hydrodynamic and phosphorus transport modeling in lake...
TRANSCRIPT
Hydrodynamic and Phosphorus Hydrodynamic and Phosphorus Transport Modeling in Lake ErieTransport Modeling in Lake Erie
Dmitry BeletskyDmitry BeletskyUniversity of MichiganUniversity of Michigan
Ann Arbor, MIAnn Arbor, MI
David J. SchwabDavid J. SchwabNOAA Great Lakes Environmental Research LaboratoryNOAA Great Lakes Environmental Research Laboratory
Ann Arbor, MIAnn Arbor, MI
Joseph Joseph DePintoDePintoLimnotechLimnotech
Ann Arbor, MIAnn Arbor, MI
Dave DolanDave DolanUniversity of WisconsinUniversity of Wisconsin
Green Bay, WIGreen Bay, WI
The Problem:
- Excessive nutrient loading in the 1960’s led to massive algal blooms, oxygen depletion, and diminished water quality in Lake Erie.
- 1972 Water Quality Agreement between the US and Canada limited P loads from municipal, industrial, and agricultural sources.
- With controls, P levels decreased to acceptable levels and water quality improved.
- In recent years, P levels in Lake Erie appear to be increasing, despite controls.
The Problem:
- Excessive nutrient loading in the 1960’s led to massive algal blooms, oxygen depletion, and diminished water quality in Lake Erie.
- 1972 Water Quality Agreement between the US and Canada limited P loads from municipal, industrial, and agricultural sources.
- With controls, P levels decreased to acceptable levels and water quality improved.
- In recent years, P levels in Lake Erie appear to be increasing, despite controls.
Our Approach:
- Incorporate phosphorus transport and fate dynamics into high resolution (hourly time scale, 2 km horizontal resolution) hydrodynamic model of Lake Erie as a first step toward spatially explicit model of entire lower food web
Princeton Ocean Model(Great Lakes version implemented by D.Schwab and
K.Bedford in the early 90-s)
• Fully 3D nonlinear Navier-Stokes equations• Flux form of equations• Boussinesq and hydrostatic approximations• Free upper surface with barotropic (external) mode• Baroclinic (internal) mode• Mellor-Yamada turbulence model for vertical mixing• Terrain following vertical coordinate (sigma-coordinate)• Generalized orthogonal horizontal coordinates• Smagorinsky horizontal diffusion• Arakawa-C staggered grid
Lake Erie Marine Observation Network
2 km model grid and NWRI moorings in 2004
Water level at Buffalo, NY
Lake Erie 1994 TP simulationHydrodynamics- Great Lakes version of POM- 20 vertical levels, 2 km horizontal grid (~6500 cells)- Hourly meteorology (1994, JD 1-365)- Realistic tributary flows- Accounts for ice cover
Mass balance for P- POM hydrodynamics (2d for now)- Realistic P loading- Constant settling velocity (for now)
Tributary
Total Load (MT yr-1)
Average Flow (m3 s-1)
Avg. Conc. (ug l-1) (Load/Flow)
Detroit 3137 5712 17 Huron (MI) 31 18 55 *Raisin 104 17 194 Ottawa 32 6 169 *Maumee 1482 102 461 Portage 208 13 507 *Sandusky 380 28 430 Huron (OH) 107 6 565 Vermilion 105 6 555 Black 57 9 201 Rocky 59 9 208 *Cuyahoga 337 26 411 Chagrin 203 16 402 *Grand (OH) 174 25 221 Ashtabula 19 7 86 Conneaut 187 40 148 Cattaraugus 86 27 101 Buffalo 46 29 50 Grand (ON) 228 60 120 Big 10 14 23 Otter 40 14 91 Kettle 33 14 75 Total 7065 6198
Lake Erie Tributaries and l994 Phosphorus Loads (Dolan, 2004)
Total Phosphorus:Atmospheric: 718 MTTributaries: 7065 MT1994 Total: 7783 MT
*daily concentration data available
Computer animation of model results:-Starts in January, 1994-Uses 2d currents from hydrodynamic model-Time dependent P loads-Combination Lax-Wendroff and upwind advection scheme-No horizontal diffusion-Initial condition: C = 10 ug/L-Settling velocity = 6.8E-7 m/s (21 m/yr)
Modeled 1994 average TP concentration
Normalized VarianceAverage
CCIW Lake Erie Sampling Stations 1994 (courtesy of M. Charlton)
Cruise Dates, Stations Avg Conc Model1 4/26-4/30 50 21 192 7/5-7/8 20 12 133 7/19-7/22 20 14 144 7/25-7/29 52 9 125 8/30-9/1 18 14 116 10/4-10/7 10 16 137 10/11-10/14 49 16 11
Total 219 15 14
Time series of phosphorus concentration at selected locations
6/4/2001
4/4/2002
MODIS Imagery
Conclusions:Conclusions:-- The model slightly underestimates P concentrations in The model slightly underestimates P concentrations in 1994, especially in the fall. The model matches 1994, especially in the fall. The model matches observations better in western Lake Erie.observations better in western Lake Erie.-- The model shows significant spatial and temporal P The model shows significant spatial and temporal P variability in western Lake Erie, consistent with variability in western Lake Erie, consistent with observations. observations.
Future Work:Future Work:-- Sediment dynamics model Sediment dynamics model -- quantify internal P recyclingquantify internal P recycling-- Additional years, pre and post ZMAdditional years, pre and post ZM-- Calibrate with field surveysCalibrate with field surveys-- Ice model Ice model -- assimilation/thermodynamicsassimilation/thermodynamics