consistent. Finally, of the potential barriers identified using the model, the field measurements indicate the seawall along Broderick Park, along the U.S. shoreline, is the most substantial impediment to upstream movement, as velocities within several shiner body-lengths of the shoreline remain between 0.83 and 1.18 m/s for nearly 200 m. The results of this study, in conjunction with empirical estimates of passage success, swimming tests on shiners native to the Niagara River, and additional field measurements will help determine if remediation work is warranted to facilitate upstream passage at the identified impediments.

1.2 Introduction

Over the past century, several stressors have been placed on the emerald shiner population (Notropis atherinoides) in the Niagara River ecosystem including damming, habitat modifications, and the introduction of non-native species (Coordinating Committee on Great

Lakes Basic Hydraulic and Hydrologic Data, 1998). At the head of the Niagara River, extensive human alterations have occurred over the past two centuries. Several notable alterations include the expansion and armoring of Union Island, construction of the Peace Bridge and International Railways bridges, implementation of the Broderick Park seawall by 1949, expansion and armoring of Nicholls Marine between 1949 and 2004, and consistent dredging and river sweeping to maintain the navigability of the river (see Fig. 1.1 for the locations of the aforementioned features; Coordinating Committee on Great Lakes Basic Hydraulic and Hydrologic Data, 1998). The impact of these alterations on the footprint of the Niagara River channel can be clearly seen in the historical navigation maps maintained by the National Oceanic and Atmospheric Administration. These alterations have increased water velocities along the shorelines used by emerald shiners to move between the river and Lake Erie placing, yet, another stressor on this small minnow species (Coordinating Committee on Great Lakes Basic Hydraulic and Hydrologic Data, 1998). In fact, Poff and Zimmerman (2009) found that there is an increased risk of ecological change with increasing alteration of hydrodynamic conditions of water bodies. The physical modifications to the Niagara River have effectively channelized the river, nearly eliminating the presence of natural roughness elements that would disrupt the flow and pro- duce valuable hydraulic cover for aquatic organisms (Coordinating Committee on Great Lakes Basic Hydraulic and Hydrologic Data, 1998; Kleinschmidt Associates and Riversedge Associates, 2005; Mudroch and Williams, 1989). Hamm and Hannes (2012) field observations found large congregations of emerald shiners within slackwater areas of the shoreline in which the emerald shiners appear to have trouble navigating upstream. The hydrodynamic conditions of the river, between the Peace Bridge and International Railroad Bridge (Fig. 1.1), have the potential to restrict fish movement between the Niagara River and Lake Erie. The natural movement habits of the emerald shiner are not fully understood, as of yet; however, emerald shiners have been observed to attempt to move upstream along the shorelines of the Niagara River.

Fig. 1.1. Google Earth image (Google, 2013) of the head of the Niagara river. The flow direction is from south to north (bottom to top). Locations identified in this Figure include; (1) Buffalo water level gauge, (2) Fort Erie water level gauge, (3) Peace Bridge water level gauge,

(4) Buffalo, NY, U.S.A.(5) Fort Erie, Ontario, Canada, (6) Extent of the hydrodynamic model domain, (7) International Railway Bridge, (8) Frenchman’s Creek water level gauge, (9) Grand Island, NY, U.S.A. (10) Peace Bridge, (11) Canadian vertical seawall, (12) Bird Island Pier, (13) Nicholls Marine, (14) Broderick Park, (15) Broderick Park vertical bulkhead shoreline recess, and (16) Unity Island.

In the past 20 years, numerical modeling has gained traction as a useful compliment to ecological studies, allowing investigators to simulate numerous hydraulic variables of ecological importance. In fact, studies such as Rodríguez et al. (2006) and Grantham (2013) have shown that hydrodynamic modeling techniques can be used to predict passage success of migratory fish.

