- Email: [email protected]
Impact of inland shipping emissions on elemental carbon concentrations near waterways in The Netherlands
Accepted Manuscript Impact of inland shipping emissions on elemental carbon concentrations near waterways in the Netherlands M.P. Keuken , M. Moerman , J. Jonkers , J. Hulskotte , H.A.C. Denier van der Gon , G. Hoek , R.S. Sokhi PII:
S1352-2310(14)00454-3
DOI:
10.1016/j.atmosenv.2014.06.008
Reference:
AEA 13032
To appear in:
Atmospheric Environment
Received Date: 23 January 2014 Revised Date:
2 June 2014
Accepted Date: 3 June 2014
Please cite this article as: Keuken, M.P, Moerman, M, Jonkers, J, Hulskotte, J, Denier van der Gon, H.A.C, Hoek, G, Sokhi, R.S, Impact of inland shipping emissions on elemental carbon concentrations near waterways in the Netherlands, Atmospheric Environment (2014), doi: 10.1016/ j.atmosenv.2014.06.008. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT 1
Impact of inland shipping emissions on elemental carbon concentrations near
2
waterways in the Netherlands
3
Keuken, M.P.1, Moerman, M.1, Jonkers, J1, Hulskotte, J.1, Denier van der Gon, H.A.C.1,
4
Hoek, G2, Sokhi, R.S3
RI PT
5 (1) Netherlands Organization for Applied Research (TNO), Utrecht, the Netherlands
7
(2) Institute for Risk Assessment Sciences (IRAS), University of Utrecht, the Netherlands
8
(3) Centre for Atmospheric and Instrumentation Research (UH-CAIR), University of Hertfordshire,
9
Hatfield, United Kingdom
10
Corresponding author: [email protected]
SC
6
ABSTRACT
12
This study aims to quantify the impact of black carbon from inland shipping on air quality,
13
expressed as elemental carbon (EC) near inland waterways in the Netherlands. Downwind
14
measurements of particle numbers and EC were used to establish emission factors for EC
15
from inland shipping using inverse modelling. These emission factors were combined with
16
data on energy consumption to derive annual average emissions rates for all Dutch
17
waterways. A line source model was applied to compute the contribution of inland
18
shipping to annual average EC concentrations for around 140 000 people living within 200
19
m of busy waterways in the Netherlands. The results showed that they are exposed to
20
additional EC concentrations of up to 0.5 µg EC per m3 depending on the shipping volume
21
and distance from the waterway. In view of the envisaged growth in water transport, this
22
underlines the need to reduce combustion emissions from inland shipping. Targeting
23
“gross” polluters may be the most effective approach since 30% of ships cause more than
24
80% of the total emissions.
25
Keywords: black carbon, elemental carbon, inland shipping, emission factors
26
1.
27
The transport of goods over Europe’s 37,000 kilometres of waterways amounted to 7% of
28
the total inland goods transport in Europe in 2010 (EC, 2012). What is more, this share is
29
increasing because CO2 emissions per ton-kilometres over water are lower than those for
30
land transport by a factor of six (EC, 2012). Research into emissions from shipping and the
31
likely wider impact on air quality and climate change has been mainly directed at sea-
AC C
EP
TE D
M AN U
11
INTRODUCTION
ACCEPTED MANUSCRIPT going ships (Eyring et al., 2010; Bond et al., 2013). This impact relates to gaseous
33
emissions of CO2, SO2 and NOx, and particulate matter (PM) consisting of elemental and
34
organic carbon, sulphates and ash (Petzold et al., 2008; Moldanová et al., 2009). In urban
35
areas near harbours too, the likely impact on air quality has been the subject of attention
36
(Hulskotte and Denier van der Gon, 2010a; Tzannatos, 2010; Keuken et al., 2012a). Less
37
research has been conducted into emissions from inland waterways, probably because their
38
contribution to total shipping emissions is limited. However, inland shipping emissions
39
may be significant for air quality and the related health impact for people living nearby.
40
This may be particularly relevant in the Netherlands where 40% of goods were transported
41
over inland waterways in 2008 and 2009 (EC, 2012). The Netherlands has the largest
42
inland freight fleet in Western Europe with 5,815 vessels out of 11,546 (IVR, 2013).
43
PM emissions from inland shipping fuelled by diesel oil are lower by about a factor of five
44
per kWh (Trozzi and De Lauretis, 2013) than those from seagoing vessels, which are
45
mainly fuelled by residual fuel oil. However PM emission factors per kWh from inland
46
ships are considerably higher than diesel truck engines due to less stringent emission
47
standards for inland ships. For the Netherlands, it has been estimated that PM emissions
48
per ton-kilometre (i.e. one ton of goods transported one kilometre) from water transport are
49
typically five times higher than from road transport (Hulskotte and Denier van der Gon,
50
2010b).
51
Most studies into PM emissions from shipping have been performed by making
52
measurements on-board or in ship exhaust plumes of seagoing ships, very often fuelled by
53
residual fuel oil (Petzold et al., 2008; Moldanová et al., 2009).
54
This paper presents a study to model the impact of inland shipping on air quality and health
55
near inland waterways in the Netherlands. This requires emission factors for inland ships
56
as an input for the model. However, these are not established on a routine basis, since there
57
are no regulatory obligations in this area, such as those that apply to road transport.
58
Emission factors for inland ships were derived from measurements using test banks of ship
59
engines. This resulted in an average emission factor of 0.5g of PM per kWh in 2000 and
60
0.4g of PM per kWh in 2010 (Klein et al., 2007; Hulskotte and Denier van der Gon,
61
2010b). However, there are some reservations about applying these emission factors in
62
dispersion modelling. Firstly, average emission factors may not be representative because
63
shipping emissions may vary considerably according to actual shipping conditions.
64
Secondly, black or elemental carbon (EC) is a more sensitive indicator for combustion
65
particles than PM due to the relatively high background concentrations of PM (Keuken et
AC C
EP
TE D
M AN U
SC
RI PT
32
ACCEPTED MANUSCRIPT 66
al., 2012b). In our study, emission factors for elemental carbon from inland shipping were
67
derived by measuring the black carbon in shipping plumes and using inverse modelling.
68
Finally, the contribution to annual average EC concentrations within 200m of busy inland
69
waterways in the Netherlands was estimated.
70
2.
