overview of models jan09(hauck) - watershed planningwatershedplanning.tamu.edu/media/5554/overview...
TRANSCRIPT
1
Presentation II 1
Overview of Models for Estimating
Pollutant Loads & Reductions
(Handbook Chapter 8.3–8.5)
Texas Watershed Planning Short Course
Wednesday, Jan 14, 2009Larry Hauck
Valley Mills (BO100) Monthly Average daily streamflow
0
20
40
60
80
100
120
Apr-93
Oct-93
Apr-94
Oct-94
Apr-95
Oct-95
Apr-96
Oct-96
Apr-97
Oct-97
Month
m3 /s
measured predicted
E = 0.66%E = - 12.0M = 14.4P = 12.7
Upper Oyster Creek - Lower portion (8/9/2004) Mainstem
02468
10121416182022
0.05.010.015.020.025.0
DO (mgO2/L) Avg(P) DO (mgO2/L) Avg(O) DO (mgO2/L) Min(P)DO (mgO2/L) Max(P) DO (mgO2/L) Min(O) DO (mgO2/L) Max(O)DO sat(P)
A Humorous View or a Reality Check?Watershed Monitoring and Modeling is
the Art and Science of• Collecting data in systems we cannot adequately sample • Using methods developed by committees of technically
unqualified participants • For organisms and pollutants we know very little about• In order to form concepts about processes we do not
fully understand • That we represent as mathematical abstractions that we
cannot precisely analyze • To determine their responses to indeterminate stresses
we cannot accurately predict now let alone in the future • All in such a way that society at large is given no reason
to suspect the extent of our ignorance.
Adapted from a slide by Dr. Thom Hardy, Utah State Univ.
Presentation II 3
Why Use Watershed Modeling?
“Models provide another approach, besides monitoring data and export coefficients, for
estimating loads, providing source load estimates, and evaluating various
management alternatives.”
Presentation II 4
What Are Models?
Mathematical models are analytical abstractions of the real world and as such represent an approximation of the real system.
In the context of watershed planning, mathematical models are computer based, simplified representations of landscape and water quality processes that govern the fate and transport of one or more pollutants.
Presentation II 5
• Empirical Model• Mechanistic Model
Two Basic Types of Models
Presentation II 6
• A model where the structure is determined by the observed relationship among experimental data.
• These models can be used to develop relationships for forecasting and describing trends.
• These relationships and trends are not necessarily mechanistically relevant.
Source: EPA website, glossary of frequently used modeling terms.
Empirical Model:
2
Presentation II 7
• Investigating the relationship of inflowing nutrients in a lake to algal biomass production (eutrophication).
• Most early (circa 1970) lake eutrophication models based on statistical relationships between mass loading of nutrients and average algal biomass (e.g., Vollenweider models with numerous adaptations by others)
• Applied to PL-566 reservoirs in North Bosque River Watershed
An Example of an Empirical Model:
Presentation II 8
Annual mean summer chlorophyll-a concentration as a function of predicted total-P for years 1993-1998 from PL-566 reservoirs. N=25
Presentation II 9
• A model that has a structure that explicitly represents an understanding of biological, chemical, and/or physical processes.
• These models attempt to quantify phenomena by their underlying casual mechanisms.
Source: EPA website, glossary of frequently used modeling terms and our Handbook
Mechanistic Model:
Presentation II 10
Presentation II 11
• HSPF – Hydrologic Simulation Program - FORTAN
• SWAT – Soil & Water Assessment Tool• Both models are comprehensive
watershed loading, hydrologic, water quality models.
• Both models are intensive in use of resources and skills to operate the models.
• More on these a little later.
Two Common Mechanistic Models Used in Watershed Modeling in Texas:
Presentation II 12
• Steady State Model• Dynamic Model
Types of Mechanistic Models
3
Presentation II 13
A mathematical model of fate and transport of waterborne constituents using constant input parameters to predict constant values of receiving water quality concentrations (typically under low-flow conditions)
Source: Our Handbook
Steady-State Models:
Presentation II 14
An Example: QUAL2K
QUAL2K is a stream water quality model. It is one-dimensional* and operates under steady-state flow.
All water quality variables are simulated on a diurnal (24-hour) time scale, including dissolved oxygen.
