Effects of urbanization on stream water quality in the ... · Effects of urbanization on stream...
19
HYDROLOGICAL PROCESSES Hydrol. Process. 23, 2860–2878 (2009) Published online 13 August 2009 in Wiley InterScience (www.interscience.wiley.com) DOI: 10.1002/hyp.7373 Effects of urbanization on stream water quality in the city of Atlanta, Georgia, USA † Norman E. Peters* US Geological Survey, Georgia Water Science Center, 3039 Amwiler Rd., Suite 130, Atlanta, Georgia 30360-2824, USA Abstract: A long-term stream water quality monitoring network was established in the city of Atlanta, Georgia during 2003 to assess baseline water quality conditions and the effects of urbanization on stream water quality. Routine hydrologically based manual stream sampling, including several concurrent manual point and equal width increment sampling, was conducted ¾12 times annually at 21 stations, with drainage areas ranging from 3Ð7 to 232 km 2 . Eleven of the stations are real-time (RT) stations having continuous measures of stream stage/discharge, pH, dissolved oxygen, specific conductance, water temperature and turbidity, and automatic samplers for stormwater collection. Samples were analyzed for field parameters, and a broad suite of water quality and sediment-related constituents. Field parameters and concentrations of major ions, metals, nutrient species and coliform bacteria among stations were evaluated and with respect to watershed characteristics and plausible sources from 2003 through September 2007. Most constituent concentrations are much higher than nearby reference streams. Concentrations are statistically different among stations for several constituents, despite high variability both within and among stations. Routine manual sampling, automatic sampling during stormflows and RT water quality monitoring provided sufficient information about urban stream water quality variability to evaluate causes of water quality differences among streams. Fecal coliform bacteria concentrations of most samples exceeded Georgia’s water quality standard for any water-usage class. High chloride concentrations occur at three stations and are hypothesized to be associated with discharges of chlorinated combined sewer overflows, drainage of swimming pool(s) and dissolution and transport during rainstorms of CaCl 2 , a deicing salt applied to roads during winter storms. One stream was affected by dissolution and transport of ammonium alum [NH 4 Al(SO 4 ) 2 ] from an alum-manufacturing plant; streamwater has low pH (<5), low alkalinity and high metals concentrations. Several trace metals exceed acute and chronic water quality standards and high concentrations are attributed to washoff from impervious surfaces. Published in 2009 by John Wiley & Sons, Ltd. KEY WORDS urbanization; stream water quality; nutrients; bacteria; major ions; weathering; monitoring Received 20 October 2008; Accepted 7 May 2009 INTRODUCTION Human activities have the potential to cause changes to the environment. Alteration of the land surface for vari- ous uses including light and heavy industry, urbanization and suburban development has changed water pathways and induced changes to natural processes (Booth and Jackson, 1997; Poff et al., 1997). Human activities can be sources of elements and compounds to the landscape and receiving waters through various pathways, including atmospheric deposition, solid and liquid waste disposal and a combination of diffuse and point-source distribution (Peters and Meybeck, 2000). In addition, types of con- taminants and mechanisms for waste disposal vary with time, economy, technology and culture. Resource man- agement has targeted environmental improvement result- ing from the understanding of deleterious effects of vari- ous elements and compounds on the environment (Peters, 2008). A major deficiency in evaluating stream water quality is that there are no detailed records of the temporal * Correspondence to: Norman E. Peters, US Geological Survey, Georgia Water Science Center, 3039 Amwiler Rd., Suite 130, Atlanta, Georgia 30360-2824, USA. E-mail: [email protected] † This article is a US Government work and is in the public domain in the USA. and spatial variations in waste disposal and other sources of solutes to streams. For example, little data exist for factors such as construction-caused changes, alteration of hydrological pathways and solute transport along those pathways. Many studies have been conducted to assess urban stream water quality (Driver and Troutman, 1989; Make- peace et al., 1995; Mulliss et al., 1996; Deletic, 1998; Duda et al., 1998; Ellis, 2004; Carle et al., 2005; Soller et al., 2005; Brilly et al., 2006; Hudak and Banks, 2006). These studies indicate elevated but highly variable con- centrations of most constituents in urban streams, with several constituents exceeding public health standards. Other urban runoff studies focused on the water quality during storms of specific hydrologic pathways including temporal characteristics, such as first flush of solutes, to assess the cause of elevated concentrations in urban streams. For example, metal concentrations and parti- tioning between dissolved and particulate phases from various sources including roads, parking lots, roofs and atmospheric deposition (Lara-Cazenave et al., 1994a,b; Sansalone and Buchberger, 1997; Gromaire-Mertz et al., 1999; Davis et al., 2001; Mosley and Peake, 2001; Gnecco et al., 2005; Lee et al., 2005; Sansalone et al., 2005; Brown and Peake, 2006), bacteria concentrations Published in 2009 by John Wiley & Sons, Ltd.
Transcript of Effects of urbanization on stream water quality in the ... · Effects of urbanization on stream...
HYDROLOGICAL PROCESSESHydrol. Process. 23, 2860–2878 (2009)Published online 13 August 2009 in Wiley InterScience(www.interscience.wiley.com) DOI: 10.1002/hyp.7373
Effects of urbanization on stream water quality in the cityof Atlanta, Georgia, USA††
Norman E. Peters*US Geological Survey, Georgia Water Science Center, 3039 Amwiler Rd., Suite 130, Atlanta, Georgia 30360-2824, USA
Abstract:
A long-term stream water quality monitoring network was established in the city of Atlanta, Georgia during 2003 to assessbaseline water quality conditions and the effects of urbanization on stream water quality. Routine hydrologically based manualstream sampling, including several concurrent manual point and equal width increment sampling, was conducted ¾12 timesannually at 21 stations, with drainage areas ranging from 3Ð7 to 232 km2. Eleven of the stations are real-time (RT) stationshaving continuous measures of stream stage/discharge, pH, dissolved oxygen, specific conductance, water temperature andturbidity, and automatic samplers for stormwater collection. Samples were analyzed for field parameters, and a broad suite ofwater quality and sediment-related constituents. Field parameters and concentrations of major ions, metals, nutrient species andcoliform bacteria among stations were evaluated and with respect to watershed characteristics and plausible sources from 2003through September 2007. Most constituent concentrations are much higher than nearby reference streams. Concentrations arestatistically different among stations for several constituents, despite high variability both within and among stations. Routinemanual sampling, automatic sampling during stormflows and RT water quality monitoring provided sufficient informationabout urban stream water quality variability to evaluate causes of water quality differences among streams. Fecal coliformbacteria concentrations of most samples exceeded Georgia’s water quality standard for any water-usage class. High chlorideconcentrations occur at three stations and are hypothesized to be associated with discharges of chlorinated combined seweroverflows, drainage of swimming pool(s) and dissolution and transport during rainstorms of CaCl2, a deicing salt applied toroads during winter storms. One stream was affected by dissolution and transport of ammonium alum [NH4Al(SO4)2] froman alum-manufacturing plant; streamwater has low pH (<5), low alkalinity and high metals concentrations. Several tracemetals exceed acute and chronic water quality standards and high concentrations are attributed to washoff from impervioussurfaces. Published in 2009 by John Wiley & Sons, Ltd.
KEY WORDS urbanization; stream water quality; nutrients; bacteria; major ions; weathering; monitoring
Received 20 October 2008; Accepted 7 May 2009
INTRODUCTION
Human activities have the potential to cause changes tothe environment. Alteration of the land surface for vari-ous uses including light and heavy industry, urbanizationand suburban development has changed water pathwaysand induced changes to natural processes (Booth andJackson, 1997; Poff et al., 1997). Human activities canbe sources of elements and compounds to the landscapeand receiving waters through various pathways, includingatmospheric deposition, solid and liquid waste disposaland a combination of diffuse and point-source distribution(Peters and Meybeck, 2000). In addition, types of con-taminants and mechanisms for waste disposal vary withtime, economy, technology and culture. Resource man-agement has targeted environmental improvement result-ing from the understanding of deleterious effects of vari-ous elements and compounds on the environment (Peters,2008). A major deficiency in evaluating stream waterquality is that there are no detailed records of the temporal
* Correspondence to: Norman E. Peters, US Geological Survey, GeorgiaWater Science Center, 3039 Amwiler Rd., Suite 130, Atlanta, Georgia30360-2824, USA. E-mail: [email protected]† This article is a US Government work and is in the public domain inthe USA.
and spatial variations in waste disposal and other sourcesof solutes to streams. For example, little data exist forfactors such as construction-caused changes, alteration ofhydrological pathways and solute transport along thosepathways.
Many studies have been conducted to assess urbanstream water quality (Driver and Troutman, 1989; Make-peace et al., 1995; Mulliss et al., 1996; Deletic, 1998;Duda et al., 1998; Ellis, 2004; Carle et al., 2005; Solleret al., 2005; Brilly et al., 2006; Hudak and Banks, 2006).These studies indicate elevated but highly variable con-centrations of most constituents in urban streams, withseveral constituents exceeding public health standards.Other urban runoff studies focused on the water qualityduring storms of specific hydrologic pathways includingtemporal characteristics, such as first flush of solutes,to assess the cause of elevated concentrations in urbanstreams. For example, metal concentrations and parti-tioning between dissolved and particulate phases fromvarious sources including roads, parking lots, roofs andatmospheric deposition (Lara-Cazenave et al., 1994a,b;Sansalone and Buchberger, 1997; Gromaire-Mertz et al.,1999; Davis et al., 2001; Mosley and Peake, 2001;Gnecco et al., 2005; Lee et al., 2005; Sansalone et al.,2005; Brown and Peake, 2006), bacteria concentrations
Published in 2009 by John Wiley & Sons, Ltd.
EFFECTS OF URBANIZATION ON STREAM WATER QUALITY 2861
in non-point and point source runoff (Duda et al., 1998;Ellis, 2004; O’Keefe et al., 2005; Garcia-Armisen andServais, 2007), water quality of combined sewer systems(Gupta and Saul, 1996; Barco et al., 2008) and waterquality of runoff from golf courses (Mallin and Wheeler,2000).
The study herein focuses on stream water quality of theAtlanta region. Previous studies in the region providedsome background on anticipated water quality character-istics. Rose (2007) assessed the impacts of urbanizationon base flow hydrochemistry of the Chattahoochee RiverBasin from a synoptic sampling of 35 streams duringApril and May 2005. The study results indicate thaturbanized watersheds were associated with higher con-centrations of sulfate, chloride, alkalinity (ALK), potas-sium and sodium than rural watersheds. Furthermore,a high correlation between urbanization and sodium-potassium-chloride concentrations was attributed to awidespread input of electrolytes present in human wasteand wastewater.
A study with similar but more extensive data collec-tion in the Atlanta region than that of the Rose (2007)study was conducted by Gregory and Calhoun (2007)during the 2003 water year (WY2003, 1 October, 2002,through 30 September, 2003). Water samples were col-lected during spring and summer base flow at 30 stationsand bimonthly at 10 of the 30 stations for nutrients, pesti-cides, chloride, sulfate, dissolved and particulate organiccarbon, particulate nitrogen, suspended sediment (SS),turbidity (TURB) and Escherichia coli (EC). In addi-tion, hydrology, stream habitat and algal, macroinverte-brate and fish communities were evaluated. Gregory andCalhoun (2007) reported that specific conductance (SC)and concentrations of chloride, sulfate and pesticidesincreased as urbanization increased. However, nutrientconcentrations did not correlate with urbanization, butnegatively correlated with percentage forest cover of thewatersheds. Even with this relatively short study period,Gregory and Calhoun (2007) reported statistically sig-nificant relations between urbanization and flashiness ofhydrographs and shorter periods of high flow.
Stormflow was also reported to be flashy in urbanwatersheds compared with agricultural or forested water-sheds in the West Georgia Piedmont by Schoonoveret al. (2006). These results are consistent with hydro-logic analysis of seven north Georgia watersheds for a38-year period of record, including a highly urbanizedAtlanta watershed, by Rose and Peters (2001). In addi-tion to increased flashiness associated with urbanization,the base flow recession constant (a measure of the rate atwhich groundwater is released from storage to the streamas base flow from May to September) and base flow weremuch less than for rural streams. Rose and Peters (2001)also reported a decrease in groundwater levels for wellsconcurrent with increasing urbanization and hypothesizedthat the decrease was caused by decreased groundwaterrecharge in the urban watersheds due to increased imper-viousness and related rapid storm runoff.
The water quality studies of watersheds of the Atlantaregion by Rose (2007) and Gregory and Calhoun (2007)targeted chronic water quality conditions, i.e. during baseflow. The studies also included watersheds that covereda broad range of land use from rural to highly urban.The results clearly indicate effects of urbanization onwater quality during base flow, but did not show waterquality among highly urbanized watersheds over a largestreamflow range.
Finally, Rose et al. (2001) evaluated dissolved tracemetal concentrations during 1998 and 1999 of streamsin the Atlanta region. Rose et al. (2001) also sampledurban street, suburban street and parking lot runoff.The highest zinc concentrations were collected from anurban street runoff having median concentrations 2 ordersof magnitude higher than base flow from non-urbanstreams. Zinc concentration in urban stream stormflowwas higher than base flow and each was higher thannon-urban stream base flow. Rose et al. (2001) analyzedconcentration hysteresis during storms of several urbanstreams and assessed a zinc mass balance, concludingthat much of the zinc mobilized during storms, i.e. fromimpervious traffic areas, is lost and likely adsorbed tosediment.
