- Email: [email protected]
Modeling Lampricide Transport in the St. Marys River
J. Great Lakes Res. 29 (Supplement 1):694–705 Internat. Assoc. Great Lakes Res., 2003
Modeling Lampricide Transport in the St. Marys River Hung Tao Shen*, Qing Xu, and Poojitha D. Yapa Department of Civil and Environmental Engineering Clarkson University Potsdam, New York 13699-5710 ABSTRACT. This paper describes use of an integrated chemical/oil spill computer model to study lampricide transport in the St. Marys River. The model is developed based on a depth-averaged two-dimensional advection-diffusion equation using a Lagrangian discrete parcel method. The hydrodynamics in the model is computed using a depth-averaged two-dimensional finite element model. The model is used to analyze the effectiveness of alternative lampricide application strategies. This analysis showed that releasing TFM from both the Great Lakes Power Plant and the Edison Sault Power Station is more effective than releasing from the Great Lakes Power Plant only, in terms of kill-per-dollar. However, the total cost for two-site releases are higher than single-site releases. The simulations also showed that the optimum duration of TFM release is between 14 and 16 hours. Simulations of bottom velocity distribution showed that the area over which the granular Bayluscide application could cover 70 to 90% of the sea lamprey (Petromyzon marinus) population in the St. Marys River, depending on the total river discharge. INDEX WORDS:
Hydrodynamic model, Bayluscide, lampricide, St. Marys River, TFM.
INTRODUCTION The sea lamprey (Petromyzon marinus) was a major cause of the severe damage done to the populations of lake trout (Salvelinus namaycush), whitefish (Coregonus clupeaformis), and chubs (Coregonus spp.), all commercially valuable species in the Great Lakes. The Great Lakes Fisheries Commission initiated a successful program in 1958 to use the lampricide TFM (3-trifluoromethyl4-nitrophenol) to reduce sea lamprey populations. TFM selectively kills sea lamprey larvae when applied in carefully regulated dosages. Field studies conducted by the Great Lakes Fishery Commission showed that a certain critical concentration of lampricide needs to be maintained for a specific minimum duration to provide a lethal dose. A regular program of applying TFM to streams with larval populations has reduced sea lamprey to 10% of their previous abundance in most of the lakes. The St. Marys River is the connecting waterway between Lake Superior and Lake Huron. The upper portion of the river extends approximately 24 km from Whitefish Bay to the St. Marys Rapids at Sault Ste. Marie. This portion has a series of hydraulic structures such as the compensating gates, *Corresponding
shipping locks, and power generating facilities to control the outflow of Lake Superior. The lower portion of the river, about 76 km in length, consists of several channels, three large and numerous small islands, and lake-like areas. Total surface area of the river is about 732 km 2 and mean annual discharge for the river is 2,140 m3/s. The St. Marys River is known to harbor large numbers of sea lamprey larvae and is the only such tributary to Lake Huron that is not treated with the lampricide. A computer model was developed in this study to investigate the feasibility of and develop an effective plan for lampricide applications in the upper reach of the lower St. Marys River (Fig. 1). A computer model that simulates lampricide transport can provide critical information needed to plan lampricide treatment by predicting the cost-effectiveness of different possible treatment plans. Mathematical models provide an inexpensive and useful tool for studying the behavior of chemical transport in a river. Many mathematical models have been developed for simulating the transport and fate of toxic chemicals in rivers and lakes (e.g., Ambrose et al. 1988, McCorquodale et al. 1986, Halfon and Brueggemann 1990). Most of these models are either one-dimensional or simple boxtype models, in which the stream or lake is consid-
author. E-mail: [email protected]
694
Modeling Lampricide Transport in the St. Marys River ∂(Ch) ∂ ∂ + (uCh) + (vCh) = ∂x ∂y ∂t ∂C ∂ ∂C ∂ hEx + hEy ∂y ∂y ∂x ∂x
695
(1)
in which, C = chemical concentration; h = depth of flow; x, y, t = space and time variables; Ex, Ey = diffusion coefficients in x and y directions; and u, v = components of depth-averaged velocity in x and y directions, respectively. Equation 1 can be written in the Lagrangian form as h
FIG. 1. Model domain and dye sampling cross sections on St. Marys River.
ered as a series of interconnected well-mixed volume segments. In a recent study, a two-dimensional multilayer computer model for chemical and oil spills in rivers was developed (Shen et al. 1995). This integrated chemical/oil spill model, RSPILL, considers chemical transport, transformation, and kinetic processes in both river water and bed sediment. A stream-tube approximation was used in simulating the hydrodynamics. In this paper, the RSPILL model is modified to study the transport and spreading of the lampricide in the St. Marys River. The modified model was first calibrated with existing dye tests data, and then used to assist the planning of a comprehensive dye test experiment. The data from the second dye test were used to further validate the model. The validated model was then used to analyze effectiveness of TFM and granular Bayluscide application scenarios in the river. MODEL FORMULATION Since TFM is soluble in water, the chemical transport model considers that the lampricide is well mixed over the depth of the water. The chemical concentration is represented by the depth-averaged value. The governing equation of the chemical concentration in the water column can be written as
∂C dC ∂ ∂C ∂ = + hEy hEx ∂y dt ∂x ∂x ∂y
(2)
Since the water current affects the advection and spreading of the chemical, analysis of the transport of the lampricide in a river requires water velocity and flow depth distributions. A two-dimensional finite element model RMA-2V is used to simulate the current velocity and flow depth. RMA-2V is a finite-element solution of the Reynolds form of the Navier-Stokes equations for turbulent shallow water flow (Ariathurai et al. 1977, Thomas and McAnally 1985, King et al. 1997). The movement of the lampricide in the river as described by Eq. 2 is governed by the advection and diffusion processes. In the present model, a Lagrangian discrete-parcel method (Shen and Yapa 1988, Shen et al. 1995) is used with some modifications. In the Lagrangian discrete-parcel algorithm, the chemical in the flow is represented as an ensemble of a large number of small parcels. Each parcel has a set of time dependent spatial coordinates, and an associated mass. The movement of each parcel in the river is affected by the water current and the concentration of surrounding parcels. During each time step, all the parcels are first displaced according to the current velocity and a turbulent fluctuation component applied at their respective locations. The turbulent fluctuation component representing the diffusion terms is simulated using the random walk method (Fischer et al. 1979). If a large number of parcels are released in the river, and the path and mass of each parcel are followed and recorded as functions of time relative to a grid system, then the concentration distribution of the chemical can be computed at all the grid points. The Lagrangian discrete parcel approach requires an efficient bookkeeping procedure rather than the solution of a large matrix associated with a conventional Eulerian finite-difference or finite-element
696
Shen et al.
