T. GalléMichael Bayerle, Denis Pittois, Andreas Krein
Luxembourg Institue of Science and Technology
Contact: [email protected]
Emission inventories for priority
substances at catchment levels:
Solving the PAH source conundrum
with an array of in‐stream tools
• All surveillance sites in Luxembourg exceed EQS for high-molecular weight PAH
• Same trends in the neighboring regions (Rhine-Mosel Commission)
• PAH often considered as ubiquitous with important atmospheric immission
• Fatalistic attitude upon improving the situation (little concrete measures in RBMP)
• Scarce efforts to investigate spatial differentitation and sources more thoroughly
2
BAD CHEMICAL STATUSPAH EQS exceedance most common reason
• Exports from catchments > current atmospheric deposition (urban areas)
• Street deposits major source
• Soils often secondary source
• Accumulation in soils in vicinity of traffic
• Building up of stocks in sewers (first flushs)
• Role of combustion derived carbonaceous particles
• Contaminated industrial sites (historical)
3
PAH SOURCES AND DYNAMICSWhat the literature says
One scheme for all compounds?
EMISSION INVENTORIES
4 • Is this adapted to secondary pollutants with diffuse sources and a strong affinity for solids?
• Substance flow analysis in different catchment
• Establishment of Q-C relationships
• Calculation of catchment loads
• Characterization of pollution level in different hydrological situations
• Contribution of WWTPs and urban runoff
5
0 20 40 60 80 100 120
0
100
200
300
400
500
Data: Data1_B
Model: Allometric1
Equation:
y = a*x^b
Weighting:
y No weighting
Chi^2/DoF = 1240.40378
R^2 = 0.76735
a 0.92937 ±0.34971
b 1.32497 ±0.09034
Susp. matter [mg/L]
Su
sp
. m
atte
r [m
g/L
]
Discharge [m3/s]
Ettelbruck
0 5 10 15 20 25 30 35 40 45 50 55
-1000
0
1000
2000
3000
4000
5000
6000
7000
y = a*x^b
R^2 = 0.98238
a 0.02843 ±0.03588
b 3.11534 ±0.32406
Discharge [m3/s]
P
AH
[n
g/l]
Sum PAH
Ettelbruck
SUBSTANCE FLOW ANALYSISRegionalized emission balances
6
0 100 200 300 400
0
5
10
15
20
25
30 Ettelbruck
P
AH
so
lid [m
g/k
g]
Susp. M. [mg/l]
Sum PAH solid > 15 mg/L susp. m
0 50 100 150 200 250 300 350
10
20
30
Steinsel
P
AH
so
lid [m
g/k
g]
Susp. M. [mg/l]
PAH solid > 15 mg/L Susp M
0 50 100 150 200 250 300 350
0
5
10
15
20
25
30
35
Hunnebour
P
AH
solid
[m
g/k
g]
Susp. M. [mg/l]
Sum PAH solid > 15 mg/L susp. matter
SUSPENDED MATTER POLLUTIONSource discrimination
• Solid contamination levels allow for objective comparison in different hydrological situations and catchments (SPM as main carrier)
• High levels of SPM indicate strong catchment wide erosion and background levels of (alluvial) contamination
Steinsel
2002-2003
Ettelbrück
2002-2003
Steinsel
2003-2004
Ettelbrück
2003-2004
