Identifying Changes to Stream Condition
caused by Urbanization
How understanding the responses can improve ecological risk characterization
----------------------------------------
James Coles – USGSThomas Cuffney – USGS
Cornell Rosiu – EPA
Denver
Portland
Dallas
Birmingham Atlanta
Raleigh
Milwaukee
BOSTON
Salt Lake City
Boston = NECBNew England Coastal Basins
Seattle
Sacramento
National Water-Quality Assessment (NAWQA) Program Urban Stream Studies
Setting up the Study Design
• Define homogeneous environmental setting.– minimize natural variability
• Determine watershed indicators of urbanization.– related to human presence
• Select sites to represent urban intensity gradient.– minimal to high levels of urbanization
• Establish consistent sampling reaches– 150 meters, riffles, riparian cover, stable channel
New England Study Sites
BOSTON
Develop an Index to measure
Urban Intensity
High Urban (100)Low Urban (0)
and which also corresponds to stream condition
Original (a priori) Index of Urban Intensity (UII) used 24 landuse and
demographic variables to characterize an urbanizing landscape.
• Infrastructure (Dams, Roadways, TRI sites)
• Land cover (Forest, Developed land)
• Census statistics (Population, Socioeconomic factors)
Perhaps an Index of Urban Intensity (UII)could be derived that is as effective,
but uses fewer than 24 variables
Revised UII
• Road Density– Expansion of Infrastructure
• Percentage of Buffer in Forest– Encroachment along riparian zone
• Percentage of Developed land– Change from natural landscape
• Population Density– Human presence on landscape
Comparing theUrban Intensity Indices (UII)
revised UII (4 variables)
a prioi UII (24 variables)
Comparing responses of Indicator Variables
between versions of the UII
Indicator VariableCategory
MetricsCompared
a priori UIIaverage
correlation
Revised-UIIaverage
correlation
• Water Chemistry 11 0.829 0.830
• Invertebrates 24 0.837 0.845
• Benthic Algae 12 0.756 0.743
• Fish 6 0.804 0.785
• Habitat 7 0.788 0.773
Overall Average 60 0.810 0.807
Impervious Surface Area (ISA) compared to
Revised and the a priori UII
revised - UII
NO
AA
IS
A
a priori UII
r2=0.982 r2=0.948
Impervious Surface Areacompared to
Urban Intensity (revised UII)
0 20 40 60 80 100
Urban Intensity
0
10
20
30
40
Percentage ISAY= 0.37X + 0.91R2 = 0.982
Invertebrate responses to urbanization (Urban Intensity scaled 0 to 100)
A B C D
0 20 40 60 80 100 Urban Intensity
30
40
50
60
70
80
90
100
Tot
al R
ich
0 20 40 60 80 100 Urban Intensity
0
10
20
30
40
EP
T R
ich
0 20 40 60 80 100 Urban Intensity
0
10
20
30
40
%R
ich
No
n In
sect
0 20 40 60 80 100 Urban Intensity
0
10
20
30
%R
ich
Mol
lusk
+C
rust
E F
0 20 40 60 80 100 Urban Intensity
0
10
20
30
40
GC
Ric
h
0 20 40 60 80 100 Urban Intensity
3
4
5
6
7
Ta
xa T
ole
ran
ce In
dex
Response of
Functional Group and Taxa Tolerance
Threshold Response Linear Response
Response of Water Quality to Urbanization
Parameters in WQ index:
- TKN- pH- Conductance- Alkalinity
0 20 40 60 80 100Urban Intensity
0
1
2
3
4
WQ
Inde
x (e
nric
hm
ent)
Improving Ecological Risk Assessments
The value of knowing the urban intensity of a site where contamination occurs.
(Is it urban intensity or the Superfund-site
that is affecting stream condition?)
Eastland Woolen Mill Site A Superfund Site in Corinna, ME.
-1 0 1 2 3 4 CA - Axis 1
-1
0
1
2
3
CA
- A
xis
2
25
2726
22
24
23
Reference(Lake)
Reference(River)
Correspondence Analysis, interpretation of the biplot :
The two sets of reference sites are distinct from the group that has the contaminated sites.