method. The algorithm is inherently stable and provides perfect mass conservation, although the time step should be compatible with the grid size and velocity for numerical accuracy (Shen and Yapa 1988). To simulate flow condition in rivers with complex geometry like the St. Marys River, the model is implemented with a triangular finite-element discretization of the domain. The water velocity and water level at the nodes of the triangular mesh system are generated by the two-dimensional hydrodynamics model. The advection velocity and water depth at parcel locations is interpolated from the solutions at the finite element nodes. Along land boundaries, the method of images is used to simulate the condition of zero concentration gradient normal to the boundaries (Tompson and Dougherty 1988). MODEL CALIBRATION The hydrodynamic model is calibrated by first adjusting Manning’s bed roughness coefficients for a discharge of 1,614 m3/s to match available water level data along the river. The hydraulic model is then further validated for the velocity distribution for a discharge of 3,109 m 3/s, which is the flow condition of field velocity measurement carried out by the Detroit District of the U.S. Army Corps of Engineers using ADCP, during 27–28 August 1996 (J. Koschik, Personal Communication, Great Lakes Hydraulics and Hydrology Branch, U.S. Army Corps of Engineers, Detroit, MI). The calibrated bed roughness coefficient ranges from 0.03 in the main channel to 0.06 in shallow vegetated areas (Xu et al. 1998). The chemical transport model was calibrated using the data from dye tests in the St. Marys River during 7 to 10 December 1981 (Great Lakes Fishery Commission 1981). This included a series of four dye tests. The purpose of the tests was to observe the general dispersal pattern of dye released from various sites. A sketch of the maximum concentration pattern was given for each test. The results were generally indicative of the maximum concentration that existed during the progression of the dye cloud. The distributions of concentration with respect to time are not available. The dye used was fluorescent Rhodamine WT, 20% aqueous. The dye injection sites were Edison Sault Electric Generation Station, U.S. Army Corps of Engineers (COE) Power Station, COE Compensating Works, and the Canadian Lock, respectively (Fig. 1).
A comprehensive dye test was planned after the calibration to further validate the model. This test was carried out during 10–12 August 1996. A total of 460 liters of dye was released over a period of 14 hours from a railway bridge immediately upstream of the Great Lake Power Plant (GLP) (Schleen et al. 2003). The total water discharge during this test was about 2,200 m 3 /s, including the locks. The water discharge from the GLP was maintained at 1,018 m 3 /s, which gave an average upstream boundary concentration of 9 ppb. Inflow discharge at other locations were 110.96 m 3 /s at the COE Compensating Works, 316.88 m 3 /s at the COE Power Station, and 739.84 m3/s at the Edison Sault Power Station. This dye study covered a much larger area than the 1981 dye study. Time dependent dye concentration distributions at selected crosssections were obtained. Figure 1 shows the location of these cross sections. The comparisons of observed and simulated dye concentrations are shown in Figure 2. The dye released was split into two parts as it moved downstream. A small portion of the dye moved down the main shipping channel south of Sugar Island. A large portion of the dye moved down the North Channel on the Canadian side of Sugar Island. The simulated result agrees reasonably well with the observed data. Figure 3 shows the simulated maximum concentration distributions, and Figure 4 shows the simulated durations in hours of exceeding a lethal level of 4.67 ppb. In Figure 5 the boundary of the area over which the lethal level has exceeded by 9 hours or longer, as estimated from the field dye test data, is shown. SCENARIO SIMULATIONS FOR TFM APPLICATIONS The calibrated model is used to analyze different TFM application scenarios. Twenty simulations with different combinations of flow and lampricide input conditions are made and the numbers of lamprey killed are calculated. These twenty scenarios are composed of combinations of five different flow conditions, i.e., cases 1 to 5, and four different lampricide input conditions, i.e., cases A to D. The flow conditions are shown in Table 1 and the lampricide input conditions are shown in Table 2. The numbering of simulated scenarios is the combination of flow condition and lampricide input condition. For example, the scenario with flow condition 5 and input condition D is designated as scenario 5D. The discharge for all 20 scenarios is the same, at 1,699.2 m3/s. Reducing the total discharge will
Modeling Lampricide Transport in the St. Marys River
FIG. 2.
697
Comparisons of concentration distributions at cross sections 3 to 6.
save lampricide cost to achieve the same lampricide concentration in the river, but 1,699.2 m3/s is the minimum discharge at which the power stations and other facilities can satisfy their minimum operating conditions. The five flow conditions differ in the distribution of flow at GLP, COE Compensating Works (Rapid), Corps of Engineers Power Station (COE), Edison Sault Power Plant, and other facilities. The discharge of GLP in flow condition 1 is 623 m3/s. It is increased by an increment of 56.6 m3/s to 679.7 m3/s in flow condition 2, 736.3 m3/s in flow condition 3, 793.0 m3/s in flow condition 4, and 849.6 m 3/s in flow condition 5. To keep the total discharge at 1,699.2 m 3/s, the discharge at COE Compensating Works is adjusted in each of the flow conditions. The discharge of Edison Sault Power Plant is required not to exceed 764.6 m3/s for the normal operation of the plant. It is fixed at 708 m3/s in all simulations. Among the four different lampricide input conditions, lampricide is released only on the GLP side in cases A and B. It is released both on the GLP side and the Edison Sault
side in cases C and D. The initial concentrations of these scenarios are 2.4 ppm, and the lethal level is 1.6 ppm, i.e., 2⁄3 of the initial concentration. The input duration is 16 hours for input condition A, 20 hours for input condition B, 14 hours for input condition C, and 16 hours for input condition D. The model simulates the diffusion and advection of lampricide in the river and generates the lampricide concentration distribution hourly for a 48-hour simulation period. Areas where the concentrations exceed the lethal level for a duration longer than 9 hours are considered to be “Lethal Areas.” The number of lamprey larvae killed by lampricide is calculated for each scenario by superimposing the simulated lethal area plots on the larval density distribution map shown in Figure 6 (Fodale et al. 2003). Table 3 shows the summary of the simulation results. Scenario 4C is the most effective scenario in terms of the number of kills per dollar. Scenario 5D achieves the maximum total kill, but the kill-perdollar value of scenario 5D is lower than scenario
698
Shen et al.
FIG. 3.
Simulated maximum concentration in ppb.
FIG. 4. Simulated duration (hrs) of dye concentration exceeds the lethal level of 4.67 ppb, for the 1996 dye test.
Modeling Lampricide Transport in the St. Marys River
FIG. 5.
Lethal area estimated from the 1996 dye test data.
4C. Table 3 also shows the two-site release scenarios, C and D, are more effective than single-site release scenarios, A and B, in term of kills per dollar. However, the total costs of two-site release scenarios are much higher than single-site release scenarios although the total kills are higher. In the single TABLE 1.
input site scenarios, the lower discharge of GLP gives the better kill-per-dollar, as shown in Figure 7. The reason is that more discharge from GLP means more lampricide input for the same initial concentration, and more diluted lampricide on the south side of the channel. It therefore reduces the
Flow conditions from different sites for scenario simulations.
Flow Condition→ Sites ↓ 1 GLP 623.0 Edison 708.0 Rapid 99.1 COE 254.9 Lock 14.2 Total 1,699.2
TABLE 2.
699
2 679.7 708.0 99.1 198.2 14.2 1,699.2
Discharge, m3/s 3 4 736.6 793.0 708.0 708.0 42.5 42.5 198.2 141.6 14.2 14.2 1,699.2 1,699.2
5 849.6 708.0 42.5 85.0 14.2 1,699.2
Lampricide input conditions for scenario simulations.
Input Condition→ Sites ↓ A GLP 2.4 ppm/16 hour Edison —
Concentration/Duration B C 2.4 ppm/20 hour 2.4 ppm/14 hour — 2.4 ppm/14 hour
D 2.4 ppm/16 hour 2.4 ppm/16 hour
700
Shen et al.