The objectives of this study are the following: (1) todetermine the baseline water quality characteristics ofcity of Atlanta (COA) streams over a wide dischargerange; (2) to evaluate water quality differences amongCOA sampling stations with respect to field parametersand concentrations of major ions, minor and trace metals,nutrients and bacteria; (3) to compare COA stream waterquality for a subset of parameters to those at nearby refer-ence stations including a small (0Ð4 km2) relatively undis-turbed forested watershed and a larger (10 km2) low-density residential watershed; and (4) to evaluate someof the water quality differences among and within COAsampling stations with respect to land-use characteristicsand other environmental factors including streamflow asa means of identifying potential causes of impairment.Even in the best case where sources are clearly defined,it is problematic to quantitatively link source contribu-tions to stream water quality. However, the data analy-sis of stream water quality variability using three typesof monitoring approaches identified some major differ-ences among streams, which indicated links to primarycausative factors.
Study area characteristics and stream water qualitymonitoring program
The COA encompasses 343 km2 and is approxi-mately centred within the ten-county Atlanta RegionalCommission (ARC) planning area (4780 km2). All butfour of the Long-Term Watershed Monitoring Program(LTWMP) sampling stations are within the COA bound-ary (Figure 1). The average elevation of the COA is320 m (NAD27), which is the highest average eleva-tion of any major city east of Denver. The northern andwestern parts of COA drain to the Chattahoochee River,which drains to the Gulf of Mexico. The southeastern part
Published in 2009 by John Wiley & Sons, Ltd. Hydrol. Process. 23, 2860–2878 (2009)DOI: 10.1002/hyp
2862 N. E. PETERS
P
Weather station4 LUL15 WOO16 PEA57 PEA48 PEA39 PEA2
10 PEA1
11 PRO212 PRO313 PRO1
14 SAN1
15 UTO316 UTO217 UTO1
18 INT1
19 SOU320 SOU221 SOU1
1 NAN3
3 NAN12 NAN2
22 PMRW23 MTNC
Base from U.S. Geological 1:24,000-scale digital data
Riv
er
Chattahooch
ee
South River
City of
Metropolitan
Area of map above
Stream-sampling stations
Combined sewer overflow/regulator
Routine with continuous dischargeand water quality (real time, RT)
Routine
Reference
Subwatershed boundary forroutine sampling stationAtlanta city limit
Watershed
Nancy Creek
Peachtree Creek
Proctor Creek
Sandy Creek
Intrenchment Creek
South River
Utoy Creek
Major River Basin Boundary
EXPLANATION
Reference Streams
1
3
2
4
56
7
8910
11
1213
14
15
16
17
181920
21
22
23
Chattachoochee River Basin
Altamaha River Basin
0 6MILES
6 KILOMETERS
N3
0 3
Atlanta
Atlanta
Figure 1. City of Atlanta, Georgia (USA) and reference stream monitoring stations
of COA drains to the South River and subsequently tothe Ocmulgee River, which drains to the Atlantic Ocean(Figure 1; Table I).
Urbanized areas, forest and agriculture are the domi-nant land use and land cover in the Atlanta region, whichis undergoing rapid development and population growth.Atlanta has been one of the fastest growing metropolitanarea in the nation. The ten-county Atlanta metropolitanarea is a sprawling urbanized and sub-urbanized complexin which the population has increased from 1Ð5 million in1970 to 4Ð1 million people in 2008 with a 19Ð5% popu-lation increase from 2000 to 2008 (ARC, 2008). Most ofthe population increase has been in the counties aroundthe COA, but population density also is very high in COAwith an average of 1510 people km�2. Urban land use inthe COA watersheds ranges from 69 to 96% (Table II),and most of the remaining land use is either low-densityresidential or forest (ARC, 2004).
The COA is in the Piedmont physiographic provinceand is underlain by medium- to high-grade metamor-phic and igneous rocks (Higgins and Crawford, 2006).On average, the study area receives 1340 mm precip-itation annually, generally distributed uniformly duringthe year, at least with respect to monthly totals (Roseand Peters, 2001). From April through October, rain-storms are primarily convective (high intensity and shortduration), with intermittent occurrence through Novem-ber and beginning in March. During the remainder ofthe year, precipitation is dominated by synoptic-scaleweather systems (low intensity and long duration). Therunoff coefficient (RC; runoff as a percentage of pre-cipitation) of suburban to urban watersheds ranges fromapproximately 30 to 40%; the highest RCs are in water-sheds with the highest percentages of impervious area(Rose and Peters, 2001). Stream base flow varies season-ally; the lowest flow occurs during the summer growingseason when evapotranspiration is the highest and the
highest base flow occurs during the winter dormant sea-son. Furthermore, the distribution of hourly streamflowat a long-term gauging station, which was one of theCOA stream stations (PEA2), was evaluated during thedesign of the LTWMP (Figure 2). The distribution ofhourly streamflow indicates that stormflow is flashy withvery short times to peak (Rose and Peters, 2001) and thatmajor storms typically occur during non-working hoursfrom 16 : 00 through the night until about 6 : 00. Figure 2also indicates that stormflows during summer are almostentirely affected by convective storms, and most of thehighest stormflows occur during nighttime.
METHODS
The Clean Water Atlanta Program developed a LTWMPin December 2002 (Horowitz and Hughes, 2006;Horowitz et al., 2007; COA, 2009) and the US Geolog-ical Survey (USGS) initiated stream water quality sam-pling during 2003. The LTWMP consists of a networkof 21 regularly sampled stream stations with watershedsranging in size from 3Ð7 to 232 km2. For station locationsand station name designation, see Figure 1 and Table I.The analyses include the following: (1) field parame-ters (temperature [T], SC, pH, dissolved oxygen [DO]and TURB); (2) major ions (calcium [Ca], magnesium[Mg], sodium [N], potassium [K], chloride [Cl], sulfate[SO4], silica [SiO2] and ALK); (3) nutrients (ammo-nium [NH4N], nitrite [NO2N], nitrate [NO3N], phosphate[PO4P] and total phosphorus [TP]); (4) indicator coliformbacteria (fecal coliform [FC], total coliform [TC] and EC,reported as most probable number (MPN) of coloniesdl�1); (5) minor and trace metals (aluminum [Al], cop-per [Cu], cadmium [Cd], hexavalent chromium [Cr], iron[Fe], lead [Pb], manganese [Mn] and zinc [Zn]); and SS.Laboratory analytical detection limits and method report
Published in 2009 by John Wiley & Sons, Ltd. Hydrol. Process. 23, 2860–2878 (2009)DOI: 10.1002/hyp
EFFECTS OF URBANIZATION ON STREAM WATER QUALITY 2863
Tabl
eI.
Cit
yof
Atl
anta
and
refe
renc
est
ream
wat
erqu
alit
ym
onit
orin
gst
atio
ns,
Geo
rgia
Maj
orri
ver
basi
nSt
ream
wat
ersh
edS
tati
onde
sign
atio
nU
SG
eolo
gica
lSu
rvey
stat
ion
num
ber
Stre
amna
me
and
stat
ion
loca
tion
Dra
inag
ear
ea(k
m2)
Ele
vati
on(m
abov
ese
ale
vel)
Rel
ief
(m)
Ave
rage
Min
imum
Max
imum
Cha
ttaho
oche
eN
ancy
Cre
ekN
AN
302
336
360
Nan
cyC
reek
atR
icke
nbac
ker
Dri
ve.
67Ð1
293Ð5
235Ð4
352Ð5
117Ð1
Cha
ttaho
oche
eN
ancy
Cre
ekN
AN
202
336
380
Nan
cyC
reek
atR
anda
llM
illR
oad
88Ð2
293Ð5
235Ð4
352Ð5
117Ð1
Cha
ttaho
oche
eN
ancy
Cre
ekN
AN
102
336
410
Nan
cyC
reek
atW
est
Wes
ley
Roa
d95
Ð429
1Ð122
9Ð835
2Ð512
2Ð7C
hatta
hooc
hee
Peac
htre
eC
reek
LU
L1
0233
622
8L
ullw
ater
Cre
ekat
Lul
lwat
erPa
rkw
ay3Ð7
295Ð8
269Ð8
319Ð3
49Ð4
Cha
ttaho
oche
ePe
acht
ree
Cre
ekW
OO
102
336
313
Woo
dall
Cre
ekat
DeF
orrs
Ferr
yR
oad
6Ð726
6Ð923
5Ð130
2Ð667
Ð6C
hatta
hooc
hee
Peac
htre
eC
reek
PEA
502
336
240
Sout
hFo
rkPe
acht
ree
Cre
ek,
John
son
Roa
d70
Ð829
9Ð324
6Ð734
1Ð194
Ð4C
hatta
hooc
hee
Peac
htre
eC
reek
PEA
402
336
120
Nor
thFo
rkPe
acht
ree
Cre
ek,
Buf
ord
Hig
hway
90Ð2
293Ð1
245Ð5
341Ð6
96Ð1
Cha
ttaho
oche
ePe
acht
ree
Cre
ekPE
A3
0233
626
7Pe
acht
ree
Cre
ekat
Pied
mon
tR
oad
176Ð9
293Ð6
240Ð6
341Ð9
101Ð3
Cha
ttaho
oche
ePe
acht
ree
Cre
ekPE
A2
0233
630
0Pe
acht
ree
Cre
ekat
Nor
thsi
deD
rive
220Ð5
290Ð5
232Ð7
341Ð9
109Ð2
Cha
ttaho
oche
ePe
acht
ree
Cre
ekPE
A1
0233
631
1Pe
acht
ree
Cre
ekat
Boh
ler
Roa
d22
7Ð728
9Ð622
8Ð934
1Ð911
3Ð0C
hatta
hooc
hee
Peac
htre
eC
reek
PRO
202
336
517
Proc
tor
Cre
ekat
Hor
tens
eW
ay19
Ð829
0Ð923
0Ð332
3Ð693
Ð3C
hatta
hooc
hee
Peac
htre
eC
reek
PRO
30
2336
521
8Pr
octo
rC
reek
Tri
bat
Spri
ngR
oad
7Ð827
8Ð723
5Ð332
2Ð186
Ð8C
hatta
hooc
hee
Peac
htre
eC
reek
PRO
102
336
526
Proc
tor
Cre
ekat
Jack
son
Park
way
36Ð2
282Ð2
252Ð7
323Ð6
69Ð9
Cha
ttaho
oche
eSa
ndy
Cre
ekSA
N1
0233
664
4Sa
ndy
Cre
ekat
Bol
ton
Roa
d8Ð8
269Ð6
236Ð1
301Ð0
64Ð9
Cha
ttaho
oche
eU
toy
Cre
ekU
TO
302
336
658
Nor
thU
toy
Cre
ekat
Peyt
onR
oad
17Ð2
297Ð0
253Ð1
321Ð8
68Ð7
Cha
ttaho
oche
eU
toy
Cre
ekU
TO
202
336
706
Sout
hU
toy
Cre
ekat
Chi
ldre
ssD
rive
24Ð
289Ð7
244Ð7
324Ð7
79Ð9
Cha
ttaho
oche
eU
toy
Cre
ekU
TO
102
336
728
Uto
yC
reek
atG
reat
Sout
hwes
tPa
rkw
ay89
Ð27
7Ð222
4Ð932
4Ð799
Ð7O
cmul
gee
Sout
hR
iver
INT
102
203
700
Intr
ench
men
tC
reek
atC
onst
itut
ion
Ave
nue
27Ð4
288Ð8
234Ð8
320Ð5
85Ð7
Ocm
ulge
eSo
uth
Riv
erSO
U3
0220
360
3So
uth
Riv
erat
Spri
ngda
leR
oad
6Ð129
7Ð925
9Ð032
3Ð164
Ð1O
cmul
gee
Sout
hR
iver
SOU
202
203
620
Sout
hR
iver
atM
acon
Dri
ve13
Ð528
9Ð324
6Ð332
3Ð176
Ð8O
cmul
gee
Sout
hR
iver
SOU
102
203
655
Sout
hR
iver
atFo
rest
Park
Roa
d58
Ð828
3Ð123
7Ð232
3Ð185
Ð9O
cmul
gee
Sout
hR
iver
PMR
W02
203
970
Mou
ntai
nC
reek
Tri
buta
ryat
Pano
laM
ount
ain
Stat
ePa
rk0Ð4
224
279
55O
cmul
gee
Sout
hR
iver
MT
NC
0220
3967
5M
ount
ain
Cre
ekne
arPa
nola
Mou
ntai
nSt
ate
Park
10Ð
210Ð3
Published in 2009 by John Wiley & Sons, Ltd. Hydrol. Process. 23, 2860–2878 (2009)DOI: 10.1002/hyp
2864 N. E. PETERS
Tabl
eII
.Pe
rcen
tage