Contaminated
Urban Intensity used to define the expected condition
0 20 40 60 80 100
Urban Intensity
3
4
5
6
7
Ec
olo
gic
al
Co
nd
itio
n
T
axa
To
lera
nce
Ind
ex
This quadrant is the Problem Zone
Large dot represents the Nyanza
Superfund site. The Ecological
Condition metric indicates that
the cleanup was successful.
Developing Applications of the Urban intensity Index
Addressing the details • Variables used in an urban index can be somewhat
esoteric: Socioeconomic factors• The Urban Intensity Index is standardized over the
range of study sites, scaled 0 -100.• Results from a single study of 30 sites may be
tentative. Other data could be used to corroborate the Urban Intensity Index and the ecological responses.
Find regression coefficients to express the UII as equation
y = 12.5x - 13.6
0
20
40
60
80
100
0 5 10
Road Density (road km/ basin km2)
Ro
ad
De
ns
ity
(s
ca
led
)
y = 1.74x - 30.5
0
20
40
60
80
100
0 20 40 60 80
% Buffer (120 m Riparian Zone)
% B
uff
er
(scale
d)
y = 8.09x - 2.0
0
20
40
60
80
100
0 5 10 15
Population Density (# / HA)
Po
p D
en (
scal
ed)
y = 1.53x - 2.2
0
20
40
60
80
100
0 20 40 60 80% Developed Landcover
% D
evel
op
ed (
scal
ed)
Set up the equation to include the Landuse Variables
Urban Intensity = (ROAD*12.5)+(BUFF%*1.74)+(DEV%*1.53)+
(POP *8.09))*0.25)
• ROAD = road density [road length (km) / watershed area (km2)]
• BUFF% = percentage stream buffer not in forest landcover [MRLC level 1]
• DEV% = percentage watershed in developed landcover [MRLC level 1]
• POP = population density, people per hectare [U.S. census data]
Urban Intensity Indexmodeled vs. scaled
0 20 40 60 80 100
Urban intensity (scaled)
0
20
40
60
80
100
120
Urban Intensity modelY = 0.97X + 13.6
Intercept
EPA Region 1
New England WadeableStreamsProject
Regressions between response variables and the urban intensity index
Response RegressionVariable Coefficient (r2)_________________________________________________________________________________________________________
EPT Richness (-) 0.743
% Richness non-Insects (+) 0.837
Taxa Tolerance Index (+) 0.824
WQ Index (+) 0.834
__________________________________________________________
0 20 40 60 80 100Urban intensity
0
1
2
3
4
5
6
Taxa Tolerance(Expected)
Expected stream condition based on Urban Intensity of watershed
0 20 40 60 80 100Urban intensity
0
1
2
3
4
5
6
Taxa Tolerance(Expected)
Expected stream condition based on Urban Intensity of watershed
Acme Chemical Site
• Urban Intensity = 40
0 20 40 60 80 100Urban intensity
0
1
2
3
4
5
6
Taxa Tolerance(Expected)
Expected stream condition based on Urban Intensity of watershed
Acme Chemical Site
• Urban Intensity = 40• Stream Condition = 2.5
0 20 40 60 80 100Urban intensity
0
1
2
3
4
5
6
Taxa Tolerance(Expected)
Expected stream condition based on Urban Intensity of watershed
Acme Chemical Site
• Urban Intensity = 40• Stream Condition = 2.5
CONCLUSIONNo unacceptable ecological risk
0 20 40 60 80 100Urban intensity
0
1
2
3
4
5
6
Taxa Tolerance(Expected)
Expected stream condition based on Urban Intensity of watershed
Acme Chemical Site
• Urban Intensity = 40• Stream Condition = 4.5
CONCLUSIONStream condition may be affected bysite contamination.
Concluding Points
• Study currently being funded by EPA Region I (New England) to use this approach, which is expected to improve Ecological Risk Characterizations.
• Even a general awareness of the urban intensity of a basin is useful in making an interpretation of stream condition at a site.