1. Options for Identifying & Quantifying Pollutant Loads

2. Presentation Overview
Goals of pollutant load estimation
Options for quantifying current loads or conditions
Data-driven approaches
Models
Modeling, as it relates to environmental systems
Types of models
Models typically used for load estimation
Data needs
Example Application of Simple Model

3. Why is Pollutant Load Estimation Necessary?
Identify relative magnitude of contributions from different sources
Determine whether locations of sources are critical
Evaluate timing of source loading
Target future management efforts
Plan restoration strategies
Project future loads under changing conditions
Develop a mechanism for quantifying potential improvement

4. Pollutant Load Estimation Approaches
Has it already been done?
Total Maximum Daily Loads (TMDLs)
Clean Lakes Studies
Other local and regional studies
If not…
Data-driven approaches
Best when detailed monitoring data available
Models
Provide greater insight into impact of sources (temporally and spatially)
Readily allow for evaluation of future conditions

5. Data-driven Approaches
Estimate source loads using:
Monitoring data
Periodic water quality concentrations and flow gauging data
Facility discharge monitoring reports
Literature
Loading rates, often by landuse (e.g., lbs/acre/year)
Typical facility concentrations and flow

6. Is a Data-driven Approach Appropriate?
Monitoring data
Does it represent most conditions that occur (low flow, storms, etc.)?
Are spatial and source variability well-represented?
Have all parameters of interest been monitored?
Is there a clear path to a management strategy?

7. Load Estimates – Monitoring Data
In simplest terms…
load = flow x concentration
Load duration curves
Flow-based presentation
Statistical techniques
Relationships between flow and concentration to “fill in the blanks” when data aren’t available
Examples include:
Regression approach
FLUX

8. Load Duration Curves Rank daily flow and generate flow duration curve
Multiply water quality concentrations by corresponding flow values
Flow “curve” represents water quality target

9. Regression Approach
Develop a regression equation by plotting flow vs. corresponding water quality concentration
Use the relationship to predict water quality concentration for days when flow data exist
Note: limited applicability to data that is heavily storm-driven and spans orders of magnitude (e.g., sediment)
should consider log transform regression approach
Minimum Variance Unbiased Estimator (MVUE) recommended by USGS for bias correction (

10. Regression Approach - Example

11. FLUX Interactive computer program
Developed for U.S. Army Corps of Engineers
“Maps” flow-concentration relationship from available data onto entire flow record
Calculates total mass, streamflow, and error statistics
Can stratify data into groups based on flow
Six available estimation algorithms

12. FLUX – Data Requirements
Constituent concentrations, ideally collected weekly to monthly for at least a year
Date each sample was collected
Corresponding flow measurements (instantaneous or daily mean)
Complete flow record (daily mean) for the period of interest

13. Load Estimates – Literature
Landuse-specific loading rates (typically annual)
Multiply loading rate by area:
loadall = (arealu1 x loading ratelu1)+ (arealu2 x loading ratelu2) +…
Generally for landuse or watershed-wide analysis
Many sources: Lin (2004); Beaulac and Reckhow (1982), etc.
Use with caution (need correct representation for your local watershed)
Pollution sources
Climate
Soils

14. Example Load Estimation Based on Literature Values

15. Limitations of Data-driven Approaches
Monitoring data
Reflect current/historical conditions (limited use for future predictions)
Insight limited by extent of data (usually water quality data)
Often not source-specific
May reflect a small range of flow conditions
Literature
Not reflective of local conditions
Wide variation among literature
Often a “static” value (e.g., annual)

16. If a Data-driven Approach Isn’t Enough…Models are Available
What is a Model?
A theoretical construct,
together with assignment of numerical values to model parameters,
incorporating some prior observations drawn from field and laboratory data,
and relating external inputs or forcing functions to system variable responses
* Definition from: Thomann and Mueller, 1987

17. Nuts and Bolts of a Model
Input
Model
Algorithms
Output
Factor 1
Rainfall Event
System
Land use
Response
Factor 2
Soil
Pollutant Buildup
Stream
Factor 3
Pt. Source
Others

