overview - cdn.ymaws.com · 0 % - less than 0.06 (blackland praire clay urbanization curves) 33 % -...

15
5/25/2017 1 US Army Corps of Engineers BUILDING STRONG® HYDROLOGIC MODEL (HEC-HMS) CALIBRATION FOR FLOOD RISK STUDIES C. Landon Erickson, P.E.,CFM Water Resources Engineer USACE, Fort Worth District April 27 th , 2017 BUILDING STRONG® Overview Flood Risk Study Components/Parameters Parameter Development Parameter Calibration Calibrated Model Results vs Flood Frequency Analysis Results 2 BUILDING STRONG® Flood Risk Study Components Hydrology How much water? Hydraulics How deep will the water get and how far will it spread? Outcome or Impacts 3

Upload: others

Post on 18-Jun-2020

2 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Overview - cdn.ymaws.com · 0 % - Less than 0.06 (Blackland Praire Clay Urbanization Curves) 33 % - 0.06 to 0.2 66 % - 0.2 to 0.6 100 % - 0.6 to 2.0 (Cross Timbers Sandy Loam Urbanization

5/25/2017

1

US Army Corps of Engineers

BUILDING STRONG®

HYDROLOGIC MODEL (HEC-HMS)

CALIBRATION FOR FLOOD RISK

STUDIESC. Landon Erickson, P.E.,CFM

Water Resources Engineer

USACE, Fort Worth District

April 27th, 2017

BUILDING STRONG®

Overview

Flood Risk Study Components/Parameters

Parameter Development

Parameter Calibration

Calibrated Model Results vs Flood

Frequency Analysis Results

2

BUILDING STRONG®

Flood Risk Study Components

Hydrology

►How much water?

Hydraulics

►How deep will the water

get and how far will it spread?

Outcome or Impacts

3

Page 2: Overview - cdn.ymaws.com · 0 % - Less than 0.06 (Blackland Praire Clay Urbanization Curves) 33 % - 0.06 to 0.2 66 % - 0.2 to 0.6 100 % - 0.6 to 2.0 (Cross Timbers Sandy Loam Urbanization

5/25/2017

2

BUILDING STRONG®

Parameters and Common Methods

Losses – Initial Abstraction-Constant Loss

Rate

Transform – Snyder Unit Hydrograph

Routing – Modified-Puls

Baseflow – None or Recession

Rainfall – USGS (2004), NOAA Atlas 14

coming soon (2018).

4

BUILDING STRONG®

Snyder Synthetic Unit Hydrograph

(UH) Method UH methodology (L.K. Sherman 1932)

“Synthetic Unit-Graphs” by Franklin F. Snyder (1938)

Synthetic UH Parameters for Ungaged Watersheds.

Watersheds studied in Appalachian Highlands

Drainage areas from 10-10,000 square miles

5

BUILDING STRONG®

Snyder Synthetic Unit Hydrograph

(UH) Method Unit Hydrograph Definition

► “The basin outflow resulting from one unit of direct runoff generated uniformly over the drainage area at a uniform rainfall rate during a specified period of

rainfall duration.” (Sherman)

𝑡𝑝 = 𝐶𝑡 𝐿𝐿𝑐0.3

𝑈𝑝

𝐴= 640

𝐶𝑝𝑡𝑝

Where

𝑡𝑝 = lag time, (hour)

Cp = UH peaking coefficient (Range seen from 0.4 to 0.8, USACE)

L = Length of the main-stem from the outlet to the drainage divide, (miles)

Lc = Length of the main-stem from the outlet to the point nearest watershed centroid, (miles)

Up = Peak of standard UH, (cubic feet per second)

A = Area, (square miles)

𝐶𝑡 = Basin coefficient accounting for slope and storage (Range seen from 0.4 to 8.0, USACE)

6

Page 3: Overview - cdn.ymaws.com · 0 % - Less than 0.06 (Blackland Praire Clay Urbanization Curves) 33 % - 0.06 to 0.2 66 % - 0.2 to 0.6 100 % - 0.6 to 2.0 (Cross Timbers Sandy Loam Urbanization

5/25/2017

3

BUILDING STRONG®

Snyder Synthetic Unit Hydrograph

(UH) Method Limitations and Conclusions

► Method estimates surface-runoff time distribution and not amount.

► Actual lag tends to increase with storm size.

► Synthetic UH departs from actual UH as basin departs from a typical fan

shape.