land
use
ofba
sin
area
for
city
ofA
tlan
tast
ream
wat
erqu
alit
ym
onit
orin
gst
atio
ns(d
eriv
edfr
omA
tlan
taR
egio
nal
Com
mis
sion
,20
04)
Lan
d-us
ety
peW
ater
shed
Nan
cyC
reek
Peac
htre
eC
reek
Proc
tor
Cre
ekSa
ndy
Cre
ekU
toy
Cre
ekSo
uth
Riv
er
NA
N1
NA
N2
NA
N3
LU
L1
PEA
4PE
A5
PEA
3PE
A2
PEA
1W
OO
1PR
O2
PRO
3PR
O1
SAN
1U
TO
3U
TO
2U
TO
1SO
U3
SOU
2SO
U1
INT
1
Urb
anC
omm
erci
al14
Ð515
Ð416
Ð98Ð6
14Ð7
10Ð7
12Ð9
14Ð8
14Ð5
1Ð613
Ð56Ð6
9Ð36Ð5
5Ð36Ð3
4Ð512
Ð216
Ð611
Ð814
Ð5In
dust
rial
and
com
mer
cial
1Ð11Ð1
1Ð50
7Ð26Ð8
6Ð96Ð8
6Ð652
Ð812
Ð20
7Ð50
3Ð53Ð2
4Ð811
Ð19Ð4
12Ð1
6Ð2
Indu
stri
al2Ð2
2Ð43Ð1
00Ð8
00Ð4
0Ð60Ð5
00
00Ð2
00Ð1
00Ð2
12Ð9
6Ð71Ð7
2Ð1In
stit
utio
nal,
exte
nsiv
e0Ð9
1Ð01Ð3
01Ð4
4Ð42Ð5
21Ð9
00
1Ð92
04Ð2
7Ð94Ð0
00
00
Inst
itut
iona
l,in
tens
ive
4Ð23Ð3
3Ð90Ð6
2Ð63Ð2
2Ð73Ð5
3Ð50
9Ð44Ð2
6Ð93Ð6
0Ð81Ð6
1Ð25Ð7
3Ð53Ð5
5Ð1
Lim
ited
acce
sshi
ghw
ay2Ð9
2Ð73Ð2
03Ð1
1Ð92Ð7
2Ð82Ð9
00Ð9
1Ð60Ð8
10Ð4
0Ð61Ð4
1Ð50Ð5
6Ð84Ð0
3Ð4
Res
iden
tial
,m
ediu
mde
nsit
y52
Ð454
Ð050
Ð826
Ð652
Ð747
Ð849
Ð745
Ð946
Ð75Ð0
25Ð9
60Ð
32Ð5
60Ð2
27Ð7
52Ð4
39Ð0
9Ð818
Ð821
Ð426
Ð6
Res
iden
tial
,hi
ghde
nsit
y0Ð9
1Ð01Ð3
410
5Ð43Ð3
4Ð84Ð6
021
Ð50
11Ð9
031
Ð92Ð1
6Ð731
Ð623
Ð116
Ð413
Ð4
Res
iden
tial
,m
obil
eho
mes
00
00
00
00
00Ð7
00
00Ð7
00
01Ð0
0Ð70Ð3
0
Res
iden
tial
,m
ulti
-uni
t7
7Ð38
5Ð710
Ð97Ð7
109Ð5
9Ð64Ð3
5Ð25Ð6
5Ð13Ð7
1Ð17Ð6
4Ð66Ð3
4Ð94Ð4
5Ð2
Tra
nsit
iona
l0Ð6
0Ð70Ð9
00Ð1
0Ð40Ð3
0Ð50Ð5
0Ð70Ð5
02Ð3
00
0Ð40Ð9
0Ð80Ð4
1Ð50Ð7
Tra
nspo
rtat
ion
0Ð91
1Ð31Ð3
2Ð10Ð6
1Ð41Ð2
1Ð214
Ð11Ð9
05Ð6
0Ð30Ð4
1Ð20Ð8
2Ð41Ð4
1Ð01Ð8
Urb
an,
othe
r0Ð4
0Ð40Ð5
4Ð60Ð6
0Ð30Ð6
0Ð80Ð8
4Ð43Ð4
02Ð4
1Ð80Ð1
1Ð40Ð8
00
1Ð29Ð6
Urb
ansu
btot
al88
Ð090
Ð092
Ð688
Ð396
Ð289
Ð293
Ð393
Ð393
Ð382
Ð984
Ð579
Ð886
Ð487
Ð275
Ð685
Ð469
Ð094
Ð492
Ð379
Ð388
Ð5
Published in 2009 by John Wiley & Sons, Ltd. Hydrol. Process. 23, 2860–2878 (2009)DOI: 10.1002/hyp
EFFECTS OF URBANIZATION ON STREAM WATER QUALITY 2865
Tabl
eII
.(C
onti
nued
)
Lan
d-us
ety
peW
ater
shed
Nan
cyC
reek
Peac
htre
eC
reek
Proc
tor
Cre
ekSa
ndy
Cre
ekU
toy
Cre
ekSo
uth
Riv
er
NA
N1
NA
N2
NA
N3
LU
L1
PEA
4PE
A5
PEA
3PE
A2
PEA
1W
OO
1PR
O2
PRO
3PR
O1
SAN
1U
TO
3U
TO
2U
TO
1SO
U3
SOU
2SO
U1
INT
1
Oth
er Agr
icul
tura
l0Ð4
0Ð50
00
0Ð20Ð1
0Ð10Ð1
00
00
00
00
00
0Ð10
Cem
eter
y0
00
00
0Ð80Ð3
0Ð30Ð3
7Ð30
7Ð51Ð8
011
Ð90Ð4
2Ð40
0Ð40Ð4
0Ð9Fo
rest
2Ð92Ð8
3Ð32Ð0
2Ð45Ð6
3Ð93Ð5
3Ð48Ð9
3Ð110
Ð68Ð7
9Ð99Ð5
9Ð720
Ð05Ð1
6Ð615
Ð17Ð9
Gol
fco
urse
s1Ð4
1Ð52
7Ð20
0Ð90Ð4
0Ð40Ð5
00
00
00
1Ð60Ð4
00
1Ð20
Park
land
,ex
tens
ive
1Ð11Ð2
02Ð6
0Ð81Ð6
1Ð01Ð4
1Ð50
00
0Ð82Ð4
2Ð62Ð0
2Ð50
01Ð7
0
Park
s0Ð7
0Ð71Ð0
00Ð3
0Ð20Ð2
0Ð30Ð3
01Ð2
2Ð11Ð1
0Ð50Ð4
0Ð70Ð5
00
0Ð91Ð9
Qua
rrie
s0
00
00
00
00
0Ð31Ð1
01Ð3
00
00
00
00
Res
ervo
irs
0Ð30Ð3
0Ð40
0Ð20
0Ð10Ð2
0Ð20
00
00
00
0Ð40
00Ð1
0Ð2R
esid
enti
al,
low
dens
ity
5Ð12Ð7
0Ð70
01Ð2
0Ð50Ð4
0Ð40
00
00
00Ð3
4Ð50Ð5
0Ð80Ð7
0Ð7
Wet
land
0Ð00Ð1
0Ð10
0Ð10Ð2
0Ð10Ð1
0Ð10
00
00
00
0Ð30
00Ð6
0
Wat
ersh
edar
ea(k
m2)
95Ð4
88Ð2
67Ð1
3Ð790
Ð270
Ð817
6Ð922
0Ð522
7Ð76Ð7
19Ð8
7Ð836
Ð28Ð8
17Ð2
2489
6Ð113
Ð558
Ð827
Ð4
Published in 2009 by John Wiley & Sons, Ltd. Hydrol. Process. 23, 2860–2878 (2009)DOI: 10.1002/hyp
2866 N. E. PETERS
10-2
10-1
100
101
102
Run
off (
mm
d-1
)
Hour of Day0...........12..........23
Winter Spring Summer Fall
Figure 2. Seasonal distribution of hourly discharge for a 5 year periodof record at PEA2 from 1997 to 2001, i.e., prior to initiation of theCOA stream water quality sampling. Seasons are consecutive three month
periods and Winter includes December, January, and February
limits have improved (are lower) for most nutrient speciessince the study began. Also, FC analyses were ceasedbeginning in March 2007 because of program changes.Eleven of the sampling stations were instrumented withreal-time (RT) water quality monitors for field parameters(T, SC, pH, DO and TURB) using standard USGS pro-cedures for operating and maintaining continuous waterquality monitors (Wagner et al., 2006). These 11 stationsalso were instrumented with continuous stage monitorsfor estimating discharge (Q) using standard USGS pro-cedures for measuring and computing discharge (Rantz,1982). Ten RT stations were instrumented during 2003.Station WOO1 was instrumented in June 2005; UTO3was deactivated at the end of 2006; and the RT equip-ment moved from UTO3 to SOU3 (only UTO3 shownas RT station on Figure 1). Sampling consisted of equalwidth increment (EWI) and manual point or grab (GRAB)sampling at each station (USGS, variously dated) andautomated Teledyne ISCO Model 6712FR refrigeratedpump sampling (ISCO) from a point intake at the RTstations for stormflow sampling.
Most stations were manually sampled at least 30 timessince the study began and most of the RT stationswere automatically sampled more than 150 times. Thesamples were processed and preserved using standardUSGS water quality field sampling protocols (USGS,variously dated) and submitted to USGS laboratories foranalyses following techniques and procedures describedby Fishman and Friedman (1989). To monitor streamoutflow from the COA, seven stations (NAN1, PEA1,PRO1, SAN1, UTO1, INT1 and SOU1) are located at themost downstream location of the seven study watersheds(Figure 1, Table I). To monitor stream inflow to the COA,three stations (NAN3, PEA4 and PEA5) are located at themost upstream location near the COA boundary; thesetributary streams have substantial drainage areas outsidethe city limits (Figure 1).
Two nearby stations within the South River Basinto the southeast and outside the COA were used asreferences for COA water quality (Figure 1, Table I). Onereference station, Mountain Creek Tributary in the Panola
Mountain Research Watershed (PMRW), is in a small(0Ð41 km2), relatively undisturbed, forested watershed inthe Panola Mountain State Conservation Park, which hasbeen monitored since 1985. The other reference station,Mountain Creek (MTNC), is in a larger (10 km2) low-density residential watershed, which has been monitoredsince 1991. Stream sampling was conducted weekly, andin PMRW, automatic sampling augmented the weeklymanual grab sampling during rainstorms, similar to theRT COA stations. For the reference stations, water qualitysampling and sample processing differed from those ofthe COA streams. At PMRW and MTNC, the manual(GRAB) samples were collected at a point and at PMRW,the automatic samples (ISCO) were collected at the endof the sampler tubing located in the pool above the weirplate. Samples were not filtered, but were chilled afterretrieval during transport and then stored in a refrigeratorprior to analysis. The concentrations of the major ionsand nutrients at PMRW were compared among filtered(0Ð1 and 0Ð45 µm) and non-filtered samples during thefirst couple of years of the study (1985–1986) and nostatistically significant differences were detected amongthem. The laboratory analytical procedures were thesame as those used for the COA samples. Some of theCOA water quality parameters were not analyzed at thereference stations, as noted by the absence of data forthese stations in the Results and Discussion.
RESULTS AND DISCUSSION
Many water quality data were collected from COAstreams. The results and discussion have been organizedto help simplify and focus the presentation. The hydro-logic characteristics are presented first, which is followedby the stream water quality. General characteristics of thestream water quality are compared among stations, and todo this, the water quality constituents are separated intogroups including major ions, nutrients, bacteria and DOand metals. Where differences among stations occur for aconstituent or group of constituents, other constituents areincluded as the combination of behaviours alludes to pro-cess or cause and effect. The last section on water qualityaddresses the general effects of streamflow on water qual-ity and does not address groups of constituents per se.
Rainfall and runoff characteristics
Three long-term weather stations are located in or nearthe COA (Figure 1). Two of the stations are operatedby the National Weather Service (NWS). One NWS sta-tion with continuous data since WY1931 is located tothe south at Atlanta Hartsfield-Jackson International Air-port and the other NWS station (Atlanta-Bolton) withgenerally continuous data since WY1957 is located tothe west of COA (http://www7.ncdc.noaa.gov/CDO/cdo).The third station with continuous data since WY1986 is atthe PMRW, southeast of the COA (Figure 1). The long-term average WY precipitation ranged from 1240 mmin the southeast to 1360 mm in the northwest. During
Published in 2009 by John Wiley & Sons, Ltd. Hydrol. Process. 23, 2860–2878 (2009)DOI: 10.1002/hyp
EFFECTS OF URBANIZATION ON STREAM WATER QUALITY 2867
the study, precipitation was highest during WY2005with each station reporting approximately 1600 mm,and lowest during WY2007 with a range from 850 to950 mm from southeast to northwest. Precipitation at thethree weather stations averaged 1300, 1600, 1270 and890 mm for WY2004, WY2005, WY2006 and WY2007,respectively.
The annual RC (runoff/precipitation) of the 10 long-term RT stations averaged 0Ð43, 0Ð51, 0Ð35 and 0Ð32for WY2004–2007, respectively. In contrast, the RC forforested PMRW reference station was 0Ð27, 0Ð40, 0Ð30and 0Ð25, which is generally consistent with the inter-annual COA results. The urban streams generated morerunoff than the reference watershed; COA RCs are higherthan PMRW RCs, which also is consistent with the resultsreported by Rose and Peters (2001). High RCs at theCOA stations are attributed to more impervious area, arti-ficial channels and conveyances in the urbanized streams,which are more effective at transferring water to COA
streams than natural infiltration through soils followedby groundwater discharge to the streams.
