Exacerbation of Flooding Responses Due toLand Cover/Land Use Change:

A Comparative Study*

Keith N. Eshleman1, Phil Townsend2, Todd Lookingbill1, and Mykola Zalogin3

1Appalachian Laboratory, University of Maryland Center for Environmental Science, Frostburg, MD2Department of Forest Ecology and Management, University of Wisconsin, Madison, WI3ISDU, Kiev, Ukraine

*Response to NASA-NEWS 2004 RFA

Flooding on the Central Appalachian Plateau (CAP): interaction among precipitation, topography,and land use change

Outline

• Prior Research• Science Questions• Technical Approach• Progress-to-Date• On-going Activities

Surface mining and reclamation practices on the Central Appalachian Plateau (CAP)

• Most “extreme” LCLUC?• Changes in topography, soil properties and

vegetation• Quasi-permanent conversions from natural forests to

pasture (grass) land• Extensive and intensive

Land cover change in eastern U.S. ecosystems (1973-2000)

Loveland et al. (USGCRP) 2003

Prior Research

ObjectiveCompare hydrologic responses of surface mined/reclaimed lands to forested lands at multiple scales

small watershed scaleplot scaleriver basin scale

ROCA Watersheds

EBNR(104 ha)

MAT1(29 ha)

NEF1(3 ha)

Field Hydrologic Measurements

Continuous discharge from Montana flumes at NEF1 and MAT1 since 10/99

Continuous hourly precipitation from Belfort weighing gage since 12/99

Infiltration capacity measurements using double-ring infiltrometers

Annual Runoff

0.210.19

0.17

0.69

0.62

0.26 0.27

0.22

0.60 0.59

0.00

0.15

0.30

0.45

0.60

0.75

2000 2001 2002 2003 2004Water Year

Ann

ual R

unof

f (m

)

TNEF

TMAT

Average annual runoff for Georges Creek at Franklin, MD(1909-present) = 0.39 m

• Similar water balances over very different hydroclimatological conditions

Mean Storm Responses+

-264CENTROID LAG, hr

0.8*0.5

(0.0-1.6)1.3

(0.0-3.6)PEAK RUNOFF,

mm hr-1

4.7*3.1

(0.0-5.1)7.8

(0.0-20.3)TOTAL RUNOFF,

mm

0.14*0.09

(0.00-0.29)0.23

(0.00-0.63)RUNOFF COEF.

DIFFERENCENEF1MAT1

++ Fifteen most intense rain-storms (April – November, 2000-2002) for which data from both sites were available

* Statistically significant difference based on one-tailed, paired t-test (p < 0.01)

Negley and Eshleman 2006

• Storm runoff response at MAT1 is, on average, 2.5 times higher than at NEF1

Negley and Eshleman 2006

• Similar unit hydrographs

0

0.1

0.2

0.3

0.4

0.5

0 1 2 3 4 5 6 7 8 9 10 11 12

Hours

Dis

char

ge

(mm

per

mm

input)

TRIB MAT1 2000TRIB NEF1 2000TRIB MAT1 2001TRIB NEF1 2001

Plot-Scale Measurements: Infiltration Capacity

TRIB NEF-1

0

10

20

30

40

50

60

70

80

90

100

0 2 4 6 8 10 12Time from start of infiltration (min)

Cum

ulat

ive

dept

h of

wat

er in

filtra

ted

(mm

)

Plot 1Plot 2Plot 3

TRIB MAT-1

0.0

0.5

1.0

1.5

2.0

2.5

3.0

0 10 20 30 40 50Time from start of infiltration (min)

Cum

ulat

ive

dept

h of

wat

er in

filtra

ted

(mm

Plot 1Plot 2Plot 3

NEF1 MAT1

• Infiltration capacity at NEF1 is ~30 times greater than at MAT1

ROCA watersheds

10 km

Georges Creek

(187 km2)

Savage River

(127 km2)

Maryland

Savage River forested watersheds

1997 Land Use/Land Cover

Annual Maximum Series (1949-2002)

0.0

1.0

2.0

3.0

4.0

5.0

6.0

7.0

8.0

0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0

Savage River maximum (mm/hr)

Geo

rges

Cre

ek m

axim

um (m

m/h

r)

1996

1954

1950

1980

1986

• Long-term USGS daily data suggest higher peak floods in SR than at GC

Annual Maximum Series

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6

1.8

2.0

Watershed

Inst

anta

neou

s di

scha

rge

(mm

/hr)

Savage River near Barton, MDGeorges Creek at Franklin, MDWills Creek near Cumberland, MD

Annual Maximum Series

0.0

2.0

4.0

6.0

8.0

10.0

1925 1935 1945 1955 1965 1975 1985 1995 2005

Year

Inst

anta

neou

s di

scha

rge

(mm

/hr)

Savage River near Barton, MDGeorges Creek at Franklin, MDWills Creek near Cumberland, MD

Direct Runoff HydrographsHurricane Fran (September 1996)

0

2000

4000

6000

8000

0 12 24 36 48

Hours after 1000 EST on 9/06/1996

Hou

rly d

isch

arge

(cfs

)

Savage River near Barton, MD (total DR = 5.84 cm)

Georges Creek at Franklin, MD (total DR = 4.21 cm)

Adjusted NEXRAD Storm RainfallSeptember 6, 1996

Direct Runoff HydrographsHurricane Fran (September 1996)

0

2000

4000

6000

8000

10000

12000

14000

0 12 24 36 48

Hours after 1000 on 9/06/1996

Hou

rly

disc

harg

e (c

fs)

Savage River near Barton, MD (total DR = 5.84 cm)

Georges Creek at Franklin, MD (total DR = 4.21 cm)

Georges Creek near Franklin, MD* (total DR = 5.38 cm)

+

= ?

