Transcript Water Quality - Swarthmore Watersheds

Professor Art McGarity, Zach Eichenwald
Assisted by Markia Collins, Sophia Richardson,
Richard Scott, Pete Cosfol
The Team (minus Sophie)
Little Crum Creek
 Watershed – the area
from which surface water
drains into a particular
body of water after an
event (rainfall)
1.
Monitor
•
Collect and test water samples to represent stream quality
with data
2. Model
•
Simulate stream flow and pollutant transport to help pinpoint
locations for stormwater management technology
3. Low Impact Development
•
Stormwater management technology and practices to reduce
runoff volume and nonpoint pollution
Collecting
Samples
ISCO Sampler
 Triggered by rain or stream depth, samples at certain
intervals throughout an event
 Stores flow data
 Velocity
 Depth
 Rainfall
 Flow
 Captures up to 24 samples
Gathering
Data
Testing for Pollutants
 Nitrates (NO3) and phosphates (PO4)
 Excess plant nutrients cause algae blooms
(eutrophication) whose decay depletes oxygen
 TSS (Total Suspended Solids)
 Sediments can clog creek beds
 Carry other pollutants, including heavy metals, along
with it
The Tests
 Hach colorimeters quantify pollutant levels by the
amount of absorbance of light
 In the TSS test solids are filtered from a 100 mL sample
and weighed to calculate concentration
Other Tests and Calculations
 Standard Additions
 Turbidity
 Turbidity vs. TSS
 Pollutant Load- an estimation of the total PO4, NO3,
and solids flowing throughout a specific interval
during an event
 L = CQ∆t
 Event Mean Concentration
 Σ(CtQt)
Σ(Qi)
A4 Standard Aditions
y = 0.95x + 0.3
R² = 0.9967
3.5
Concentration (mg/L)
3
2.5
y = 0.811x + 0.276
R² = 0.9639
2
1.5
1
NO3(mg/L)
0.5
PO4(mg/L)
0
-2
-1
0
1
2
3
-0.5
Linear (NO3(mg/L))
Linear (PO4(mg/L))
-1
Standard Addition
Sample
Turbidity (fau)
TSS (mg/L)
NO3 (mg/L)
Abs %
PO4 (mg/L)
Abs%
A1
18
17
2.1
46.28
0.34
79.7
A2
12
-76
2.1
46.3
1.57
35.24
A3
169
533
0.3
89
0.39
76.93
A4
506
853
0.3
89.08
0.42
75.87
A5
280
490
0.4
86.26
0.47
73.1
A6
142
210
0.9
72.47
0.28
83.18
A7
112
147
0.9
70.87
0.29
82.62
The Sonde
 Remotely and continuously monitors:
 pH/ORP
 Dissolved oxygen
 Nitrate
 Conductivity
 Temperature
 Turbidity
 Depth
Why Model?
 We can’t observe the entire
watershed
 We aren’t able to observe all
possible weather events
 The model allows us to see the
response of the watershed to any
possible input, including large
storm events that occur
infrequently
 We can experiment with different
development and storm water
reduction scenarios
Modeling the (Big) Watershed
 Previous work: StormWISE (StormWater Investment
Strategy Evaluator)
 Optimization program developed by Professor Arthur
McGarity


Uses RUNQUAL (Penn State) to develop water quality
parameters
Placement of Best Management Practices (BMPs) optimized
using linear programming techniques.
 Locations for BMPs are not site specific
Zooming in
 Summer work involves developing a more site specific
version of StormWISE
 Water quality and quality are modeled using EPA’s
SWMM (StormWater Management Model)
 Model will be able to identify site specific locations for
BMPs and model the effects of implementation
SWMM
 Dynamic rainfall-runoff simulation
 Can be used for single event or long term simulation of
storm water runoff quantity and quality
 Is used to develop a simulated hydrograph and
pollutograph given rainfall input
 Can model the transport of Nitrate, Phosphate, and TSS
The SWMM Model
Subcatchments
Conduits
Nodes
SWMM Parameters
 SWMM requires (a few) basic parameters about each
subcatchment, node, and conduit
Subcatchments
SCS CN, amount of
impervious surface (%),
slope (%), hydraulic
length
Nodes
Invert elevation, initial
depth, maximum depth
Conduits
Length, roughness, size, type
Basic Hydrology
(SWMM uses this!)
Source: Louisiana DEQ http://www.deq.louisiana.gov/portal/Default.aspx?tabid=1979
Infiltration
 Not all precipitation enters the stream
 Must calculate effective precipitation (precipitation that
is converted to runoff) using an infiltration model
 Many infiltration models have been developed
 One common model is the SCS Method (USDA’s Soil
Conservation Service, now Natural Resource Conservation Service [NRCS])