Stream water quality variations among stations
Water quality also varied markedly within and amongstations. The distribution of concentrations of SO4, nutri-ents, bacteria and metals at both COA and reference sta-tions was positively skewed, i.e. asymmetrical with mostvalues clustered at the lower end of the scale (Figures 3),which has been reported elsewhere for most of theseconstituents (van Buren et al., 1997). Urbanization hasa marked effect on water quality. The concentrations ofmost constituents at each station were statistically signif-icantly higher than those of the two reference stations.Also, ISCO sampling at the RT stations provided a muchmore thorough characterization of the chemical condi-tions of streamwater during stormflow than at stationswhere only manual samples were collected; differences
K (
mg
l-1)
0
5
10
NA
N3
NA
N1
NA
N2
LUL1
WO
O1
PE
A5
PE
A4
PE
A3
PE
A2
PE
A1
PR
O2
PR
O3
PR
O1
SA
N1
UT
O3
UT
O2
UT
O1
INT
1S
OU
3S
OU
2S
OU
1P
MR
WM
TN
C
(e)
pH
456789
10 (a)
Ca
(mg
l-1)
0
20
40
60(b)
Mg
(m
g l-1
)
0
2
4
6
8 (c)
Na
(mg
l-1)
0
50
100
150
200(d)
AL
K (
mg
l-1 C
aCO
3)
0
50
100(f)
SO
4 (m
g l-1
)
0
50
100
150(h)
SC
(µ
S c
m-1
)
0
500
1000
NA
N3
NA
N2
NA
N1
LUL1
WO
O1
PE
A5
PE
A4
PE
A3
PE
A2
PE
A1
PR
O2
PR
O3
PR
O1
SA
N1
UT
O3
UT
O2
UT
O1
INT
1S
OU
3S
OU
2S
OU
1P
MR
WM
TN
C
(j)
NH
4N (
mg
l-1)
0
1
2
3
4
Median75%90%
25%10%Outlier
Outlier (k)
Cl (
mg
l-1)
0
100
200
300(g)
NO
2N (
mg
l-1)
0
.5
1
1.5 (l)
NO
3N (
mg
l-1)
0
1
2
3 (m)
Si (
mg
l-1)
0
10
20
30(i)
PO
4P (
mg
l-1)
0
.2
.4
.6
.8(n)
TP
(m
g l-1
)
0
.5
1
1.5
2
NA
N3
NA
N2
NA
N1
LUL1
WO
O1
PE
A5
PE
A4
PE
A3
PE
A2
PE
A1
PR
O2
PR
O3
PR
O1
SA
N1
UT
O3
UT
O2
UT
O1
INT
1S
OU
3S
OU
2S
OU
1P
MR
WM
TN
C
(o)
Figure 3. Concentration distributions of (a) pH, (b) Ca, (c) Mg, (d) Na, (e) K, (f) ALK, (g) Cl, (h) SO4, (i) Si, (j) SC, (k) NH4N, (l) NO2N, (m)NO3N, (n) PO4P, (o) TP, (p) Al, (q) Mn, (r) Fe, (s) DO, (t) DO Saturation, (u) Cd, (v) Cr, (w) Cu, (x) Pb, (y) Zn, (z) FC, (aa) EC, (ab) TC, (ac)TURB, and (ad) SS for all water quality samples of streams in the city of Atlanta and two reference streams, WY2004–WY2007. The chronic andacute standards for Cd, Cu, Pb and Zn shown on the figure the values for a sample with a hardness of 50 mg l�1 as CaCO3; the actual number of
samples that exceed the computed standard from the sample hardness is listed in Table III
Published in 2009 by John Wiley & Sons, Ltd. Hydrol. Process. 23, 2860–2878 (2009)DOI: 10.1002/hyp
2868 N. E. PETERSA
l (µg
l-1)
0
2000
4000
6000(p)
Mn
(µg
l-1)
0
1000
2000
3000(q)
Fe
(µg
l-1)
0
1000
2000(r)
DO
(m
g l-1
)
0
5
10
15 (s)
FC
(C
ount
s 10
0ml-1
)
100101102103104105106107 (z)
DO
(%
Sat
urat
ion)
40
80
120
160 (t)
NA
N3
NA
N2
NA
N1
LU
L1
WO
O1
PE
A5
PE
A4
PE
A3
PE
A2
PE
A1
PR
O2
PR
O3
PR
O1
SA
N1
UT
O3
UT
O2
UT
O1
INT
1S
OU
3S
OU
2S
OU
1P
MR
WM
TN
C
Cd
(µg
l-1)
(u)
100
10-1
10-2
101
102
Cr
(µg
l-1)
(v)101
100
10-1
10-2 EC
(C
ount
s 10
0ml-1
)
(aa)
100101102103104105106
AcuteChronic
Cu
(µg
l-1)
(w)
100
101
10-1
102
TC
(C
ount
s 10
0ml-1
)
(ab)
100101102103104105106107108
TU
RB
(F
NU
)
(ac)
100
101
102
103
Acute: 30 µg l-1
Pb
(µg
l-1)
(x)101
100
10-1
10-2
Acute = Chronic
Zn
(µg
l-1)
(y)
NA
N3
NA
N2
NA
N1
LU
L1
WO
O1
PE
A5
PE
A4
PE
A3
PE
A2
PE
A1
PR
O2
PR
O3
PR
O1
SA
N1
UT
O3
UT
O2
UT
O1
INT
1S
OU
3S
OU
2S
OU
1P
MR
WM
TN
C
100
101
102
103
SS
(m
g l-1
)
(ad)
NA
N3
NA
N2
NA
N1
LU
L1
WO
O1
PE
A5
PE
A4
PE
A3
PE
A2
PE
A1
PR
O2
PR
O3
PR
O1
SA
N1
UT
O3
UT
O2
UT
O1
INT
1S
OU
3S
OU
2S
OU
1P
MR
WM
TN
C
100
101
102
103
Figure 3. (Continued )
were most pronounced when concentrations increasedwith increasing streamflow.
Concentrations of many constituents can vary spatiallyin a stream cross-section, particularly large streams andrivers, which could affect interpretation of results gen-erated from point sampling compared with an integratedsample of the cross-section. The concentration variationsfor EWI and GRAB sampling were evaluated for pairedsamplings of the COA stations, which were collectedover a wide range of streamflows, and results indicatethat there is no statistically significant concentration dif-ference between EWI and GRAB sampling for the con-stituents analyzed herein; this result also was reported fortotal SS, total dissolved solids, dissolved nutrients andtotal metals in streamwater of a nearby county (Knaakand Ankcorn, 2003).
Major ions. In general, SC and major ion concen-trations varied markedly within samples from a singlestation. The SC and major ion concentrations also werehigher and more variable than those of the two refer-ence stations, despite the low-density residential devel-opment of MTNC. On average, SC, Ca, Mg, Na, K,ALK and Si concentrations are 5, 8, 4, 4, 4, 10 and 2times, respectively, higher at the COA stations than at the
PMRW reference station; concentrations at MTNC were,on average, two times higher than at PMRW except for Si,which is comparable; for some constituents such as SO4,average concentrations at some COA stations were morethan 30 times average PMRW concentrations. At each RTstation and PMRW, concentrations of most of major ionswere highly correlated with each other and negativelycorrelated with discharge (Spearman rho; p < 0Ð0001),with a few exceptions.
Major ion concentrations in natural systems are pri-marily controlled by weathering and the residence timeof water along hydrologic pathways, which provides con-tact with the soils and rocks in the drainage basin (Drever,1997). Concentrations of weathering products in ground-water typically increase with residence time. The primarynatural weathering process involves H2CO3 breakdownof minerals producing base cations (Ca, Mg, Na and K),ALK (primarily bicarbonate) and Si. The concentrationsamong these major ions at COA stations and PMRWwere highly linearly correlated and regression slope dif-ferences likely reflect variations in mineralogy of thebedrock and soils. For the COA RT stations and PMRW,the negative correlation of major ion/field property con-centrations with discharge is attributed to the dilution
Published in 2009 by John Wiley & Sons, Ltd. Hydrol. Process. 23, 2860–2878 (2009)DOI: 10.1002/hyp
EFFECTS OF URBANIZATION ON STREAM WATER QUALITY 2869
of more concentrated long-residence time groundwater,which dominates streamflow during base flow.
The relation between ALK and Si for the stationsis quite revealing (Figure 4). Differences in the relationbetween ALK and Si among stations can reflect the rela-tive contributions of carbonate and aluminosilicate min-eral weathering; carbonate mineral weathering producesALK only, whereas aluminosilicate mineral weatheringproduces both ALK and Si. Furthermore, carbonate min-erals tend to weather faster than aluminosilicate minerals,producing ¾10 times more Ca and ALK in streamwa-ter (Meybeck, 1994). Linear regression relations of ALKon Si for most COA streams were highly significant
0
50
100
AlL
K (
mg
l-1 C
aCO
3)
0 10 20 30
Si (mg l-1)
(a)
(b)
0
10
20
30
40
50
Ca
(mg
l-1)
0 10 20 30
Si (mg l-1)
(c)
06 PEA507 PEA408 PEA309 PEA210 PEA1
12 PRO3
14 SAN115 UTO316 UTO217 UTO118 INT1
01 NAN302 NAN203 NAN104 LUL1
19 SOU320 SOU221 SOU1
05 WOO1
11 PRO2
13 PRO1
22 PMRW23 MTNC
0
10
20
30
40
50
Ca
(mg
l-1)
0 50 100
ALK (mg l-1 CaCO3)
Figure 4. Concentration relations between (a) alkalinity and Si, (b) Caand alkalinity and (c) Ca and Si of streams in the city of Atlanta and two
reference streams, WY2004–WY2007
(p < 0Ð0001) and the slopes were approximately 2,except for PRO1, PRO2 and WOO1, which were larger(3Ð3, 2Ð5 and 4Ð6, respectively) and intercepts wereapproximately 10 or less, except for PRO2 (21). Theslopes of the COA streams are much larger than the refer-ence sites (0Ð55 and 0Ð85 for PMRW and MTNC, respec-tively). The larger slopes of the COA streams comparedwith the reference streams indicate a carbonate-mineralweathering source. The reason for the larger ALK ver-sus Si slopes for the Proctor Creek stations than COAstations is not evident from the geology or land use char-acteristics (Table II). For WOO1, the temporal variationof other parameters/constituents provides some additionalinformation about sources as discussed below. In contrastto the large slopes at most of the COA stations, ALK wasmore poorly correlated with Si, at two headwater stationson the South River (SOU2 and SOU3) than those of otherstations and although highly variable, the average ratioof ALK to Si was much lower. The likely cause of thisresult for these South River stations also is discussed inmore detail at the end of this section
The bedrock composition is generally comparableacross the study area and does not have any lithologicunits dominated by carbonate minerals that could havea major affect the release rates of major ions (Higginsand Crawford, 2006). However, concrete is a major com-ponent of urbanized watersheds supporting the buildingand transportation infrastructure, and typically containscarbonate minerals (Matschel et al., 2007). Furthermore,high Ca content of the concrete may dominate over therelease of products generated from the natural or back-ground mineral weathering of the soils and bedrock inwatersheds. Assuming this to be the case, it is not sur-prising that Ca was highly correlated with ALK amongmost stations (Figure 4b), whereas the relation betweenCa and Si should vary depending on the relative weath-ering contributions of basin materials and other potentialsources (Figure 4c). Although Ca concentrations corre-late with Si, linear regression slopes differ among sta-tions. This result combined with the highly significant(p < 0Ð0001) and similar linear relation between Ca andALK among all COA stations also indicates a carbonatemineral source, which is most likely concrete.
Some other water quality differences are noted amongCOA monitoring stations, e.g. see pH in Figure 3a. Twostations had markedly different pH values compared withother stations; the pH of several samples at WOO1 weregreater than 8Ð5 and at SOU3 were less than 6, whichdefines the range of the Georgia water quality standard(Georgia Environmental Protection Division [GA EPD],2005).
The high pH of WOO1 was associated with high Ca,Mg and ALK concentrations, as discussed previously andsuggests a carbonate source based on the these major ionrelations. The WOO1 watershed (Woodall Creek) con-tains the highest percentage of commercial-plus-industrialland use (52Ð8%) and transportation, communicationsand utilities (14Ð1%) of any of the COA watersheds
Published in 2009 by John Wiley & Sons, Ltd. Hydrol. Process. 23, 2860–2878 (2009)DOI: 10.1002/hyp
2870 N. E. PETERS
pH
pHSC
TURB
7.0
7.5
8.0
8.5
9.0
9.5
10.
Dis
char
ge (
cm d
-1)
Discharge
0
1
2
3
SC
(µS
cm
-1),
TU
RB
(N
TU
)
0
100
200
300
400
18:00 11-Oct 2006 00:00 12-Oct 2006 6:00 12-Oct 2006
Figure 5. Real-time variations of pH, specific conductance, turbidity anddischarge at WOO1 during a storm in October 2006
(Table II). An examination of the RT constituent vari-ations of WOO1 reveals a consistent pattern of high-pH values (>9) during stormflow with pH increasesoccurring consistently after peak discharge, suggestingmobilization from a source distant from the gauging sta-tion (Figure 5). The high pH cannot be explained bythe dissolution of carbonate minerals that would bufferthe stream to a pH <9, and is more likely associatedwith leaching of a liquid Ca base or dissolution of Caoxide (lye) or a Ca–Mg oxide, which might explain thehigh correlation among ALK, Ca and Mg concentrations(Stumm and Morgan, 1996). The higher percentage ofcommercial-plus-industrial, transportation, communica-tions and utilities land use is indicative of more impervi-ous surface than found in other watersheds (Table II). TheCOA Department of Watershed Management deployed aportable pH monitor on Woodall Creek in early 2007. Themonitor has been systematically moved upstream and theresults have narrowed the source to a headwater tribu-tary, but their efforts have been hampered by the near
absence of frequently recurring stormflow due to a long-term drought, which has affected the Atlanta Region sincespring 2007.