FIG. 6.
Observed sea lamprey larvae density distribution in larvae per square meter.
effectiveness on the south side of the channel, although the total kill increases. However, in the twoinput-site scenarios as shown in Figure 8, the kill-per-dollar increases as the discharge of GLP increases from flow condition 1 to condition 4, then decreases as flow discharge of GLP further increases to flow condition 5. The improvement of kill effectiveness from flow condition 1 to 4 is due to the reduction of the gap between lampricide clouds. When the input of GLP increases from flow condition 4 to 5, the two clouds overlap each other and decrease the kill-per-dollar. In the single input site scenarios, increasing the input will increase the total kill, but it will also decrease the kill-per-dollar. Table 3 shows that the effectiveness of short input duration scenarios is better than that of long input duration scenarios for the same flow conditions. To further understand the relationship between input duration and the kill effectiveness, 12 different input durations ranging from 9 hours to 20 hours are tested using flow condition 1. The simulation results, as shown in Figure 9, demonstrate that increasing input duration from 8 hours to 14 hours will improve kill-effectiveness. Further increasing input duration will increase total kill, but the kill-
per-dollar will decrease slightly. The best input duration lies in the range of 14 hours to 16 hours. SCENARIO ANALYSIS FOR GRANULAR BAYLUSCIDE APPLICATION Granular Bayluscide has been found to be an effective lampricide, which can be used instead of TFM for sea lamprey control (Fodale et al. 2003). Granular Bayluscide has the advantage over TFM, since it can be applied to localized targeted areas. However, the velocity at the bottom of the river channel should not exceed a certain level for granular Bayluscide application to be effective. The general guideline obtained from a preliminary field test is that the flow velocity within a vertical distance of 2.5 cm from the channel bottom should be lower than 8 cm/s (R. Bergstedt, U.S. Geological Survey, Millersburg, MI, personal communication). The effectiveness of granular Bayluscide application on the upper St. Marys River is analyzed using the hydrodynamic model. The vertical velocity profile for turbulent channel flows can be described by a logarithmic function as (Fischer et al. 1979):
Modeling Lampricide Transport in the St. Marys River TABLE 3.
701
Comparison of the effectiveness of lampricide applications for different scenarios.
No. 1A 2A 3A 4A 5A
TFM Input (m3) GLP Edison 86.57 93.89 101.72 0 109.54 117.37
1B 2B 3B 4B 5B
108.21 117.37 127.15 136.93 146.71
1C 2C 3C 4C 5C
75.75 82.16 89.00 95.85 102.70
1D 2D 3D 4D 5D
86.57 93.89 101.72 109.54 117.37
0
85.58
97.80
V ( y) = V +
Input Duration (hrs)
TFM Input (103 lb) 195 211 229 246 264
Total Kill (103) 1,270.7 1,317.0 1,371.0 1,426.6 1,420.7
Kill/lb (1/lb) 6.64 6.35 6.10 5.89 5.48
Kill/Vol (103/m3) 14.68 14.03 13.48 13.02 12.10
TFM Cost (million $) 5.84 6.33 6.86 7.39 7.92
Kill/$ (1/$) 0.218 0.208 0.200 0.193 0.179
20
243 264 286 308 330
1,334.0 1,420.7 1,449.0 1,459.5 1,467.3
5.58 5.48 5.16 4.82 4.53
12.33 12.10 11.40 11.15 10.00
7.30 7.92 8.57 9.23 9.89
0.183 0.179 0.169 0.158 0.148
14
363 377 392 408 423
2,329.0 2,563.4 2,827.0 2,991.8 3,064.8
6.53 6.92 7.33 7.46 7.37
14.44 15.28 16.19 16.49 16.28
10.9 11.3 11.8 12.2 12.7
0.214 0.227 0.240 0.245 0.241
16
414 431 449 466 483
2,618.2 2,776.4 2,914.3 3,122.4 3,177.9
6.43 6.55 6.61 6.81 6.68
14.20 14.48 14.61 15.06 14.77
12.4 12.9 13.5 14.0 14.5
0.211 0.215 0.216 0.223 0.219
16
u* y ln + 1 κ h
(3)
where V(y) is the— velocity at vertical distance y from the bottom, V is depth-averaged velocity, u* is shear velocity, κ is von-Karman’s constant, and h is the flow depth. The depth-averaged velocity is calculated in the two-dimensional hydrodynamic —— model. The shear velocity is defined as u* = √ghS , where g is gravity, and S is channel slope. It is found that the boundary between the turbulent flow region and the transient region in the St. Marys River is in the range of 2.5 cm above the bottom except for in some dead zones, where the flow is nearly stagnant. In these dead zones, the mean velocity is much lower than 8 cm/s; therefore the velocity within the 2.5 cm bottom layer is always less than 8 cm/s. To simplify the analysis, the logarithmic profile is assumed for the whole river channel. Bottom velocity distributions are calculated for three flow conditions: flow conditions 1, 2, and 3, as described in Table 1. Figure 10 shows the velocity at 2.5 cm from the river channel bottom for flow condition 1, as an example. It is shown in the figure
that most of the area is covered by a bottom flow velocity lower than 8 cm/s, except for the upstream part and in the main shipping channel. The area where velocity is less than 8 cm/s increases when the bottom layer thickness decreases from 2.5 cm to 1.25 cm. It decreases when the bottom layer thickness increases to 5 cm. The result is sensitive to the criterion of selecting bottom layer thickness. Simulation results for different flow conditions show that the area covered by 8 cm/s or higher velocities will increase as the total discharge increases. However, this increase in the area of higher bottom velocity in the south channel is confined to the main shipping channel. Two additional high discharge flow conditions are simulated to show the effect of total discharge on the velocity distribution at the bottom of the St. Marys River. One flow condition has a total discharge of 2,183.5 m3/s. The allocation of total discharge among U.S. Locks, Compensating Works, COE Power Station, Edison Sault Power Station, and Great Lakes Power Plant was 17.0 m3/s, 99.1 m3/s, 311.5 m3/s, 736.3 m3/s, and 1,019.5 m3/s, respectively. Another flow condition has an even higher total discharge of 3,115 m3/s, of which 17.0
702
Shen et al.
FIG. 7. Kill per dollar and total kill of single site input scenarios A and B.
FIG. 8. Kill per dollar and total kill of two-site input scenarios C and D.
FIG. 9. Effectiveness of different input duration corresponding to flow condition 1, in kill per dollar.
m 3/s comes from the U.S. Locks, 1,030.85 m 3/s from the Compensating Works, 311.5 m3/s from the COE Power Station, 736.3 m3/s from the Edison Sault Power Station, and 1,019.5 m 3 /s from the Great Lakes Power Plant. Table 4 shows the effect on the lamprey population if granular Bayluscide is applied to areas where the critical condition that is required for it to be effective is satisfied, i.e., in the area where the bottom layer velocity is 8 cm/s or lower. For flow condition 1, if the bottom layer thickness for the critical velocity is 2.5 cm, the percentage of sea lamprey population killed will be 88.2%. If the bottom layer thickness is changed to 1.25 cm, the lamprey population covered will increase to 93.5%. If the bottom layer thickness is changed to 5 cm, the lamprey population covered by velocities of 8 cm/s or lower will decrease to 73.5%. The differences between flow conditions 1, 3, and 5 are small. The lamprey populations covered by velocities of 8 cm/s or lower are around 88.5%. Higher river flow will reduce the area covered. In the dye test case, the coverage is 79.5%. However, if the total flow
Modeling Lampricide Transport in the St. Marys River
FIG. 10.