Generally, studies utilizing numerical modeling techniques for ecological purposes have been contained to small spatial scales in order to simulate hydraulic variables on scales ecologically relevant to the species of interest (Crowder and Diplas, 2000). The extension of ecologically oriented numerical techniques to a large spatial scale is feasible, especially when the results are not directly tied to the design of fish pas- sage structures. Large scale hydrodynamic modeling, when coupled with small-scale, highly targeted, and ecologically relevant field measurements, allows for the evaluation of fish passage in large ecological systems such as the Niagara River, ultimately making it possible to identify potential barriers across a wide range of hydro- dynamic conditions.

Hydrodynamic barriers to fish passage can take many forms, with common examples including dams, culverts, and other hydraulic control structures (Jones et al., 1974; Zheng et al., 2009).

While barriers may not be identical in the manner in which they limit fish passage, they are similar in that they reduce or limit upstream fish passage rates. For instance, some barriers, like dams, create physical impediments to passage, while others, like culverts, create hydraulic or behavioral deterrents (Jones et al., 1974; Zheng et al., 2009). If a fish is unable to jump over a dam or traverse large water velocities (as is assumed in human modified rivers such as the Niagara River) natural patterns of movement can be interrupted, negatively impacting the fish population (Castro-Santos and Haro, 2010). The extent, location, and function of a barrier in limiting fish passage are products of many variables related to the fish species of interest and the surrounding environment. The variable that is recognized as the most important when evaluating performance at barriers is fish swimming capacity. Swimming capacity has long been a design parameter in fishway construction and evaluations (Castro-Santos et al., 2009), and has been extensively studied (Beamish, 1979; Brett, 1964; Hammer, 1995).

The swimming capacity of a fish can be described in terms of its performance in three main swimming modes: sustained, pro- longed, and burst (Beamish, 1979; Hammer, 1995). Brett (1964) showed that swimming performance could be approximated by swim-fatigue curves, which allow the prolonged and burst swimming modes of a given fish to be quantified with mathematical relationships. Empirical relationships based on Brett’s (1964) work have been used to approximate passage success and facilitate the design of fishways and other hydraulic control structures by identifying the flow velocities required to allow successful entrance and passage of fish (Peake et al., 1997). According to the distinction of swimming modes outlined in Brett (1964) and reviewed in Beamish (1979), when oncoming flow speeds allow emerald shiners to swim within their sustainable swimming mode, their upstream progress will be unimpeded because they can sustain these speeds for very long times (>200 min). However, when oncoming flow velocities force emerald shiners to swim in either prolonged or burst modes, their upstream progress is jeopardized by fatigue. Jones et al. (1974) and Young (2001) published an average critical swim speed values for emerald shiners of 0.59 m/s for a 10 min test increment, and 0.61 m/s for a 5 min test increment, respectively, and these values are assumed to be most closely related to a prolonged swim speed for emerald shiners. A burst swim speed for emerald shiners was calculated by Leavy and Bonner (2009) as an absolute swimming speed of 0.81 m/s for shiners tested over a fatigue time of 10 s. Finally, unpublished swim tests were conducted by the U.S. Army Corps of Engineers Research and Development Center in Vicksburg, Mississippi, U.S.A., giving a sustained swim speed for emerald shiners of 0.4 m/s; water velocity magnitudes above this value are likely to fatigue the fish. The methods of Peake et al. (1997) and the critical swim speed of 0.4 m/s were used in this study in order to conservatively estimate the location and nature of barriers to emerald shiner passage at the head of the Niagara River. It is recognized that there are many biological (size, sex, species, etc.) and environmental (water temperature, dissolved oxygen concentration, light levels, predation pressure, etc.) factors that can alter the sustained, prolonged, and burst swimming speeds achieved by any fish (Beamish, 1979; Hammer, 1995), and thus may influence their theoretical performance at hydraulic barriers. However, it is important to determine the existence of potential barriers in order to assess the impact of problematic areas to fish passage.