Yearly discharge [Mm3] 183 401 99 210
Yearly load PAH [kg] 241 824 56 123
Yearly average
concentration PAH [µg/L]
1.32 2.05 0.58 0.58
Yearly average
concentration SPM [mg/L]
95 160 33 37
7
0
50
100
150
200
250
01/04/2002 01/08/2002 01/12/2002 01/04/2003
0
5
10
15
20
25
Da
ily d
isch
arg
e [M
m3]
Ave
rag
e s
olid
PA
H 1
6 [m
g/k
g]
Daily discharge [Mm3]
Daily average PAH solid [mg/kg]Ettelbruck
Daily PAH 16 load [kg]
Da
ily P
AH
16
lo
ad
[kg
]
2002-2003
Discharge 401 Mm3
PAH 16 load 824 kg
Average PAH 16 conc. 2.05 g/L
0
50
100
150
200
250
01/04/2003 01/08/2003 01/12/2003 01/04/2004
0
5
10
15
20
25
2003-2004
Discharge 210 Mm3
PAH 16 load 123 kg
Average PAH 16 conc. 0.58 g/L
Da
ily P
AH
16
lo
ad
[kg
]
Da
ily d
isch
arg
e [M
m3]
Ave
rag
e s
olid
PA
H 1
6 [m
g/k
g]
Daily discharge [mM3]
Daily average PAH solid [mg/kg] Ettelbruck
DAily PAH 16 load [kg]
CATCHMENT BALANCESYearly variability - uncertainties
• Yearly loads are governed by SPM yield – read: hydrological events
• SPM pollution levels is the more objective measure
• Contribution of WWTPs:
medians of measurement
campaigns
• Contribution of surface runoff
• Combined sewers
Difference between
effective precipitation
on impervious surfaces
and discharge of WWTPs
• Separative sewers: Effective
precipitation
• Median concentrations:
measurements/literature
8
A
Schifflange
B
Bettembourg
C
Bonnevoie
D
Beggen
E
Mersch
Beringen
0
5
10
15
20
25
WWTP outlet SPM
PA
H s
olid
[m
g/k
g]
A
Schifflange
B
Bettembourg
C
Bonnevoie
D
Beggen
E
Mersch
0
100
200
300
400
500
PA
H 1
5 [n
g/l]
Schifflange
Bettembourg
Bonnevoie
Beggen
Mersch
WWTP outlets
CONTRIBUTION BY URBAN AREASWWTPs vs runoff pollution
9
3.2 (1.75%)
5.7 (3.12%)
31.9 (17.45%)
142 (77.68%)
Alzette by Difference
WWTP
CSO
Stormwater
Steinsel discharge [Mm3/y]
2002-2003
133 (12.14%)
249 (22.72%)138 (12.59%)
576 (52.55%)
Alzette by Difference
WWTP
CSO
Stormwater
Steinsel Cutot
loads [kg/y]
2002-2003
11.65 (6.37%)
15.9 (8.69%)
5.52 (3.02%)149.93 (81.93%)
Alzette by Difference
WWTP
CSO
Stormwater
Steinsel PAH loads [kg/y]
2002-2003
Steinsel PAH loads [kg/y]
2003-2004
8.76 (15.64%)
12.56 (22.43%)5.34 (9.54%)
29.34 (52.39%)
Alzette by Difference
WWTP
CSO
Stormwater
CONTRIBUTION PATTERNSRunoff as the main source?
• The contribution of WWTPs and urban runoff is much smaller for PAH than for metals
• The contribution is largely dependent on the hydrological season (wet year vs. dry year)
• Follow-up project on
urban runoff through in-
stream balances
• 3 triggerd autosamplers
in a longitudinal profile
from industrial region to
strongly urbanized
segments
• Event based balancing
and peak analysis in
flood waves
10
FLOOD EVENTSSource mobilisation and transport dynamics
• Can we depict fast urban runoff contributions in chemographs?
