hydromorphology of two urbanizing watershedsase.tufts.edu/water/pdf/melissang_2010symposium.pdf ·...

Hydromorphology of Two Urbanizing Watersheds

Melissa NgMaster’s Candidate

Civil and Environmental EngineeringTufts University

Hydromorphology

The study of the changes in the hydrologic system.

Stationary systems are“natural systems that fluctuate within an unchanging envelope of variability”-Milley 2008

Non-stationarity

“The stationarity assumption has long been compromised by human disturbances” – Milley, 2008

Non-stationarity: Urban Watershed

Aberjona River Flow Duration Curve

0.1

1

10

100

1000

0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00

Exceedence Probability

Flow

(cfs

)

1940-19691970-2007

Aberjona River Streamflow

Increase in flow regime

Understand the impact that climate and human activities have on streamflow

Develop planning tool for dealing with non-stationarity of streamflow

Study Goals

Case Studies: Aberjona River WatershedNeponset River Watershed

Population Changes 1940-2008

0

4

8

12

16

1940 1950 1960 1970 1980 1990 2000 2010Year

Pop

ulat

ion

(tho

usan

ds)

Neponset River watershedAberjona River watershed

Case Studies: Aberjona River WatershedNeponset River Watershed

Aberjona River Watershed Neponset River Watershed

• Traditional approach assumes stationarity• Generates possible realizations of future

streamflowQt = f(Qt-k)

Qt = streamflow at time tQt-k = streamflow at time t-kk = lag

• Water Resources planningWater quality modeling, Reservoir Planning, Irrigation planning,

Hydropower, etc.

Current Planning Tools:Stochastic Streamflow Modeling

Current software packages: HEC-4, SPIGOT, SAMS

Application

Sveinsson, et al., 2003

Aberjona River Flow Duration Curve

0.1

1

10

100

1000

0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00

Exceedence Probability

Flow

(cfs

)

1940-19691970-2007

Aberjona River Streamflow

Increase in flow regime

Neponset River Flow Duration Curve

1

10

100

1000

0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00Exceedence Probability

Stre

amflo

w (c

fs)

1940-19691970-2007

Neponset River Streamflow

Increase in high flowsDecrease in low flows

Causes of Nonstationarity

30

40

50

60

70

1985 1990 1995 2000 2005Year

Prec

ipita

tion

(in) Aberjona River Basin

Neponset River Basin

0

500

1000

1500

2000

1935 1945 1955 1965 1975 1985 1995 2005Year

Lan

d U

se (A

cres

)

Aberjona River basinNeponset River basin

2,000

2,500

3,000

3,500

1985 1990 1995 2000 2005Year

With

draw

al (m

gal)

1,000

1,500

1985 1990 1995 2000 2005

Year

With

draw

al (m

gal

Aberjona River watershed withdrawals Neponset River watershed withdrawals

Precipitation Land Use (Impervious Area)

• Models include exogenous variables• Exogenous variables: Precipitation, land use, water

withdrawal

ln Qt = a0 + Σb1lnQ(t-k) + Σ c 1lnX (t-k) + εt

Qt = streamflow at time tx(t-k) = exogenous variable at time t-kk = lag

Multivariate Non-stationary Stochastic Streamflow model

Model Results: Physical implications

ABERJONA RIVER MODEL

ln Q(m,y) = 11.045 + 0.521ln Q(m-1,y) + 0.225ln Q(m,y-1)

+ 0.872ln P(m,y) - 0.156ln P(m,y-1) – 0.578ln W(m,y)

Monthly streamflow:

-increases with increasing previous streamflow and contemporaneous precipitation

-decreases with increasing water withdrawal

Model Results: Physical implications

NEPONSET RIVER MODEL

ln Q(m,y) = 11.045 + 0.614ln Q(m-1,y) + 0.196ln Q(m,y-1)

+ 0.747ln P(m,y) - 0.175ln P(m,y-1)

-0.917ln LU(y) – 0.761ln W(m,y) – 0.576ln W(m,y-1)

Monthly streamflow:

-increases with increasing previous streamflow and contemporaneous precipitation

-decreases with increasing land use and water withdrawal

Model Results: Interaction of variables improves explanatory power of the model

ABERJONA RIVER MODEL

Variables Adjusted R2

Streamflow 31.50

Streamflow,Precipitation 79.00

Streamflow,Precipitation,Withdrawal 80.90

NEPONSET RIVER MODELVariables Adjusted R2

Streamflow 47.60

Streamflow,Precipitation 79.80

Streamflow,Precipitation,Land use 79.80

Streamflow,Precipitation,Withdrawal 83.40

streamflow,precipitation,Land use, Withdrawal 83.70

1. Historical streamflow is not adequate alone, to forecast future streamflow

2. Precipitation, landuse and water withdrawal cause changes in streamflow

3. Interactions among water balance components are important

4. Stationarity is Dead in an urban environment!

Summary

Thesis committee:

Dr. John Durant (Tufts University, CEE)

Dr. Chuck Kroll (SUNY-ESF)

Dr. Rich Vogel (Tufts University, CEE)

Dr. Jeff Zabel (Tufts University, Dept of Economics)

Acknowledgements