The impacts of water velocity stressors on the emerald shiner population in the Lake Erie and Niagara River ecosystem are of concern because it is a vital species in the region from both an ecological and economic standpoint. Although, without pre-industrialization field fish surveys of the region, it is impossible to determine the full impact of water velocity stressors on the emerald shiner populations. However, it has been established that the emerald shiner is a key prey species in the Lake Erie Niagara River food web as they are one of the predominant forage fish species in the eastern basin of Lake Erie (Lake Erie Forage Task Group, 2012). Emerald shiners are known to be a major food source for the New York State threatened common tern (Sterna hirundo), an avian species that exploits the habitat available in the Niagara River ecosystem for various life processes. In addition, the Niagara River flyway is an internationally designated Important Bird Area (IBA; in the same category for conservation purposes as the Everglades and the Galapagos Islands) supporting the largest and most diverse concentration of gulls in the world (Bird Life International, 2016). Economically, the emerald shiner is valuable as a commercially and recreationally harvested bait fish for bass, muskellunge, and other sportfish that attract anglers and other outdoor enthusiasts to the region.

The objective of this study was to use a combination of computational fluid dynamics shallow-water model simulations, least-cost path approximations, field measurements, and published and unpublished data about the swimming capability of emerald shiners to develop informed, scientific examinations of hydrodynamic barriers to fish passage in large scale, human-modified rivers. The system used in this study is the upstream passage of emerald shiners from the Niagara River to eastern Lake Erie. Specific to the Niagara River, our main goal is to quantify the horizontal velocity profiles present in the river for a range of typical flow conditions to identify locations that may be potential barriers to fish passage. The results from this study will not only help improve our understanding of emerald shiner movement in the upper Niagara River but will also provide a framework for future studies in other river systems.

1.3. Methods

1.3.1 Computational fluid dynamics modeling

1.3.1.1 Modeling scenarios

Computational fluid dynamics (CFD) modeling techniques were used to quantify how water velocities change as hydrodynamic conditions change, and ultimately to determine how those changes influence the potential barriers to upstream passage. The water levels at the Fort Erie gauge, located approximately one kilometer south of the Peace Bridge (Fig. 1.1), were chosen as an indicator of the hydrodynamic conditions in the river such as average velocity, assuming that the water level in the river is controlled mainly by the water level in Lake Erie (URS Corporation, Gomez and Sullivan Engineers, & E/PRO Engineering & Environmental Consulting, 2005). Water level was chosen in favor of discharge because discharge estimates in the Niagara River for purposes of power generation and regulation are traditionally estimated by approximate stage-discharge relationships. Three flow scenarios were chosen as representative of typical conditions in the river. The flow scenarios corresponded to the 10, 50, and 90 annual percent exceedance water levels at the Fort Erie gauge based on a 12 year record of data. The exceedance values, 10, 50, and 90, were chosen to eliminate non-typical or extreme water level events. For a sense of the approximate discharges that correspond to the range of water levels evaluated, the high water level scenario is estimated to correspond to a discharge between 6,500 and 7,000 m3/s, while the low water level scenario is estimated to correspond to a dis- charge between 5,000 and 5,500 m3/s, depending on the stage discharge relationship being used (NYPA and Ontario Hydro, 1988; USACE, 1989). In all, there is not a large fluctuation in the water level observed in the Niagara River because of the heavy regulations on Great Lakes water levels, and because of the heavy restriction on the amount of flow that must be going over Niagara Falls at any given point in time.

1.3.1.2 Model geometry

Flow-3D, a general-purpose computational fluid dynamics (CFD) software package (Flow Science Inc., 2014a) was used to create the 2-D model for this study. Flow-3D uses numerical techniques to solve the equations of fluid motion, and was used to simulate the depth-averaged velocities in the river. Key physical components in the river including the Peace Bridge Piers, Canadian and U.S. shoreline seawalls, Nichols Marine, Bird Island Pier, and Broderick Park (see Fig. 1.1) were created within the Flow-3D geometry editor using primitive shapes and approximate features sizes from Google Earth (Google Earth, 2013). The dimensions of the Broderick Park seawall were confirmed by computer aided design drawings available from the U.S. Army Corps of Engineers. The riverbed geometry, the final geometric component, was imported as a stereolithographic file (.STL). The riverbed stereolithographic file was created by digitizing bathymetric contour data available from a freely accessible navigation map (Map No.