11
0 1 2 3 4 5
0
50
100
150
Model: Allometric1
Equation:
y = 63,96336*x^-0,23224
R^2 = 0.87077
EM
C C
u s
olid
[m
g/k
g]
Q tot [Mm3]
Livange
0 1 2 3 4 5
0
10
20
30
40
Data: EMC_J
Model: Allometric1
Equation:
y = a*x^b
Weighting:
y No weighting
Chi^2/DoF = 13.59298
R^2 = 0.77278
a 20.9788 ±1.73531
b -0.19328 ±0.05202
Sum 16 PAH [mg/kg]
Eve
nt
me
an
co
nce
ntr
atio
n [
mg
/kg
]
Q tot [Mm3]
Livange
0 1 2 3 4 5 6 7
0
50
100
150
200
Model: Allometric1
Equation:
y = 73,93306*x^-0,34051
R^2 = 0.70805
EM
C C
u s
olid
[m
g/k
g]
Q tot [Mm3]
Hesperange
0 1 2 3 4 5 6
0
10
20
30
40
Eve
nt
me
an
co
nce
ntr
atio
n [
mg
/kg
] Sum 16 PAH [mg/kg]
Q tot [Mm3]
Hesperange
0 1 2 3 4 5 6 7 8
0
50
100
150
200
250
Cu solid [mg/kg]
Eve
nt
me
an
co
nce
ntr
atio
n [
mg
/kg
]
Q tot [Mm3]
Pfaffenthal
0 1 2 3 4 5 6 7 8
0
10
20
30
40
Data: EMC_J
Model: Allometric1
Equation:
y = a*x^b
Weighting:
y No weighting
Chi^2/DoF = 14.56994
R^2 = 0.84231
a 22.75079 ±2.07217
b -0.28274 ±0.08227
Sum 16 PAH [mg/kg]
Eve
nt
me
an
co
nce
ntr
atio
n [
mg
/kg
]
Q tot [Mm3]
Pfaffenthal
EVENT MEAN CONCENTRATIONSFingerprinitng the sources
• EMC for Copper and PAH show higher solid contamination in small events (Cu > PAH)
• Small events have higher contributions of first-flushes vs catchment erosion
12
Liv
an
ge
He
sp
era
ng
e
Pfa
ffe
nth
al
0
50
100
150
200
EM
C C
u s
olid
[m
g/k
g]
Liv
an
ge
He
sp
era
ng
e
Pfa
ffe
nth
al
0
10
20
30
40
EM
C 1
6 P
AH
[m
g/k
g]
Eve
nt 1
Eve
nt 2
Eve
nt3
Eve
nt 4
+5
Eve
nt 6
Eve
nt 7
Eve
nt 8
Eve
nt 9
Eve
nt 1
0
Eve
nt 1
1
0
50
100
150
200
EM
C C
u s
olid
[m
g/k
g]
Livange
Hesperange
Pfaffenthal
Eve
nt 1
Eve
nt 2
Eve
nt3
Eve
nt 4
+5
Eve
nt 6
Eve
nt 7
Eve
nt 8
Eve
nt 9
Eve
nt 1
0
0
5
10
15
20
25
30
35
EM
C P
AH
16
so
lid [m
g/k
g]
Livange
Hesperange
Pfaffenthal
SPATIAL VARIABILITYLimited outreach of PAH pollution
• PAH pollution and Copper pollution behave differently in longitudinal profile
• Sources seem to be diverse and variably mobilisable
13
0
500
1000
1500
2000
2500
3000
0 50 100 150 200 250
0
1000
2000
3000
4000
5000
6000
7000
Dis
ch
arg
e [m
3/1
5 m
in]
time steps [15 min]
Discharge
Su
sp
. M
. [k
g/1
5 m
in]
Livange Wave 2 Real load ssp 129.90 t
FF load ssp 18.45 t (15%)
BG load ssp 97.14 t (77%)
FF2 load ssp 10.10 t (8%)
Sum load ssp 127.7 t (fit 0.97)
MASS FLOW COMPONENTSDeconvoluting flood waves
0
50
100
150
200
0 50 100 150 200 250
0
1000
2000
3000
4000
5000
6000
7000
Dis
ch
arg
e [m
3/1
5 m
in]
time step [15 min]
Discharge
Particle solid concentrations
[mg/kg]:
FF: 122
BG: 44
FF2: 161
Livange Wave 2
Cu
lo
ad
[g
/15
min
]
Real load Cu 9.07 kg
FF load Cu 2.26 kg (28%)
BG load Cu 4.27 kg (52 %)
FF2 load Cu 1.63 kg (20%)
Sum load Cu 8.16 kg (fit 0.9)
0
20
40
60
0 50 100 150 200 250
0
1000
2000
3000
4000
5000
6000
7000
Dis
cha
rge
[m
3/1
5 m
in]
time step [15 min]
Discharge
Particle solid concentrations
[mg/kg]:
FF: 16
BG: 25
FF2: 27
Livange Wave 2
PA
H lo
ad
[g
/15
min
]
Real load PAH 3.36 kg
FF load PAH 0.29 kg (10%)
BG load PAH 2.44 kg (81 %)
FF2 load PAH 0.27 kg (9%)
Sum load PAH 3.00 kg (fit 0.89)
• First we identify and fit SPM loads (turbidity signal)
• Then we allocate a pollution to each SPM source
14
Event
load
PAH
[kg]
First
Flush
load
fraction
[%]
FF
conc.
