nationwide study of the benefits of green infrastructure .../media/files/a/...region of influence...

Nationwide study of the benefits of green

infrastructure (GI) for flood loss avoidance

in the United States

2013 International Low Impact Development Symposium, Minnesota, USA

Atkins Lectures

2

Daniel Medina, senior engineer,

Water Resources 21 August 2013

Background

Objective: Estimate flood losses avoided by the implementation of green infrastructure (GI) for new development and redevelopment.

Upcoming stormwater rule proposal in the USA:

● Based on GI

● Capture and retain on site a high percentile storm

● Example capture standard:

– 90th percentile for new development

– 85th percentile for redevelopment

● Assumed to start in 2020; snapshot at 2040

● Environmental Protection Agency is not proposing GI for flood control; these are side-benefits to water quality benefits.

Retention standard definition

Xth percentile storm: The event in which precipitation depth is greater than or equal to X% of all storm events over a given period of record.

The retained volume must be infiltrated, evapotranspired or harvested for beneficial use.

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

4.5

0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%

Ra

infa

ll d

ep

th (

in)

Percentile

Arapahoe, CO

Miami, FL

Baltimore, MD

Quillayute, WA

Study plan

Rationale: smaller run-off volume leads to smaller floodplains and thus fewer flood damages

Evaluate 20 Hydrologic Unit Code (HUC) 8 watersheds with and without GI-based retention

Estimate monetary flood losses for each scenario

Benefits = losses without GI – losses with GI

Scale results nationwide.

Methodology overview

Sample of twenty HUC8 watersheds

Publicly available datasets

Stream gauge hydrology

Hydraulic modelling

Hazus damage estimation

Nationwide scale-up.

Sample watersheds

Datasets

US Geological Society (USGS) streamflow records

USGS National Elevation Dataset (NED) (10-metres)

National Hydrography Dataset (NHD) Plus hydrography

National Land Cover Dataset (NLCD)

STATSGO2 soils

Census 2000 economic activity

Integrated climate and land use scenarios economic growth projections.

Procedure

1. Peak flow distribution from USGS gauge data

2. Adjust peak flows to 2040 hydrology without GI

3. Hydraulic modelling to estimate flood depths without GI

4. Estimate monetary damages without GI

5. Adjust peak flows to 2040 hydrology with GI

6. Hydraulic modelling to estimate flood depths with GI

7. Estimate monetary damages with GI

8. Damages avoided = damages without GI – damages with GI.

Hydrology

Flood frequency analysis with USGS’s PeakFQ software

Region of Influence (RoI) technique for spatial interpolation of peak flows (Eng et al., 2005)

Peak flows at any location in HUC8, existing conditions

Estimate future run-off volumes to approximate future peak flows, with and without GI.

Estimation of future hydrology

Use run-off volume ratios to adjust peak flows (MMSD, 2005)

Run-off volume from TR-55 methodology (curve number)

Future conditions (2040), no GI

Future conditions, with GI

Example: d80 = 80th percentile depth

Hydraulic modelling

Rapid Flood Delineation (RFD) model

As accurate as HEC-RAS

High speed hydraulic profile calculation (6,000 miles per CPU hour)

Automatic cross sections

Depth grids

Flood damage estimation

● FEMA’s methodology for estimating potential losses from disasters

● GIS-based

● Physical damage

● Economic loss

● Social impacts

– Shelter requirements

– Displaced households

– Population exposed to scenario floods, earthquakes and hurricanes.

Vulnerability curves

Federal Insurance Administration (FIA)

USACE

-10

-5

0

5

10

15

20

25

30

0% 20% 40% 60%

Flo

od

ing

dep

th (

ft)

Percent structural damage

Vulnerability curve

One-story house, no basement

Flood damage computation

Hazus uses General Building Stock (GBS)

Assumes uniformly distributed assets on Census blocks

Results:

Sample damage distribution

Middle James, without GI

Upper San Antonio, without GI

Sample damage distribution

Average

Annualized

Losses (AAL) =

area under

damage curve

Middle James, without GI

Upper San Antonio, without GI

Flood losses avoided

0

200

400

600

800

1,000

1,200

0 20 40 60 80 100

Dam

age

s (m

illi

on

s)

Return period (years)

with GI

Without GI

Damages avoided

Zero-damage threshold

Damages begin to occur when:

● Flood waters enter the floodplain, and

● Water reaches exposed assets.