14832) from the National Oceanic and Atmospheric Administration (NOAA) in ArcGIS (ESRI, 2013). The NOAA navigation map had a vertical resolution of six feet. All of the components in the model geometry were considered solid, non-porous media and represented a non-permeable boundary to flow. The component surfaces have designated drag coefficients that are used to calculate the shear stress at the fluid-component interfaces. The drag coefficients used for each component were calculated using the following equation (Flow Science Inc., 2014b) and are provided in Table 1, 𝐶𝐶𝑑𝑑 = 𝑛𝑛2𝑔𝑔 ℎ 3𝑎𝑎2

(1) where n is the Manning’s roughness coefficient (determined from the material of each component from Chow (1959)), g is the gravitational acceleration, h is the approximate water depth where the Manning’s roughness coefficient was determined, and a is a constant equivalent to 1 in S.I. units (Flow Science Inc., 2014b). The final riverbed drag coefficient (Table 1) was calibrated using water levels at the Peace Bridge water level gauge (Fig. 11).

Table 1.1. Properties for the geometric components of the water velocity model domain.

Component Surface Type

Manning’s Roughness (n)

Final Cd

Riverbed Bedrock 0.021 0.0031 Peace Bridge Piers Concrete 0.018 0.00209 Old Water Intake Structure

Concrete 0.018 0.00209

Broderick Park Seawall Wood 0.012 0.00093 Bird Island Pier Rip-Rap Rip-rap 0.08 0.09379 Bird Island Pier Concrete 0.018 0.00405

1.3.1.3 Computational mesh and boundary conditions

The grid extent (Fig. 1.1) and mesh size for the Flow-3D water velocity model of this study (hereafter referred to as WV Model) was based on the area of interest for evaluating passage, a spatial resolution that could be supplemented by field measurements, and the location of water level gauge data used for inflow and outflow boundary conditions. Therefore, a rectangular, non-conforming, nested mesh was chosen which included a 4 m by 4 m base resolution (total domain) and a nested 2 m by 2 m grid (between the Peace Bridge and the downstream end of Broderick Park; Fig. 1.1). The inflow and outflow boundaries of the course grid were fixed by selecting hydrostatic pressure boundaries with maximum fluid levels set at the water levels at the Fort Erie gauge (inflow) and the Frenchman’s Creek gauge (outflow; Fig. 1.1). The riverbed and the two river shorelines were set as zero-fluid flux boundaries and the river surface was set as an atmospheric pressure boundary for the course grid. The boundary conditions for the nested grid are defined automatically as inter-block boundaries and the values are interpolated using the coarse mesh grid values at the boundary intersections (Flow Science Inc., 2014b).

1.3.1.4 Runtime and initial conditions

The initial conditions of the simulations included a global fluid elevation equal to the Fort Erie gauge water level (173.86, 174.23, 174.61 m for the 10, 50 and 90 percent exceedance water level scenarios, respectively, referenced to the International Great Lakes Datum (IGLD85)), and a global fluid velocity of 1.5 m/s in the downstream direction. The simulations were run to steady-state under the assumption that the steady-state solution was sufficient to capture the magnitudes of the velocities. Steady-state is deemed to be achieved when the difference in the numerical solution of one of the simulation variables (magnitude of the total fluid mass, average mean kinetic energy, average mean turbulent energy, or average mean turbulent dissipation) is below one per- cent for two consecutive time steps. On average, the simulations took approximately 3,000 s of simulated time to reach their steady- state solution.

1.3.1.5 Model validation

The WV model was validated by comparing the simulated velocities to field velocity data collected in the Niagara River on September 4, 2014 with a 600 kHz Teledyne RDI Workhorse, therefore, the mean water level scenario was used for validation.