High Cl concentrations occurred at LUL1, SOU1 andINT1 (Figure 3g). The high Cl concentrations at INT1and LUL1 were associated with high Na concentrations(Figure 6a). INT1 is downstream from two combinedsewer overflow (CSO) facilities including the following:(1) the Custer Street Regulator, a combined conveyanceof stormwater to INT1 and treatment facility, which dis-charges treated stormwater to Intrenchment Creek beforeand after major storm runoff; and (2) a larger CSO facil-ity downstream, which receives and treats CSO from theCuster Street Regulator and elsewhere during stormflow.The high Cl concentrations at INT1 were likely causedby chlorination of the CSO stormwater with sodiumhypochlorite and subsequent conversion of the chlorineto chloride (O’Shea and Field, 1992). Most of the high Clconcentrations at INT1 also were associated with the lowcoliform bacteria concentrations (e.g. EC in Figure 6b),which is consistent with this hypothesis. The timing ofoutflow discharges from the National Pollutant DischargeElimination System monthly Combined Sewer OverflowMonitoring Report for the Intrenchment Creek CSO wasevaluated with respect to the USGS water quality mon-itoring data. High Na and Cl and low coliform bacteriaconcentrations in INT1 samples are associated with CSOdischarges. In contrast, background stream-water quality,i.e. collected when the CSO was not discharging, wasmore enriched in Mg and other weathering products (seeMg relation with Cl in Figure 6c) and may be simply dueto the relative enrichment of the CSO with by-productsof the chlorination process, i.e. Na and Cl.
22 PMRW23 MTNC
01 NAN302 NAN203 NAN1
05 WOO106 PEA507 PEA408 PEA309 PEA210 PEA111 PRO2
12 PRO313 PRO114 SAN115 UTO316 UTO217 UTO1
19 SOU320 SOU2
04 LUL1
18 INT1
21 SOU1
Na
(mg
l-1)
0
50
100
150
0 50 100 150 200 250 300
(a)
0 50 100
EC
(C
ount
s 10
0 m
l-1)
100
101
102
103
104
105
Cl (mg l-1)
(b)
Mg
(mg
l-1)
0
1
2
3
4
5
0 10 20 30 40 50
CSO DischargeNo CSO Discharge
(c)
Ca
(mg
l-1)
0
10
20
30
40
50
0 50 100 150 200 250 300Cl (mg l-1)
(d)
Figure 6. Concentration relations between (a) Na, (b) E. Coli, (c) Ca, (d) Mg and Cl of streams in the city of Atlanta and two reference streams for(a), (b) and (d) and INT1 for (c) during WY2004–WY2007
Published in 2009 by John Wiley & Sons, Ltd. Hydrol. Process. 23, 2860–2878 (2009)DOI: 10.1002/hyp
EFFECTS OF URBANIZATION ON STREAM WATER QUALITY 2871
0
50
100
150
200
250
300
350
0.0
0.2
0.4
0.6
0.8
1.0
1.2
06:00 2/25/2007 09:00 2/25/2007 12:00 2/25/2007 15:00 2/25/2007 18:00 2/25/2007
SC
(µS
cm
-1),
TU
RB
(N
TU
)
Q (
cm d
-1)
(b)
0
50
100
150
200
250
0.0
0.2
0.4
0.6
0.8
1.0
1.2
12:00 2/13/2007 18:00 2/13/2007 00:00 2/14/2007
SC
(µS
cm
-1),
TU
RB
(N
TU
)
Q (
cm d
-1)
Q
TURB
SC
(a)
ISCO Sample
Figure 7. Real-time discharge, specific conductance and turbidity forSOU1, the South River at Forrest Park Road, for two rainstorms on
(a) 13 February, 2007, and (b) 25 February, 2007
LUL1 is a small residential watershed, and a golfcourse covers a large percentage of the drainage area(Table II). The cause for the high Na and Cl concen-trations is not as easily determined for LUL1 as it wasfor INT1. A possible cause was the periodic drainage ofswimming pools, which also are chlorinated using sodiumhypochlorite. The samples containing the high Na and Clconcentrations were collected during two different timeperiods, September 2003 and June 2004, which excludesthe possibility that the cause was a single event.
The high Cl concentrations for samples collected atSOU1 occurred during two rainstorms in February 2007;the first rainstorm on 13 February followed a major winterstorm that hit Atlanta on 1 February. Freezing conditionswere anticipated for each rainstorm, but while majorsnowstorms affected the northeastern United States, only¾12 mm rainfall accompanied each rainstorm in theAtlanta Region. The high Cl concentrations at SOU1were associated with high Ca concentrations (Figure 6d).The RT data showed a marked increase in SC duringeach event but with slight differences in the timing(Figure 7a and b for the rainstorms on 13 and 25February, respectively). The high SC occurred at orslightly after peak streamflow during the 13 Februaryrainstorm. During the 25 February rainstorm, the SCmaximum was higher than that of the previous rainstormand began to decrease prior to the peak streamflow.Decreases in SC below pre-event values occurred duringthe recession of each rainstorm. The likely source wasdissolution of CaCl2, which is a deicing salt. However,
deicing salt usage records were not available from theCOA Office of Transportation to confirm this hypothesis.
The water quality of SOU3 was highly degradedcompared with the other COA streams and the referencestreams. The pH and ALK were the lowest of allsteams and SC, SO4, NH4N, NO3N, Al, Mn, Cd andZn concentrations were higher to much higher thanother stations (Figure 3). In addition, it is likely thatstream transport of these contaminants could explainthe relatively high streamwater concentrations of someconstituents at downstream stations. The pronouncedchemical differences were attributed to transport ofleachate, derived from the dissolution of residual alumwaste, through contaminated groundwater contributionto the stream. An alum-manufacturing plant upstreamfrom the COA boundary in East Point generates thewaste, but it no longer uses surface impoundments forthe alum, which likely had a direct impact on the stream(Tracy Hillick, COA, oral communication, 2006). Alumis an aluminum sulfate compound containing K, Na orNH4N. The later form, ammonium alum [NH4Al(SO4)2],is used in water treatment for coagulation, which causessettling of suspended particles and was the main productproduced by the alum plant. Alum is relatively solubleand on dissolution, produces an acidic solution high inAl and SO4, which would be partly neutralized by theALK produced from weathering. The acid reaction withthe ALK likely explains the non-characteristic relationsbetween ALK and Si, and Ca and ALK (Figure 4a and b).The relatively higher NO3N concentrations at SOU3 thanat other stations would likely result from the oxidation ofNH4 released from the dissolution of ammonium alum,which is another acidifying reaction.
Nutrients. The nutrient species data for many sam-ples were below the method-reporting limit. The num-ber of samples affected by censoring has decreased withthe improved detection/reporting limits as the study pro-gressed. The distribution of nutrient concentrations forNH4N, NO2N, NO3N, PO4P and TP in Figure 3k–o is forsamples with concentrations above the method-reportinglimit and include 54, 68, 100, 30 and 63% of all of thesamples analyzed, respectively. Nutrient concentrations,on average, were higher at the COA stations than atthe reference stations. N species concentrations generallywere much less than 10 mg l�1 NO3N, the US Environ-mental Protection Agency (US EPA) drinking water stan-dard (US EPA, 1986) and Georgia standard for wastewa-ter disposal (GA EPD, 2005). Furthermore, less than 10%of the samples at each station exceeded 0Ð1 mg l�1 PO4P,the US EPA recommended concentration limit for streamsto prevent excessive algal growth (US EPA, 1986).
Bacteria and DO.Some stations exceeded Georgia water quality stan-
dards (GA EPD, 2005). Samples collected at four sta-tions (PEA2, PRO3, SOU1 and UTO1) were less thanthe minimum DO standard (4 mg l�1), although the val-ues were few, i.e. no more than three samples at any
Published in 2009 by John Wiley & Sons, Ltd. Hydrol. Process. 23, 2860–2878 (2009)DOI: 10.1002/hyp
2872 N. E. PETERS
TURB (FNU)
EC
(C
ount
s 10
0 m
l-1)
100
101
102
103
104
105
100 101 102 103
(a)
TC (MPN dl-1)
EC
(C
ount
s 10
0 m
l-1)
100
101
102
103
104
105
100 101 102 103 104 106 107105
(b)
01 NAN302 NAN203 NAN104 LUL105 WOO106 PEA507 PEA408 PEA309 PEA210 PEA111 PRO2
12 PRO313 PRO114 SAN15 UTO316 UTO217 UTO118 INT1
19 SOU320 SOU221 SOU122 PMRW23 MTNC
SS (mg l-1)
TU
RB
(F
NU
)
100
101
102
103
100 101 102 103
(c)
Figure 8. Relations between (a) E. Coli and turbidity, (b) E. Coli andtotal coliform and (c) turbidity and suspended sediment concentration ofstreams in the city of Atlanta, WY2004–WY2007, and two reference
streams from April 2001 through March 2002
station (Figure 3s). In the worst case, only a few per-centage of the 15-min data were less than 4 mg l�1 atthe RT stations. The streams are generally well aeratedand streamwater was more than 75% saturated with DOfor more than 75% of the measurements at each station(Figure 3t).
In some cases, the skew of the concentration distribu-tion was pronounced, e.g. coliform bacteria (EC, TC andFC), TURB and SS (Figure 3z–ad). The monitoring pro-gram was not designed to compute monthly geometricmeans for FC concentrations, the procedure for deter-mining acceptability of stream water quality (GA EPD,
2005). Low bacteria concentrations (FC, TC and EC)were reported at INT1, which is affected by CSO chlori-nation, and SOU3, which is extremely acidic and has hightrace metal concentrations, and for which no fish weredetected during surveys conducted during 2001, 2003 or2005 (Chrissy Thom, CH2M HILL, written communi-cation, 2006). The degraded water quality conditions forSOU3 are similar to those of acidified streams due to acidmine drainage or acidic atmospheric deposition, wherethe low pH and high metal concentrations, particularly Al,are toxic to fish (Driscoll et al., 2001). For the remainderof the stations, instantaneous FC concentrations of the25th percentile of the sampling at the most rural stations,i.e. Sandy, Nancy and Utoy Creeks and 10th percentileat the other COA stations exceeded the Georgia waterquality standard of 200 MPN dl�1 for any usage class.Furthermore, the FC concentrations of all high stream-flow samples at these stations exceeded the Georgia waterquality standard. In general, coliform bacteria concentra-tions were significantly correlated among each other (TC,FC and EC) and with TURB and SS (see Figure 8 forrelations between (1) EC and TC, (2) EC and TURB and(3) TURB and SS). The TC, FC and EC concentrationsat the COA stations were high, but were comparableto those reported for other urban streams (Makepeaceet al., 1995; Arienzo et al., 2001; Ellis, 2004). Theseresults also are consistent with results for the Chatta-hoochee River from the BacteriALERT Program, whichwas designed to provide a public alert when bacteria lev-els in the Chattahoochee River National Recreation Areaexceed US EPA criteria (Lawrence, 2006). Although theconcentrations are somewhat lower at the reference sta-tions (from sampling conducted from April 2001 throughMarch 2002), the relations among EC, TC and TURB aresimilar to those for the COA streams.