EXPERIMENTAL METHODS
RI PT
71
2.1 Monitoring periods
73
Concentrations of black carbon and the total number of particles (PN) were monitored near
74
two major Dutch waterways, as well as the wind speed and direction. The two sites are
75
shown in Fig. 1, which also shows other important waterways in the Netherlands and
76
Europe. Also shown is a typical inland water vessel in the Netherlands. The average length
77
of these vessels is 110 m with a width of 11 m and a loading capacity of 3,000 tons (IVR,
78
2013).
M AN U
SC
72
79 80
Figure 1: Important inland waterways in Europe and the Netherlands and the two
81
measurement sites: east of the Amsterdam-Rhine Canal (1) and north of the river Waal (2);
82
a typical inland water vessel on the Amsterdam-Rhine Canal.
TE D
83
The first monitoring campaign took place at a site east of the Amsterdam-Rhine Canal,
85
four kilometres north of the town Breukelen. The water is kept at a fixed level and there is
86
hardly any water flow. Consequently, the average speed (“speed over ground”) of ships is
87
similar in both directions. The date of this monitoring campaign was selected because of
88
forecasted favourable meteorological conditions for inverse modelling (see: section 2.3):
89
wind directions which carried exhaust plumes perpendicularly across the canal, wind
90
speeds with limited temporal variability and no rainfall. The monitoring equipment was
91
located directly on the side of the canal with the sampling inlet about 4.5m above the
92
surface of the water. The distance down-wind of the exhaust plume and the height of the
93
exhaust pipe were measured manually on-site with a laser. The speed of each ship was
94
retrieved on-line from a website (www.marinetraffic.com).
95
The second monitoring campaign was at a site north of the river Waal about seven
96
kilometres east of the city of Nijmegen. The Waal is the main of the three distributaries
97
that the river Rhine splits into after crossing the Dutch border. Unlike the Amsterdam-
98
Rhine Canal, the Waal has a current and consequently ships use more power going
AC C
EP
84
ACCEPTED MANUSCRIPT upstream than vice versa. The second monitoring campaign was from 25 to 28 July 2013.
100
The measurements were taken directly on the side of the river with the sampling inlet
101
about 6.5 m above the water surface. In the first monitoring campaign the shipping data
102
were measured on-site manually, which is laborious and costly. Therefore, in the second
103
monitoring campaign, the relevant data from passing ships were retrieved and stored on-
104
line from the Automatic Identification System (AIS). The AIS is compulsory for all ships
105
over 300 gross tonnage and reports the GPS position and speed of a ship every second.
106
Passing ships were detected automatically, which enabled exhaust plumes from a large
107
number of ships to be measured unattended. The meteorological data and shipping data
108
were matched after the monitoring campaign to select passing ships during favourable
109
meteorological conditions. Only on 28 July were the meteorological conditions favourable
110
for downwind plume detection at the monitoring site. Due to the relatively large amount of
111
shipping traffic on the Waal exhaust plumes of passing ships mixed regularly and therefore
112
from the AIS data, only ships were selected where exhaust plumes of could be detected
113
separately. The details of the two monitoring campaigns are summarized in Table 1.
114
M AN U
SC
RI PT
99
2.2 Monitoring equipment
116
The concentration of the total number of particles (PN) were measured with a condensation
117
particle counter – CPC (TSI, type 3775) with a 50% cut-off at 4 nm, a measurement
118
frequency of 1 s and an accuracy of ± 10%. Black carbon (BC) concentrations were
119
measured with a multi-angle absorption photometer - MAAP (Thermo Scientific, model
120
5012) with a temporal resolution of 1 minute and a detection limit of 0.1 µg m-3. The BC
121
data were converted to elemental carbon (EC) with a conversion factor of 0.75. This
122
correction was applied because the MAAP overestimates EC when compared to
123
measurements with the reference thermal-optical method (Petzold and Schönlinner, 2004;
124
Chow et al., 2009; Keuken et al., 2013). The meteorological parameters were measured
125
with a Vaisala, weather transmitter WXT 510 (www.vaisala.com) for wind speed and
126
direction with a response time of 1 s. The distance between the monitoring equipment and
127
passing ships was measured manually with a laser during the first monitoring campaign
128
(Leica, CRF 1200) with an accuracy of ± 1 m up to 350 m. Hence, the uncertainty due to
129
measuring the distances and height of the exhaust pipe is less than 1%. The AIS data on
130
passing ships were collected during the second monitoring campaign with an AIS receiver
131
(COMAR systems). All the data collected were stored on a laptop computer.
AC C
EP
TE D
115
ACCEPTED MANUSCRIPT 132
2.3 Inverse modelling
133
The well-known Gaussian plume model provides the following relationship between the
134
concentration (C) of a pollutant directly downwind at ground level and the emission factor
135
(Q) from a stack emission:
136
Q = (C * π * σy * σz * v)/(exp[-1/2 (H/σz)2])
(1)
RI PT
137 138
with Q in µg s-1, C in µg m-3 and σy and σz the lateral and vertical plume standard
140
deviations in m, v the wind speed at height H in m s-1 and H the effective height of the
141
plume emission in metres.
142
The above equation (1) was used to derive emission factors of EC from the PN and BC
143
concentrations measured in the shipping plumes at the monitoring sites during the two
144
monitoring campaigns. The background concentrations measured before and after a
145
shipping plume passed over the monitoring sites were averaged and subtracted from the
146
measured concentrations. The effective plume height was the sum of the exhaust pipe on
147
the ship and the plume rise. These were estimated manually by a laser for each ship during
148
the first monitoring campaign. The averages were 2.5 m (exhaust pipe) and 4 m (plume
149
rise). These figures were also applied to all ships during the second monitoring campaign
150
since the fleet composition is similar on both waterways. The plumes downwind of the
151
ships may be characterized by the standard deviations of the plume concentrations in the
152
lateral and vertical directions (σy and σz). These are a function of the distance downwind of
153
the ship and atmospheric stability. The latter can be defined by the Pasquill Stability
154
Classes relating to wind speed and cloud cover during daytime (Pasquill, 1971). The
155
meteorological conditions (i.e. daytime, half cloud cover and wind speed over 2 m s-1)
156
during both monitoring campaigns were similar and classified as class B. The related
157
plume standard deviations may be computed from the following equations (2) and (3):
159
M AN U
TE D
EP
AC C
158
SC
139
σy = a * X * exp(b) + c
(2)
σz = α * X * exp(β)
(3)
160 161 162 163
with a: 0.038, b: 1.149, c: 3.3, α: 0.275, β: 0.903 and X the downwind distance from the
164
ship to the monitoring site (Pasquill, 1971). The results from the two monitoring sites are
165
presented in Table 2.