1.3.2 Least-cost path analysis and barrier identification

A least-cost path analysis is an ArcGIS procedure that identifies the path between two geographic points that accumulates the least amount of total cost with respect to one or more cost variables (ESRI, 2013). This study assumes that an emerald shiner moving from a downstream to upstream location will closely follow a path where the water velocities are consistently the smallest. This assumption is expected to be particularly true in locations where fish passage is potentially restricted by large water velocities. Least-cost path analyses have been used previously by Grantham (2013) to evaluate salmon passage using flow depth as the cost variable.

For this study, a least-cost path was created to connect points at either end of the 2 m by 2 m model domain along both shorelines of the Niagara River for the 10, 50, and 90 percent exceedance water level scenarios.

The least cost-pathways were used to identify the locations where upstream emerald shiner movement may be restricted based on the water velocity. Along the paths, nodes were defined as sections where the water velocity in consecutive computational cells was below 0.4 m/s, and potential barriers were defined as sections where the water velocity in consecutive computational cells was above 0.4 m/s, the conservative sustained swimming speed of emerald shiners.

1.3.3 Barrier verification with field measurements

To verify that the WV model and least-cost path analysis of this study accurately narrows down the locations of potential barriers to fish passage, a few sets of more precise field velocities were collected along the shorelines of the Niagara River (Delavan et al., 2017). Using a Swoffer 3000 water velocity meter, measurements were collected every 30 m along the U.S. and Canadian shorelines in 2012 and 2014. A 30 m sampling interval was chosen because the collection area spanned several thousand meters of the U.S. and Canadian shorelines. Due to the variation of shoreline types (vertical seawall, rip-rap, shallow bedrock) within the study area, different sampling designs had to be employed to accurately mea- sure the shoreline flow velocities. On November 6th, 2012, from near the Peace Bridge abutment to the southern end of Broderick Park along the rip-rap shoreline of Bird Island Pier, measurements were taken approximately 0.3 m from the shoreline at a depth of approximately 0.3-0.6 m. In contrast, the measurements recorded along the vertical seawall of Broderick Park were taken as close to the wall as possible without the propeller of the current meter hit- ting the wall (in order to best capture the velocities that emerald shiners must overcome), at 0.3, 0.6 and 0.9 m depth increments, and averaged over the three measurement depths. The daily average water level at the Buffalo water level gauge was 173.78 m (IGLD 85 m), approximately equal to the 90 percent exceedance model scenario (173.86 m), on the day of collection along the U.S. shoreline. Three measurements were collected along the Canadian shoreline on May 28, 2014 at each station along the shoreline at 0.3, 0.9 and 1.5 m from the shoreline at depth of approximately 0.3-0.6 m and averaged. The daily average water level at the Buffalo water level gauge was 174.32 m on the day of collection along the Canadian shoreline, which is between the 50 and 10 percent exceedance scenarios evaluated in the WV model (174.23 m and

174.61 m, respectively).

1.4 Results and discussion

1.4.1 Compu1tational fluid dynamics model validation

The results of the WV model validation are shown in Fig. 1.2, which presents a comparison of the field and simulated water velocities at corresponding locations in the river and the model domain. For smaller velocity magnitudes, below 1 m/s, the WV model tended to over-predict the velocity magnitude. Above a velocity of 1 m/s, the WV model consistently under-predicted the velocity magnitude. The field velocity measurements capture non-steady, instantaneous conditions in the river. The WV model, on the other hand, simulates steady-state, or consistent, conditions. A greater difference between the WV model and field velocity magnitudes would occur near the shorelines where there are smaller turbulent vortices (Pavlov et al., 2000) and more complex flow patterns that cause more fluctuation in the instantaneous velocity from the time averaged value. Hence, the large differences between the model and the field values at smaller velocities are expected as they represent velocity magnitudes near the shorelines. At greater velocities, typically occurring toward the center of the river, the conditions would deviate less from steady-state, as is true for this study (Fig. 1.2). This is especially true in the Niagara River where the majority of the flow is occurring in the downstream direction as it discharges from Lake Erie and where the bottom is nearly uniform in depth and composed predominantly of bedrock (Mudroch and Williams, 1989). The weaker performance of the model at lower velocities, those that emerald shiners could exploit for passage, suggests that conclusions drawn from the model should be conservative. Furthermore, ADCP velocity measurements within the presumed swimming capability of the emerald shiner were found, but no continuous stretches were identified that would suggest a potential pathway to navigate the Niagara River. The weaker performance at lower velocity and presence of navigable velocities based on the ADCP results suggests that additional field data or a more refined model grid near the shorelines would be beneficial given that field observations show emerald shiner movement is likely within close proximity of the shorelines.