► Extreme care should be used in applying UH parameters derived from

ordinary storms to less frequent storms

► The equations and coefficients given are based mainly on fairly

mountainous watersheds and may need adjustment in flatter

areas.

► This procedure can be applied in studies of flood-control, flood-routing,

and flood-forecasting.

► Regional analysis recommended

7

BUILDING STRONG®

USACE Regional Study – Synthetic

Unit Hydrograph Parameters

Nelson, T.L. “Synthetic Unit Hydrographs Relationships, Trinity River

Tributaries, Fort Worth-Dallas Urban Area”, Seminar on Urban Hydrology, Davis, CA, 1970.

Rodman, P.K. “Effects of Urbanization on Various Frequency Peak Discharges”, USACE Water Resources Meeting, Albuquerque, NM,

1977

8

BUILDING STRONG®

Regional Study (Nelson, 1970)

Study Purpose: Develop method for coming up with reasonable

estimates for present and future peak discharges for flood plain management studies for ungaged areas, where funds or study time

are both in short supply.

► Rapid development during 1960s (Pre-NFIP)

► National Flood Insurance Act of 1968

► Large number USACE SWF Flood Studies

9

Page 4: Overview - cdn.ymaws.com · 0 % - Less than 0.06 (Blackland Praire Clay Urbanization Curves) 33 % - 0.06 to 0.2 66 % - 0.2 to 0.6 100 % - 0.6 to 2.0 (Cross Timbers Sandy Loam Urbanization

5/25/2017

4

BUILDING STRONG®

Regional Study (Nelson, 1970)

8 DFW Watersheds

(8-130 sq. mi.)

Urbanization

estimates ranged

between 0-100%

USGS Special

Urban Hydrology

Study (circa [1961-1979])

10

BUILDING STRONG®

Regional Study (Nelson, 1970)

Utilized methods in EM 1110-2-1405 “Flood Hydrograph Analyses

and Computations.”

Correlated lag time to measurable watershed characteristics such as

longest flow length, flow length to centroid, and weighted watershed

slope.

Literature review indicated several studies that concluded that lag

time (tp) correlated well with 𝐿𝐿𝑐𝑎 𝑆0.5

Where

𝑡𝑝 = lag time, (hour)

L = Length of the main-stem from the outlet to the drainage divide, (miles)

Lca = Length of the main-stem from the outlet to the point nearest watershed centroid,

(miles)

𝑆 = stream slope over reach between 10% and 85% of L (feet per mile)

11

BUILDING STRONG®

Regional Study (Nelson, 1970)

Literature Review of

Existing Studies► Slope of line is

consistent for

different areas ( So.

California, Louisville, Houston).

► Scatter in study data was explained by

differences in terrain

type and urbanization

amounts.

► Lag time decreased

as urbanization

amount increased.

Southern California Study

(Linsley, USACE)

Louisville, Kentucky Study

Houston, Texas Study

12

Page 5: Overview - cdn.ymaws.com · 0 % - Less than 0.06 (Blackland Praire Clay Urbanization Curves) 33 % - 0.06 to 0.2 66 % - 0.2 to 0.6 100 % - 0.6 to 2.0 (Cross Timbers Sandy Loam Urbanization

5/25/2017

5

BUILDING STRONG®

Regional Study (Nelson, 1970)

Final DFW

area curves

Houston, Texas Study

0

0

0

35

10Percent Urban Values of Study Watersheds

55

100

70

13

BUILDING STRONG®

Regional Study (Nelson, 1970)

Conclusions

► The method accounts for differences in urban development on

adjacent areas and may be used to predict the effect that urban

development might have on a given area.

► Additional analyses will be required to improve the relationships

developed, to isolate the effect that channel improvement produces,

and to develop an improved method of determining the unit

hydrograph peak.

► For the stated purpose of developing flood peaks for flood plain

management studies for ungagged areas, where funds or study time

are both in short supply, the present method appears to offer

reasonably accurate results.

14

BUILDING STRONG®

Regional Study (Rodman, 1977)

Study Purpose: Develop a method that offers a quick, relatively

consistent and reasonable procedure for estimating the effect of urbanization on the unit hydrograph for a specific watershed in a

specific urban area.

► D-FW Clay Urbanization Curves (22 watersheds, 1-130 sq. mi.)

• 8 watersheds from Nelson (1970) with 14 additional watersheds.