ACCEPTED MANUSCRIPT 166
Because wind directions during the monitoring campaigns were almost perpendicular to
168
the two waterways (75o for the Amsterdam-Rhine Canal and 60o for the Waal), the ship
169
plumes passed over the monitoring sites with the speed of the passing ships. Consequently,
170
the concentrations of PN and BC were measured over a cross-section of the passing plume.
171
In this cross-section, the concentrations follow a normal distribution in accordance to the
172
Gaussian plume model. The width of this cross-section may be estimated from four times
173
the lateral dispersion (σy) of a plume at the monitoring locations. The factor “four” takes
174
into consideration the fact that 95% of the observations in a normal distribution are
175
between the average and plus-and-minus two times the standard deviation. From Table 1, it
176
was derived that the width of the plumes during the first monitoring campaign are in the
177
order of 48 m: four times the average σy of 9 m and 14 m for northbound and southbound
178
ships respectively. For the second monitoring campaign, the width of the plumes was in the
179
order of 104 m: four times the average σy of 23 and 28 m for ships moving at the river
180
Waal downstream and upstream respectively. Considering the speed of the ships during the
181
first and second monitoring campaign – of around 3 and 4 m s-1 respectively – the time for
182
the passage of the plumes over the first and second monitoring sites was in the order of 16
183
and 26 s respectively.
184
The PN measurements were performed at a frequency of once every second, while the BC
185
measurements, which were converted to EC concentrations (see: section 2.2) had a
186
frequency of once every minute. Hence, the average concentration of EC in the plume may
187
be measured but the maximum concentration in the cross-section of the shipping plume
188
passing over the monitoring equipment was missing. Since the latter is required for inverse
189
modelling with equation (1), the maximum concentration per second of EC (EC-max) was
190
inferred from the ratio between the maximum PN peak per second (PN-max) and the
191
average PN concentration (PN-avg) by equation 4:
193
SC
M AN U
TE D
EP
AC C
192
RI PT
167
EC-max = (PN-max/PN-avg) * EC-avg
(4)
194 195
with EC-max and EC-avg in µg m-3, and PN-max and PN-avg in # cm-3. The ratio of the
196
maximum and average concentration in the cross-section of a plume is equal for all
197
pollutants. Because, this ratio is a function of atmospheric dilution during atmospheric
198
transport of the exhaust plume. Dilution (“the size of the plume”) is characterized by sigma
199
y and sigma z. It depends on atmospheric conditions and the downwind distance of the
ACCEPTED MANUSCRIPT 200
plume from the exhaust pipe. The concentration of a pollutant in a plume may also change
201
by other processes such as condensation or deposition. This is relevant for PN but it is
202
assumed that these additional processes both affect PN-max and PN-avg and therefore with
203
a limited effect on their ratio. After determining EC-max in each shipping plume, inverse
204
modelling was applied
205
waterways have been estimated.
206
2.4 Shipping emissions and dispersion modelling
207
The average emission factors at the two waterways (see: Section 2.3) were combined with
208
the number of ships per year to derive emission rates for EC in µg per meter and per
209
second for each of the two waterways. In the next step, available data for the energy
210
consumption at all waterways in the Netherlands were used to determine emission rates of
211
EC for these waterways. Finally, the annual contribution to EC concentrations within 200m
212
of the waterways was estimated with a line source model. Further details on the method
213
and the uncertainty are provided in the Sections 3.2-3.6.
214
3.
M AN U
SC
RI PT
and emission factors of EC for individual ships at the two
RESULTS AND DISCUSSION
215
3.1 PN and BC monitoring data
217
The PN and BC concentrations were measured during the two monitoring campaigns. The
218
average one-minute results are presented in Figure 2 with the results for EC as converted
219
from the BC concentrations, multiplied by ten for presentational reasons. The passage of
220
ships during the monitoring campaigns are also indicated..
EP
221
TE D
216
Figure 2: One minute average number of particles (#/cm3) (“PN”) and elemental carbon
223
(10*ng/m3) (“EC”) during the first (Amsterdam-Rhine Canal) and second (Waal)
224
monitoring campaign and passing ships in both directions.
225
AC C
222
226
Fig. 2 illustrates that the plume concentrations near the Amsterdam-Rhine Canal are higher
227
than near the Waal mainly because the down-wind distances of the exhaust plume to the
228
monitoring site was shorter(see: Table 1). In the first monitoring campaign all passing
229
ships were included in the data analysis but in the second campaign specific ships were
230
selected from the AIS data. This was required as the shipping intensity at the Waal resulted
231
in mixing of exhaust plumes and not all exhaust peaks could be detected due to the larger
232
distances to the ships and variability in the wind direction. The data indicate that emissions
ACCEPTED MANUSCRIPT 233
of black carbon and PN vary considerably between ships, as well as the ratio of PN
234
emissions to BC emissions. These measurements have been used to determine the EC-max
235
per shipping plume as presented in Table 3.
236 237
3.2 EC emission factor per ship
238
The PN and EC measurements were combined with shipping data to link the plume
239
concentrations to specific ships. After subtracting
240
measured before and after a plume passage over the monitoring equipment, the maximum
241
and average PN concentration and the average EC concentrations in each plume were
242
calculated. Subsequently, the maximum EC concentration in each plume was inferred, as
243
described in section 3.1. The maximum EC concentration and the plume characteristics σy
244
and σz (see: Section 2.3) for each plume were used to calculate the emission factor per ship
245
of EC in µg s-1, using equation (1). The average values are presented in Table 4 for each
246
waterway and each shipping direction.
247 248
The average emission factors in both directions at the Amsterdam-Rhine Canal are very
249
similar. This was expected because there is hardly any flow in the canal. The average
250
emission factor for ships going downstream on the Waal was in line with ships on the
251
Amsterdam-Rhine Canal, but by contrast ships travelling upstream on the Waal have
252
higher emission factors which is attributed to the use of more engine power.