703
Velocity at 2.5 cm from channel bottom for flow condition 1.
rate is 3,115 m3/s , the coverage is only 70.5%. The lower part of Table 4 shows the percentage of coverage that will result from applying granular Bayluscide to areas in the North Channel only. SUMMARY This paper describes the development and application of a lampricide transport model for rivers. The model is applied to the St. Marys River to evaluate the effectiveness and feasibility of alternative lampricide application strategies. The hydrodynamics necessary for the model is computed based on a depth-averaged two-dimensional finite element model. The hydrodynamic model is calibrated and verified using the ADCP based field velocity data. The model for transport and mixing of lampricide is developed based on a depth-averaged two-dimensional advection-diffusion equation, using a Lagrangian discrete parcel tracking method. To calibrate and validate the model, a number of
dye test cases were simulated numerically. Comparison of numerical results with the observed data from dye tests show that the model can be used adequately to simulate scenarios. The model is used to evaluate alternative lampricide application strategies by simulating various TFM application plans. The effectiveness of granular Bayluscide applications is also analyzed using the hydrodynamic model and a plausible critical bottom velocity criterion. In TFM application simulations, twenty scenarios were considered subjected to the constraints of the flow operating conditions. These scenarios consisted of combinations of five flow conditions and four TFM application conditions. The scenario simulation results showed that two-site releases are more effective than single-site release, in terms of kill-per-dollar. However, the total cost for two-site releases are higher than single-site releases. The 16hour release achieves the maximum total kill, but the kill-per-dollar is best at 14-hour release. Further
704
Shen et al.
TABLE 4. Comparsion of lamprey population before and after application of granular Bayer under different flow conditions.
Flow Flow rate condition m3/s Apply to The Entire River 1 1,699.2 1 1,699.2 1 1,699.2 3 1,699.2 5 1,699.2 Dye Study 2,183.5 High 3,115.2 Apply to North Channel Only 1 1,699.2 1 1,699.2 1 1,699.2 3 1,699.2 5 1,699.2 Dye Study 2,183.5 High 3,115.2
Total lamprey population 1,000
Bottom distance cm
Critical velocity cm/s
1.25 2.5 5 2.5 2.5 2.5 2.5
8 8 8 8 8 8 8
4,768.7 4,768.7 4,768.7 4,768.7 4,768.7 4,768.7 4,768.7
1.25 2.5 5 2.5 2.5 2.5 2.5
8 8 8 8 8 8 8
2,326 2,326 2,326 2,326 2,326 2,326 2,326
simulations showed that the optimum duration of TFM release is between 14 and 16 hours. Simulations of bottom velocity distribution showed that the area over which the granular Bayluscide application can be effective covers 70 to 90% of the lamprey population in the St. Marys River, depending on the total river discharge. ACKNOWLEDGMENTS This study was funded by the Great Lakes Fishery Commission. The authors would like to thank members of the St. Marys River Control Task Force for their input and valuable suggestions during the course of this study. REFERENCES Ambrose, R.B., Woll, T.A., Connolly, J.P., and Schang, R.W. 1988. WASP4, A Hydrodynamic and Water Quality Model—Model Theory, Users Manual, and Programmer’s Guide. EPA/600/3-87/039, USEPA, Athens, GA. Ariathurai, R., MacArthur, R.C., and Krone, R.B. 1977. Mathematical Model of Estuarial Sediment Transport. Technical Report D-77-12, U.S. Army Engineer Waterways Experiment Station, Vicksburg, MS. Fischer, H., List, E., Koh, R., Imberger, J., and Brooks, N. 1979. Mixing in Inland and Coastal Waters. New York: Academic Press.
Lamprey population survived 1,000
Lamprey population killed 1,000
% of lamprey killed %
311.8 562.1 1,263.3 552.6 525.5 977.1 1,406.4
4,456.9 4,206.6 3,505.4 4,216.1 4,243.2 3,791.7 3,362.3
93.46% 88.21% 73.51% 88.41% 88.98% 79.51% 70.51%
178.6 391.8 615 382.3 355.2 535.1 707.5
2,147.3 1,934.2 1,710.9 1,943.7 1,970.7 1,790.8 1,618.4
92.32% 83.16% 73.56% 83.56% 84.72% 76.99% 69.58%
Fodale, F.M., Bergstedt, R.A., Cuddy, D.W., Adams, J.V., and Stolyarenko, D.M. 2003. Planning and executing a lampricide treatment of the St. Marys River using georeferenced data. J. Great Lakes Res. 29 (Suppl. 1):706–716. Great Lakes Fishery Commission. 1981. Report of a study of current patterns in the Sault Ste. Marie harbor area of St. Marys River using a tracer dye. Unpublished Report, Ann Arbor, MI. Halfon, E., and Brueggemann, R. 1990. Simulation of Disulfoton Fate in the River with the Toxfate Model. The Science of Total Environment 97/98: 385–394. King, I., Donnell, B., Finnie, J., Letter, J., McAnally, W., Roig, L., and Thomas, W. 1997. Users Guide To RMA2 WES Version 4.3. U.S. Army Corps of Engineers—Waterways Experiment Station, Hydraulics Laboratory, Vicksburg, MS. McCorquodale, J.A., Ibrahim, K., and Hamdy, Y. 1986. Fate and transport modelling of Perchloroethylene in the St. Clair River. Water Poll. Res. J. Canada 21(3): 398–410. Schleen, L.P., Christie, G.C., Heinrich, J.W., Bergstedt, R.A., Young, R.J., Morse, T.J., Lavis, D.S., Bills, T.D., Johnson, J.E., and Ebener, M.P. 2003. Development and implementation of an integrated program for control of sea lampreys in the St. Marys River. J. Great Lakes Res. 29 (Suppl. 1):677–693. Shen, H.T., and Yapa, P.D. 1988. Oil Spill Transport in Rivers. J. Hydraulic Engineering ASCE, 114(5): 529–543. ———, Yapa, P.D., and Zhang, B. 1995. A simulation
Modeling Lampricide Transport in the St. Marys River model for chemical spills in the upper St. Lawrence River. J. Great Lakes Res. 21:652–664. Thomas, W., and McAnally, W. 1985. Open-Channel Flow and Sedimentation TABS-2. Instruction Report HL-85-1, Waterways Experiment Station, U.S. Army Corps of Engineers, Vicksburg, MS. Tompson, A.F.B., and Dougherty, D.E. 1988. On the use of particle tracking methods for solute transport in porous media, In Computational Methods in Water Resources, 2, Numerical Methods for Transport and
705
Hydrologic Processes, ed. M. Celia et al. New York: Elsevier. Xu, Q., Shen, H.T., and Yapa, P.D. 1998. A Model for Lampricide Transport in the St. Marys River. Report 98-8, Department of Civil and Enviorn. Engineering, Clarkson University, Potsdam, NY. Submitted: 21 December 2000 Accepted: 20 March 2002 Editorial handling: John W. Heinrich
Modeling Lampricide Transport in the St. Marys River Hung Tao Shen*, Qing Xu, and Poojitha D. Yapa Department of Civil and Environmental Engineering Clarkson University Potsdam, New York 13699-5710 ABSTRACT. This paper describes use of an integrated chemical/oil spill computer model to study lampricide transport in the St. Marys River. The model is developed based on a depth-averaged two-dimensional advection-diffusion equation using a Lagrangian discrete parcel method. The hydrodynamics in the model is computed using a depth-averaged two-dimensional finite element model. The model is used to analyze the effectiveness of alternative lampricide application strategies. This analysis showed that releasing TFM from both the Great Lakes Power Plant and the Edison Sault Power Station is more effective than releasing from the Great Lakes Power Plant only, in terms of kill-per-dollar. However, the total cost for two-site releases are higher than single-site releases. The simulations also showed that the optimum duration of TFM release is between 14 and 16 hours. Simulations of bottom velocity distribution showed that the area over which the granular Bayluscide application could cover 70 to 90% of the sea lamprey (Petromyzon marinus) population in the St. Marys River, depending on the total river discharge. INDEX WORDS:
Hydrodynamic model, Bayluscide, lampricide, St. Marys River, TFM.