[mg/kg]
BG
conc.
[mg/kg]
Livange 3.36 29 % 16 (27) 25
Hesperange 3.61 22 % 33 (29) 18
Pfaffenthal 6.85 37% 48 (13) 21
0
20
40
60
80
100
120
0 50 100 150 200 250
0
2000
4000
6000
8000
10000
12000
Pfaffenthal wave 2
Solid concentration PAH:
FF: 48 mg/kg
BG: 21 mg/kg
FF2: 13 mg/kg
Dis
ch
arg
e [m
3/1
5 m
in]
time steps [15 min]
Discharge
PA
H s
olid
co
nc. [m
g/k
g]
Real load PAH 6.85 kg
FF load PAH 1.49 kg (21%)
BG load PAH 4.38 kg (63%)
FF2 load PAH 1.14 kg (16%)
Sum load PAH 7.01 kg (fit 1.02)
MASS FLOW COMPONENTSDeconvoluting flood waves
• First flush contributions can be below background contamination
• Contamination levels of first flush vs. background are variable downstream
• Country wide sampling with
sediment nets at low-flow
• Large variety of catchment
properties (land use)
15
Chier
s-Ath
us
Alzet
te-S
chiff
lang
e
Alzet
te-R
oese
r
Alzet
te-B
egge
n
Alzet
te-E
ssen
Mam
er-S
choe
nfels
Eisch
-Hun
nebu
r
Atte
rt-Bisse
n
Sur
e-Bigon
ville
Clerv
e-Kau
tenb
ach
Wiltz-
Kau
tenb
ach
Ern
z-blan
che-
Reisd
Sur
e-W
asse
rbilli
g
Syr
-Mer
tert
0
5
10
15
20
1
6 P
AH
[m
g/k
g]
Sum 16 PAH [mg/kg]
LARGE SCALE PICTURECatchment properties and PAH pollution
16
Chiers-Athus
Alzette-Schifflange
Alzette-Roeser
Alzette-Beggen
Mamer-SchoenfelsEisch-Hunnebur
Alzette-Essen
Attert-Bissen
Sure-Bigonville
Wiltz-Kautenbach
Clerve-Kautenbach
Ernz-blanche-Reisd
Sure-Wasserbillig
Syr-Mertert
0 10 20 30 40
0
20
40
60
80
100200
400
600
Su
sp
. m
att
er
co
nte
nt
[mg
/kg
]
Impermeable surface [%]
Lead [mg/kg]
Linear Fit of DATA1_D
Y = A + B * X
Parameter Value Error
------------------------------------------------------------
A 16.09032 6.36148
B 1.78357 0.35721
------------------------------------------------------------
R SD N P
------------------------------------------------------------
0.83298 14.30142 13 4.0696E-4
------------------------------------------------------------
Chiers-AthusAlzette-Schifflange
Alzette-Roeser
Alzette-Beggen
Mamer-Schoenfels
Eisch-Hunnebur
Alzette-EssenAttert-Bissen
Sure-Bigonville
Wiltz-Kautenbach
Clerve-Kautenbach
Ernz-blanche-Reisd
Sure-WasserbilligSyr-Mertert
0 10 20 30 40 50
0
5
10
15
20
Susp.
matt
er
conte
nt
[mg/k
g]
Impermeable surface [%]
Sum PAH 16 [mg/kg]
Linear Fit of DATA1_I
LOW FLOW PATTERNThe random nature of PAH sources
• Metals correlate well with impermeable surfaces while PAH do not at all
• Alluvial contaminated sites as probable sources for PAH
• Combined passive sampler campaign (SPM
nets & Empore disks for dissolved fraction)
• Longitudinal stretch of 4 km length with 3
monitored sites
• Different pollution sources
• Urban + historical background (Livange)
• Gas station (Berchem)
• WWTP with known PAH pollution (Roeser)
17
TRACKING THE SOURCESHow small scale is the problem?