253
The data in Table 4 show a large range in emission factors between ships. For both
254
directions on the two waterways, it was found that around 30% of all ships emit over 80%
255
of the total emissions. This may be related to parameters such as engine power, the age of
256
the engine, maintenance, and loaded versus unloaded vessels. In the Netherlands, the
257
BIVAS database has been established which combines traffic and freight data and includes
258
shipping activities on all inland waterways in the Netherlands related to the number of
259
ships and engine power (http://bivas.chartasoftware.com). Therefore, it is known that the
260
number of unloaded ships is considerably less than two-thirds of all vessels, which would
261
indicate that “loaded versus unloaded” ships is unlikely to explain the variation in emission
262
factors completely. Engine type and maintenance are other likely important factors in
263
explaining the variation in emission factors. This indicates that one out of three ships is a
264
“gross” polluter. This conclusion has implications for measures to reduce shipping
265
emissions by targeting a limited number of ships rather than the whole fleet.
266
AC C
EP
TE D
M AN U
SC
RI PT
the background concentrations
ACCEPTED MANUSCRIPT 3.3 EC emission rates for each waterway
268
The impact of shipping emissions on air quality near waterways was modelled with a line-
269
source model analogous to the dispersion of traffic emissions on air quality near
270
motorways (Wesseling et al., 2003; Beelen et al., 2010; Keuken et al., 2012a). As well as
271
meteorological parameters, the emission rate for air pollutants is required as an input for
272
the model. The emission rate was derived from the product of the emission factor for each
273
ship and the number of ships on the waterway during a specific time period. The emission
274
rate (E) may be computed from equation 5:
RI PT
267
275
E = (N/3600 * EF)/V
(5)
SC
276 277
with E in µg m-1 s-1, N the number of ships per hour, EF the emission factor per ship in µg
279
s-1 and V the speed over ground of each ship in m s-1. From the BIVAS data base, it was
280
known that in 2011 the average numbers of ships on the Amsterdam-Rhine Canal and the
281
Waal were 105,000 and 175,000 ships, respectively or 12 and 20 ships per hour.
282
Subsequently, the average EC emission rates (EEC) at the monitoring sites near the
283
Amsterdam-Rhine Canal and the Waal in each shipping direction have been calculated
284
using equation 5. The emission factor for each ship (EF in equation 5) was computed as
285
described in Section 3.2. The average EC emission rates (EEC) are presented in Table 5.
TE D
286
M AN U
278
The next step was to generalize these derived emission rates at the monitoring sites near
288
the Amsterdam-Rhine Canal and the Waal in Table 5 to all Dutch waterways. The BIVAS
289
data base provided the annual energy consumption for a specific length of a waterway: at
290
the monitoring sites near the Amsterdam-Rhine Canal 0.024 TJ m-1 and near the Waal 0.05
291
TJ m-1 respectively. Considering the average EC emission rates at both locations in Table
292
5, a conversion factor may be derived between EC emission rates and energy consumption.
293
At the monitoring sites near the Amsterdam-Rhine Canal and the Waal, the conversion
294
factors are 0.0074 and 0.0058 µg EC per TJ energy consumption per s. This provides an
295
average conversion factor of 0.0066 between annual EC emission rates and energy
296
consumption on Dutch inland waterways.
AC C
EP
287
297 298
3.4 Uncertainty of the established emission factors and rates
299
The emission factors have been derived from measurements during the two monitoring
300
campaigns of about 2% of the total number of Dutch inland vessels. Whether the size of
ACCEPTED MANUSCRIPT 301
this sample was representative for the fleet has been assessed as follows. During the first
302
monitoring campaign, twenty seven and twenty eight
303
southbound, respectively at the Amsterdam-Rhine Canal were measured. The ships going
304
in both directions are assumed to be similar in fleet composition. Also, it is expected that
305
the engine load in both directions is similar as there is no current in the canal and the wind
306
direction was almost perpendicular to the shipping direction. The similarity in established
307
emission factors for both directions at the Amsterdam-Rhine Canal was tested with two
308
nonparametric statistical tests: the Wilcoxon’s T-test and Spearman’s correlation
309
coefficient. These nonparametric tests have been applied, as the large range in emission
310
factors presented in Table 4 suggests that the emission factors may not be normally
311
distributed. The result of the Wilcoxon test with a T value of 0.939 at a 95 percentage
312
confidence interval, indicates that the emission factors for both shipping directions at the
313
Amsterdam-Rhine Canal are similar. Also, the Spearman’s correlation coefficient of 0.999
314
shows that the established emission factors in both shipping directions are highly
315
correlated. Both statistical tests confirm that there is no difference in emission factors
316
derived from two samples of twenty seven and twenty eight
317
concluded that the number of fifty five
318
adequate to estimate shipping emissions at the Amsterdam-Rhine Canal.
319
The uncertainty to apply the established emission factors at other waterways than the
320
Amsterdam-Rhine Canal has been investigated as follows.
321
Canal, the annual energy consumption is 0.024 TJ m-1 (http://bivas.chartasoftware.com)
322
and the average emission rate is 3.2 µg EC m-1.s-1 (see: Table 5). This provides a ratio of
323
0.0074 between the energy consumption and the emission rate of EC. This ratio has been
324
applied to derive an emission rate of EC for the Waal from its energy consumption of 0.05
325
TJ m-1 near the monitoring location. This results in an emission rate of 6.8 µg EC m-1 s-1
326
which is 78% of the established emission rate of 8.7 µg EC m-1.s-1 (see: Table 5). From the
327
conversion ratios at the Amsterdam-Rhine Canal and the Waal, an average conversion ratio
328
was established of 0.0066 with a relative standard deviation of 17%. The latter provides an
329
estimate of the uncertainty to apply the average conversion factor to all Dutch waterways.
ships. It is therefore
ships in the first monitoring campaign was
At the Amsterdam-Rhine
AC C
EP
TE D
M AN U
SC
RI PT
ships going northbound and
330 331
3.5 Contribution from inland shipping emissions to EC concentrations near waterways
332
The conversion factor was applied to the BIVAS data base on energy consumption to
333
derive the annual average EC emission rates from inland waterways in the Netherlands.
334
These emission rates were projected at two line sources for each waterway as an input for
ACCEPTED MANUSCRIPT the line-source model (Wesseling et al., 2003; Beelen et al., 2010; Keuken et al., 2012a). It
336
is based on a Gaussian plume model which takes into account vehicle-induced turbulence,
337
the upwind roughness of the terrain and atmospheric stability. This model for road traffic
338
was applied to model the dispersion of black carbon shipping emissions but the vehicle-
339
induced turbulence was not included in the model. For canals, the emission rates of the two
340
line sources are similar while for rivers the emission rates differ for upstream and
341
downstream ships in accordance with the results for the Waal. The annual contribution to
342
EC concentrations were modelled for the Amsterdam-Rhine Canal and the Waal in 2011.