INTRODUCTION The sea lamprey (Petromyzon marinus) was a major cause of the severe damage done to the populations of lake trout (Salvelinus namaycush), whitefish (Coregonus clupeaformis), and chubs (Coregonus spp.), all commercially valuable species in the Great Lakes. The Great Lakes Fisheries Commission initiated a successful program in 1958 to use the lampricide TFM (3-trifluoromethyl4-nitrophenol) to reduce sea lamprey populations. TFM selectively kills sea lamprey larvae when applied in carefully regulated dosages. Field studies conducted by the Great Lakes Fishery Commission showed that a certain critical concentration of lampricide needs to be maintained for a specific minimum duration to provide a lethal dose. A regular program of applying TFM to streams with larval populations has reduced sea lamprey to 10% of their previous abundance in most of the lakes. The St. Marys River is the connecting waterway between Lake Superior and Lake Huron. The upper portion of the river extends approximately 24 km from Whitefish Bay to the St. Marys Rapids at Sault Ste. Marie. This portion has a series of hydraulic structures such as the compensating gates, *Corresponding
shipping locks, and power generating facilities to control the outflow of Lake Superior. The lower portion of the river, about 76 km in length, consists of several channels, three large and numerous small islands, and lake-like areas. Total surface area of the river is about 732 km 2 and mean annual discharge for the river is 2,140 m3/s. The St. Marys River is known to harbor large numbers of sea lamprey larvae and is the only such tributary to Lake Huron that is not treated with the lampricide. A computer model was developed in this study to investigate the feasibility of and develop an effective plan for lampricide applications in the upper reach of the lower St. Marys River (Fig. 1). A computer model that simulates lampricide transport can provide critical information needed to plan lampricide treatment by predicting the cost-effectiveness of different possible treatment plans. Mathematical models provide an inexpensive and useful tool for studying the behavior of chemical transport in a river. Many mathematical models have been developed for simulating the transport and fate of toxic chemicals in rivers and lakes (e.g., Ambrose et al. 1988, McCorquodale et al. 1986, Halfon and Brueggemann 1990). Most of these models are either one-dimensional or simple boxtype models, in which the stream or lake is consid-
author. E-mail: [email protected]
694
Modeling Lampricide Transport in the St. Marys River ∂(Ch) ∂ ∂ + (uCh) + (vCh) = ∂x ∂y ∂t ∂C ∂ ∂C ∂ hEx + hEy ∂y ∂y ∂x ∂x
695
(1)
in which, C = chemical concentration; h = depth of flow; x, y, t = space and time variables; Ex, Ey = diffusion coefficients in x and y directions; and u, v = components of depth-averaged velocity in x and y directions, respectively. Equation 1 can be written in the Lagrangian form as h
FIG. 1. Model domain and dye sampling cross sections on St. Marys River.
ered as a series of interconnected well-mixed volume segments. In a recent study, a two-dimensional multilayer computer model for chemical and oil spills in rivers was developed (Shen et al. 1995). This integrated chemical/oil spill model, RSPILL, considers chemical transport, transformation, and kinetic processes in both river water and bed sediment. A stream-tube approximation was used in simulating the hydrodynamics. In this paper, the RSPILL model is modified to study the transport and spreading of the lampricide in the St. Marys River. The modified model was first calibrated with existing dye tests data, and then used to assist the planning of a comprehensive dye test experiment. The data from the second dye test were used to further validate the model. The validated model was then used to analyze effectiveness of TFM and granular Bayluscide application scenarios in the river. MODEL FORMULATION Since TFM is soluble in water, the chemical transport model considers that the lampricide is well mixed over the depth of the water. The chemical concentration is represented by the depth-averaged value. The governing equation of the chemical concentration in the water column can be written as
∂C dC ∂ ∂C ∂ = + hEy hEx ∂y dt ∂x ∂x ∂y
(2)
Since the water current affects the advection and spreading of the chemical, analysis of the transport of the lampricide in a river requires water velocity and flow depth distributions. A two-dimensional finite element model RMA-2V is used to simulate the current velocity and flow depth. RMA-2V is a finite-element solution of the Reynolds form of the Navier-Stokes equations for turbulent shallow water flow (Ariathurai et al. 1977, Thomas and McAnally 1985, King et al. 1997). The movement of the lampricide in the river as described by Eq. 2 is governed by the advection and diffusion processes. In the present model, a Lagrangian discrete-parcel method (Shen and Yapa 1988, Shen et al. 1995) is used with some modifications. In the Lagrangian discrete-parcel algorithm, the chemical in the flow is represented as an ensemble of a large number of small parcels. Each parcel has a set of time dependent spatial coordinates, and an associated mass. The movement of each parcel in the river is affected by the water current and the concentration of surrounding parcels. During each time step, all the parcels are first displaced according to the current velocity and a turbulent fluctuation component applied at their respective locations. The turbulent fluctuation component representing the diffusion terms is simulated using the random walk method (Fischer et al. 1979). If a large number of parcels are released in the river, and the path and mass of each parcel are followed and recorded as functions of time relative to a grid system, then the concentration distribution of the chemical can be computed at all the grid points. The Lagrangian discrete parcel approach requires an efficient bookkeeping procedure rather than the solution of a large matrix associated with a conventional Eulerian finite-difference or finite-element
696
Shen et al.
