Ecological Forecasting for the Great Lakes

1
Ecological Forecasting for the Great Lakes

• Regional Data Exchange Workshop
• University at Buffalo
• May 15, 2008
• Joseph Atkinson
• Great Lakes Program
• University at Buffalo

2
Use of models in management/decision making
3
Management/modeling Issues

• Water quantity and flows (hydrologic model)
• Hydropower, shipping, recreational boating
• Controls at Lake Superior and Lake Ontario
• Diversions
• Changes in habitat (wetlands), fisheries
• Pollution, eutrophication (hydrodynamic,
  nutrients)
• Algal blooms and HABs
• Invasive species (ecological model)
• Persistent toxic chemicals (water quality model)
• Organics, metals, etc. bioaccumulation
• Contaminated sediments (IJC areas of
  concern)(sediment transport model)
• Climate change (multiple concerns, models)

Focus on integrated modeling approaches
4
Other issues/features

• International waters
• Closed basin circulation
• Coastal flows, upwelling and downwelling
• River/lake interactions
• Vertical suspended solids structure (benthic
  nepheloid layer)
• Cycling of organics
• Vertical and horizontal (thermal bar)
  stratification
• Water/sediment interactions
• Atmospheric deposition and exchange
• Point and non-point source loads

5
What is a model?

• Idealized representation of the real system
• Conceptual
• Simple analytical
• Physical
• Mathematical (numerical)
• Expressed in terms of governing equations
• Differential equations describing conservation
  statements(mass, momentum, energy, etc.)
• Constitutive relations (equation of state,
  coefficients)
• Incorporate approximations --- all models are
  wrong
• Scale and resolution (time and space)
• Processes to be considered
• Numerical approximations (computer solutions)

6
What are models used for?

• Integrate and synthesize data
• ex water level regulation in Lake Ontario
• Simulate the real world
• Demonstrate understanding of system
• Allow experimentation, evaluation of what if
  scenarios
• Convey results
• Graphics, tables, etc.
• Management support, options, risk

7
Model application
Processes to consider, Resolution
Management, scientific questions
Conceptual framework
Problem statement
Model formulation
iteration
Solution method
Scenarios (test management options)
Confirmation (system understanding)
Calibration
Risk and uncertainty
Data
8
Examples

• Algal bloom monitoring and modeling (MERHAB)
• Source locations and resource sheds
• Integrated coastal ecosystem model
• New York Ocean and Great Lakes Ecosystem
  Conservation
• Sediment transport

9
Hydrodynamic and particle tracking tools

• Three-dimensional hydrodynamic model (Princeton
  Ocean Model, POM)
• Uses actual or historic meteorological data
• Forecasting based on actual, current conditions
• Current applications using surface velocity field
• Any level can be used
• POM produces velocity and diffusion fields

10
Hydrodynamic and particle tracking tools (cond)

• Lagrangian (particle tracking) approach
• Random walk algorithm
• Conservative, passively transported particles
  (like a water molecule)
• Gridless model, but interpolates from POM grid
  values

11
Random walk algorithm
Particle movement deterministic component
stochastic component
In x direction,
(similar for y direction)
deterministic
stochastic

• Deterministic component real velocity pseudo
  velocity
• Stochastic component random walk based on
  diffusivity

Iterative approach used to account for changes in
velocity and diffusivity values at initial and
final location
12
Application to Lake Erie

• Forward and backward tracking
• August and May conditions
• General circulation
• Source areas
• One-day, one-week and one-month resource shed
  simulations
• Connection with watershed model

13
Forward tracking
Particles move with predicted water flow
General circulation
Point release (bloom tracking)
14
Source regions - Western Basin Lake Erie
15
Long-term vision - MERHAB-LGL
project(Monitoring and Event Response for
Harmful Algal Blooms)

• Provide predictions of algal bloom growth and
  movement, with certainty estimates, to predict
  potential impacts in Great Lakes basin
• Early warning system/management tool
• Focus on Lakes Erie and Ontario

16
Approach

• Run hydrodynamic model (POM) continuously
• Maintain initial conditions for forecast runs
• Click on map of lake, or enter location (web
  based application)
• Run hydrodynamic model for desired forecast
  period (several days to several weeks)
• Historical or forecast meteorological data
• Produce velocity and diffusivity fields
• Run particle tracking/population model
• Different modes possible
• Multiple particles
• Backtracking

17
Basic system arrangement (web-based modeling
interface)
18
Resource sheds - overview

• Resource sheds in coastal waters (Great Lakes)
• Motivation
• What are they?
• Hydrodynamic and particle tracking tools
• Application to Lake Erie
• Integration with watershed model

19
Motivation

• Determine source of materials (resources) to a
  particular area
• Zebra mussels
• Algae blooms
• Understand physical connectivity among
  different areas of the lake

20
What are they?(how are they calculated?)

• Particle tracking, used in combination with
  hydrodynamic model, to illustrate circulation and
  flow patterns
• backtracking
• Single release all locations from which
  materials originate at a common time
• One day, one week, one month, etc.
• Pathlines full trajectories over time period of
  interest
• Continuous release - particle positions
  plotted for continuous release to fill in all
  locations that may be contributing to a location
  of interest during the chosen time period

21
One-day backtracks (August)
22
One-week
23
One-week (May)
24
One-month
25
(No Transcript)
26
Density plots
27
Example Resource Shed Distributions Defined with
Particle Backtracking (in Western Central Lake
Erie)
Central Basin Site 311 August 31
max
1 day
1 week
2 weeks
3 weeks
0
1 month
28
Example Resource Shed Distributions Defined with
Particle Backtracking (in Western Central Lake
Erie)
Western Basin Site 835 August 31
max
1 day
1 week
2 weeks
3 weeks
0
1 month
29
General components coastal ecosystem
model(intensive monitoring study in Lake Ontario
summer 2008)

• Want to test biological filtering, or
  near-shore shunt hypothesis
• Include interactions with shore and with open
  water
• Combined physical/chemical/biological structure
• Synthesize data, evaluate system responses to
  various stressors, provide predictive
  capabilities (hypothesis testing)

30
Considerations

• Define state variables
• Desired temporal and spatial resolution
• Nested model?
• Same resolution for all components?
• Data availability
• Match watershed model(s) with lake model
• Time period of simulation

31
Data needs

• Meteorological (wind speed and direction, air
  temp., dew point, etc.)
• Point, non-point sources
• Flows, temperatures, concentrations, .
• Benthic conditions
• Sediment, algae, .
• In-lake currents and temperatures,
  concentrations, .
• Desired level of detail in time and space

32
Possible approaches (model team)

• Existing models
• POM (hydrodynamic)
• Saginaw Bay model (food web interactions,
  bioaccumulation)
• Particle tracking
• LOTOX (water quality)
• Delft/Elcom (hydrodynamics, water quality)
• Cladophora growth
• Watershed (?) SWAT, other
• Others (?)
• Canada/US 3 focus areas each (proposed)

33
Proposed model
Input data Geometry, bathymetry, topography Land
use, soil type Initial conditions Meteorology
Output Tributary flows, loadings Lake
circulation, water temperature, bottom shear P
concentrations, biomass
Coastal zone ecosystem model
Hydrodynamics
Chemical fate and transport
Particle tracking
Watershed, hydrological
Sediment transport
Ecological (nutrients, lower food web)
Cladophora growth
34
Simple 2 - box model
Off-shore region
inflows
outflows
transport
Near-shore region
35
Basic model

• Mass balance for near-shore (NS) region
• or
• Mass balance for off-shore (OS) region
• or

36
Sample results
37
Conclusions

• Were ready