343
The results are presented as a function of the distance to the middle of the waterway in
344
Figure 3. For the Amsterdam-Rhine Canal, the line-sources are situated at 35 m and 60 m
345
from the eastern bank of the waterway and for the Waal at 165 m (downstream) and 195 m
346
(upstream) from the northern bank of the waterway. The reason that the down-wind
347
distances of the exhaust pipe to the monitoring stations in Table 2 are longer than the
348
distances of the line sources to the board of the waterway, is the angle between the wind
349
direction and the orientation of the waterway was not exactly perpendicular during the two
350
campaigns (see: Table 1) but 75o (Amsterdam-Rhine Canal) and 60o (Waal),
M AN U
SC
RI PT
335
351
Figure 3: The annual average contribution of shipping emissions to EC concentrations in
353
µg m-3 on both sides of the Waal (left = north; right = south) and the Amsterdam-Rhine
354
Canal (left = west; right = east) in 2011; straight (Waal) and dotted (Amsterdam-Rhine
355
Canal) lines indicate the banks of both waterways.
EP
356
TE D
352
Fig. 3 illustrates that the contribution of shipping emissions to air quality is not identical on
358
both sides of the two waterways as a result of prevailing wind directions and wind speeds
359
in the Netherlands. Accordingly, the impact on air quality is different north and south of
360
the Waal, and east and west of the Amsterdam-Rhine Canal. Fig. 3 also shows that the
361
concentration of shipping emissions beyond 200 m of the banks of the waterways is less
362
than 5% of the concentrations in the middle of the waterway. Hence, the contribution of
363
shipping emissions beyond 200 m from the waterway to the annual average concentrations
364
of EC may be regarded as insignificant. Using the line-source model, the contribution of
365
shipping emissions to annual average EC at each residential address within 200 m of the
366
inland waterways was calculated for the year 2011. The results at a spatial resolution of
367
25*25 m for each household are shown in Figure 4 for busy Dutch waterways. Busy
368
waterways in this study being defined as having - according to the BIVAS database - 78%
AC C
357
ACCEPTED MANUSCRIPT 369
of the total emissions from inland shipping in the Netherlands. These waterways represent
370
8% of the total length of waterways and 140 000 people live within 200 m of these
371
waterways. The black carbon emissions at these busy waterways are over 3 µg EC per s
372
per km, as derived from this study.
373
Figure 4: Additional annual average EC concentrations in µg m-3 for the population living
375
within 200 m of busy Dutch waterways. and the measurement sites near the Amsterdam-
376
Rhine Canal (1) and the Waal (2).
RI PT
374
377
Fig. 4 illustrates that near the Waal the additional on annual average EC concentrations is
379
in the range of 0.2 to 0.3 µg m-3 while further downstream and near the Amsterdam-Rhine
380
Canal the contribution is in the range of 0.01 to 0.1 µg m-3 EC. In Figure 5, the percentages
381
of the population are presented in accordance with different levels of additional EC
382
concentrations due to shipping emissions.
M AN U
SC
378
383 384
Figure 5: The percentage of households (%) within 200 m from busy waterways in the
385
Netherlands exposed to additional annual average EC (µg m-3) due to shipping emissions.
386
4.
387
A method has been presented to derive EC emission factors for individual ships and EC
388
emission ratesfrom inland waterways. The method is based on downwind measurements
389
of PN and black carbon which were converted to EC concentrations in shipping plumes
390
near waterways followed by inverse modelling. In this approach, it is assumed that
391
atmospheric dilution of PN and BC is similar in the few minutes of transport time from the
392
ship’s exhaust pipe to the monitoring location. Cooling of the exhaust plume will result in
393
particle formation (Fernández-Camacho et al., 2010) and change in PN concentration other
394
than dilution. However, it is assumed that this process does not affect significantly the
395
ratio of the maximum and average PN concentrations which enables to infer the maximum
396
EC concentration in the plume from PN and BC measurements.
397
The method was applied near the Amsterdam-Rhine Canal and the Waal in the Netherlands
398
during two monitoring periods under specific meteorological conditions (e.g. wind speeds
399
of over 2 m s-1 and a wind direction almost perpendicular to the waterway). Despite the
400
more labour-intensive strategy in the first campaign to collect on-site shipping data, it is
401
preferred over the automatic AIS system in the second campaign near a waterway with
AC C
EP
TE D
CONCLUSION
ACCEPTED MANUSCRIPT intense shipping traffic. The emission factors varied from 0.3 to 23.5 mg EC per s between
403
ships which is mainly attributed to differences in engine age and maintenance. The average
404
emission factors were 4.1 and 6.6 mg EC per s for ships going downstream and upstream
405
on the Waal and 3.6 and 3.8 mg EC per s for southbound and northbound ships on the
406
Amsterdam-Rhine Canal. The reproducibility of the method was better than 96% as
407
estimated from the emission factors derived separately for southbound and northbound
408
ships on the Amsterdam-Rhine Canal. The accuracy was 78% based on comparison of the
409
emission rates for the Waal by the method presented and by energy consumption.
410
The energy consumption data for busy Dutch waterways was converted into annual
411
average emission rates for EC. These emission rates were used as an input for a line source
412
model to calculate the contribution
413
concentrations up to 200 m from busy waterways. The shipping volume and the distance of
414
ships to the bank of a waterway are important parameters for the air quality near
415
waterways. The Waal, with about 175,000 ships per year, is the busiest waterway in the
416
Netherlands with an EC emission rate similar to that of a motorway with more than
417
120,000 vehicles per day. However, the distance of the ships from the bank of the river is
418
in the order of 135 m and therefore shipping emissions are considerably dispersed before
419
reaching the side of the waterway. The contribution to air quality near a waterway ranged
420
from 0.2 to 0.3 µg m-3 EC for a busy but wide waterway (360 m) such as the river Waal,
421
and from 0.01 to 0.1 µg m-3 EC for a less busy but relatively narrow waterway (110 m)
422
such as the Amsterdam-Rhine Canal.
423
The emissions factors measured in this study were compared to the emission factors used
424
in the official Netherlands emission inventory where an average emission rate of 36 kg
425
PM10 per TJ is applied for inland shipping (BIVAS). With an average contribution of 40%
426
EC in exhaust PM in diesel engines for road traffic, this would result in an emission rate of
427
9 kg EC per TJ or 14.4 µg EC m-1.s-1 at the Waal. This is about 50% higher than in this
428
study. PM emissions are known to be uncertain and a 50% discrepancy is not extreme.