18. Is a Model Necessary? It depends what you want to know…
Probably Not
What are the loads associated with individual sources?
Where and when does impairment occur?
Is a particular source or multiple sources generally causing the problem?
Will management actions result in meeting water quality standards?
Which combination of management actions will most effectively meet load targets?
Will future conditions make impairments worse?
How can future growth be managed to minimize adverse impacts?
Probably
Models are used in many areas…
TMDLs, stormwater evaluation and design, permitting, hazardous waste remediation, dredging, coastal planning, watershed management and planning, air studies

19. Types of Models Landscape models Receiving water models
Runoff of water and materials on and through the land surface
Receiving water models
Flow of water through streams and into lakes and estuaries
Transport, deposition, and transformation in receiving waters
Watershed models
Combination of landscape and receiving water models
Site-scale models
Detailed representation of local processes, for example Best Management Practices (BMPs)

20. Types of Models Landscape/Site-scale models
Crops
Pasture
Urban
Crops
Pasture
Urban
Landscape/Site-scale models
Landscape/Site-scale models
Receiving water models
Receiving water models
Watershed models
Watershed models

21. Model Basis Empirical formulations Deterministic models
mathematical relationship based on observed data rather than theoretical relationships
Deterministic models
mathematical models designed to produce system responses or outputs to temporal and spatial inputs (process-based)

22. Review of Commonly Used Models
Landscape and Watershed models
Simple models
Mid-range models
Comprehensive watershed models
Field-scale models

23. Simple Models Limitations: Minimal data preparation
Loading Rate
Simple Method
USLE / MUSLE
USGS Regression
PLOAD
STEPL
Minimal data preparation
Landuse, soil, slope, etc.
Good for long averaging periods
Annual or seasonal budgets
No calibration
Some testing/validation is preferable
Comparison of relative magnitude
Limitations:
Limited to waterbodies where loadings can be aggregated over longer averaging periods
Limited to gross loadings

24. Mid-range Models More detailed data preparation
AGNPS
GWLF P8
SWAT ( + receiving water)
More detailed data preparation
Meteorological data
Good for seasonal/event issues
Minimal or no calibration
Testing and validation preferable
Application objectives
Storm events, daily loads
Limitations:
Daily/monthly load summaries
Limited pollutants simulated
Limited in-stream simulation and comparison with standards

25. Comprehensive Watershed Models
HSPF/LSPC
SWMM
Accommodate more detailed data input
Short time steps and finer configuration
Complex algorithms need state/kinetic variables
Ability to evaluate various averaging periods and frequencies
Calibration is required
Addresses a wide range of water and water quality problems
Include both landscape and receiving water simulation
Limitations:
More training and experience needed
Time-consuming (need GIS help, output analysis tools, etc.)

26. Source of Additional Information on Model Selection
EPA 1997, Compendium of Models for TMDL Development and Watershed Assessment. EPA841-B
Review of loading and receiving water models
Ecological assessment techniques and models
Model selection

27. Example of Simple Model Application
Spreadsheet Tool for Estimating Pollutant Load (STEPL)
Employs simple algorithms to calculate nutrient and sediment loads from different land uses
Also includes estimates of load reductions that would result from the implementation of various BMPs
Data driven and highly empirical
A customized MS Excel spreadsheet model
Simple and easy to use

28. STEPL Users?
Basic understanding of hydrology, erosion, and pollutant loading processes
Knowledge (use and limitation) of environmental data (e.g., land use, agricultural statistics, and BMP efficiencies)
Familiarity with MS Excel and Excel Formulas

29. Process STEP 1 STEP 2 STEP 3 STEP 4 Sources Cropland Urban Pasture
Forest
Feedlot
Others
Runoff
Erosion/
Sedimentation
BMP
Load after BMP
Load before BMP
STEP 1
STEP 2
STEP 3
STEP 4

30. STEPL Web Site
Link to on-line
Data server
Link to download
setup program to
install STEPL program and documents
Temporary URL: until moved to EPA server

32. STEPL Main Program
Run STEPL executable program to create and customize spreadsheet dynamically
Go to demonstration

33. Conclusions Many tools are available to quantify pollutant loads
Approach depends on intended use of predictions
Simplest approaches are data-driven
Watershed modeling is more complex and time-consuming
provides more insight into spatial and temporal characteristics
useful for future predictions and evaluation of management options
One size doesn’t fit all!