Metals. Trace metal concentrations were evaluatedwith respect to GA EPD standards (GA EPD, 2005).The chronic and acute criteria correspond to the USEPA definition for Criteria Continuous Concentrationand Criteria Maximum Concentration, respectively; theyare defined as the highest concentration of a pollutantto which aquatic life can be exposed for an extendedperiod of time (4 days) and a short period of time (1-h average) without deleterious effects (US EPA, 2009).The chronic and acute values of Cd, Cu, Pb and Zn varywith hardness with the percentage of values exceeding thestandard being determined by comparing sample valueswith the computed standard (GA EPD, 2005); the chronicand acute results are summarized for each sample typeand station in Table III. The chronic and acute valuesshown for these metals on Figure 3 are for a hardness of50 mg l�1 as CaCO3. Few Cd concentrations exceededthe acute standard except at the South River stations,whereas noted previously the low pH likely enhancedmetal mobility (Figure 3u and Table III). Concentrationsof hexavalent Cr were below the acute and chronicstandards for all but one sample at PEA4 (Figure 3v).The Cu and Zn concentrations of many samples at most
Published in 2009 by John Wiley & Sons, Ltd. Hydrol. Process. 23, 2860–2878 (2009)DOI: 10.1002/hyp
EFFECTS OF URBANIZATION ON STREAM WATER QUALITY 2873
Tabl
eII
I.Pe
rcen
tage
and
num
ber
ofsa
mpl
esfo
rw
hich
conc
entr
atio
nsof
Cd,
Cu,
Pban
dZ
nex
ceed
GA
EPD
stan
dard
s,w
here
the
stan
dard
conc
entr
atio
nsw
ere
com
pute
dfr
omth
esa
mpl
eha
rdne
ssby
equa
tions
deve
lope
dby
the
US
EPA
and
adop
ted
byG
AE
PD(G
AE
PD,
2005
)
Sta
tion
Type
aC
dC
uP
bZ
n
EW
IG
RA
BIS
CO
Tota
lE
WI
GR
AB
ISC
OTo
tal
EW
IG
RA
BIS
CO
Tota
lE
WI
GR
AB
ISC
OTo
tal
(%)
(#)
(%)
(#)
(%)
(#)
(%)
(#)
(%)
(#)
(%)
(#)
(%)
(#)
(%)
(#)
(%)
(#)
(%)
(#)
(%)
(#)
(%)
(#)
(%)
(#)
(%)
(#)
(%)
(#)
(%)
(#)
NA
N3
A0
(8)
NA
(0)
0(8
3)0
(91)
46Ð3
(41)
15Ð8
(19)
91Ð7
(108
)72
(168
)28
Ð2(3
9)14
Ð3(1
4)68
Ð3(1
23)
55Ð1
(176
)65
Ð9(4
1)31
Ð6(1
9)82
Ð1(1
23)
73Ð2
(183
)B
0(8
)N
A(0
)1Ð2
(83)
1Ð1(9
1)34
Ð1(4
1)5Ð3
(19)
76Ð4
(123
)59
Ð6(1
83)
0(3
9)0
(14)
0(1
23)
0(1
76)
65Ð9
(41)
31Ð6
(19)
82Ð1
(123
)73
Ð2(1
83)
NA
N2
A0
(9)
0(3
)N
A(0
)0
(12)
35Ð5
(31)
16Ð7
(18)
NA
(0)
28Ð6
(49)
24Ð2
(33)
15Ð4
(13)
NA
(0)
21Ð7
(46)
32Ð4
(37)
22Ð2
(18)
NA
(0)
29Ð1
(55)
B0
(9)
0(3
)N
A(0
)0
(12)
27(3
7)16
Ð7(1
8)N
A(0
)23
Ð6(5
5)0
(33)
0(1
3)N
A(0
)0
(46)
32Ð4
(37)
22Ð2
(18)
NA
(0)
29Ð1
(55)
NA
N1
A0
(12)
0(5
)1Ð3
(77)
1Ð1(9
4)35
Ð9(3
9)25
Ð0(2
0)95
Ð9(9
8)72
(157
)17
Ð6(3
4)26
Ð7(1
5)67
Ð3(1
04)
52Ð3
(153
)23
Ð1(3
9)25
Ð0(2
0)82
Ð7(1
04)
61Ð3
(163
)
B0
(12)
0(5
)2Ð6
(77)
2Ð1(9
4)23
Ð1(3
9)25
Ð0(2
0)82
Ð7(1
04)
61Ð3
(163
)0
(34)
0(1
5)0
(104
)0
(153
)23
Ð1(3
9)25
Ð0(2
0)82
Ð7(1
04)
61Ð3
(163
)
LU
L1
A0
(16)
0(3
)N
A(0
)0
(19)
20Ð6
(34)
21Ð1
(19)
NA
(0)
20Ð8
(53)
12Ð1
(33)
11Ð8
(17)
NA
(0)
12Ð0
(50)
29Ð7
(37)
21Ð1
(19)
NA
(0)
26Ð8
(56)
B0
(16)
0(3
)N
A(0
)0
(19)
13Ð5
(37)
15Ð8
(19)
NA
(0)
14Ð3
(56)
0(3
3)0
(17)
NA
(0)
0(5
0)29
Ð7(3
7)21
Ð1(1
9)N
A(0
)26
Ð8(5
6)
WO
O1
A0
(21)
0(2
3)0
(89)
0(1
33)
50Ð0
(18)
60Ð9
(23)
86Ð5
(89)
76Ð9
(130
)19
Ð0(2
1)47
Ð8(2
3)59
Ð6(8
9)51
Ð1(1
33)
90Ð0
(20)
91Ð3
(23)
95Ð5
(89)
93Ð9
(132
)
B0
(21)
0(2
3)0
(89)
0(1
33)
36Ð8
(19)
56Ð5
(23)
77Ð5
(89)
67Ð9
(131
)0
(21)
0(2
3)0
(89)
0(1
33)
90Ð0
(20)
91Ð3
(23)
95Ð5
(89)
93Ð9
(132
)
PEA
5A
0(1
8)0
(14)
2Ð9(7
0)2Ð0
(102
)36
Ð4(3
3)45
Ð5(3
3)93
(100
)72
Ð3(1
66)
22Ð2
(36)
45Ð2
(31)
70Ð6
(102
)55
Ð6(1
69)
47Ð2
(36)
63Ð6
(33)
95Ð1
(102
)78
Ð9(1
71)
B0
(18)
0(1
4)5Ð7
(70)
3Ð9(1
02)
27Ð8
(36)
45Ð5
(33)
88Ð2
(102
)67
Ð3(1
71)
0(3
6)0
(31)
1(1
02)
0Ð6(1
69)
47Ð2
(36)
63Ð6
(33)
95Ð1
(102
)78
Ð9(1
71)
PEA
4A
0(1
7)0
(7)
3(1
00)
2Ð4(1
24)
34Ð2
(38)
25(2
0)85
Ð8(1
20)
68(1
78)
22Ð9
(35)
0(1
8)62
Ð5(1
28)
48Ð6
(181
)56
Ð4(3
9)38
Ð1(2
1)90
Ð6(1
28)
77Ð7
(188
)
B0
(17)
0(7
)3
(100
)2Ð4
(124
)25
Ð6(3
9)10
(20)
68Ð8
(128
)53
Ð5(1
87)
0(3
5)0
(18)
0(1
28)
0(1
81)
56Ð4
(39)
38Ð1
(21)
90Ð6
(128
)77
Ð7(1
88)
PEA
3A
6Ð7(1
5)N
A(0
)N
A(0
)6Ð7
(15)
47Ð7
(44)
0(1
4)N
A(0
)36
Ð2(5
8)38
Ð6(4
4)0
(13)
NA
(0)
29Ð8
(57)
47Ð7
(44)
7Ð1(1
4)N
A(0
)37
Ð9(5
8)
B6Ð7
(15)
NA
(0)
NA
(0)
6Ð7(1
5)40
Ð9(4
4)0
(14)
NA
(0)
31(5
8)0
(44)
0(1
3)N
A(0
)0
(57)
47Ð7
(44)
7Ð1(1
4)N
A(0
)37
Ð9(5
8)
PEA
2A
0(2
6)0
(17)
1Ð1(8
7)0Ð8
(130
)40
(40)
40Ð9
(22)
98(9
8)75
Ð6(1
60)
26Ð2
(42)
27Ð3
(22)
82Ð7
(98)
60Ð5
(162
)56
Ð1(4
1)59
Ð1(2
2)95
Ð9(9
8)80
Ð7(1
61)
B0
(26)
0(1
7)2Ð3
(87)
1Ð5(1
30)
31Ð7
(41)
40Ð9
(22)
94Ð9
(98)
71Ð4
(161
)0
(42)
0(2
2)0
(98)
0(1
62)
56Ð1
(41)
59Ð1
(22)
95Ð9
(98)
80Ð7
(161
)
PEA
1A
0(2
7)0
(9)
NA
(0)
0(3
6)35
Ð3(3
4)6Ð7
(15)
NA
(0)
26Ð5
(49)
18Ð4
(38)
6Ð7(1
5)N
A(0
)15
Ð1(5
3)39
Ð5(3
8)26
Ð7(1
5)N
A(0
)35
Ð8(5
3)
B0
(27)
0(9
)N
A(0
)0
(36)
28Ð9
(38)
6Ð7(1
5)N
A(0
)22
Ð6(5
3)0
(38)
0(1
5)N
A(0
)0
(53)
42Ð1
(38)
26Ð7
(15)
NA
(0)
37Ð7
(53)
PRO
2A
0(3
4)0
(23)
NA
(0)
0(5
7)25
(32)
20Ð8
(24)
NA
(0)
23Ð2
(56)
20Ð6
(34)
41Ð7
(24)
NA
(0)
29Ð3
(58)
51Ð5
(33)
54Ð2
(24)
NA
(0)
52Ð6
(57)
B0
(34)
0(2
3)N
A(0
)0
(57)
18Ð2
(33)
16Ð7
(24)
NA
(0)
17Ð5
(57)
0(3
4)0
(24)
NA
(0)
0(5
8)51
Ð5(3
3)54
Ð2(2
4)N
A(0
)52
Ð6(5
7)
PRO
3A
0(2
3)0
(7)
NA
(0)
0(3
0)32
Ð4(3
4)10
Ð5(1
9)N
A(0
)24
Ð5(5
3)17
Ð1(3
5)5Ð6
(18)
NA
(0)
13Ð2
(53)
41Ð7
(36)
31Ð6
(19)
NA
(0)
38Ð2
(55)
B0
(23)
0(7
)N
A(0
)0
(30)
22Ð9
(35)
5Ð3(1
9)N
A(0
)16
Ð7(5
4)0
(35)
0(1
8)N
A(0
)0
(53)
41Ð7
(36)
31Ð6
(19)
NA
(0)
38Ð2
(55)
PRO
1A
0(3
3)0
(35)
1Ð9(1
08)
1Ð1(1
76)
32Ð4
(34)
54Ð3
(35)
87(1
08)
70Ð1
(177
)20
(35)
37Ð1
(35)
75Ð9
(108
)57
Ð3(1
78)
38Ð9
(36)
62Ð9
(35)
88Ð9
(108
)73
Ð7(1
79)
B0
(33)
0(3
5)1Ð9
(108
)1Ð1
(176
)22
Ð2(3
6)37
Ð1(3
5)83
Ð3(1
08)
62Ð0
(179
)0
(35)
0(3
5)1Ð9
(108
)1Ð1
(178
)38
Ð9(3
6)62
Ð9(3
5)88
Ð9(1
08)
73Ð7
(179
)SA
N1
A0
(24)
0(9
)N
A(0
)0
(33)
41Ð7
(36)
36Ð8
(19)
NA
(0)
40Ð0
(55)
17Ð1
(35)
29Ð4
(17)
NA
(0)
21Ð2
(52)
43Ð2
(37)
47Ð4
(19)
NA
(0)
44Ð6
(56)
B0
(24)
0(9
)N
A(0
)0
(33)
21Ð6
(37)
21Ð1
(19)
NA
(0)
21Ð4
(56)
0(3
5)0
(17)
NA
(0)
0(5
2)43
Ð2(3
7)47
Ð4(1
9)N
A(0
)44
Ð6(5
6)
UT
O3
A0
(14)
0(5
)5Ð3
(75)
4Ð3(9
4)35
Ð9(3
9)20
(20)
90Ð4
(83)
65Ð5
(142
)32
Ð4(3
7)17
Ð6(1
7)83
Ð1(8
3)61
Ð3(1
37)
43Ð6
(39)
25Ð0
(20)
86Ð7
(83)
66Ð2
(142
)
Published in 2009 by John Wiley & Sons, Ltd. Hydrol. Process. 23, 2860–2878 (2009)DOI: 10.1002/hyp
2874 N. E. PETERSTa
ble
III.
(Con
tinu
ed)
Sta
tion
Type
aC
dC
uP
bZ
n
EW
IG
RA
BIS
CO
Tota
lE
WI
GR
AB
ISC
OTo
tal
EW
IG
RA
BIS
CO
Tota
lE
WI
GR
AB
ISC
OTo
tal
(%)
(#)
(%)
(#)
(%)
(#)
(%)
(#)
(%)
(#)
(%)
(#)
(%)
(#)
(%)
(#)
(%)
(#)
(%)
(#)
(%)
(#)
(%)
(#)
(%)
(#)
(%)
(#)
(%)
(#)
(%)
(#)
B0
(14)
0(5
)12
(75)
9Ð6(9
4)27
Ð5(4
0)10
(20)
84Ð3
(83)
58Ð0
(143
)0
(37)
0(1
7)1Ð2
(83)
0Ð7(1
37)
43Ð6
(39)
25Ð0
(20)
88(8
3)66
Ð9(1
42)
UT
O2
A7Ð7
(13)
0(1
1)N
A(0
)4Ð2
(24)
25Ð0
(32)
33Ð3
(21)
NA
(0)
28Ð3
(53)
11Ð5
(26)
13Ð3
(15)
NA
(0)
12Ð2
(41)
18Ð2
(33)
33Ð3
(21)
NA
(0)
24Ð1
(54)
B7Ð7
(13)
0(1
1)N
A(0
)4Ð2
(24)
9Ð4(3
2)28
Ð6(2
1)N
A(0
)17
Ð0(5
3)0
(26)
0(1
5)N
A(0
)0
(41)
18Ð2
(33)
33Ð3
(21)
NA
(0)
24Ð1
(54)
UT
O1
A0
(24)
0(1
6)1Ð2
(81)
0Ð8(1
21)
41(3
9)32
(25)
95(1
01)
72Ð7
(165
)37
Ð1(3
5)21
Ð7(2
3)74
Ð3(1
01)
58Ð5
(159
)97
Ð4(3
9)10
0(2
5)10
0(1
01)
99Ð4
(165
)
B0
(24)
0(1
6)2Ð5
(81)
1Ð7(1
21)
35(4
0)24
(25)
89Ð1
(101
)66
Ð3(1
66)
0(3
5)0
(23)
0(1
01)
0(1
59)
97Ð4
(39)
100
(25)
100
(101
)99
Ð4(1
65)
INT
1A
2Ð9(3
4)6Ð5
(31)
1Ð7(1
17)
2Ð7(1
82)
41Ð7
(36)
48Ð5
(33)
89Ð9
(109
)72
Ð5(1
78)
32Ð4
(37)
48Ð5
(33)
73Ð5
(117
)61
(187
)55
Ð6(3
6)75
Ð8(3
3)85
Ð5(1
17)
78(1
86)
B2Ð9
(34)
6Ð5(3
1)1Ð7
(117
)2Ð7
(182
)33
Ð3(3
6)45
Ð5(3
3)76
Ð9(1
17)
62Ð9
(186
)0
(37)
0(3
3)1Ð7
(117
)1Ð1
(187
)55
Ð6(3
6)75
Ð8(3
3)85
Ð5(1
17)
78(1
86)
SOU
3A
55Ð0
(40)
82Ð4
(17)
14Ð3
(28)
47Ð1
(85)
95Ð0
(40)
94Ð1
(17)
100
(28)
96Ð5
(85)
67Ð5
(40)
70Ð6
(17)
57Ð1
(28)
64Ð7
(85)
100
(40)
100
(17)
100
(28)
100
(85)
B52
Ð5(4
0)82
Ð4(1
7)21
Ð4(2
8)48
Ð2(8
5)90
Ð0(4
0)94
Ð1(1
7)10
0(2
8)94
Ð1(8
5)0
(40)
0(1
7)0
(28)
0(8
5)10
0(4
0)10
0(1
7)10
0(2
8)10
0(8
5)
SOU
2A
10Ð3
(39)
23Ð5
(17)
NA
(0)
14Ð3
(56)
55Ð3
(38)
41Ð2
(17)
NA
(0)
50Ð9
(55)
24Ð1
(29)
7Ð7(1
3)N
A(0
)19
Ð0(4
2)10
0(3
9)10
0(1
7)N
A(0
)10
0(5
6)
B7Ð7
(39)
11Ð8
(17)
NA
(0)
8Ð9(5
6)39
Ð5(3
8)35
Ð3(1
7)N
A(0
)38
Ð2(5
5)0
(29)
0(1
3)N
A(0
)0
(42)
100
(39)
100
(17)
NA
(0)
100
(56)
SOU
1A
0(4
2)0
(18)
0(1
18)
0(1
78)
39Ð5
(43)
5Ð6(1
8)92
Ð6(1
21)
71Ð4
(182
)12
Ð2(4
1)5Ð6
(18)
76Ð9
(121
)55
Ð0(1
80)
100
(42)
100
(18)
100
(121
)10
0(1
81)
B0
(42)
0(1
8)1Ð7
(118
)1Ð1
(178
)34
Ð9(4
3)5Ð6
(18)
86(1
21)
65Ð9
(182
)0
(41)
0(1
8)3Ð3
(121
)2Ð2
(180
)10
0(4
2)10
0(1
8)10
0(1
21)
100
(181
)
Not
e:E
WI,
equa
lw
idth
incr
emen
t;G
RA
B,
man
ual
poin
tsa
mpl
e;IS
CO
,au
tom
ated
poin
tsa
mpl
e.a
AD
chro
nic,
BD
acut
e.