429
However, it suggests that our exposure calculations will be a lower estimate, as the
430
emission factors used were lower than currently in-use emission factors. Therefore,
431
monitoring EC concentrations near waterways, as well as additional emission factor
432
measurements are recommended to verify the modelling results in this study.
433
Emission standards for shipping still lag behind the strict standards in place for road
434
transport. Such measures could be the introduction of LNG and low(er) sulphur content
435
fuels in this sector. In view of the relatively limited number of ships that likely cause more
SC
RI PT
402
AC C
EP
TE D
M AN U
of shipping emissions on annual average EC
ACCEPTED MANUSCRIPT 436
than 80% of the EC emissions, identifying and targeting “gross” polluters may be another
437
successful mitigation option. Follow-up research is recommended to detect these gross
438
polluters and understand the cause for their relatively high emissions. The long-term
439
monitoring of EC concentrations in cities near busy waterways with docking locations for
440
inland ships is also recommended to determine the actual exposure of people.
RI PT
441
5.
ACKNOWLEDGEMENTS
443
This work was supported by The Netherlands Ministry of Infrastructure and Environment
444
and the 7th European Framework project: TRANSPHORM directed to emissions,
445
dispersion and health impact of transport-related PM (Grant Agreement No. 243406).
446
We gratefully thank Ernst Bolt of the Ministry for Infrastructure and Environment
447
(Rijkswaterstaat Water, Verkeer en Leefomgeving) for access to the BIVAS data.
SC
442
M AN U
448
AC C
EP
TE D
449
ACCEPTED MANUSCRIPT 450
6.
REFERENCES
451 452
Bond, T.C., Doherty, S.J., Fahey, D.W., Forster, P.M., et al., 2013. Bounding the role of
453
black carbon in the climate system: a scientific assessment. Journal of Geophysical
454
Research 10.1002/jgrd.50171
RI PT
455
Beelen, R., Voogt, M., Duyzer, J., Zandveld, P., Hoek, G., 2010. Comparison of the
457
performances of land-use regression modelling and dispersion modelling in estimating
458
small-scale variations in long-term air pollution concentrations in a Dutch urban area.
459
Atmospheric Environment 44, 4614-4621
SC
456
460
Chow, J.C., Watson, J.G., Doraiswamy, P., Antony Chen, L-W, Sodeman, D.A.,
462
Lowenthal, D.H., Park, K., Arnott, W.P., Motallebi, N., 2009. Aerosol light
463
absorption, black carbon, and elemental carbon at the Fresno Supersite, California.
464
Atmospheric Environment 93, 874-887
M AN U
461
465
Eyring, V., Köhler, H.W., van Aardenne, J., Lauer, A., 2005. Emissions from international
467
shipping. The last 50 years. Journal of Geophysical Research 110, D17305.
468
doi:10.1029/2004jD005619
TE D
466
469
Eyring, V., Isaksen, I.S.A., Berntsen, T., Collins, W.J., Corbett, J.J., Endresen, O.,
471
Grainger, R.G., Moldanova, J., Schlager, H., Stevenson D.S., 2010. Transport impacts
472
on atmosphere and climate; shipping. Atmospheric Environment 44, 4735-4771
474 475 476
AC C
473
EP
470
EC, 2012. Energy, transport and environment indicators. Eurostat Pocketbooks ISSN 17254566. (http://epp.eurostat.ec.europa.eu/portal/page/portal/publications/collections/pocketbooks)
477
Fernández-Camacho, R., Rodríguez, S., de la Rosa, J., Sánchez de la Campa, A.M., Viana,
478
M., Alastuey, A., Querol, X., 2010. Ultrafine particle formation in the inland sea
479
breeze airflow in Southwest Europe. Atmopsheric Chemistry and Physics, 10, 9615-
480
9630
481
ACCEPTED MANUSCRIPT 482
Hulskotte, J., Denier van der Gon, H.A.C., 2010a. Fuel consumption and associated
483
emissions from seagoing ships at berth derived from an on-board survey. Atmospheric
484
Environment 44, 1229-1236
485 486
Hulskotte, J., Denier van der Gon, H.A.C., 2010b. Methodologies for estimating shipping emissions in the Netherlands. RIVM, Bilthoven, the Netherlands. BOP-report
488
http://www.rivm.nl/bibliotheek/rapporten/500099012.pdf
489
491
IVR,
2013.
IVR
–
Total
Western
Europe
Fleet
(April
2013).
http://www.ivr.nl/fileupload/statistieken/2013/Total_fleet_2012-new.pdf
SC
490
RI PT
487
492
Keuken, M.P., Henzing, J.S., Zandveld, P., Elshout van den, S., Karl, 2012a. Dispersion of
494
particle number and elemental carbon from road traffic, a harbour and an airstrip in
495
the Netherlands. Atmospheric Environment 54, 320-327
M AN U
493
496
Keuken, M.P., Jonkers, S., Zandveld, P., Voogt, M., Elshout van den, S., 2012b. Elemental
498
carbon as an indicator for evaluating the impact of traffic measures on air quality and
499
health. Atmospheric Environment 61, 1-8
TE D
497
500
Keuken, M.P., Zandveld, P., Jonkers, S., Moerman, M., Jedynska, A.D., Verbeek, R.,
502
Visschedijk, A., Elshout van den, S., Panteliadis, P., Velders, G.J.M., 2013. Modelling
503
elemental carbon at regional, urban and traffic locations in the Netherlands.