18
0
5
10
15
20
25
30
25/06/2014 09/07/2014 23/07/2014
0
1
2
3
4
5
6
Dis
cha
rge
[m
3/s
]
Discharge [m3/s]
Turbidity [NTU]
Suspended matter 16 PAH [mg/kg]
Livange
Berchem
Roeser
1
6 P
AH
[m
g/k
g]
0
500
1000
1500
25/06/2014 09/07/2014 23/07/2014
0
1
2
3
4
5
6
Dis
cha
rge
[m
3/s
]
Discharge [m3/s]
Empore disk TWA [ng/L]
Turbidity [NTU]
Livange
Berchem
Roeser
1
6 P
AH
dis
solv
ed
[n
g/L
]
SPATIAL DISCRIMINATIONSources reveal under low flow
• Differences in SPM contamination can only be observed prior to the floodwave
• Empore disk TWA show highest input by dissolved PAH downstream of the fresh
sources (WWTP, Gas station)
19
lit value 30.6. 07.07. 16.07. 31.07.
2
3
4
5
6
7
8
9
10
Lo
g K
OC [L
/kg
OC
]
Livange
Berchem
Roeser
Fluoranthene
FAR FROM EQUILIBRIUMApparent log Koc higher than literature values
• Apparent log Koc calculated with SPM and Empore disks are 3 orders of magnitude
higher than expected from literature regressions
• Differences between the 3 sites are < 1 log unit and largest under low-flow conditions
20
SOURCE DISCRIMINATIONThe potential of log Koc to reveal aged contamination
• The south of Luxembourg is a historical steel working area
• SPM contamination and log Koc are decreasing downstream of the source
0
5000
10000
15000
20000
25000
30000
35000
40000
PAH Sum SPMf
(ng/g)
Alzette Longitudinal Profile
7.0
7.2
7.4
7.6
7.8
8.0
8.2
8.4
Log KOC (l/kg OC)
Alzette Longitudinal Profile
21
0 50 100 150 200
0.0
0.2
0.4
0.6
0.8
1.0
Fra
ctio
n in
wa
ter
[-]
Suspended matter conc. [mg/L]
log Koc =4
log Koc =6
log Koc =8
dissolved concentration
particulate conc.total extractable whole water sample
0 50 100 150 200 250 300 350 400 450 500
0.1
1
10
100
1000
Flu
ora
nth
en
e c
on
ce
ntr
atio
n [n
g/l]
Suspended matter conc. [ mg/L]
AA-EQS water
MAC-EQS water
predicted dissolved concentration
predicted whole water concentration
Fluoranthene measured
Fluoranthene
Cs 2 mg/kg
log Koc
field 8.5
foc
0.06
IMPLICATIONS FOR MOINTORINGRelevance of log Koc for whole water extraction?
• With field Koc up to 3 orders higher than expected suspended matter concentration
impacts EQS-evaluation heavily
• Are these EQS relevant under these conditions (bioavailability)?
• PAH EQS exceedences are one of the main reasons of chemical status failure
• Several investigations with different approaches suggest that urban runoff is not the main
source of PAH in Luxembourgish catchments
• Instead, suspended sediment profiles under low-flow suggest very localized
contaminations
• The outreach of these pollutions is very limited in longitudinal profiles (-> erratic
conclusions on upstream situation)
• Apparent log Koc of suspended sediments are 3 orders of magnitude higher than literature
values (implications for whole water sampling, SPM impact)
• Log Koc have the potential to discern fresh from old PAH pollution sources
• Longitudinal profiles at low flow with combined SPM and Empore disk sampling can be
used to pinpoint pollution sources
22
SUMMARY & CONCLUSIONS