evaluation of hydrologic models for alternative covers › work › 09 › 2073363.pdfpredicted and...
TRANSCRIPT
-
Evaluation of Hydrologic Models forAlternative Covers
Charles D. ShackelfordColorado State University
Craig H. BensonUniversity of Wisconsin-Madison
-
What are Alternative Covers?
Soil covers consisting of one or more layers used inplace of a prescribed cover and having equivalenthvdroloaic performance.
Replace clay caps & composite caps (RCRA C & D)
Significant cost savings (~$50 to $75k per acre)
Employ natural materials and principles... fit well withnature and likely are to have a long service life.
-
How do alternative covers work?
Exploit (i) natural water storage capacity of finer soils and (ii)water removal capabilities of evaporation and transpiration.
c
oQ.CD
LU
Storage capacity of cover
O)C
's_Q.
CO
0£Ezs
CO
(00)-l-»
o>c
"l_Q.
CO
COO
0)«-+CD
O)r-+*OS
CQCD
Key: Design for sufficient storage capacity to retain water thataccumulates during winter with limited drainage
-
Conventional Alternative Cover Profiles
Capillary Barrier
150mmTopsoJI
460 mm Si It
600 mm Sand
Interim Cover
Monolithic Barrier
150mmTopsoil
1200 mm Sandy Clay
300 mm Interim Cover
-
Design Methodology
Characterize unsaturated hydraulicproperties of borrow soils and growingcharacteristics of plants
Preliminary sizing of cover profile by handcalculations
Demonstrate performance standard is met(e.g., < 1 mm/yr percolation) by numericalmodeling or field demonstration.
-
Advantages of Modeling
Assess a wide range of scenarios that mayaffect cover during service life
Fast and inexpensive (relative to field studies)
Disadvantages of Modeling
Accuracy of models has not been assessed.,critical since performance criteria limitpercolation to small quantities (~ 1 mm/yr)
Potential to mislead by careful input selection
-
Models Being Evaluated
Name
HELP
HYDRUS-
Vadose/W
LEACHM
Source
USEPA
Type
Unit GradientBucket Model
UNSAT-H PNNL/USDOE Richards' Equation
USDASalinityLaboratoryMEND and
Geo-Slope Ltd.Cornell Univ.
Richards' Equation
Richards' Equation
Richards' Equation
-
Tasks
I: Baseline Assessment of Hydrologic Models
: Comparison of Field Data and ModelPredictions
: Model Improvement
IV: Algorithms for Long-Term PerformanceAssessment
-
I: Baseline Assessment of Hydrologic Models
- Comparison of model predictions for a suite ofcontrasting climates (Las Vegas, NV; Richland, WA;Albuquerque, NM; Denver, CO; Columbus, OH; andSavannah, GA)
- Identify range of predictions obtained for a given set ofinput conditions.
- Identify algorithms having greatest impact on predictions
: Comparison of Field Data and Model Predictions
- Unbiased assessment of model accuracy
- Identify changes needed to improve accuracy
- Guides work in Task
-
Sources of Field Data
University of Wisconsin Field Study (Khire etal. 1997, 1999)
Sandia Alternative Landfill CoverDemonstration (Steve Dwyer of Sandia)
USEPA's Alternative Cover AssessmentProgram (Bill Albright of Desert ResearchInstitute and Craig Benson of UW)
-
ACAP Test Sites:
• Twenty-four test covers at eleven sites in sevenstates.
• Ten conventional covers (seven composite andthree clay)
• Fourteen alternative covers (eight monolithicbarriers and six capillary barriers) — used in thisstudy
• Eight sites with side-by-side comparison ofconventional and alternative covers
-
ACAP Field SitesPolson.MT
Boardman, OR
Sacramento, CA
Altamont, CA
Monterey.CA
Apple Valley, CA Albany.GA
DRIDesert fteewth totirte WtSCXXdN
-
20
SRODiversjon
Berm
20
Alldimensionsin meters
0.6
10
DownSlope
-0.6
ACAPTest Section
Plan View
Large bathtubso with cover soil
instruments
-
Typical Lysimater Cross-Section
20
InterimCover Soil
LLDPECutoff
Earthen^Berm—^ LLDPE
Geomembrane'ercolationPipe
Existing Slope (>2%)
Geocomposite Drain
arthenBerm
-
dozer at Altamont CA site.