Assigns a curve number (CN) to many different land use
categories
CN range from 0 – 100 (completely pervious to completely
impervious). Pavement is 98.
SCS Method
 Develops an empirical relationship between effective
precipitation and actual precipitation:
(P  I a )2
Q
P  Ia  S
Ia = initial abstraction (in)
P = the observed precipitation (in)
S = maximum potential retention (in)
Q = effective precipitation (in)
SCS Method
 The CN describes the maximum possible retention,
where
S
1000
 10
CN
 We assume Ia = 0.2S, determined from a study of many
small watersheds by SCS
SCS Curve Number
Source: USDA NRCS TR-55
SCS Curve Number
 Adjustments are made for antecedent moisture
conditions
 CN(II) is for average moisture conditions
 CN(I) and CN(III) are for dry and moist conditions,
respectively
4.2CN(II)
10  0.058CN(II)
23CN(II)
CN(III)
10  0.13CN(II)
CN(I) 
SCS Curve Number
 An analysis of rainfall-runoff relationships for Little
Crum Creek has found a strong correlation between
antecedent moisture and effective precipitation
0.8
R² = 0.9237
0.7
Runoff Fraction
0.6
0.5
0.4
0.3
0.2
0.1
0
0
0.5
1
5-Day Antecedent Rainfall (in)
1.5
2
SCS Curve Number
Source: USDA NRCS TR-55
Problems with SCS
 Developed by USDA for use on agricultural land types
 Attempts to apply the SCS CN method to the Little
Crum Creek watershed result in underestimates of the
effective precipitation
 Not terribly useful for envisioning the effects of
numerous parking lots, storm sewer drainage systems,
etc.
Problems with SCS
 We calculated the theoretical CN(II) for one section of
the watershed to be 88.8
 Underestimates total runoff
 Analysis of observed rain events shows that the actual
CN is closer to 96
 Solutions (Easy and Hard):
 Account for roads (Easy)
 Find a new relationship between S and Ia (Hard)
Other Parameters
 Average impervious percentage, slope, conduit length,
and elevations are determined from GIS analysis
 Elevations are from a Digital Elevation Map (DEM)
 Impervious percent is from a raster dataset that
classifies land use into 5 categories
Land Use and Impervious Percent
Putting it all together
 Model currently built for a section of the watershed
Little Crum
Creek Park
Girard
Close…
70
60
Flow (cfs)
50
40
Simulated
30
Actual
20
10
0
0
0.05
0.1
0.15
0.2
Time (hours)
0.25
0.3
0.35
0.4
Still close…
200000
180000
y = 1.2215x - 69045
R² = 0.9213
160000
Actual Volume (cf)
140000
120000
100000
80000
60000
40000
20000
0
0
50000
100000
150000
Simulated Volume (cf)
200000
250000
Preliminary Results
 Simulated results either underestimate or overestimate
the amount of flow
 This difference is sometimes quite pronounced,
depending on the nature of the storm event
 Simulation results typically exhibit a time lag
What’s Next
 Adjust parameters to get a better fit to actual data
 Add capability to model Nitrate, Phosphate, and TSS
to the model
 Model the implementation of BMPs and LID within
the watershed
Modeling Low Impact
Development and BMPs
 A completed model allows BMP and LID alternatives
to be compared
 A benefit-cost analysis can be performed to determine
the most economically efficient method of reducing
runoff
Types of BMPs/LIDs
 Many ways to reduce runoff, including:
 Green roof (we have one on the roof of Alice Paul and David Kemp)
 Constructed Wetland
 Cisterns and rain barrels
 Permeable pavement surfaces
Preliminary BMP
Recommendations
Site
BMP
Springfield Square, Springfield, PA
Green Roof
Farmhouse Circle, Springfield, PA
Constructed Wetland
See http://watershed.swarthmore.edu/littlecrum for ongoing
recommendations for all four municipalities: Springfield, Swarthmore,
Ridley Township, Ridley Park
Springfield Square
Green Roof on Swarthmore’s Alice Paul Hall
(image: Meghan Whalen)
Farmhouse Circle
Constructed Wetland at Ridley High School
http://watershed.swarthmore.edu