TSSR

TSNR

Science Questions

• How has LCLUC affected flooding responses in watersheds in two predominantly-forested, mountainous regions of the world?– Central Appalachian Plateau (CAP) in eastern U.S.

• Focus is on surface mining & reclamation– Carpathian Mountain (CM) region of northern

Eurasia• Focus is on deforestation

• At what scale are flooding responses affected?– Small watershed scale– River basin scale

Central Appalachian Plateau (CAP) Ecoregion

300 0 300 600 Kilometers

N

EW

S

POLAND

HUNGARY

CZECH REPUBLIC

ROMANIA

UKRAINE

SLOVAKIA

AUSTRIA

Carpathian Mountain (CM) Region

Carpathian Mountain (CM) Region:Zakarpatska Oblast in Ukraine

Technical Approach (CAP; CM similar)• Task 1. Quantify ~decadal-scale LCLUC over period 1965-present for 8

gaged river basins using various LANDSAT images (MSS, TM, ETM+) or aerial photography

– Classification requires discrimination of reclaimed mines (grasses and forests)– Rules reference digital surface mining permit data

Technical Approach (CAP; CM similar)• Task 1. Quantify ~decadal-scale LCLUC over period 1965-present for 8

gaged river basins using various LANDSAT images (MSS, TM, ETM+) or aerial photography

– Classification requires discrimination of reclaimed mines (grasses and forests)– Rules reference digital surface mining permit data

• Task 2. Compare flooding responses of 4 pairs of gageddisturbed/reference watersheds as measured by differences in storm runoff and peak discharge for paired low frequency events

– Utilize NEXRAD Stage II/IV (daily rainfall) products in interpretation

Technical Approach (CAP; CM similar)• Task 1. Quantify ~decadal-scale LCLUC over period 1965-present for 8

gaged river basins using various LANDSAT images (MSS, TM, ETM+) or aerial photography

– Classification requires discrimination of reclaimed mines (grasses and forests)– Rules reference digital surface mining permit data

• Task 2. Compare flooding responses of 4 pairs of gageddisturbed/reference watersheds as measured by differences in storm runoff and peak discharge for paired low frequency events

– Utilize NEXRAD Stage II/IV (daily rainfall) products in interpretation• Task 3. Test the ability of calibrated hydrologic models to predict

exacerbation of flooding responses attributable to LCLUC in realwatersheds

– NRCS runoff model: CN = f(LC, HSG)– HSPF (decadal-scale LCLUC implemented via “special actions”); parameter

estimation based on zero-order catchment studies/PEST – RHEESys ?

Technical Approach (CAP; CM similar)

• Task 1. Quantify ~decadal-scale LCLUC over period 1965-present for 8 gaged river basins using various LANDSAT images (MSS, TM, ETM+) or aerial photography

– Classification requires discrimination of reclaimed mines (grasses and forests)– Rules reference digital surface mining permit data

• Task 2. Compare flooding responses of 4 pairs of gageddisturbed/reference watersheds as measured by differences in storm runoff and peak discharge for paired low frequency events

– Utilize NEXRAD Stage II/IV (daily rainfall) products in interpretation• Task 3. Test the ability of calibrated hydrologic models to predict

exacerbation of flooding responses attributable to LCLUC in realwatersheds

– NRCS runoff model: CN = f(LC, HSG)– HSPF (decadal-scale LCLUC implemented via “special actions”); parameter

estimation based on zero-order catchment studies/PEST – RHEESys ?

• Task 4. Exercise the models to examine the effects of spatial scale, spatial pattern, extent, and intensity of LCLUC on flooding responses

– 25-50 years of observations and predictions for 8 basins– “Threshold” LCLUC

Progress-to-Date (CAP: 6 months)

• Obtained LANDSAT images for CAP• Obtained digital mining-permit data from WV, PA, and MD• Completed selection of 8 gaged river basins that the project will focus on

Downloaded historical USGS daily data for 8 river basins• Developed method for de-archiving NWS WSR88D (NEXRAD) Stage II/IV

daily rainfall products• Established 1 new precipitation station at TSSR• Produced a continuous record of runoff/precipitation data for 7 small

catchments in western MD (2005-present)• Preliminary calibration of WIN-HSPF for Georges Creek• PEST operational

On-going Activities

• CAP– Classifying land cover (Wisconsin)– Using intra-annual and inter-annual NDVI variations as an index of mineland

hydrologic function (Wisconsin)– De-archiving NEXRAD products (Wisconsin)– Monitoring rainfall and runoff (AL)– Comparing rainfall-runoff responses and calibrating the models (AL)– Archiving data and technology transfer (AL, Wisconsin)– Upgrading project website/data warehouse (AL): http://roca.al.umces.edu

• CM– Identifying CM collaborators/river basin datasets (ISDU, AL)– Assisting with CM digital river basin & catchment data acquisition (ISDU)– Obtaining imagery for CM river basins (ISDU)– Classifying land cover (Wisconsin, ISDU)– Comparing and modeling flooding responses (AL)