*fully-mixed in the vertical and lateral directions
Presentation II 15
Upper Oyster Creek - Lower Reach - May 2004 Verification Case (5/7/2004) Mainstem
0
2
4
6
8
10
12
14
16
18
20
0.05.010.015.020.025.0
DO (mgO2/L) Avg(P) DO (mgO2/L) Avg(O) DO (mgO2/L) Min(P) DO (mgO2/L) Max(P)DO (mgO2/L) Min(O) DO (mgO2/L) Max(O) DO sat(P)
Portion of Oyster Creek – Dissolved Oxygen (DO) Results for Verification Period
1207712075 12074
17690
Presentation II 16
A mathematical model of fate and transport of waterborne constituents formulated to describe the physical behavior of a system and its temporal variability
Dynamic Models:
Presentation II 17
SWAT Prediction for a Subbasin of North Bosque River
(Streamflow)SC020 Monthly Average daily streamflow
0.0
0.1
0.2
0.3
0.4
0.5
0.6
Jan-95 Jun-95 Nov-95 Apr-96 Sep-96 Feb-97 Jul-97 Dec-97
month
m3 /s
measured predicted
E = 0.30%E = - 29.0M = 0.11P = 0.08
Presentation II 18
SWAT Prediction for a Subbasin of North Bosque River(Total Phosphorus)
SC020 Monthly Total TP
0
200
400
600
800
1,000
1,200
1,400
1,600
Jan-95 Jun-95 Nov-95 Apr-96 Sep-96 Feb-97 Jul-97 Dec-97month
kgs
measured predicted
E = -8.7%E = 199M = 88P = 263
4
Presentation II 19
• Both supported by EPA BASINS (Better Assessment Science Integrating Point & Nonpoint Sources)
• Include processes of:• Rainfall/runoff• Erosion & sediment transport• Pollutant loading• Stream transport• Management practices
HSPF & SWAT – Two Dynamic Watershed Models:
Presentation II 20
Model efficiency (E) for measured and predicted daily (d) and monthly (m) flow, total suspended solids (TSS), and nutrient loadings during (a) calibrations and (b) verification, at the outlet of the UNBRW.
Presentation II 21
• Both models are complete watershed and water quality models with several state variables (runoff, streamflow, bacteria, nutrients, dissolved oxygen, toxics).
• HSPF has an edge in predicting hydrology• SWAT more user friendly• SWAT more powerful in representing
agricultural practices• HSPF more powerful in urban environment• HSPF has an edge with instream fate and
transport processes
HSPF & SWAT General Impressions by Presenter
Presentation II 22
• Field-scale: application focused at the subbasin or smaller level; often on a single land use
• EPIC – Environmental Policy Integrated Climate
• APEX – Agricultural Prediction Policy/Environmental eXtender
Developer: Dr. Jimmy Williams, BREC
EPIC & APEX – Two Dynamic Field-Scale Models:
Presentation II 23
Field plot and rain gauge locations within the Goose Branch microwatershed.
Presentation II 24
APEX Application: Measured versus predicted storm runoff volume for a) all values and b) for values less than 50 cubic meters per hectare.
5
Presentation II 25
APEX ApplicationComparison of measured and predicted storm loads of total nitrogen.
Presentation II 26
APEX predicted soil extractable P levels (0-15 cm) for Coastal bermudagrass field.
Presentation II 27
• Not a model• Method of data organization and
presentation that assists in understanding and refining water quality assessments (data requirements: streamflow and pollutant data)
• Applicable to stream systems• Explained in previous session.
Load Duration Curves
Presentation II 28
Load duration curves including MOS, historical data, and flow regimes - Station 10938, West Fork Trinity River
─ 75th percentile─ geo. mean
1.E+08
1.E+09
1.E+10
1.E+11
1.E+12
1.E+13
1.E+14
1.E+15
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
Percent of days loading exceeded
E. c
oli
(cfu
/day
)
Allowable Load Geometric Mean (CFU / Day) Allowable Load Single Sample (CFU / Day)Measured Load Geomean75th Percentile
─ 75th percentile ─ geo. mean
Presentation II 29
• Will management actions achieve desired goals?• Which sources are the main contributors to pollutant
load?• What are the loads associated with individual sources?• Which combination of management actions will most
effectively meet identified goals?• When does impairment occur?• Will the loading or impairment get worse under future
conditions?• How can future growth be managed?
What Questions Can Models Assist in Answering?