Fig. 1.2. Comparison of the velocity magnitudes for the field measurements and WV model simulations for the mean water level scenario. The black line is a 1:1 line; the dashed line is the linear regression of the water velocity comparison.

1.4.2 Horizontal velocity profiles

The steady-state, depth-averaged, horizontal velocity profiles for the simulated water level scenarios are displayed in Fig. 1.3. The results in Fig. 1.3 indicate that even at the lowest water level (representing the 10th percentile annual water level), nearly the full width of the river is characterized by velocities in excess of the critical swim speed of emerald shiners. Because maxi- mum burst swimming speeds of emerald shiners do not likely far exceed 0.81 m/s (Leavy and Bonner, 2009) and the sustainable swimming speeds of 1 or 2 year-old emerald shiners are within 0.4-0.6 m/s (Young, 2001; Jones et al., 1974), the water velocities for the central 80% of water level conditions (between the 10 and 90 percent exceedance scenarios), are likely to be impassable for the emerald shiner. In all water level scenarios, the velocity profiles can be divided into three distinct regions with similar characteristics (Fig. 1.3). Region 1 is an area of small velocity magnitudes south of the Peace Bridge. The average velocities between the 90, 50, and 10 percent exceedance water level scenarios in Region 1 were 1.07, 1.13, and 1.18 m/s, respectively. Region 2 is a large magnitude velocity region between the Peace Bridge and Broderick Park. The average velocities in Region 2 were 2.04 and 2.33 m/s for the 90-10 percent exceedance water level scenarios, respectively. In all three water level scenarios the minimum predicted velocity was 0 m/s in the stagnant areas behind Nicholls Marine and the Broderick Park recess, while the maximum velocities ranged from 3.02 to 3.49 m/s in the three scenarios.

Finally, Region 3 is north of Broderick Park and is characterized by small velocities. The average velocities in Region 3 were 1.11, 1.22, and 1.34 m/s in the 90, 50, and 10 percent exceedance water level scenarios, respectively. The high flows of Region 2 are of particular concern because of the large congregations of emerald shiners that are observed immediately downstream of Region 2 (Hamm and Hanes, 2012; Delavan et al., 2017). In this region, emerald shiner passage would be limited to within very close proximity of the shorelines, consistent with observations that have been made in the field (Delavan et al., 2017). However, Regions 1 and 3 seem to have larger areas near the shorelines where the velocities would not hinder the movement of emerald shiners as is seen in more natural rivers and along the Canadian shoreline of the upper Niagara River. The Canadian shoreline of this river has been modified less significantly than the U.S. side which is a likely explanation for the larger areas of lower velocity magnitudes seen along the shorelines in Region 3. In general, emerald shiner passage is likely to be less hindered along the Canadian shoreline than the U.S. shoreline for the area modeled here.

Future restoration efforts that focus on returning the shorelines to more natural bathymetry and vegetated states would likely increase the areas available for the free movement not only of emerald shiners but even less fusiform fish species.

Fig. 1.3. WV model predictions of depth-averaged velocity magnitude for the total model domain for: (a) 90 percent exceedance water level scenario, (b) 50 percent exceedance water level scenario, and (c) high water level scenario. Geographic features: (1) the Canadian seawall, (2) the Peace Bridge, (3) Nicholls Marine, and (4) Broderick Park.

The differences in the flow characteristics between the 90, 50,and 10 percent exceedance water level scenarios has the potential to change the location, extent, or magnitude of the barriers to fish passage. For the upper Niagara River, Fig.