method. The algorithm is inherently stable and provides perfect mass conservation, although the time step should be compatible with the grid size and velocity for numerical accuracy (Shen and Yapa 1988). To simulate flow condition in rivers with complex geometry like the St. Marys River, the model is implemented with a triangular finite-element discretization of the domain. The water velocity and water level at the nodes of the triangular mesh system are generated by the two-dimensional hydrodynamics model. The advection velocity and water depth at parcel locations is interpolated from the solutions at the finite element nodes. Along land boundaries, the method of images is used to simulate the condition of zero concentration gradient normal to the boundaries (Tompson and Dougherty 1988). MODEL CALIBRATION The hydrodynamic model is calibrated by first adjusting Manning’s bed roughness coefficients for a discharge of 1,614 m3/s to match available water level data along the river. The hydraulic model is then further validated for the velocity distribution for a discharge of 3,109 m 3/s, which is the flow condition of field velocity measurement carried out by the Detroit District of the U.S. Army Corps of Engineers using ADCP, during 27–28 August 1996 (J. Koschik, Personal Communication, Great Lakes Hydraulics and Hydrology Branch, U.S. Army Corps of Engineers, Detroit, MI). The calibrated bed roughness coefficient ranges from 0.03 in the main channel to 0.06 in shallow vegetated areas (Xu et al. 1998). The chemical transport model was calibrated using the data from dye tests in the St. Marys River during 7 to 10 December 1981 (Great Lakes Fishery Commission 1981). This included a series of four dye tests. The purpose of the tests was to observe the general dispersal pattern of dye released from various sites. A sketch of the maximum concentration pattern was given for each test. The results were generally indicative of the maximum concentration that existed during the progression of the dye cloud. The distributions of concentration with respect to time are not available. The dye used was fluorescent Rhodamine WT, 20% aqueous. The dye injection sites were Edison Sault Electric Generation Station, U.S. Army Corps of Engineers (COE) Power Station, COE Compensating Works, and the Canadian Lock, respectively (Fig. 1).
A comprehensive dye test was planned after the calibration to further validate the model. This test was carried out during 10–12 August 1996. A total of 460 liters of dye was released over a period of 14 hours from a railway bridge immediately upstream of the Great Lake Power Plant (GLP) (Schleen et al. 2003). The total water discharge during this test was about 2,200 m 3 /s, including the locks. The water discharge from the GLP was maintained at 1,018 m 3 /s, which gave an average upstream boundary concentration of 9 ppb. Inflow discharge at other locations were 110.96 m 3 /s at the COE Compensating Works, 316.88 m 3 /s at the COE Power Station, and 739.84 m3/s at the Edison Sault Power Station. This dye study covered a much larger area than the 1981 dye study. Time dependent dye concentration distributions at selected crosssections were obtained. Figure 1 shows the location of these cross sections. The comparisons of observed and simulated dye concentrations are shown in Figure 2. The dye released was split into two parts as it moved downstream. A small portion of the dye moved down the main shipping channel south of Sugar Island. A large portion of the dye moved down the North Channel on the Canadian side of Sugar Island. The simulated result agrees reasonably well with the observed data. Figure 3 shows the simulated maximum concentration distributions, and Figure 4 shows the simulated durations in hours of exceeding a lethal level of 4.67 ppb. In Figure 5 the boundary of the area over which the lethal level has exceeded by 9 hours or longer, as estimated from the field dye test data, is shown. SCENARIO SIMULATIONS FOR TFM APPLICATIONS The calibrated model is used to analyze different TFM application scenarios. Twenty simulations with different combinations of flow and lampricide input conditions are made and the numbers of lamprey killed are calculated. These twenty scenarios are composed of combinations of five different flow conditions, i.e., cases 1 to 5, and four different lampricide input conditions, i.e., cases A to D. The flow conditions are shown in Table 1 and the lampricide input conditions are shown in Table 2. The numbering of simulated scenarios is the combination of flow condition and lampricide input condition. For example, the scenario with flow condition 5 and input condition D is designated as scenario 5D. The discharge for all 20 scenarios is the same, at 1,699.2 m3/s. Reducing the total discharge will
Modeling Lampricide Transport in the St. Marys River
FIG. 2.
697
Comparisons of concentration distributions at cross sections 3 to 6.
save lampricide cost to achieve the same lampricide concentration in the river, but 1,699.2 m3/s is the minimum discharge at which the power stations and other facilities can satisfy their minimum operating conditions. The five flow conditions differ in the distribution of flow at GLP, COE Compensating Works (Rapid), Corps of Engineers Power Station (COE), Edison Sault Power Plant, and other facilities. The discharge of GLP in flow condition 1 is 623 m3/s. It is increased by an increment of 56.6 m3/s to 679.7 m3/s in flow condition 2, 736.3 m3/s in flow condition 3, 793.0 m3/s in flow condition 4, and 849.6 m 3/s in flow condition 5. To keep the total discharge at 1,699.2 m 3/s, the discharge at COE Compensating Works is adjusted in each of the flow conditions. The discharge of Edison Sault Power Plant is required not to exceed 764.6 m3/s for the normal operation of the plant. It is fixed at 708 m3/s in all simulations. Among the four different lampricide input conditions, lampricide is released only on the GLP side in cases A and B. It is released both on the GLP side and the Edison Sault
side in cases C and D. The initial concentrations of these scenarios are 2.4 ppm, and the lethal level is 1.6 ppm, i.e., 2⁄3 of the initial concentration. The input duration is 16 hours for input condition A, 20 hours for input condition B, 14 hours for input condition C, and 16 hours for input condition D. The model simulates the diffusion and advection of lampricide in the river and generates the lampricide concentration distribution hourly for a 48-hour simulation period. Areas where the concentrations exceed the lethal level for a duration longer than 9 hours are considered to be “Lethal Areas.” The number of lamprey larvae killed by lampricide is calculated for each scenario by superimposing the simulated lethal area plots on the larval density distribution map shown in Figure 6 (Fodale et al. 2003). Table 3 shows the summary of the simulation results. Scenario 4C is the most effective scenario in terms of the number of kills per dollar. Scenario 5D achieves the maximum total kill, but the kill-perdollar value of scenario 5D is lower than scenario
698
Shen et al.
FIG. 3.
Simulated maximum concentration in ppb.
FIG. 4. Simulated duration (hrs) of dye concentration exceeds the lethal level of 4.67 ppb, for the 1996 dye test.
Modeling Lampricide Transport in the St. Marys River
FIG. 5.
Lethal area estimated from the 1996 dye test data.
4C. Table 3 also shows the two-site release scenarios, C and D, are more effective than single-site release scenarios, A and B, in term of kills per dollar. However, the total costs of two-site release scenarios are much higher than single-site release scenarios although the total kills are higher. In the single TABLE 1.
input site scenarios, the lower discharge of GLP gives the better kill-per-dollar, as shown in Figure 7. The reason is that more discharge from GLP means more lampricide input for the same initial concentration, and more diluted lampricide on the south side of the channel. It therefore reduces the
Flow conditions from different sites for scenario simulations.
Flow Condition→ Sites ↓ 1 GLP 623.0 Edison 708.0 Rapid 99.1 COE 254.9 Lock 14.2 Total 1,699.2
TABLE 2.
699
2 679.7 708.0 99.1 198.2 14.2 1,699.2
Discharge, m3/s 3 4 736.6 793.0 708.0 708.0 42.5 42.5 198.2 141.6 14.2 14.2 1,699.2 1,699.2
5 849.6 708.0 42.5 85.0 14.2 1,699.2
Lampricide input conditions for scenario simulations.
Input Condition→ Sites ↓ A GLP 2.4 ppm/16 hour Edison —
Concentration/Duration B C 2.4 ppm/20 hour 2.4 ppm/14 hour — 2.4 ppm/14 hour
D 2.4 ppm/16 hour 2.4 ppm/16 hour
700
Shen et al.