-
Anti-Seep CoiSar Detail
Root Barrier
Wires
Cable Tire
Bentonite
PolyethyleneExcavation
Anti-Seep Collar(see Detail)
Heat Dissipation UnitThermocoupleWater Content Reflectometer
Heat Dissipation UnitThermocoupleWater Content Reffectometer
GeocompositeDrain
Geornembrane
-
weatherstation &
dataloggereel
-
cover SOIL «^
-
Sample of below ground
density distribution...
0.0
0.2
0.4
f 0-6CDQ
0.8
1.0
1.2
Relative Root Density
0.0 0.2 0.3 0.5 0.6
A •
R = 0.61 e'107z +0.007
j»-» Kiefer Landfill 'Sacramento, CA
June '01
• Sample A* Sample 8
-
Aerial view of completed test sections at Kiefer
-
-
Kiefer Site - Sacramento, CAWarm semi-arid climate (precipitation ~ 434 mm/yr)
1080-mm-thick monolithic covers
E 1500
co
"co•|_ 1200
c£•5Q.
g 900LU"OCCO
-g 600CO-^'g."oCD
CD
^»_co15EIDO
300
0
T T I 1 1 1 1 1 T 1 T
(a) Thin AlternativePrecipitation
Evapotranspiration
SoilWater Storage
Percolation
Surface Runoff
f
• ' •
MissingData
i . . .
500
400 V)o
o o
73
-
HYDRUS-2D
Uncoupled mass & heat transfer. For liquidflow (Richards' Eq.):
eedt
d
dx.,K(K A
I ij ^dXj
where KjjAand KizA are anisotropy factors
-
Boundary ConditionsVariable flux — upper boundary foratmospheric interaction (evaporation)
Prescribed head (constant or variable)
Prescribed flux
Seepage face (requires saturation) - lowerboundary for lysimeter
Deep drainage (unit gradient) - conventionallower boundary for cover modeling
-
Evaporation in HYRUS-2D
Potential Flux = Precipitation - Potential Evaporation (PE)
Precipitation < PE,calculate actual evaporation (AE)
Precipitation > PE,No evaporation
Solve Richards' Equationw/ interchangeable surface head/flux BC
Flux BC based onhydraulic gradient
Head BC if> hsat)
-
Transpiration in HYDRUS-2D
Read Input forPotential Transpiration
(PT)
Read Input forRoot Length Density Function
(RLD)
Obtain Plant Stress factor (a) corresponding toPressure heads in the root zone
Calculate Actual Transpiration (AT) as a sink termf(PT,a,RLD)
-
Example of HYDRUS-2DPredictions: Marina, CA sit
120mmTopsoi I
Costal sub-humid climate with466 mm/yr. Seasonal, but nosnow or freezing
-. H™ Capillary barrier with ~ 1.2 m1100mm , ,T / , , .. rSandy Clay °^fme textured soil for storage
300mmSand
Four primary water balancequantities: runoff,evapotranspiration, soil waterstorage, and percolation
-
odel Predictions and Field Data300
200
•ocCD
.0
100Q.'o0
T I
Marina Alternative Cover
Precipitation
Runoff
QL5/1/00 7/1/00 9/1/00 11/1/00 1/1/01 3/1/01 5/1/01 7/1/01
300
£200cg
Q.(A
IOQ.