GBS uniform distribution of assets on Census blocks:

● Some assets appear at risk when they are not

● Damages can be overestimated.


Flood event at which damages begin to occur:

1. No assets exist in the 2-year floodplain



masks

Distribution of avoided losses Year 2040 development (2006 dollars)

0

500

1,000

1,500

2,000

2,500

0 20 40 60 80 100

Dam

age

s (m

illio

ns)


Upper Chattahoochee HUC8

Original Hazus estimate

No assets in the 2-year floodplain



100-year event no GI

100-year event with GI

Two-year event no GI

Two-year event with GI


$0

$10

$20

$30

$40

$50

$60

$70

$80

$90

$100

0 20 40 60 80 100

Loss

es

avo

ide

d (

mill

ion

s)


2-year mask01100004, Quinnipiac, New Haven, CT

02030201+02030202, Southern Long Island, Long Island, NY

02040205, Brandywine-Christina, Northern DE

03150201, Upper Alabama, Montgomery, AL

05120208, Lower East Fork White, Bloomington, IN

12040104, Buffalo-San Jacinto, Houston, TX

10190004, Clear, West Denver, CO

12080005, Johnson Draw, West Texas - Odessa, TX

12080007, Beals, West Texas - Big Spring, TX

10190003, Middle South Platte-Cherry Creek, East Denver, CO

05140205, Tradewater, West KY

16040101, Upper Humboldt, Northeast NV

02050306, Lower Susquehanna, North of Baltimore in PA

12090205, Austin-Travis Lakes, Austin, TX

02080201, Upper James, Southern WV

03130001, Upper Chattahoochee, Northeast of Atlanta, GA

04080203, Shiawassee, Near Flint, MI

07010102, Leech Lake, Northern MN

12100301, Upper San Antonio River, San Antonio, TX

02080205, Middle James River, Near Richmond, VA

$0

$10

$20

$30

$40

$50

$60

$70

$80

$90

$100

0 20 40 60 80 100

Loss

es

avo

ide

d (

mill

ion

s)























Average annualized losses avoided

(AALA) AALA = AAL without GI – AAL with GI

Year 2040 development

Nationwide scale-up

Regression of AALA vs. watershed properties

● Exposure

● Climate

● Development forecast

Extrapolation to HUC8s not modelled

Relationship with watershed

properties

AALA2040

E

AN + AR 0.45

A

Losses avoided (two-year mask)

Avoided losses in 2040 = $730 million

Present value (2020-2040) = $5 billion

Losses avoided (five-year mask)


Present value (2020-2040) = $2.3 billion

Losses avoided (10-year mask)


Present value (2020-2040) = $0.8 billion

Validation tests

Diagnostic case studies, not calibration

Stream gauge approach vs. hydrologic modelling


NED terrain vs. LiDAR terrain

GBS vs. user-defined facilities (UDF)

Conclusions

When applied watershed wide, GI is effective at reducing

● Peak flows for large events

● Flood elevations

● Flood losses

Benefits can be quantified by the average annualized losses avoided (AALA)

GI is necessary for water quality/stream health

GI adds community resiliency and environmental protection

GI reduces future federal expenditure and protects existing federal investments in flood control

Need to improve messaging about the link between run-off volume reduction and flood loss avoidance.

For more information contact: Dan Medina

[email protected]

Celebrating 75 years of

design, engineering and

project management

excellence.

www.atkinsglobal.com

nationwide study of the benefits of green infrastructure .../media/files/a/...region of influence...

Documents