FIG. 6.
Observed sea lamprey larvae density distribution in larvae per square meter.
effectiveness on the south side of the channel, although the total kill increases. However, in the twoinput-site scenarios as shown in Figure 8, the kill-per-dollar increases as the discharge of GLP increases from flow condition 1 to condition 4, then decreases as flow discharge of GLP further increases to flow condition 5. The improvement of kill effectiveness from flow condition 1 to 4 is due to the reduction of the gap between lampricide clouds. When the input of GLP increases from flow condition 4 to 5, the two clouds overlap each other and decrease the kill-per-dollar. In the single input site scenarios, increasing the input will increase the total kill, but it will also decrease the kill-per-dollar. Table 3 shows that the effectiveness of short input duration scenarios is better than that of long input duration scenarios for the same flow conditions. To further understand the relationship between input duration and the kill effectiveness, 12 different input durations ranging from 9 hours to 20 hours are tested using flow condition 1. The simulation results, as shown in Figure 9, demonstrate that increasing input duration from 8 hours to 14 hours will improve kill-effectiveness. Further increasing input duration will increase total kill, but the kill-
per-dollar will decrease slightly. The best input duration lies in the range of 14 hours to 16 hours. SCENARIO ANALYSIS FOR GRANULAR BAYLUSCIDE APPLICATION Granular Bayluscide has been found to be an effective lampricide, which can be used instead of TFM for sea lamprey control (Fodale et al. 2003). Granular Bayluscide has the advantage over TFM, since it can be applied to localized targeted areas. However, the velocity at the bottom of the river channel should not exceed a certain level for granular Bayluscide application to be effective. The general guideline obtained from a preliminary field test is that the flow velocity within a vertical distance of 2.5 cm from the channel bottom should be lower than 8 cm/s (R. Bergstedt, U.S. Geological Survey, Millersburg, MI, personal communication). The effectiveness of granular Bayluscide application on the upper St. Marys River is analyzed using the hydrodynamic model. The vertical velocity profile for turbulent channel flows can be described by a logarithmic function as (Fischer et al. 1979):
Modeling Lampricide Transport in the St. Marys River TABLE 3.
701
Comparison of the effectiveness of lampricide applications for different scenarios.
No. 1A 2A 3A 4A 5A
TFM Input (m3) GLP Edison 86.57 93.89 101.72 0 109.54 117.37
1B 2B 3B 4B 5B
108.21 117.37 127.15 136.93 146.71
1C 2C 3C 4C 5C
75.75 82.16 89.00 95.85 102.70
1D 2D 3D 4D 5D
86.57 93.89 101.72 109.54 117.37
0
85.58
97.80
V ( y) = V +
Input Duration (hrs)
TFM Input (103 lb) 195 211 229 246 264
Total Kill (103) 1,270.7 1,317.0 1,371.0 1,426.6 1,420.7
Kill/lb (1/lb) 6.64 6.35 6.10 5.89 5.48
Kill/Vol (103/m3) 14.68 14.03 13.48 13.02 12.10
TFM Cost (million $) 5.84 6.33 6.86 7.39 7.92
Kill/$ (1/$) 0.218 0.208 0.200 0.193 0.179
20
243 264 286 308 330
1,334.0 1,420.7 1,449.0 1,459.5 1,467.3
5.58 5.48 5.16 4.82 4.53
12.33 12.10 11.40 11.15 10.00
7.30 7.92 8.57 9.23 9.89
0.183 0.179 0.169 0.158 0.148
14
363 377 392 408 423
2,329.0 2,563.4 2,827.0 2,991.8 3,064.8
6.53 6.92 7.33 7.46 7.37
14.44 15.28 16.19 16.49 16.28
10.9 11.3 11.8 12.2 12.7
0.214 0.227 0.240 0.245 0.241
16
414 431 449 466 483
2,618.2 2,776.4 2,914.3 3,122.4 3,177.9
6.43 6.55 6.61 6.81 6.68
14.20 14.48 14.61 15.06 14.77
12.4 12.9 13.5 14.0 14.5
0.211 0.215 0.216 0.223 0.219
16
u* y ln + 1 κ h
(3)
where V(y) is the— velocity at vertical distance y from the bottom, V is depth-averaged velocity, u* is shear velocity, κ is von-Karman’s constant, and h is the flow depth. The depth-averaged velocity is calculated in the two-dimensional hydrodynamic —— model. The shear velocity is defined as u* = √ghS , where g is gravity, and S is channel slope. It is found that the boundary between the turbulent flow region and the transient region in the St. Marys River is in the range of 2.5 cm above the bottom except for in some dead zones, where the flow is nearly stagnant. In these dead zones, the mean velocity is much lower than 8 cm/s; therefore the velocity within the 2.5 cm bottom layer is always less than 8 cm/s. To simplify the analysis, the logarithmic profile is assumed for the whole river channel. Bottom velocity distributions are calculated for three flow conditions: flow conditions 1, 2, and 3, as described in Table 1. Figure 10 shows the velocity at 2.5 cm from the river channel bottom for flow condition 1, as an example. It is shown in the figure
that most of the area is covered by a bottom flow velocity lower than 8 cm/s, except for the upstream part and in the main shipping channel. The area where velocity is less than 8 cm/s increases when the bottom layer thickness decreases from 2.5 cm to 1.25 cm. It decreases when the bottom layer thickness increases to 5 cm. The result is sensitive to the criterion of selecting bottom layer thickness. Simulation results for different flow conditions show that the area covered by 8 cm/s or higher velocities will increase as the total discharge increases. However, this increase in the area of higher bottom velocity in the south channel is confined to the main shipping channel. Two additional high discharge flow conditions are simulated to show the effect of total discharge on the velocity distribution at the bottom of the St. Marys River. One flow condition has a total discharge of 2,183.5 m3/s. The allocation of total discharge among U.S. Locks, Compensating Works, COE Power Station, Edison Sault Power Station, and Great Lakes Power Plant was 17.0 m3/s, 99.1 m3/s, 311.5 m3/s, 736.3 m3/s, and 1,019.5 m3/s, respectively. Another flow condition has an even higher total discharge of 3,115 m3/s, of which 17.0
702
Shen et al.
FIG. 7. Kill per dollar and total kill of single site input scenarios A and B.
FIG. 8. Kill per dollar and total kill of two-site input scenarios C and D.
FIG. 9. Effectiveness of different input duration corresponding to flow condition 1, in kill per dollar.
m 3/s comes from the U.S. Locks, 1,030.85 m 3/s from the Compensating Works, 311.5 m3/s from the COE Power Station, 736.3 m3/s from the Edison Sault Power Station, and 1,019.5 m 3 /s from the Great Lakes Power Plant. Table 4 shows the effect on the lamprey population if granular Bayluscide is applied to areas where the critical condition that is required for it to be effective is satisfied, i.e., in the area where the bottom layer velocity is 8 cm/s or lower. For flow condition 1, if the bottom layer thickness for the critical velocity is 2.5 cm, the percentage of sea lamprey population killed will be 88.2%. If the bottom layer thickness is changed to 1.25 cm, the lamprey population covered will increase to 93.5%. If the bottom layer thickness is changed to 5 cm, the lamprey population covered by velocities of 8 cm/s or lower will decrease to 73.5%. The differences between flow conditions 1, 3, and 5 are small. The lamprey populations covered by velocities of 8 cm/s or lower are around 88.5%. Higher river flow will reduce the area covered. In the dye test case, the coverage is 79.5%. However, if the total flow
Modeling Lampricide Transport in the St. Marys River
FIG. 10.