100
LU
\ \ I
ET
/ HYDRUS-2D
05/1/00 7/1/00 9/1/0011/1/001/1/01 3/1/01 5/1/01 7/1/01
CDCD
-co
500
400
300
200
100
Field Capacity-
Storage HYDRUS-2D
Field
Wilting Point' •
60
E 40
o
20
0
PercolationField
HYDRUS-2D
J I
5/1/00 7/1/00 9/1/00 11/1/00 1/1/01 3/1/01 5/1/01 7/1/01 5/1/00 7/1/00 9/1/00 11/1/00 1/1/01 3/1/01 5/1/01 7/1/01
-
Predicted and Measured Runoff: HYDRUS-2D
Prediction
Over-prediction
Under-prediction
Almostmatch
Test Site
Poison, MT
Boardman, OR(thin)
Field PredictedSRO SRO(mm) (mm)
Albany, GA 14.9
Altamont, CA 1.34
Marina, CA 0.00
Sacramento, Qn .CA(thin) yu>4
Sacramento, fiCA (thick) b -b
Cedar Rapids, oc 0IA 26'8
17.7
0.00
420
26.5
47.4
289
318
0.07
SurfaceLayer(cm/d)
0.017
0.24
0.0058
0.077
0.077
0.46
4.20
1.68
Boardman, OR(thick)
0.00 0.06 1.21
MeanIntensity
I(cm/d)
4.26
1.65
2.79
2.24
2.24
4.13
1.23
1.39
1.39
255
-
Surface Runoff Calibration bySimultaneous Adjustment of Ksat and a
300
£co
Q."o2>
Q_
200
100
0
\ \ \
Marina Alternative Cover
Field
K = 10"8 cm/s, cc=0.0004 1/cmS
K = 10~7cm/s,a=0.01 1/cm
K =10"D cm/s, a=0.04 1/cm
K =1CTcm/s, a=0.1 1/cms
,-5.Field SRO &K=10'°cm/s
Precipitation
Ks
(cm/s)-
10-8
rr*»
D" •
I.
---10....... 10'
-7
-6
300
200CO
8
I
100
05/1/00 7/1/00 9/1/00 11/1/00 1/1/01 3/1/01 5/1/01 7/1/01
Scaling to field condition is a key issue!
-
Model Calibration to Runoff400
1"
c300
I8. 200
UJ
• 100CD-̂>'g.'oo
0
Marina Alternative Cover
FieldEvapotranspiration'
HYDRUSEvapotranspiration HYDRUS
Surface RunoffField Surface Runoff
to
o 3-3 8
2 I
05/1/00 6/28/00 8/24/00 10/21/00 12/18/00 2/14/01 4/13/01 6/10/01
500
E
^300SCO
200
100
0
.-SWS-fc 9
Field Soil Water Storage
HYDRUSSoil Water Storage
-SWS-wp
200
150
JS Percolation / i
' ^/
TJ
-
Sensitivity AnalysisMost influential factors:
rCat, a and n parameters (affect infiltration & storage)
Growing season (affects water removal)
Lower boundary condition (opposite of expected)
Errors in meteorological data (precipitation errors)
Less influential factors:
Pore interaction term, hysteresis
Wilting point, root depth, and root density distribution
-
Example - Effect of n Parameter300
§200
"a.
I 100Q_
0
Marina Alternative Cover
Precipitation
Field SRO
30
20
ZJO
10
5/1/00 7/1/00 9/1/0011/1/001/1/01 3/1/01 5/1/01 7/1/01
0)•a
500
400
300
200
100
•Field Capacity
Field
Wilting Point
I I I I I I
500
400 i:
CD
300
COCD
200
100
c_o
c
400
300
20°
100
0
n-«K
= 1.3
400
3005"OO
I200 |
3
100
05/1/00 7/1/00 9/1/0011/1/001/1/01 3/1/01 5/1/01 7/1/01
60
40
o
20
0
Field
n = 1.3,1.4, and 1.5
1 J I
60
40
-
Accomplishments to DateTasks:
• Task I - 25% complete (on schedule)
Task II - 50% complete (on schedule)
Key Findings:
• Model predictions can deviate greatly from field condition(critical issue for regulatory acceptance of designs)
Hydraulic property scaling is a key issue (pedogenesis willbe critical)
Growing season is a key issues (affect long-termpredictions)
-
Near Future Work
Complete comparisons between field data and modelpredictions (HYDRUS-2D, UNSAT-H, HELP complete,Vadose/W and LEACHM in progress)
Complete sensitivity analysis with all models (HYDRUS-2D, UNSAT-H, HELP complete, Vadose/W and LEACHMin progress)
Complete baseline comparisons (all models about 25%complete)