► D-FW Sandy Loam Urbanization Curves (4 watersheds, 27-333 sq. mi.)

15

Page 6: Overview - cdn.ymaws.com · 0 % - Less than 0.06 (Blackland Praire Clay Urbanization Curves) 33 % - 0.06 to 0.2 66 % - 0.2 to 0.6 100 % - 0.6 to 2.0 (Cross Timbers Sandy Loam Urbanization

5/25/2017

6

BUILDING STRONG®

Regional Study (Rodman, 1977) Final Blackland Praire Clay Urbanization Curves

16

BUILDING STRONG®

Regional Study (Rodman, 1977) Final Cross Timbers Sandy-Loam Urbanization Curves

17

1

1

1

5

BUILDING STRONG®

Regional Study(Rodman, 1977) Final DFW Urbanization Curve Comparison

A complete change from rural to urban conditions

would reduce the lag time by about 50%.

18

Page 7: Overview - cdn.ymaws.com · 0 % - Less than 0.06 (Blackland Praire Clay Urbanization Curves) 33 % - 0.06 to 0.2 66 % - 0.2 to 0.6 100 % - 0.6 to 2.0 (Cross Timbers Sandy Loam Urbanization

5/25/2017

7

BUILDING STRONG®

Regional Study (Rodman, 1977)

Conclusions

► The urbanization curves offer a quick, relatively consistent and reasonable

procedure for estimating the effect of urbanization on the unit hydrograph

► Recommended peaking coefficient of 0.72 (Cp640=460) for DFW area.

► The urbanization curves have been verified with observed hydrographs.

Limitations

► This study assumed future urbanization practices will approximate those of

the past.

► This study did not separately calibrate all the complicating factors of

urbanization (Percent imperviousness, storm sewers, channelization)

► The data used for the Sandy-Loam Urbanization curves was very limited

and could use additional verification. Only 4 watersheds (Compared to 22

for the Clay curves) ranging from 1-5 percent urban were used to develop

these curves.

19

BUILDING STRONG®

FW District Methodology in HMS

Sand (%) based on Soil Permeability (inches per hour)

► 0 % - Less than 0.06 (Blackland Praire Clay Urbanization Curves)

► 33 % - 0.06 to 0.2

► 66 % - 0.2 to 0.6

► 100 % - 0.6 to 2.0 (Cross Timbers Sandy Loam Urbanization Curves)

Clay (0% Sand) watersheds generally praire/grassland under natural

conditions.

Sandy Loam (100% Sand) generally have more trees and brush to

slow down and attenuate hydrograph peak.

Possible for Clay watershed with heavy brush and trees may

hydrologically act as Sandy watershed.

20

BUILDING STRONG®

FW District Methodology in HMS

21

Page 8: Overview - cdn.ymaws.com · 0 % - Less than 0.06 (Blackland Praire Clay Urbanization Curves) 33 % - 0.06 to 0.2 66 % - 0.2 to 0.6 100 % - 0.6 to 2.0 (Cross Timbers Sandy Loam Urbanization

5/25/2017

8

BUILDING STRONG®

From the Component Editior

FW District Methodology in HMS

22

BUILDING STRONG®

FW District Methodology in HMS

23

From the Global Editior

BUILDING STRONG®

River Routing Parameters Storage-Discharge Values developed in HEC-RAS (Steady Flow)

Profiles computed for a range of discharges to define the relationship of storage to flow between two channel cross sections.

Subreach estimate should be determined by calibration to streamflow

gages. If unable to calibrate, can be estimated by computation based on flood wave travel time through reach. Subreach value can

approach one for very wide floodplains with heavy attenuation.

24

Page 9: Overview - cdn.ymaws.com · 0 % - Less than 0.06 (Blackland Praire Clay Urbanization Curves) 33 % - 0.06 to 0.2 66 % - 0.2 to 0.6 100 % - 0.6 to 2.0 (Cross Timbers Sandy Loam Urbanization

5/25/2017

9

BUILDING STRONG®

Loss and Baseflow Parameters Initial Abstraction-Constant Loss

► From HMS Technical Reference Manual

Recession Baseflow

25

BUILDING STRONG®

Calibration Purpose Modify initial parameter estimates to improve model’s ability to simulate

actual/physical watershed response to observed storm events.

Develop a single set of representative watershed parameters

Rapid Response (Tall, narrow hydrograph) or slow response (Flat, wide

hydrograph)?

