L
Modeling the impact of land cover change and water withdrawals on runoff and N retention in the Ipswich River, MA
Hydrological Modeling
Nitrogen Loading and Removal • DIN inputs to the river network are based on an empirical relationship with land use type and runoff conditions (Wollheim et al. 2008)• Removal based on a two-compartment nutrient spiraling model (Mulholland and DeAngelis 2000) and updated via Stewart et al. (In Review)• The output flux from a river grid cell becomes the input flux to the cell immediately downstream, and so on downstream for the sequence of grid cells to the river mouth
Conclusions• Water withdrawals offset the increase in runoff (+28.5%) due to impervious cover, septic inputs, lawns, and lawn watering at annual time scales.
• Impervious surfaces provide the greatest increase to runoff during annual time periods, and are equally assignificant as lawn watering during summer periods.
• Land cover change generally reduces DIN retention whereas water withdrawals increase apparent DIN retention at basin scales.
• Network models are necessary to better understand the counter-acting forces of urbanization on water and N fluxes to oceans.
University of New Hampshire
Water SystemsAnalysis Group
Introduction:Question: What is the impact of urbanization (specifically impervious cover, septic inputs, lawn surfaces, lawn watering, and water withdrawals) on water quantity and DIN retention in the Ipswich River?
Rationale: 1)Increases in impervious cover and lawn surfaces have been shown to increase the magnitude of runoff (thereby reducing water residence times) whereas water withdrawals can have a counter-acting effect. 2)These hydrological changes have implications for nitrogen retention because discharge and channel hydraulics influence the duration that stream water is in contact with reactive surfaces.
Approach: Use a spatially-distributed, process-based river network DIN removal model that has been populated with established hydrologic, geomorphologic, and biologic parameters to quantify runoff and DIN fluxes under various land cover scenarios.
Long Term Ecological Research
R.J. Stewart1 ([email protected]), W.M. Wollheim1, C. Polsky2, R.G. Pontius2, C.S. Hopkinson3
(1) University of New Hampshire, Durham, NH, (2) Clark University, Worcester, MA, (3) University of Georgia, Athens, GA
• River width and depth are simulated using power law relationships as a function of mean annual Q:
DepthMC = 0.45 * Q0.17
WidthMC = 9.56 * Q0.65
Surface TSRemovalSTS = 1 – exp (-ktτSTS)
Main ChannelRemovalMC = 1 – exp (-Vf/HL)
STS Transfer = (αSTS * AMC * L) / Q
HTS Transfer = (αHTS * AMC * L) / Q
Downstream DIN Flux
Upstream and Local DIN Inputs
Hyporheic TSRemovalHTS = 1 – exp (-ktτHTS)
Conceptual Diagram of a Single River Grid Cell
Model Scenarios:
Ipswich River Network
kt = 0.64 d-1
Vf = 0.08 m d-1
Precipitation
Rgw
ET
Rsurface
Impervious
*20% 80%
Rooting Zone(AWC)
Rstorm Snowpack
Snowmelt
WetlandDetention
Pool
1.
2.
3. 5.
1. Hamon method2. γ * surplus3. (1 – γ) * surplus4. β * GWpool5. Φ * Swpool
* Pellerin et al. 2007
Groundwater Detention Pool4.
Lawn
Rtotal = Rstorm + Rsurface + Rgw
ETPervious
6 Parametersγ = infiltration fractionβ = groundwater releaseΦ = surface storage releaseΩ = soil drying coefficientSnowfall = temp. thresholdSnowmelt = temp. threshold
• A modified version of the Water Balance Model (WBM, Vörösmarty et al. 1998) was applied to the Ipswich River to simulate river discharge on a daily time-step [right panel].
Urban Features
Zarriello 2002 (USGS)Archfield et al. 2009 (USGS)
USGS Ipswich
USGS Middleton
Salem – Beverly Wdl.
PeabodyWdl.
LynnWdl.
River Mouth
USGS GaugesWater Supply Wdls.Commercial Wdls.
Ipswich River Network
0 2 4 km
1.) Pristine: no urban features2.) + Impervious: impervious surfaces only3.) + Septic: imperious surfaces and septic inputs4.) + Lawns and Watering: all of the above plus lawns/watering5.) + Withdrawals: all of the above with water withdrawals
kt = time specific uptake rate [T-1]AMC = cross sectional area of MC [L2]AHTS = cross sectional area of HTS [L2]
Nutrient Spiraling Model TermsASTS = cross sectional area of STS [L2]Vf = nutrient uptake velocity [L T-1]L = grid cell length [L]
α = exchange rate [T-1]HL = hydraulic load [L T-1]τ = ATS / (α AMC) [T]
Results
• Impervious surfaces based on MassGIS data (2007).• Septic inputs based on town data and average US domestic water use (waterfootprint.org, 2001). • Lawns and watering: Data layers provided by MassGIS and Clark U. Reduced soil rooting depths were applied for lawns (AWC = 25) and lawn watering consists of 1 inch, applied once per week (June, July, August).• Water withdrawal schedules and volumes based on USGS data
Septic
Observed and Predicted Runoff at USGS Ipswich
Average Daily Runoff Summary [2000 – 2004]
Period Obs.(mm d-1)
Pristine(mm d-1)
+Imp.(mm d-1)
+Septic(mm d-1)
+Lawns(mm d-1)
+Wdls.(mm d-1)
Annual 1.41(-)
1.41 (-)
1.58 (12.1%)
1.68 (19.1%)
1.81 (28.4%)
1.36 (-3.5%)
Summer
0.84(-)
0.10 (-)
0.31 (310%)
0.41(410%)
0.65 (650%)
0.54 (540%)
Runoff depth (percent change from pristine scenario)
Observation
1.) Pristine (NS = 0.41)2.) + Impervious (NS = 0.44)
4.) + Lawns/Watering (NS = 0.41)5.) + Withdrawals (NS = 0.50)
Legend
3.) + Septic (NS = 0.43)
Scenario 5+ Wdls.
Percent of Total DIN Inputs Removed(Summer Average, 2000-2004)
Scenario 1Pristine
Scenario 2+ Impervious
Scenario 3+ Septic
Perc
ent o
f Tot
al D
IN
Inpu
ts R
emov
ed
Scenario 4+ Lawns/Watering
93.7%
82.0%81.0%
71.5%
77.4%
2000 2001 2002 2003 2004 2005
50
0.005
0.05
0.5
5
Run
off
(mm
/d)
50
0.005
0.05
0.5
5
Run
off
(mm
/d)
0 to 1 m3 s-1
Modeled(Scenario 5)
Observed
1 to 2 m3 s-1 2 to 3 m3 s-1 3 to 4 m3 s-1
Modeled(Scenario 5)
Observed Modeled(Scenario 5)
Observed Modeled(Scenario 5)
Observed
Observed and Predicted DIN Concentrations(binned based on discharge at river mouth)
DIN
(mg
L-1)
Proportion of Local InputsLeaked from Network(Summer Average, 2000-2004)
Scenario 4+
Lawns/Watering
Scenario 1Pristine
Scenario 3+ Septic
Scenario 5+ Withdrawals
Scenario 2+ Impervious
Legend
NationalScience
Foundation