Clifton Bell, P.E., P.G. Chesapeake Bay Modeling Perspectives for the Regulated Community
Feb 24, 2016
Clifton Bell, P.E., P.G.
Chesapeake Bay Modeling Perspectives for the Regulated Community
Themes
Chesapeake Bay modeling framework is an remarkable set of tools.• Impressive capabilities• Important limitations
TMDLs lead to an overreliance on models. Be prepared to advocate local achievements in
model world.
Primary purposes of the Bay modeling framework:
Identify the:
1. Nutrient and sediment loads that will meet water quality standards in tidal waters.
2. Management actions that will achieve these loads.
“The model” is actually many linked models and data processing tools
Models developed, refined over 25+ years
Models developed, refined over 25+ years
Originally used to predict “hypoxic volumes” in Bay
Estimate watershed-scale reductions (e.g., 40% reduction by 2000)
Track progress over large areas
Use of the model has also evolved
Now trying to predict water quality at very specific locations and depths
Predict ≤1% changes in attainment. Estimate local loads
Use of the model has also evolved
Some Important Strengths
Watershed model relatively well calibrated at Baywide and major tributary basin level
Water quality model relatively well calibrated for dissolved oxygen in critical deep water segments
Calibration ValidationPredictive
Management Scenarios
Modeling Process
Uncertainty
Modeling Framework is Conservative with a Implicit Margin of Safety Attainment controlled
by small area, timing. All WWTPs
discharging at full permitted load
Conservative assumptions
Conservative BMP Efficiencies
BMP Text from BMP ReportsRiparian buffers “…a 20% reduction in the effectiveness values is applied to
efficiencies from literature sources…”Urban wet ponds and wetlands
“…recommendation to use a more conservative percent removal estimate.”
Bioretention “The 10% TN concentration reduction [is] a conservative judgment…”
Vegetated open channel
“A more conservative value …was selected…”
Permeable pavement
“…a conservative approach is taken to estimating…performance.”
Infiltration basins and trenches
“…a 15% reduction in TN is used here …to be…conservative.”
Categories of Model Limitations
Limitations of the basic algorithms Calibration errors Overparameterization Scale limitations Input errors Poor model behavior Imprecision of management predictions
Limitations of Basic Algorithms
Examples from watershed model:• Groundwater component crude• No explicit simulation of stream bank erosion• No mass balance of fertilizer
Calibration issues
No calibration is perfect. Quality of Bay model calibration varies greatly
by parameter and location. Watershed model partially calibrated to another
model.
Overparameterization
Complex nutrient cycling algorithms
Overparameterization
x + y =100
Highly Empirical Regional Transport Factors
RegionalDeliveryFactors
Edge of Stream
In Stream Concentrations
21
ungaged basins
TN calibrated Factors0.25 - 0.50.5 - 0.6670.667 - 0.8330.833 - 1.21.2 - 1.51.5 - 2 2 - 4
Phase 5.0 TP Calibrated Regional Factors
Scale Issues
Watershed model lack resolution for accuracy at the local scale• Segmentation• Input data• Calibration
Hoffman County
DianeRiverBasin
STAC Peer Review: 2008
23
“[The] current [watershed model]… is not appropriate for development and implementation of TMDLs at the local watershed scale. A major barrier appears to be the scale of information built into the [model]…”
Input Errors
No benefit of agricultural nutrient management Urban land use
Poor model behavior
Many segments where the model doesn’t “behave”.
e.g., poor calibration e.g., non-intuitive
trends Often the cause and
its extent is undiagnosed.
Summary so far
The model is• Complex• Conservative• Imprecise
So how precise are model predictions of future attainment, anyway? Impossible to accurately quantify. Bay program instituted the “1% rule”. Field measurements are not this precise. Laboratory measurement are not this precise. Model is nowhere near this precise. Lowest realistic estimates:
• 5% for DO attainment.• 15% for chlorophyll-a attainment.
USEPA’s Justification for “1 % Rule”
How Will the Model be Used Post-2010?
Phase II WIPs• Quantify local loads?
Model “locked down” until 2017 Tracking progress
• Baywide• Major state tributary basin• Local level?
Community Model Scenario Builder
Phase 5.3 watershed model publically available.
Scenario Builder• Tool for creating input
to watershed model• Web version planned.• Can’t refine model
scale.
How Should Stakeholders Use the Model and Scenario Builder? Don’t
• Use current watershed model for local TMDLs.• Let current watershed model output drive Bay TMDL
implementation at local level.• Let MS4 permits base compliance on current
watershed model predictions.
Do• Track BMPs for input to watershed model.• Use current watershed to track progress at major
tributary, state, and Baywide scale.• Base MS4 permit requirements on MEP.• Use refined models for local TMDL planning.
How Should Stakeholders Use the Model and Scenario Builder
Do• Use watershed model to identify offsets and trades• Advocate new BMPs for inclusion in the Baywide
model New structural BMPs Non-structural BMPs
– Ordinances– Public education and outreach– Improved BMP maintenance
How Should Stakeholders Use the Model and Scenario Builder
Questions?