‘Art’ of Calibration in the ‘Science’ of H&H Modeling
Amit Sachan, PE, CFM, Project Manager/ Water Resource Engineer
Robert Billings, PE, PH, CFM Project Manager, Mecklenburg County
Agenda
Definitions
Case Study
Lessons Learned
Summary
Definitions
Case Study
Lessons Learned
Summary
Modeling
Representation of a process/object
Calibration
Models: “Adjustment of the parameters of a mathematical or numerical model in order to optimize the agreement between observed data and the model's predictions (source: Wikipedia)
Tools: Checking or adjusting (by comparison with a standard) the accuracy of a measuring instrument (source: TheFreeDictionary.com)
Consistent and systematic
Modeling
Representation of a process/object
Calibration
Models: “Adjustment of the parameters of a mathematical or numerical model in order to optimize the agreement between observed data and the model's predictions (source: Wikipedia)
Tools: Checking or adjusting (by comparison with a standard) the accuracy of a measuring instrument (source: TheFreeDictionary.com)
Consistent and systematic
Definitions
Calibration
Setting for known values
Accuracy vs. Precision
For Physical models
Multiple parameters
Multiple values
Classical Calibration for the Models
Sensitivity Analysis
Calibration
Validation
Calibration
Setting for known values
Accuracy vs. Precision
For Physical models
Multiple parameters
Multiple values
Classical Calibration for the Models
Sensitivity Analysis
Calibration
Validation
Definitions
Source: MedicalScale
Southern NC at the NC-SC
Border
330 miles of FEMA streams
State-of-the-art Flood warning
system
50 stream gages within LSCBC
Case Study
Case Study
Cavalier Apartments
Model accuracy & precision directly
Impact public $’s due to highly urban area
Storm Selection
High-end Storm (~100 Yr)
August 2008
Low-end Storms (~2 Yr)
September 2006
March 2007
Storm Selection
Continuous Data
Low Variability
Homogenous Response
Storm Selection
High-end Storm (~100 Yr)
August 2008
Low-end Storms (~2 Yr)
September 2006
March 2007
Storm Selection
Continuous Data
Low Variability
Homogenous ResponseRainfall Distribution during
August 25-27, 2008 (Source: USGS)
Case Study – Hydrologic Calibration
Case Study – Hydrologic Calibration
a. Sensitivity Analysis
Routing Techniques
Modified Puls
Mushkingham Cunge
Level Pool Routing
Lag time
Curve Numbers
AMC Conditions
Initial Abstraction
a. Sensitivity Analysis
Routing Techniques
Modified Puls
Mushkingham Cunge
Level Pool Routing
Lag time
Curve Numbers
AMC Conditions
Initial Abstraction
Case Study – Hydrologic Calibration
Case Study- Hydrologic Calibration
Additional Storage Areas
Case Study- Hydrologic Calibration
Change Increase Reduction
Addition of Storage Area on Edwards Branch
2% -
Addition of Storage Areas on Briar Creek 2% -
Change in Effective areas - -
Change in Lag times - 5%
‘Change & Effect’ Matrix
b. Calibration Alternatives & Iterations 14 scenarios
Gage weighting Equal Weighting AMC conditions Initial Abstraction
Aug 2008 Event Initial Abstraction Lag Time
b. Calibration Alternatives & Iterations 14 scenarios
Gage weighting Equal Weighting AMC conditions Initial Abstraction
Aug 2008 Event Initial Abstraction Lag Time
Rain event Scenario Description
Rain gages
Initial Abstraction
CNs
Sept 13, 2006
Scenario 1.1: All rain gages were weighted equally to come up with combined hyetograph.
Weighted Equally
0.05 inches AMC
II
Scenario 1.2: All rain gages were weighted equally to come up with combined hyetograph, to come up with combined hyetograph.
Weighted Equally
0.6 inchesAMC
II
March 1, 2007
Scenario 2.1: All rain gages were weighted equally to come up with combined hyetograph.
Weighted Equally
0.05 inches AMC
II
Case Study – Hydrologic Calibration
Case Study- Hydrologic Calibration
Observations Time to Peak - matches
Peak Discharges – within reasonable limits
Volume – within reasonable limits
Observations Time to Peak - matches
Peak Discharges – within reasonable limits
Volume – within reasonable limits
12:00 24:00 36:00 48:00 60:00 72:00 84:000
2,000
4,000
6,000
8,000
10,000
12,000
14,000
Little Sugar Creek @ Archdale Dr
Gage Flow HEC-HMS-2008
From Aug 25 to Aug 28 (Hours)
Flo
w (
cfs
)
Calibration – Inter relationship
Airplane example Airplane example
Source: Wikipedia
Calibration – Inter relationship
Relationship between Hydrologic & Hydraulic Models Relationship between Hydrologic & Hydraulic Models
30000 32000 34000 36000 38000 40000 42000 44000
540
550
560
570
580
LittleSugarCreek Plan: Calibration-August2008 9/24/2009
Main Channel Distance (ft)
Elev
atio
n (ft
)
Legend
WS 2008-Interpolate
Ground
OWS 2008- Interpolate
LittleSugarCreek Main
Hydrologic Model Hydraulic Model
Case Study- Hydraulic Calibration
High Water Marks Little Sugar Creek
Little Hope Creek
Briar Creek
High Water Marks Little Sugar Creek
Little Hope Creek
Briar Creek
Case Study- Hydraulic Calibration
High Water Marks Match within 0.5 ft
Quality
Parameter Changes Manning’s n
Ineffective Areas
Cont/Exp Coeff
High Water Marks Match within 0.5 ft
Quality
Parameter Changes Manning’s n
Ineffective Areas
Cont/Exp Coeff30000 32000 34000 36000 38000 40000 42000 44000
540
550
560
570
580
LittleSugarCreek Plan: Calibration-August2008 9/24/2009
Main Channel Distance (ft)
Ele
vatio
n (f
t)
Legend
WS 2008-Interpolate
Ground
OWS 2008- Interpolate
LittleSugarCreek Main
Case Study- Hydraulic Calibration
c. Validation Hydrologic
Hydraulic
c. Validation Hydrologic
Hydraulic
40000 50000 60000 70000 80000
560
580
600
620
LittleSugarCreek Plan: Calibration-August2008 9/21/2009
Main Channel Distance (ft)
Ele
vatio
n (f
t)
Legend
WS 2008-Interpolate
Ground
OWS 2008- Interpolate
LittleSugarCreek Main
Case Study- Hydraulic Calibration
…and iterations continue until reasonable matches in hydrologic AND hydraulic models
…and iterations continue until reasonable matches in hydrologic AND hydraulic models
30000 32000 34000 36000 38000 40000 42000 44000
540
550
560
570
580
LittleSugarCreek Plan: Calibration-August2008 9/24/2009
Main Channel Distance (ft)
Elev
atio
n (ft
)
Legend
WS 2008-Interpolate
Ground
OWS 2008- Interpolate
LittleSugarCreek Main
Hydrologic Model Hydraulic Model
Lessons Learned
Analysis (Science Component) Based on available data
Based on required detail
Sensitivity Analysis
Time/Budget Available
Trade offs (Art Component) Not exact matches
Engineering Judgment
Analysis (Science Component) Based on available data
Based on required detail
Sensitivity Analysis
Time/Budget Available
Trade offs (Art Component) Not exact matches
Engineering Judgment
QUESTIONS ??Thank youThank you