stations exceeded Georgia water quality chronic and acutestandards and the Pb chronic standard (Table III andFigure 3u–y) These metals are ubiquitous in urban areaswith sources ranging from highway runoff interactingwith the washing off of metals derived from tire rubberparticles to discarded and decayed metal fittings used inbuilding construction (Sansalone and Buchberger, 1997;Gromaire-Mertz et al., 1999; Davis et al., 2001; Roseet al., 2001; Sansalone et al., 2005; Hudak and Banks,2006).
Effect of streamflow on water quality
As noted above, some parameters and constituentconcentrations differ markedly among stream samplingstations. However, some of the identified differencesamong stations may be an artifact of the type of sampling.The manual sampling was not fixed-interval sampling,which would have been primarily skewed towards baseflow, but was designed to cover a range of hydrologicalconditions. Despite the attempt to provide a balancedhydrologic coverage, the manual sampling was moreskewed towards base flow than stormflow (Figure 9a). Incontrast, the ISCO samplers, which were only installed atthe RT stations, targeted stormflows. Consequently, ISCOsamples were skewed towards stormflows, particularlywhen compared with the manual sampling. The dischargeat the time of sampling for the ISCO samples wassignificantly higher than for the GRAB samples (p <0Ð005) for each station except PEA2 and NAN1.
The relations between nutrient concentrations and flowgenerally were not statistically significant and variedamong stations. However, TP and PO4P concentrationswere higher in the ISCO stormflow samples (Figure 9band c) than the GRAB samples, but TP concentrationswere not significantly correlated with PO4P. Furthermore,TP concentrations correlated with TURB and SS concen-trations, and these relations are attributed to the strongaffinity for and related adsorption of P on Fe and Aloxides and hydroxides, a ubiquitous coating on soils inthe southeastern United States. The TP concentrationswere also highly correlated with Q as was TURB andSS, but the PO4P concentrations did not correlate withQ. The cause of the generally higher PO4P concentra-tions in the ISCO samples is not known, but it is notlinked to higher discharge or sediment concentrations.In contrast, N species concentrations in GRAB sampleswere similar to ISCO samples, as shown for NO3N inFigure 9d. As mentioned previously, flow was negativelycorrelated with the concentrations of weathering productsas shown by the comparison of ISCO and GRAB sam-pling for Si (Figure 9e). Concentration distributions forEC, TURB and SS (Figure 3u–ad) show marked vari-ability among stations, but most of the among-stationdifferences are attributed to the difference in samplingmethod; SS, TURB and EC concentrations were muchhigher in the ISCO samples than in the GRAB samples(Figure 9f–h). The FC, EC and TC of most samples col-lected at SOU3 were very low until automatic sampling
Published in 2009 by John Wiley & Sons, Ltd. Hydrol. Process. 23, 2860–2878 (2009)DOI: 10.1002/hyp
EFFECTS OF URBANIZATION ON STREAM WATER QUALITY 2875
NO
3N (
mg
l-1)
0
1
2
(d) (j)
Cu
(µ
g l-1
)
101
100
10-1
102
(k)
Pb
(µ
g l-1
)
100
10-1
10-2
101
(f)
4
5
6
7
8
9
10
pH
NA
N3
NA
N2
NA
N1
LUL1
WO
O1
PE
A5
PE
A4
PE
A3
PE
A2
PE
A1
PR
O2
PR
O3
PR
O1
SA
N1
UT
O3
UT
O2
UT
O1
INT
1S
OU
3S
OU
2S
OU
1P
MR
WM
TN
C
(c)
0
.4
2.0
1.2
1.6
.8
TP
(m
g l-1
)
(e)
0
10
20
30
Si (
mg
l-1)
(i)
100
101
102
103
104
105
EC
(C
ou
nts
100
ml-1
)
(b)
PO
4P (
mg
l-1)
0
.2
.4
.6
(a)
Q (
mm
/d)
101
100
10-1
102
103GRABISCO
(g)
TU
RB
(F
NU
)
101
100
102
103
(h)
SS
(m
g l-1
)
101
100
102
103
(l)
Zn
(µ
g l-1
)
NA
N3
NA
N2
NA
N1
LUL1
WO
O1
PE
A5
PE
A4
PE
A3
PE
A2
PE
A1
PR
O2
PR
O3
PR
O1
SA
N1
UT
O3
UT
O2
UT
O1
INT
1S
OU
3S
OU
2S
OU
1P
MR
WM
TN
C
101
100
102
103
Figure 9. Distributions of (a) discharge (Q) and concentrations of (a) Q, (b) PO4P, (c) TP, (d) NO3N, (e) Si, (f) pH, (g) EC, (h) SS, (i) Cu, (j) Pb,and (k) Zn for GRAB and ISCO samples of streams in the city of Atlanta and two reference streams, WY2004–WY2007
began in 2007. The ISCO samples in 2007 had muchhigher and variable EC and TC. These samples werenot analyzed for FC and box plot of bacteria concen-trations (Figure 3z–ab) reflects this, i.e. the plots showlow concentration for FC but much higher and variableconcentrations for EC and TC.
The cause of the high concentrations of Cu, Pband Zn can be qualified somewhat by the relationswith sampling method. Cu, Pb and Zn are significantly
higher in the ISCO samples than in the base flowdominated GRAB samples (e.g. Figure 9j–l). Also, ahigher percentage of samples exceed Georgia waterquality standards for these metals in the higher flowISCO samples than in the lower flow GRAB or EWIsamples (Table III). These metals likely originate inrunoff during rainstorms, possibly from washoff of themetal that accumulates on impervious surfaces (Revittand Ellis, 1980; Ellis et al., 1986; Davis et al., 2001;
Published in 2009 by John Wiley & Sons, Ltd. Hydrol. Process. 23, 2860–2878 (2009)DOI: 10.1002/hyp
2876 N. E. PETERS
Rose et al., 2001; Sansalone et al., 2005). However, Cu,Pb and Zn concentrations although highly correlated witheach other, generally are not correlated with dischargeor TURB. Furthermore, the metal concentrations ofthe ISCO samples show hysteresis with the highestconcentrations generally occurring during the beginningof stormflow, but not for every storm. The storms withhigh concentrations at the beginning of the storm areconsistent with a first-flush phenomenon as observedin urban streams elsewhere (Sansalone and Buchberger,1997; Lee et al., 2005). However, the concentrationsdischarge relations vary with the storms and it is likelythat the variation is caused by the routing and mixingof runoff, which would be partly dependent on thespatial and temporal rainfall characteristics. Also, theconcentrations vary by site and storm, which is consistentwith results of a first-flush assessment of some Texasstreams that displayed a more consistent pattern of firstflush of metals than at the COA streams (Appel andHudak, 2001).
For the few ISCO storm samples collected at SOU3during 2007 after the RT station was installed, metalconcentrations were lower in the ISCO compared withthe GRAB samples. The pH values are also higher in theISCO samples than GRAB samples at SOU3 indicatingthat the acidification due to dissolution of alum occursduring base flow, when groundwater contributions are thehighest. These results suggest that groundwater near thealum-manufacturing plant is highly acidic and containshigh metal concentrations, a result that has not beenconfirmed.
CONCLUSIONS
A long-term stream water quality monitoring networkwas established in the COA during 2003 to assess base-line conditions and the effects of several managementoperations on stream water quality. Routine hydrologi-cally based stream sampling was conducted at 21 stations,with drainage areas ranging from 3Ð7 to 232 km2; 11 ofthe stations are RT water quality stations having contin-uous measures of stream stage (which is converted todischarge), pH, DO, SC, water temperature and TURB,and automatic samplers for stormwater collection. Thesamples are analyzed for field parameters and a broadsuite of dissolved and sediment-related constituents. Theobjectives of this study are the following: (1) to eval-uate water quality differences among COA streamwatersampling stations with respect to field parameters andconcentrations of the dissolved major ions, minor andtrace metals, nutrients and bacteria; (2) to compare COAstream water quality for a subset of parameters to thoseat nearby reference stations including a small (0Ð4 km2)relatively undisturbed forested watershed and a larger(10 km2) low-density residential watershed; and (3) toevaluate the water quality differences among and withinCOA sampling stations with respect to differences inland-use characteristics and other environmental factorsincluding streamflow.
Urbanization affects water quality. The drainage areasof the 21 COA stations are highly urbanized with 69 to96% urban land use. Concentrations of most constituentswere significantly higher at COA stations than at twonearby reference stations. Sufficient information aboutvariations in urban stream water quality derived fromroutine manual sampling, automatic sampling duringstormflows and RT water quality monitoring providedplausible hypotheses for observed differences in waterquality among streams in the COA.
For the COA stations, ALK, ALK to Si concentrationratios and Ca and Mg concentrations were much higherthan for the reference stations. These differences are con-sistent with a carbonate mineral weathering source of theCa, Mg and ALK in COA streams, which is hypothesizedto be concrete. For WOO1, the slope of a linear regres-sion of ALK on Si is much higher than the other COAstreams indicating another possible source. Also, RT pHwas very high (>9) during stormflows exceeding the pHassociated with carbonate dissolution, and the pH max-imum consistently lagged peak streamflow. The WOO1results indicate the possible dissolution and mobilizationof a Ca or Ca–Mg base or oxide such as lye from a pointsource in the headwaters.
High Cl concentrations occur at three stations andare hypothesized to be associated with discharges ofchlorinated CSO to Intrenchment Creek (INT1), drainingof swimming pool(s) in the Lullwater Creek basin(LUL1) and dissolution and transport during rainstormsof CaCl2, a deicing salt, on the South River (SOU1),which also was indicated by RT data. The headwaterstation on the South River (SOU3) is a highly degradedstream affected by the dissolution and transport ofammonium alum [NH4Al(SO4)2] from an alum plantupstream outside the COA boundary in East Point. SOU3has low pH (<5), low ALK and high concentrations ofSO4, NH4N, NO3N, Al, Mn, Cd and Zn, and no fishwere found in the stream during any survey conductedsince 2000. The effects of the alum on water qualitywere diminished, but observable at the two downstreamstations, SOU2 and SOU1.
More than 90% of the FC concentrations exceededthe US EPA standard of 200 MPN dl�1 for any waterusage class. Coliform bacteria concentrations (FC, TC,EC) were highly correlated among type and with TURBand SS concentrations. Nutrient concentrations generallywere high compared with the nearby reference stations,but were low compared with US EPA standards. Manysamples were below the analytical-method reporting limitfor PO4P, TP and NH4N. The streams are well aeratedand only a few samples showed low DO concentrations.
Concentrations of most of the major ions decreasedwith increasing streamflow as observed, but increasedwith increasing streamflow for several SS-related con-stituents such as TP, PO4P, TURB, SS and coliform bac-teria, particularly at RT stations. These relations also wereobserved in a comparison of the manual GRAB sam-ples and automated samples. The distribution of stream-flows for the GRAB samples was skewed towards base
Published in 2009 by John Wiley & Sons, Ltd. Hydrol. Process. 23, 2860–2878 (2009)DOI: 10.1002/hyp
EFFECTS OF URBANIZATION ON STREAM WATER QUALITY 2877
flow, despite attempts to collect samples over a range ofstreamflows; rainstorms typically occur at night and arehard to predict. In contrast, the automated samples tar-geted stormflows. Some trace metals (Cu and Pb) werehigher in automated storm samples than in the manualsamples indicating mobilization during rainstorms, butconcentration–discharge relations varied among storms.