Presentation II 30
Examples of questions models can answer fromNorth Bosque River (SWAT)
Upper Oyster Creek (QUAL2K)
6
Presentation II 31
Land Use and Soluble-P SourcesNorth Bosque River at Clifton
PO4-P Source Contribution
Dairy Waste Appl.49%
Pasture8%
WWTP10%
Crop 5%
Urban6%
Wood/Range22%
Land Uses
Dairy Waste Appl.4%
Pasture14%
Crop9%
Wood/Range71%
Urban2%
Presentation II 32
Scenario WWTPFlow
WWTPP Limit
DairyManureApp. Rate
DairyFeedP Diet
DairyManureHaul off
Existing 1997-98 Median Btw N&PRate
No No
Baseline 2020 Median N rate Yes NoScenario A 1997-98 1 mg/l P rate Yes NoScenario B 2020 1 mg/l P rate Yes NoScenario C 2020 1 mg/l P rate Yes Yes
BMPs Imposed on Selected Scenarios
Presentation II 33
0
20
40
60
80
100
120
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0Probability of Exceedence
Solu
ble
Phos
phor
us C
once
ntra
tion
(ppb
)
Baseline Existing Scenario A Scenario B Scenario C
Preliminary Target
Exceedence Probability for the North Bosque River at Valley Mills
Presentation II 34
Upper Oyster Creek -Lower Reach
Presentation II 35
Lower Reach Summer Critical High Temperature - Upper Oyster Cr. (Preliminary Results)
0
1
2
3
4
5
6
7
8
0510152025Location, km
Dis
solv
ed O
xyge
n, m
g/L
Present Permit LimitsCriteriaBOD=5 mg/L, NH3=2 mg/L, DO=6 mg/L
Stafford Run
Hwy 6
Steep Bank Cr.
Dam 3
Presentation II 36
Complex Watershed Modeling Systems:
• Reservoirs
• Tidal & Estuarine Systems
7
Presentation II 37
• The Soil and Water Assessment Tool (SWAT)
• applied to the Bosque River Watershed to predict In-stream PO4-P concentrations
• Used to assess various P Control Strategies
Integrated Modeling Approach for Watershed and Reservoir
Presentation II 38
Integrated Modeling Approach
CE-QUAL-W2, a two- dimensional, laterally averaged, hydrodynamic and water quality model • developed by U.S. Army Corps of
Engineers, Waterways Experiment Station, Vicksburg, MS
• applied to Lake Waco to predict in-lake PO4-P concentrations and algal response
• Used to assess various P Control Strategies
Presentation II 39Lake Waco-Bosque River Watershed Presentation II 40
CE-QUAL-W2 Segmentation of Lake Waco
Presentation II 41
CE-QUAL-W2 Calibration
Presentation II 42
02468
10121416
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0Probability of Exceedence
Solub
le Ph
osph
orus
Con
cent
ratio
n (p
pb)
Baseline Existing Scenario A Scenario B Scenario C
TargetRange
Preliminary Target
Exceedence Probability for Lake Waco
8
Presentation II 43
• Link watershed model to appropriate hydrodynamic and water quality models
• Watershed: HSPF, SWAT, etc.• Water Quality: for example, Water Quality
Analysis Simulation Program (WASP)• Hydrodynamic model: for example, DYNHYD,
RMA2• Complete H / WQ models: for example,
Environmental Fluid Dynamics Computer Code (EFDC)
Tidal Streams & Estuaries
Presentation II 44
1. Relevance: Is the approach appropriate to your situation?
2. Credibility: Has the model been shown to give valid results?
3. Usability: Is the model easy to learn and use? (If not, is the expertise available to apply a sophisticated model?) Are data available to support the model?
4. Utility: Is the model able to predict the water quality changes based on anticipated management changes?
Four Considerations That Assist in Defining Your Model Application
Presentation II 45
• See our Handbook, pages 8-18 through 8-27
• EPA. 1997. Compendium of Tools for Watershed Assessment and TMDL Development. EPA841-B-97-006
• EPA’s Council on Regulatory Environmental Modeling (http://cfpub.epa.gov/crem/)
Available Models & Model Capabilities
Presentation II 46Source: Handbook, page 8-26
Presentation II 47
Oh, No!
