Univ. of Minnesota

Modeling Nutrient and Sediment Modeling Nutrient and Sediment Losses from CroplandLosses from Cropland

D. J. MullaDept. Soil, Water, & Climate

University of Minnesota

Univ. of Minnesota

Watershed Management FrameworkWatershed Management Framework● Identify the problems and their extent● Monitor water quality● Evaluate pollution sources (modeling)● Set water quality goals (modeling)● Prioritize watersheds and

agroecoregions● Identify and implement BMPs to

improve water quality ● Evaluate progress towards goals

Univ. of Minnesota

Watershed ModelingWatershed Modeling

● Used to represent transport and fate of pollutants from the landscape to mouth of watershed

● Accuracy depends on ability of model to represent actual transport and fate processes

● Ability to evaluate effect on flow and water quality of alternative scenarios

Univ. of Minnesota

Model Selection CriteriaModel Selection Criteria

● Questions to be answered● Processes and pathways simulated● Spatial and temporal resolution needed● Complexity of model● Availability of input data● Time frame needed for results● Costs and staff expertise

Univ. of Minnesota

Simulation ModelsSimulation Models

● Export coefficient models● Statistical models● Mechanistic watershed scale models

– HSPF - EPA– SWAT, AGNPS - USDA– ADAPT - Univ. of Minnesota, Ohio State

Univ. (DRAINMOD + GLEAMS + Routing)● Mechanistic field scale models

– EPIC, RZWQM, DRAINMOD, GLEAMS

Univ. of Minnesota

Export Coefficient ModelsExport Coefficient Models

● Able to differentiate water quality impacts across broad land use classes

● Unable to account for variability caused by soil or climatic effects

● May not account for the diversity of agricultural management operations

Univ. of Minnesota

Statistical ModelsStatistical Models● Linear or non-linear regression● Most useful at the field scale● Tendency to over- or under-

parameterize● Interpretation of causes and effects may

be problematic● Statistical relationships are not

necessarily consistent with underlying transport or fate mechanisms

Univ. of Minnesota

The Universal Soil Loss The Universal Soil Loss EqtnEqtn (USLE)(USLE)

● A = R * K * L S * C * P– A is Estimated Soil Loss (tons/acre-yr) – Rainfall-Runoff Erosivity Factor (R) – Soil Erodibility Factor (K)– Slope Length and Steepness Factor (LS)– Cover Management Factor (C)– Supporting (Conservation) Practices Factor

(P)

Univ. of Minnesota

Univ. of Minnesota

Phosphorus Index Pathway Model Concept

Erosion(PP) c Soil P c

BMPsStructuresDelivery

RISK=

Rainfall Runoff(DP) c Soil P

Applied P c PracticeFactors RISK=

OverallRisk

Snowmelt Runoff(DP) c Biomass

Applied P c PracticeFactors RISK=

TransportMechanism

PhosphorusSource

ManagementEffects

Univ. of Minnesota

Mechanistic ModelsMechanistic Models

● Attempt to describe underlying processes of transport and fate

● Designed for application at different scales

● Require more detailed input data than statistical models

● Differ in degree of empiricism used to describe underlying mechanisms

Univ. of Minnesota

Mechanistic Model StrengthsMechanistic Model Strengths

● Can separate effects of point and non-point sources

● Can investigate impacts of changing climatic conditions

● Estimate both concentrations and loads (useful for setting TMDLs)

● Can identify impacts of alternative management strategies

Univ. of Minnesota

Mechanistic Watershed Scale ModelsMechanistic Watershed Scale Models

● Hydrological Simulation Program Fortran (HSPF) – USGS and Stanford

● Soil and Water Assessment Tool (SWAT) - USDA-ARS

Univ. of Minnesota

HSPFHSPF● Continuous rainfall hydrology, runoff and

water quality model linked to nationwide GIS databases

● Represents watershed as pervious and impervious areas, stream channels and reserviors

● Sediment loads based on rainfall detachment and wash off based on transport capacity and scour

● Phosphorus loads based on phosphate and organic forms using buildup and washoffcoefficients

Univ. of Minnesota

HSPFHSPF● Strengths

– Widely used and accurate for daily and monthly flows

– Well suited for urban hydrology modeling– Accounts for overland transport as well as channel

and reservoir transport

● Weaknesses– Very difficult to calibrate– Does not represent agricultural management

practices explicitly– Doesn’t explicitly estimate gully or streambank

erosion

Univ. of Minnesota

SWATSWAT● Continuous rainfall hydrology, runoff,

sediment, crop growth, nutrients, agricultural management model with channel and reservoir routing linked to nationwide GIS databases

● Sub-basins grouped based on climate, land use, soil, management, ponds, and channel

● Sediment loads based on Modified Universal Soil Loss Equation

● Phosphorus loads based on runoff partitioning and erosion loading functions

Univ. of Minnesota

SWATSWAT● Strengths

– Ability to evaluate impacts of riparian, tillage, fertilizer and manure management practices on flow and water quality

– Widely used and accurate for monthly average flows

– Accounts for overland transport as well as channel and reservoir transport

– Accounts for groundwater and tile drain flow

● Weaknesses– Many calibration parameters– Doesn’t explicitly estimate gully or streambank

erosion

Univ. of Minnesota

Model Calibration and ValidationModel Calibration and Validation

● Calibrate model using multiple years of monitoring data using measured data for input parameters wherever possible

● Need good match between model predictions and measured data for flow, sediment, phosphorus, nitrate, etc.

● Predicted contributions to flow from runoff, interflow, tile drainage must be reasonable

● Use independent data for validation

Univ. of Minnesota

Modeling OutcomesModeling Outcomes

● Pollutant concentrations and loads at mouth of watershed

● Ability to identify sources of pollutant loads– Helps assess Waste Load Allocations (point sources) and

Load Allocations (non-point sources)● Ability to estimate load reductions with various

alternative interventions – Helps assess feasibility of attaining TMDL

● Ability to estimate changes in loads in response to climatic or landuse changes– Helps set reserve capacity

Univ. of Minnesota

Modeling Time FrameModeling Time Frame● TMDL modeling involves several stages:

– Data collection– Modeling– Analysis– Outreach– Public participation– Administrative duties

● Time required increases with size of area● HSPF takes twice as long to run as SWAT

and requires more FTE than SWAT● Two years is probably the absolute minimum

needed to model portions of the upland areas in the Lake Pepin Basin

Univ. of Minnesota

Model UncertaintyModel Uncertainty

● All models have uncertainty, these are characterized during calibration and validation using measures such as standard error, root mean square error, index of agreement, etc.

● Uncertainty is partially caused by climatic variability● The impacts of uncertainty on a TMDL can be

quantitatively estimated from model results● As uncertainty increases, the loads allowed for point

and non-point sources decrease● Uncertainty decreases as the complexity of the model

increases

ConclusionsConclusions

● Modeling is an important component of integrated watershed assessment

● Ability to evaluate management alternatives depends on type and scale of model

● The type of model selected has a big impact on time required for TMDL evaluation and on model uncertainty