26

BUILDING STRONG®

Calibration Needs Initial Parameter Estimates

NEXRAD Stage III or MPE Precipitation Data► http://dipper.nws.noaa.gov/hdsb/data/nexrad/nexrad.html

[email protected]

USGS Streamflow Data

► https://maps.waterdata.usgs.gov/mapper/index.html

HEC-GridUtil (Precipitation viewing, processing, and analysis tool)

► http://www.hec.usace.army.mil/software/hec-gridutil/

27

Page 10: Overview - cdn.ymaws.com · 0 % - Less than 0.06 (Blackland Praire Clay Urbanization Curves) 33 % - 0.06 to 0.2 66 % - 0.2 to 0.6 100 % - 0.6 to 2.0 (Cross Timbers Sandy Loam Urbanization

5/25/2017

10

BUILDING STRONG®

Calibration Process Baseflow Parameters - Match the magnitude and slope of the flow in the stream before and

after the storm.

Initial Loss – Adjust initial loss to match beginning of runoff.

Constant Loss – Match the volume.

Lag Time – Match the time and magnitude of the peak.

Peaking Coefficient – Match the shape and try to keep peaking coefficient consistent among

subbasins with similar slopes.

Routing Subreaches – Match hydrograph peak attenuation.

Completed from upstream to downstream.

Parameter adjustments made uniformly to all subbasins above a gage, unless there is

strong evidence to adjust individual areas.

Goal is to match peak, shape, timing, and volume of observed hydrograph.

Keep each parameter within its reasonable, expected range. For example, do not use such a short lag time that the water would have to be traveling at > 20 ft/s to reach the outlet.

Parameter calibration is compared to initial parameter estimates and calibrated parameters

from other storms.

28

BUILDING STRONG®

Calibration Process Baseflow Parameters - Match flow in the stream before storm.

29

Initial baseflow too high

BUILDING STRONG®

Calibration Process Initial Loss – Adjust initial loss to match beginning of runoff.

30

Initial loss

too high.

Initial runoff

beginning

too late

Initial loss

too low.

Initial runoff

beginning

too soon

Page 11: Overview - cdn.ymaws.com · 0 % - Less than 0.06 (Blackland Praire Clay Urbanization Curves) 33 % - 0.06 to 0.2 66 % - 0.2 to 0.6 100 % - 0.6 to 2.0 (Cross Timbers Sandy Loam Urbanization

5/25/2017

11

BUILDING STRONG®

Calibration Process Constant Loss – Match the volume.

31

Constant loss

too high.

Constant loss

too low.

BUILDING STRONG®

Calibration Process Lag Time – Match the time and magnitude of the peak.

32

Lag time too high.

Peak is late and

hydrograph is

wider

Lag time too low.

Peak is early and

hydrograph is

narrower

BUILDING STRONG®

Calibration Process Peaking Coefficient – Match the shape.

33

Peaking coefficient

too low. Hydrograph

is flatter than

observed

Peaking coefficient is

usually set to the

same value for all

storm events

Peaking coefficient

too high. Hydrograph

is steeper than

observed

Page 12: Overview - cdn.ymaws.com · 0 % - Less than 0.06 (Blackland Praire Clay Urbanization Curves) 33 % - 0.06 to 0.2 66 % - 0.2 to 0.6 100 % - 0.6 to 2.0 (Cross Timbers Sandy Loam Urbanization

5/25/2017

12

BUILDING STRONG®

Calibration Process Routing Subreaches – Match hydrograph peak attenuation.

34

Routing subreach

values too high.

Hydrograph is steeper

than observed

Routing subreach

values too low.

Hydrograph is flatter

than observed

BUILDING STRONG®

Calibrating only to peak discharge. Many different ways to match peak.

Not comparing calibration results between different events.

Common Calibration Mistakes

35

Parameter/Storm Initial Est. 1998 2002 2004 2007 2015

Lag Time 5.1 5.1 2.4 5.5 2.2 4.7

Peaking Coefficient 0.7 0.7 0.4 0.7 0.3 0.7

Parameter/Storm Initial Est. 1998 2002 2004 2007 2015

Lag Time 5.1 5.1 4.5 5.5 5.7 4.7

Peaking Coefficient 0.7 0.7 0.7 0.7 0.7 0.7

Losses

decreased to

match peak, but

volume and

shape are off.