No Model Does Close to What Is Needed
Try load duration curve approach
or
Modify and adapt an existing model
Presentation II 48
An Example:
Modifying APEX to Better Accommodate Situations in
Forestry
Work by Drs. Jimmy Williams, BREC & Ali Saleh, TIAER
9
Presentation II 49
Modifications of APEX for forestry conditions.Presentation II 50
Simulated and measured sediment losses for (a) SHR and (b) CON treatments during 1980-1985 (average of three replications).
Presentation II 51
An Example Model Selection & Application
North Bosque River Watershed
Presentation II 52Land Use – Lake Waco Watershed
Presentation II 53
North Bosque River Watershed with SWAT subbasin delineation
Presentation II 54
Measured X-sectional AreasX-sectional areas measured at over 10 sites along the North Bosque and its tributaries.
BO095 Cross-Section
50.00
60.00
70.00
80.00
90.00
100.00
0.00 100.00 200.00 300.00 400.00
Tagline (ft)
Elev
atio
n (ft
)
10
Presentation II 55
Model ValidationThe validation process consists of model calibration and
verification.
Calibration - model parameters are adjusted within allowable limits until model output for a given time period matches measured output within some predetermined measure of model performance.
Verification - refers to running the calibrated model (i.e., holding adjustment parameters constant) during a different time period and comparing model output to measured values.
Note: Definitions & terms vary from those in Handbook.
Presentation II 56
Model Validation
Calibration and verification increases confidence that the model will accurately simulate watershed conditions for different management scenarios.
This process gives confidence that the model output during management scenarios will reasonably reflect what the true measured values would be.
Presentation II 57
Calibration and Verification Periods
VerificationCalibration
TimeSite 1993 1994 1995 1996 1997 1998 1999 2000SF020NF009NF020NF050SF075BO020 MB040BO040IC020AL040SC020GC100SP020BO070 (Hico) BO090 (Clifton)NC060BO100 (VM)
Presentation II 58
Measures of Model PerformanceNash-Suttcliffe goodness of fit (E)
% error (% E)
Presentation II 59
Measures of Model PerformancePercent Error ( + %E )
> 70> 55Unsatisfactory40 - 7030 - 55Satisfactory25 - 4015 - 30Good
< 25< 15Very GoodP, NSediment
Moriasi et al. (2007)
Presentation II 60
Calibration of Total Streamflow
Hico Monthly (BO070) Average daily streamflow
01020304050607080
Apr-93
Sep-93
Feb-94
Jul-94
Dec-94
May-95
Oct-95
Mar-96
Aug-96
Jan-97
Jun-97
Nov-97
Month
m3 /s
measured predicted
E = 0.86%E = -8.5M = 3.45P = 3.16
11
Presentation II 61
Calibration: Average daily loads
BO070 average daily load 1994-1997
0
100
200
300
400
500
600
700
800
900
OrgN OrgP NO3 MinP TP TN
kgs BO070 sim
BO070 meas
Presentation II 62
Verification Period(flow)
Valley Mills (BO100) Monthly Average daily streamflow
01020304050607080
Jan-98 Jul-98 Jan-99 Jul-99
Month
m3 /s
measured predicted
E = 0.70%E = 11.9M = 6.1P = 6.8
Presentation II 63
Verification: Average daily loads
BO070 average daily load 1998-1999
0
50
100
150
200
250
300
350
OrgN OrgP NO3 MinP TP TN
kgs BO070 sim
BO070 meas
Presentation II 64
Scenarios Evaluated
xxxxxxxxxGxxxxxxxxF
xxxxxxxxExxxxxxxxD
xxxxxxxCxxxxxxB
xxxXxA
Red. graze
WWTP .5 mg/l
75% manure
HWAF remed.
Microgyremed.
New PL-566
NRCS50% manure
New lagoon
WWTP 1 mg/l
Filter stripScen
SECITCARP
Presentation II 65
1990s and Future Baseline and Scenario A and G Annual daily-average PO4 conc. and annual PO4 loadings
At Valley Mills(PRELIMINARY RESULTS SUBJECT TO CHANGE)
0
20
40
60
80
100
120
140
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Exceedance Probability
PO4 c
once
ntra
tion
(ppb
)
full-permitted baseline 1990s baselinefuture A future G
0
20,000
40,000
60,000
80,000
100,000
120,000
140,000
160,000
180,000
200,000
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Exceedance Probability
PO4 l
oad
(kg)
full-permitted baseline 1990s baselinefuture A future G
Presentation II 66
Land Uses of North Bosque River Watershed with dairy waste application fields
12
Presentation II 67
Questions?