ACKNOWLEDGEMENTS
This study is a contribution of the city of Atlanta Long-Term Water Quality Monitoring Program and was con-ducted in cooperation with the city of Atlanta. I amgrateful to the technical oversight provided by a tech-nical advisory committee, which is chaired by UpperChattahoochee Riverkeeper and has representatives frompublic agencies including Cobb County—Marietta WaterAuthority, Gwinnett County Department of Public Util-ities, Georgia Environmental Protection Division andUS Environmental Protection Agency and universitiesincluding Georgia Institute of Technology, the Univer-sity of Georgia and Georgia State University. Any useof trade, firm or product names is for descriptive pur-poses only and does not imply endorsement by the USgovernment.
REFERENCES
Appel PL, Hudak PF. 2001. Automated sampling of stormwaterrunoff in an urban watershed, north-central Texas. Journal ofEnvironmental Science and Health Part A-Toxic/Hazardous Substances& Environmental Engineering A36(6): 897–907.
Arienzo M, Adamo P, Bianco MR, Violante P. 2001. Impact of land useand urban runoff on the contamination of the Sarno River Basin insouthwestern Italy. Water Air and Soil Pollution 131(1–4): 349–366,DOI: 10.1023/A:1011908019933.
Atlanta Regional Commission. 2004. GIS coverage of land use:LandPro2001 . Atlanta Regional Commission, Atlanta, GA, 1: 14 000scale.
Atlanta Regional Commission. 2008. Atlanta Regional Commis-sion, Atlanta, GA. Accessed online 20th February, 2009, athttp://www.atlantaregional.com/html/205.aspx.
Barco J, Papiri S, Stestrom MK. 2008. First flush in a combined sewersystem. Chemosphere 71: 827–833, DOI:10.1016/j.chemosphere.2007.11.049.
Booth DB, Jackson CR. 1997. Urbanization of aquatic systems:degradation thresholds, stormwater detection, and the limits ofmitigation. Journal of the American Water Resources Association 33:1077–1090.
Brilly M, Rusjan S, Vidmar A. 2006. Monitoring the impact ofurbanisation on the Glinscica stream. Physics and Chemistry of theEarth 31: 1089–1096.
Brown JN, Peake BM. 2006. Sources of heavy metals and polycyclicaromatic hydrocarbons in urban stormwater runoff. Science of the TotalEnvironment 359: 145–155.
Carle MV, Halpin PN, Stow CA. 2005. Patterns of watershedurbanization and impacts on water quality. Journal of the AmericanWater Resources Association 41: 693–708.
City of Atlanta. 2009. Clean Water Atlanta j Water Quality Monitoringj Long-Term Watershed Monitoring Program . Accessed online20th February, 2009, at http://www.cleanwateratlanta.org/monitoring/Improv/LTWM.htm.
Davis AP, Shokouhian M, Ni S. 2001. Loading estimates of lead, copper,cadmium, and zinc in urban runoff from specific sources. Chemosphere44: 997–1009, DOI:10.1016/S0045-6535(00)00561-0.
Deletic A. 1998. The first flush load of urban surface runoff. WaterResearch 32: 2462–2470.
Drever JI. 1997. The Geochemistry of Natural Waters: Surface andGroundwater Environments , 3rd edn. Prentice Hall: Englewood Cliffs,NJ; 438.
Driscoll CT, Lawrence GB, Bulger AJ, Butler TJ, Cronan CS, Eagar C,Lambert KF, Likens GE, Stoddard JL, Weathers KC. 2001. Acidicdeposition in the northeastern United States: sources and inputs,ecosystem effects, and management strategies. Bioscience 51(3):180–198.
Driver NE, Troutman BM. 1989. Regression models for estimating urbanstorm-runoff quality and quantity in the United States. Journal ofHydrology 109: 221–236.
Duda AM, Lenat DR, Penrose DL. 1998. Water quality in urbanstreams—what we can expect. Journal of the Water Pollution ControlFederation 54(7): 1139–1147.
Ellis JB. 2004. Bacterial sources, pathways and management strategiesfor urban runoff. Journal of Environmental Planning and Management47: 943–958.
Ellis JB, Harrop DO, Revitt DM. 1986. Hydrological controls ofpollutant removal from highway surfaces. Water Research 20(5):589–595.
Fishman MJ, Friedman LC (eds). 1989. Methods for determination ofinorganic substances in water and fluvial sediments. U.S. GeologicalSurvey Technical Water Resources Investigations , 3rd edn, Book 5,Chapter A1. U.S. Government Printing Office: Washington, DC.
Garcia-Armisen T, Servais P. 2007. Respective contributions of point andnon-point sources of E. coli and enterococci in a large urbanizedwatershed (the Seine river, France). Journal of EnvironmentalManagement 82: 512–518.
Georgia Environmental Protection Division (GA EPD). 2005. Rules andRegulations for Water Quality Control. Accessed online 22nd February,2009, at http://rules.sos.state.ga.us/docs/391/3/6/03.pdf.
Gnecco I, Berretta C, Lanza LG, Barbera PL. 2005. Storm waterpollution in the urban environment of Genoa, Italy. AtmosphericResearch 77: 60–73.
Gregory MB, Calhoun DL. 2007. Physical, chemical, and biologicalresponses of streams to increasing watershed urbanization in thePiedmont Ecoregion of Georgia and Alabama, 2003 . U.S. GeologicalSurvey Scientific Investigation Report 2006-5101-B; 104.
Gromaire-Mertz MC, Garnaud S, Gonzalez A, Chebbo G. 1999. Char-acterisation of urban runoff pollution in Paris. Water Science andTechnology 39: 1–8.
Gupta K, Saul AJ. 1996. Specific relationship for the first flush load incombined sewer flows. Water Research 30: 1244–1252.
Higgins MW, Crawford RF. 2006. Geologic Map of the Atlanta 30-minute ð 60-minute Quadrangle. The Geologic Mapping Institute andthe Atlanta Geological Society: Georgia, Map 1, ver. 4Ð1Ð0.
Horowitz AJ, Elrick KA, Smith JJ. 2007. Monitoring urban impactson suspended sediment, trace element, and nutrient fluxes withinthe city of Atlanta, Georgia, USA: program design, methodologicalconsiderations, and initial results. Hydrological Processes 22(10):1473–1496, DOI: 10.1002/hyp.6699.
Horowitz AH, Hughes WB. 2006. The U.S. Geological Survey and Cityof Atlanta water quality and water-quantity monitoring network . U.S.Geological Survey Fact Sheet 2005–3126; 4.
Hudak PF, Banks KE. 2006. Compositions of first flush and compositestorm water runoff in small urban and rural watersheds, north-centralTexas. Urban Water Journal 3: 43–49.
Knaak AE, Ankcorn PD. 2003. Quality assurance and quality con-trol procedures for an urban water quality monitoring program,Gwinnett County, Georgia. Proceedings of the 2003 Georgia WaterResources Conference. University of Georgia: Athens, Georgia,(http://cms.ce.gatech.edu/gwri/uploads/proceedings/2003/Knaak%20and%20Ankcorn.pdf).
Lara-Cazenave MB, Castetbon A, Potin-Gautier M, Astruc M. 1994a.Pollution of urban runoff waters by heavy metals. Part II: speciation.Environmental Technology 15: 1149–1159.
Lara-Cazenave MB, Levy V, Castetbon Potin-Gautier M, Astruc M,Albert E. 1994b. Pollution of urban runoff waters by heavy metals.Part I: total metal. Environmental Technology 15: 1135–1147.
Lawrence SJ. 2006. Chattahoochee River BacteriALERT . USGS GeorgiaWater Science Center: Atlanta, GA. Accessed online 29th September,2008, at http://ga2.er.usgs.gov/bacteria/SummaryAll.cfm.
Lee BC, Matsui S, Shimizu Y, Matsuda T. 2005. Characterizations of thefirst flush in storm water runoff from an urban roadway. EnvironmentalTechnology 26(7): 773–782.
Makepeace DK, Smith DW, Stanley SJ. 1995. Urban stormwater quality:summary of contaminant data. Critical Reviews in EnvironmentalScience and Technology 25: 93–139.
Published in 2009 by John Wiley & Sons, Ltd. Hydrol. Process. 23, 2860–2878 (2009)DOI: 10.1002/hyp
2878 N. E. PETERS
Mallin MA, Wheeler TL. 2000. Nutrient and fecal coliform dischargefrom coastal North Carolina golf courses. Journal of EnvironmentalQuality 29: 979–986.
Matschel T, Lothenbach B, Glasser FP. 2007. The role of calciumcarbonate in cement hydration. Cement and Concrete Research 37(4):551–558, DOI:10.1016/j.cemconres.2006.10.013.
Meybeck M. 1994. Origin and variable composition of present dayriverborne material. National Research Council (U.S.) Board on EarthSciences, Material Fluxes on the Surface of the Earth. NationalAcademies Press: Washington, DC; 61–73.
Mosley LM, Peake BM. 2001. Partitioning of metals (Fe, Pb, Cu, Zn) inurban run-off from the Kaikorai Valley, Dunedin, New Zealand. NewZealand Journal of Marine and Freshwater Research 35: 615–624.
Mulliss RM, Revitt DM, Shutes RB. 1996. The impacts of urbandischarges on the hydrology and water quality of an urban watercourse.Science of the Total Environment 189/190: 385–390.
O’Keefe B, D’Arcy BJ, Davidson J, Barbarito B, Clelland B. 2005.Urban diffuse sources of faecal indicators. Water Science andTechnology 51: 183–190.
O’Shea ML, Field R. 1992. Detection and disinfection of pathogensin storm-generated flows. Canadian Journal of Microbiology 38:267–276.
Peters NE. 2008. Water quality monitoring and process understandingin support of environmental policy and management. Proceedingsof the 9th Kovacs Colloquium on River Basins—From HydrologicalScience to Water Management/Les Bassins Versant—de la ScienceHydrologique a la gestion des eaux . International Association ofHydrological Sciences IAHS Red Book 323 (IAHS) Press: WallingfordParis, France; 93–109.
Peters NE, Meybeck M. 2000. Water quality degradation effectson freshwater availability: impacts of human activities. WaterInternational 25(2): 185–193.
Poff NL, Allan JD, Bain MB, Karr JR, Prestegaard KL, Richter BD,Sparks RE, Stromberg JC. 1997. The natural flow regime—a paradigmfor river conservation and restoration. BioScience 47: 769–784.
Rantz SE. 1982. Measurement and computation of streamflow: volume1, measurement of stage and discharge and volume 2, computation ofdischarge. U.S. Geological Survey Water-Supply Paper, 2175.
Revitt DM, Ellis JB. 1980. Rain water leachates of heavy metals in roadsurface sediments. Water Research 14(10): 1403–1407.
Rose S. 2007. The effects of urbanization on the hydrochemistry of baseflow within the Chattahoochee River Basin (Georgia, USA). Journalof Hydrology 341: 42–54, DOI:10.1016/j.hydrol.2007.04.019.
Rose S, Crean MS, Sheheen DK, Ghazi AM. 2001. Comparative zincdynamics in Atlanta metropolitan region stream and street runoff.Environmental Geology 40(8): 983–992, DOI:10.1007/s002540100285.
Rose S, Peters NE. 2001. Urbanization effects on streamflow inAtlanta Region (Georgia, USA): a comparative hydrological approach.Hydrological Processes 15(8): 1441–1457.
Sansalone JJ, Buchberger SG. 1997. Partitioning and first flush of metalsin urban roadway storm water. Journal of Environmental Engineering123(2): 134–143.
Sansalone JJ, Hird JP, Cartledge FK, Tittlebaum ME. 2005. Event-based stormwater quality and quantity loadings from elevated urbaninfrastructure affected by transportation. Water Environment Research77: 348–365.
Schoonover JE, Lockaby BG, Helms BS. 2006. Impacts of land coveron stream hydrology in the west Georgia Piedmont, USA. Journal ofEnvironmental Quality 35: 2123–2131, DOI:10.2134/jeq2006.0013.
Soller J, Stephenson J, Olivieri K, Downing J, Olivieri AW. 2005.Evaluation of seasonal scale first flush pollutant loading andimplications for urban runoff management. Journal of EnvironmentalManagement 76: 309–318, DOI:10.1016/j.jenvman.2004.12.007.
Stumm W, Morgan JJ. 1996. Aquatic Chemistry , 3rd Ed., John Wiley &Sons, New York.
US Environmental Protection Agency. 1986. Quality criteria for water1986. U.S. Environmental Protection Agency Report 440/5-86-001,Office of Water: Washington, DC.
US Environmental Protection Agency (US EPA). 2009. NationalRecommended Water Quality Criteria. Accessed online on 11th June2009, at http://www.epa.gov/waterscience/criteria/wqctable/.
US Geological Survey, variously dated, National field manual for thecollection of water quality data: U.S. Geological Survey Techniquesof Water-Resources Investigations, book 9, chaps. A1-A9. Accessedonline 11th June 2009, http://pubs.water.usgs.gov/twri9A.
van Buren MA, Watt WE, Marsalek J. 1997. Application of the log-normal and normal distributions to stormwater quality parameters.Water Research 31(1): 95–104.
Wagner RJ, Boulger RW Jr, Oblinger CJ, Smith BA. 2006. Guidelinesand standard procedures for continuous water quality monitors: stationoperation, record computation, and data reporting. U.S. GeologicalSurvey Techniques and Methods 1—D3. Accessed online 11th June2009, at http://pubs.water.usgs.gov/tm1d3.
Published in 2009 by John Wiley & Sons, Ltd. Hydrol. Process. 23, 2860–2878 (2009)DOI: 10.1002/hyp