703
Velocity at 2.5 cm from channel bottom for flow condition 1.
rate is 3,115 m3/s , the coverage is only 70.5%. The lower part of Table 4 shows the percentage of coverage that will result from applying granular Bayluscide to areas in the North Channel only. SUMMARY This paper describes the development and application of a lampricide transport model for rivers. The model is applied to the St. Marys River to evaluate the effectiveness and feasibility of alternative lampricide application strategies. The hydrodynamics necessary for the model is computed based on a depth-averaged two-dimensional finite element model. The hydrodynamic model is calibrated and verified using the ADCP based field velocity data. The model for transport and mixing of lampricide is developed based on a depth-averaged two-dimensional advection-diffusion equation, using a Lagrangian discrete parcel tracking method. To calibrate and validate the model, a number of
dye test cases were simulated numerically. Comparison of numerical results with the observed data from dye tests show that the model can be used adequately to simulate scenarios. The model is used to evaluate alternative lampricide application strategies by simulating various TFM application plans. The effectiveness of granular Bayluscide applications is also analyzed using the hydrodynamic model and a plausible critical bottom velocity criterion. In TFM application simulations, twenty scenarios were considered subjected to the constraints of the flow operating conditions. These scenarios consisted of combinations of five flow conditions and four TFM application conditions. The scenario simulation results showed that two-site releases are more effective than single-site release, in terms of kill-per-dollar. However, the total cost for two-site releases are higher than single-site releases. The 16hour release achieves the maximum total kill, but the kill-per-dollar is best at 14-hour release. Further
704
Shen et al.
TABLE 4. Comparsion of lamprey population before and after application of granular Bayer under different flow conditions.
Flow Flow rate condition m3/s Apply to The Entire River 1 1,699.2 1 1,699.2 1 1,699.2 3 1,699.2 5 1,699.2 Dye Study 2,183.5 High 3,115.2 Apply to North Channel Only 1 1,699.2 1 1,699.2 1 1,699.2 3 1,699.2 5 1,699.2 Dye Study 2,183.5 High 3,115.2
Total lamprey population 1,000
Bottom distance cm
Critical velocity cm/s
1.25 2.5 5 2.5 2.5 2.5 2.5
8 8 8 8 8 8 8
4,768.7 4,768.7 4,768.7 4,768.7 4,768.7 4,768.7 4,768.7
1.25 2.5 5 2.5 2.5 2.5 2.5
8 8 8 8 8 8 8
2,326 2,326 2,326 2,326 2,326 2,326 2,326
simulations showed that the optimum duration of TFM release is between 14 and 16 hours. Simulations of bottom velocity distribution showed that the area over which the granular Bayluscide application can be effective covers 70 to 90% of the lamprey population in the St. Marys River, depending on the total river discharge. ACKNOWLEDGMENTS This study was funded by the Great Lakes Fishery Commission. The authors would like to thank members of the St. Marys River Control Task Force for their input and valuable suggestions during the course of this study. REFERENCES Ambrose, R.B., Woll, T.A., Connolly, J.P., and Schang, R.W. 1988. WASP4, A Hydrodynamic and Water Quality Model—Model Theory, Users Manual, and Programmer’s Guide. EPA/600/3-87/039, USEPA, Athens, GA. Ariathurai, R., MacArthur, R.C., and Krone, R.B. 1977. Mathematical Model of Estuarial Sediment Transport. Technical Report D-77-12, U.S. Army Engineer Waterways Experiment Station, Vicksburg, MS. Fischer, H., List, E., Koh, R., Imberger, J., and Brooks, N. 1979. Mixing in Inland and Coastal Waters. New York: Academic Press.
Lamprey population survived 1,000
Lamprey population killed 1,000
% of lamprey killed %
311.8 562.1 1,263.3 552.6 525.5 977.1 1,406.4
4,456.9 4,206.6 3,505.4 4,216.1 4,243.2 3,791.7 3,362.3
93.46% 88.21% 73.51% 88.41% 88.98% 79.51% 70.51%
178.6 391.8 615 382.3 355.2 535.1 707.5
2,147.3 1,934.2 1,710.9 1,943.7 1,970.7 1,790.8 1,618.4
92.32% 83.16% 73.56% 83.56% 84.72% 76.99% 69.58%
Fodale, F.M., Bergstedt, R.A., Cuddy, D.W., Adams, J.V., and Stolyarenko, D.M. 2003. Planning and executing a lampricide treatment of the St. Marys River using georeferenced data. J. Great Lakes Res. 29 (Suppl. 1):706–716. Great Lakes Fishery Commission. 1981. Report of a study of current patterns in the Sault Ste. Marie harbor area of St. Marys River using a tracer dye. Unpublished Report, Ann Arbor, MI. Halfon, E., and Brueggemann, R. 1990. Simulation of Disulfoton Fate in the River with the Toxfate Model. The Science of Total Environment 97/98: 385–394. King, I., Donnell, B., Finnie, J., Letter, J., McAnally, W., Roig, L., and Thomas, W. 1997. Users Guide To RMA2 WES Version 4.3. U.S. Army Corps of Engineers—Waterways Experiment Station, Hydraulics Laboratory, Vicksburg, MS. McCorquodale, J.A., Ibrahim, K., and Hamdy, Y. 1986. Fate and transport modelling of Perchloroethylene in the St. Clair River. Water Poll. Res. J. Canada 21(3): 398–410. Schleen, L.P., Christie, G.C., Heinrich, J.W., Bergstedt, R.A., Young, R.J., Morse, T.J., Lavis, D.S., Bills, T.D., Johnson, J.E., and Ebener, M.P. 2003. Development and implementation of an integrated program for control of sea lampreys in the St. Marys River. J. Great Lakes Res. 29 (Suppl. 1):677–693. Shen, H.T., and Yapa, P.D. 1988. Oil Spill Transport in Rivers. J. Hydraulic Engineering ASCE, 114(5): 529–543. ———, Yapa, P.D., and Zhang, B. 1995. A simulation
Modeling Lampricide Transport in the St. Marys River model for chemical spills in the upper St. Lawrence River. J. Great Lakes Res. 21:652–664. Thomas, W., and McAnally, W. 1985. Open-Channel Flow and Sedimentation TABS-2. Instruction Report HL-85-1, Waterways Experiment Station, U.S. Army Corps of Engineers, Vicksburg, MS. Tompson, A.F.B., and Dougherty, D.E. 1988. On the use of particle tracking methods for solute transport in porous media, In Computational Methods in Water Resources, 2, Numerical Methods for Transport and
705
Hydrologic Processes, ed. M. Celia et al. New York: Elsevier. Xu, Q., Shen, H.T., and Yapa, P.D. 1998. A Model for Lampricide Transport in the St. Marys River. Report 98-8, Department of Civil and Enviorn. Engineering, Clarkson University, Potsdam, NY. Submitted: 21 December 2000 Accepted: 20 March 2002 Editorial handling: John W. Heinrich