504
Atmospheric Environment 73, 73-80
EP
501
505
Klein, J., Hoen, A., Hulskotte, J., Duynhoven van, N., Smit, R., Hensema, A.,
507
Broekhuizen, D., 2007. Methods for calculating the emissions of transport in the
508 509
AC C
506
Netherlands. Task Force Traffic and Transport of the National Emission Inventory, Central Bureau of Statistics, Voorburg, the Netherlands (in Dutch)
510 511
Kasper, A., Aufdenblatten, S., Forss, A., Mohr, M. and Burtscher, H., 2007. Particulate
512
emissions from a low-speed marine diesel engine. Aersol Science and Technology 41,
513
24-32
514 515
ACCEPTED MANUSCRIPT 516
Moldanová, J., Fridell, E., Popovicheva, O., Demirdjian, B., Tishkova, V., Faccinetto, A.,
517
Focsa, C., 2009. Characterisation of particulate matter and gaseous emissions from a
518
large ship diesel engine. Atmospheric Environment 43, 2632-2641
519
521
Pasquill, F., 1971. Atmospheric dispersion of pollution. Journal of the Royal Meteorological Society, 97, 369-395
RI PT
520
522
Petzold, A., Schönlinner, M., 2004. Multi-angle absorption photometry – a new method for
524
the measurement of aerosol absorption and atmospheric black carbon. Aerosol Science
525
35, 421-441
SC
523
526
Petzold, A., Hasselbach, J., Lauer, P., Baumann, R., Franke, K., Gurk, C., Schlager, H.,
528
Weingartner, E., 2008. Experimental studies on particle emissions from cruising ship,
529
their characteristic properties, transformation and atmospheric lifetime in the marine
530
boundary layer. Atmospheric Chemistry and Physics, 8, 2387-2403
M AN U
527
531
535 536 537 538 539 540 541
TE D
534
fishing and military (shipping), EMEP/EEA Guidebook Version 2013
Tzannatos, E., 2010. Ship emissions and their externalities for the port of Piraeus – Greece. Atmospheric Environment 44, 400-407
EP
533
Trozzi, C. and De Lauretis, R., 2013. International navigation, national navigation, national
Wesseling, J.P., Visser, G.Th., 2003. An inter-comparison of the TNO Traffic Model, field data and wind tunnel measurements. TNO, Utrecht, The Netherlands. Report 2003/207
AC C
532
ACCEPTED MANUSCRIPT 542 543
Table 1: Details of the two monitoring campaigns near the Amsterdam-Rhine Canal and
544
the river Waal.
545 First campaign
Second campaign Waal river
547
Waterway
Amsterdam-Rhine Canal
548
Width/Orientation
110 m/north-south
549
Current
550
Monitoring period
07-09-2012; 09-15 h
551
Wind direction/speed
240-270o/3 m s-1
552
Collection shipping data
553
Number of ships/direction
554
Shipping speed
RI PT
546
360 m/east-west 0.5-1.5 m s-1
none
27 (N-S)/28 (S-N) 3.2 m s-1
555 556
SC
185-250o/2.7 m s-1 automatic (AIS)
M AN U
manual (on-site)
28-07-2013; 09-18 h
24 (down-stream)/15 (up-stream) 4.4 m s-1 (down)/ 3.7 m s-1 (up)
Table 2: Distances of the exhaust plume downwind from the passing ships to the
558
monitoring equipment (X) in m and the lateral (σy) and vertical (σz) dispersion in m of
559
shipping plumes at the monitoring sites during the first and second monitoring campaigns.
561
X
σy
σz
(m)
(m)
(m)
563
first campaign (northbound)
51
9
7
564
first campaign (southbound)
76
14
9
565
second campaign (downstream)
233
38
23
566
AC C
562
EP
560
TE D
557
second campaign (upstream)
277
44
28
567 568 569 570
ACCEPTED MANUSCRIPT 571
Table 3: The measured maximum PN, average PN and average EC concentrations (PN-
572
max, PN-avg and EC-avg) and the inferred maximum EC (EC-max) in the shipping plumes
573
(n) near the Amsterdam-Rhine Canal (ARC) and the Waal.
574 575
average and 10th-90th percentile
576
ARC
577
(n=55)
RI PT
Waal (n=39)
578
PN-max (# cm-3)
580
PN-avg (# cm-3)
581
Ratio
582
EC-avg (µg m-3)
22,044 (8,555-41,879)
18,483 (10,431-31,525)
14,812 (5,803-27,855)
2.9 (1.2-5.4) 1.6 (0.6-2.8)
-3
EC-max (µg m )
4.2 (1.4-7.6)
584 585
1.5 (1.1-2.1)
0.7 (0.3-1.0)
M AN U
583
53,712 (14,887-133,199)
SC
579
1.0 (0.4-1.7)
586
Table 4: The average and range of emission factors for EC of ships on the Amsterdam-
587
Rhine Canal (ARC) and the Waal.
TE D
588
EC (µg s-1) per ship
589
number of ships
590
(n)
average
27
3,800
300 – 11,000
28
4,000
400 – 11,000
10th - 90th percentile
ARC (northbound)
592
ARC (southbound)
593
Waal (downstream)
24
4,100
500 – 16,400
594
Waal (upstream)
15
6,600
500 – 23,500
AC C
595
EP
591
ACCEPTED MANUSCRIPT 596
Table 5: The average number of ships N in # h-1, speed V in m s-1 and emission rates EEC in
597
µg m-1 s-1 for the Amsterdam-Rhine Canal (ARC) and the Waal in 2011.
598
N
V
(# h-1)
600
EEC
(m s-1)
601
ARC (northbound)
6
3.8
602
ARC (southbound)
6
3.6
ARC (total)
12
3.7
604
Waal (downstream)
10
4.4
605
Waal (upstream)
10
3.7
606
Waal (total)
20
4.1
1.5
AC C
EP
TE D
M AN U
SC
603
(µg m-1 s-1)
RI PT
599
1.7
3.2
2.7
6.0
8.7
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
AC C
EP
TE D
Figure 5: The percentage of households (%) within 200 m from busy waterways in the Netherlands exposed to additional annual average EC (µg m-3) due to shipping emissions.
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
AC C
EP
TE D
Figure 1: Important inland waterways in Europe and the Netherlands and the two measurement sites: east of the Amsterdam-Rhine Canal (1) and north of the river Waal (2); a typical inland water vessel on the Amsterdam-Rhine Canal.
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
Figure 2: One minute average number of particles (#/cm3) (“PN”) and elemental carbon (10*ng/m3) (“EC”) during the first (Amsterdam-Rhine Canal) and second (Waal) monitoring campaign and passing ships in both directions.
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
Figure 3: The annual average contribution of shipping emissions to EC concentrations in µg m-3 on both sides of the Waal (left = north; right = south) and the Amsterdam-Rhine Canal (left = west; right = east) in 2011; straight (Waal) and dotted (Amsterdam-Rhine Canal)
AC C
EP
lines indicate the banks of both waterways.
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
Figure 4: The impact of shipping emissions to EC concentrations in µg m-3 on the population living within 200 m of Dutch waterways with more than 3 µg EC emissions per m and per s. The measurement sites are also indicated near the Amsterdam-Rhine Canal (1) and the Waal (2).