BUILDING STRONG®

Final Parameter Selection

Can use weighting based on subbasin runoff volume or peak discharge to select representative parameter for the subbasin.

Avoid using or assign low weighting to calibration parameters that are very different from other events where the simulated hydrograph has a

poor match to the observed hydrograph.

36

Page 13: Overview - cdn.ymaws.com · 0 % - Less than 0.06 (Blackland Praire Clay Urbanization Curves) 33 % - 0.06 to 0.2 66 % - 0.2 to 0.6 100 % - 0.6 to 2.0 (Cross Timbers Sandy Loam Urbanization

5/25/2017

13

BUILDING STRONG®

Comparison with Statistical Analysis

Interagency Flood Risk Management (InFRM) Hydrology Report for the San Marcos River Basin (Sept 2016)

37

BUILDING STRONG®

Comparison with Statistical Analysis

Blanco River at Wimberley, TX (355 sq. mi., length of record – 91 years)

Peak Discharge Frequency Curves Change of 100-yr Peak Discharge Over Time

38

95 % Confidence Limit

Peak Discharges for 100-yr (1% annual chance)

Statistical Analysis – 154,000 cfs

Calibrated HMS Model – 152,600 cfs

*Wimberley systematic record began in 1925

BUILDING STRONG®

Comparison with Statistical Analysis

San Marcos River at Luling, TX (838 sq. mi., length of record – 79 years)

Peak Discharge Frequency Curves Change of 100-yr Peak Discharge Over Time

39

95 % Confidence Limit

Peak Discharges for 100-yr (1% annual chance)

Statistical Analysis – 143,600 cfs

Calibrated HMS Model – 142,400 cfs

Page 14: Overview - cdn.ymaws.com · 0 % - Less than 0.06 (Blackland Praire Clay Urbanization Curves) 33 % - 0.06 to 0.2 66 % - 0.2 to 0.6 100 % - 0.6 to 2.0 (Cross Timbers Sandy Loam Urbanization

5/25/2017

14

BUILDING STRONG®

Statistical analysis results change

significantly over

time (+33% to -38%).

Calibrated numerical

modeling results are

relatively stable over time (+6% to -12%).

95 % Confidence Limit (Statistical Resutlts)

Uncertainty in Flood Risk Estimates

40

205,000

95,000

*Wimberley systematic record began in 1925

135,000

Calibration Event Sensitivity163,000

BUILDING STRONG®

Summary

The USACE Fort Worth District has developed urbanization curves

which account for different terrain conditions and the affects of urbanization. These curves produce reasonably accurate and

consistent flood risk estimates for a given area where funds and time

are in short supply.

The uncertainty in flood risk estimates has the potential of affecting

people’s property and lives. Model calibration is an effective way of reducing the uncertainty in our flood risk estimates.

.

41

BUILDING STRONG®

References

Interagency Flood Risk Management (InFRM) Team, “Hydrology Report for San

Marcos River Basin”, Submitted to FEMA Region VI, Sept. 15, 2016.

Nelson, T.L. “Synthetic Unit Hydrographs Relationships, Trinity River Tributaries, Fort

Worth-Dallas Urban Area”, Seminar on Urban Hydrology, Davis, CA, 1970.

Rodman, P.K. “Effects of Urbanization on Various Frequency Peak Discharges”, USACE Water Resources Meeting, Albuquerque, NM, 1977.

US Army Corps of Engineers, EM1110-2-1405, “FLOOD-HYDROGRAPH ANALYSES

AND COMPUTATIONS”, August 31, 1959.

US Army Corps of Engineers, EM 1110-2-1417, “FLOOD-RUNOFF ANALYSES”,

August 31, 1994.

US Army Corps or Engineers, Hydrologic Engineering Center, “HMS Technical

Reference Manual”, Davis, CA, 2010.

42

Page 15: Overview - cdn.ymaws.com · 0 % - Less than 0.06 (Blackland Praire Clay Urbanization Curves) 33 % - 0.06 to 0.2 66 % - 0.2 to 0.6 100 % - 0.6 to 2.0 (Cross Timbers Sandy Loam Urbanization

5/25/2017

15

BUILDING STRONG®

Questions?

C. Landon Erickson, P.E.,CFM

Water Resources Engineer

(817) 886-1692 [email protected]

U.S. Army Corps of Engineers

Fort Worth District (SWF)

819 Taylor Street

Fort Worth, TX 76102