ACCEPTED MANUSCRIPT
Inland shipping important source of black carbon emissions
•
Additional annual average black carbon near busy waterways up to 0.5µg EC per m3
•
30% of the ships contribute to 80% of the total shipping emissions
AC C
EP
TE D
M AN U
SC
RI PT
•
S1352-2310(14)00454-3
DOI:
10.1016/j.atmosenv.2014.06.008
Reference:
AEA 13032
To appear in:
Atmospheric Environment
Received Date: 23 January 2014 Revised Date:
2 June 2014
Accepted Date: 3 June 2014
Please cite this article as: Keuken, M.P, Moerman, M, Jonkers, J, Hulskotte, J, Denier van der Gon, H.A.C, Hoek, G, Sokhi, R.S, Impact of inland shipping emissions on elemental carbon concentrations near waterways in the Netherlands, Atmospheric Environment (2014), doi: 10.1016/ j.atmosenv.2014.06.008. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT 1
Impact of inland shipping emissions on elemental carbon concentrations near
2
waterways in the Netherlands
3
Keuken, M.P.1, Moerman, M.1, Jonkers, J1, Hulskotte, J.1, Denier van der Gon, H.A.C.1,
4
Hoek, G2, Sokhi, R.S3
RI PT
5 (1) Netherlands Organization for Applied Research (TNO), Utrecht, the Netherlands
7
(2) Institute for Risk Assessment Sciences (IRAS), University of Utrecht, the Netherlands
8
(3) Centre for Atmospheric and Instrumentation Research (UH-CAIR), University of Hertfordshire,
9
Hatfield, United Kingdom
10
Corresponding author: [email protected]
SC
6
ABSTRACT
12
This study aims to quantify the impact of black carbon from inland shipping on air quality,
13
expressed as elemental carbon (EC) near inland waterways in the Netherlands. Downwind
14
measurements of particle numbers and EC were used to establish emission factors for EC
15
from inland shipping using inverse modelling. These emission factors were combined with
16
data on energy consumption to derive annual average emissions rates for all Dutch
17
waterways. A line source model was applied to compute the contribution of inland
18
shipping to annual average EC concentrations for around 140 000 people living within 200
19
m of busy waterways in the Netherlands. The results showed that they are exposed to
20
additional EC concentrations of up to 0.5 µg EC per m3 depending on the shipping volume
21
and distance from the waterway. In view of the envisaged growth in water transport, this
22
underlines the need to reduce combustion emissions from inland shipping. Targeting
23
“gross” polluters may be the most effective approach since 30% of ships cause more than
24
80% of the total emissions.
25
Keywords: black carbon, elemental carbon, inland shipping, emission factors
26
1.
27
The transport of goods over Europe’s 37,000 kilometres of waterways amounted to 7% of
28
the total inland goods transport in Europe in 2010 (EC, 2012). What is more, this share is
29
increasing because CO2 emissions per ton-kilometres over water are lower than those for
30
land transport by a factor of six (EC, 2012). Research into emissions from shipping and the
31
likely wider impact on air quality and climate change has been mainly directed at sea-
AC C
EP
TE D
M AN U
11
INTRODUCTION
ACCEPTED MANUSCRIPT going ships (Eyring et al., 2010; Bond et al., 2013). This impact relates to gaseous
33
emissions of CO2, SO2 and NOx, and particulate matter (PM) consisting of elemental and
34
organic carbon, sulphates and ash (Petzold et al., 2008; Moldanová et al., 2009). In urban
35
areas near harbours too, the likely impact on air quality has been the subject of attention
36
(Hulskotte and Denier van der Gon, 2010a; Tzannatos, 2010; Keuken et al., 2012a). Less
37
research has been conducted into emissions from inland waterways, probably because their
38
contribution to total shipping emissions is limited. However, inland shipping emissions
39
may be significant for air quality and the related health impact for people living nearby.
40
This may be particularly relevant in the Netherlands where 40% of goods were transported
41
over inland waterways in 2008 and 2009 (EC, 2012). The Netherlands has the largest
42
inland freight fleet in Western Europe with 5,815 vessels out of 11,546 (IVR, 2013).
43
PM emissions from inland shipping fuelled by diesel oil are lower by about a factor of five
44
per kWh (Trozzi and De Lauretis, 2013) than those from seagoing vessels, which are
45
mainly fuelled by residual fuel oil. However PM emission factors per kWh from inland
46
ships are considerably higher than diesel truck engines due to less stringent emission
47
standards for inland ships. For the Netherlands, it has been estimated that PM emissions
48
per ton-kilometre (i.e. one ton of goods transported one kilometre) from water transport are
49
typically five times higher than from road transport (Hulskotte and Denier van der Gon,
50
2010b).
51
Most studies into PM emissions from shipping have been performed by making
52
measurements on-board or in ship exhaust plumes of seagoing ships, very often fuelled by
53
residual fuel oil (Petzold et al., 2008; Moldanová et al., 2009).
54
This paper presents a study to model the impact of inland shipping on air quality and health
55
near inland waterways in the Netherlands. This requires emission factors for inland ships
56
as an input for the model. However, these are not established on a routine basis, since there
57
are no regulatory obligations in this area, such as those that apply to road transport.
58
Emission factors for inland ships were derived from measurements using test banks of ship
59
engines. This resulted in an average emission factor of 0.5g of PM per kWh in 2000 and
60
0.4g of PM per kWh in 2010 (Klein et al., 2007; Hulskotte and Denier van der Gon,
61
2010b). However, there are some reservations about applying these emission factors in
62
dispersion modelling. Firstly, average emission factors may not be representative because
63
shipping emissions may vary considerably according to actual shipping conditions.
64
Secondly, black or elemental carbon (EC) is a more sensitive indicator for combustion
65
particles than PM due to the relatively high background concentrations of PM (Keuken et
AC C
EP
TE D
M AN U
SC
RI PT
32
ACCEPTED MANUSCRIPT 66
al., 2012b). In our study, emission factors for elemental carbon from inland shipping were
67
derived by measuring the black carbon in shipping plumes and using inverse modelling.
68
Finally, the contribution to annual average EC concentrations within 200m of busy inland
69
waterways in the Netherlands was estimated.
70
2.
EXPERIMENTAL METHODS
RI PT
71
2.1 Monitoring periods
73
Concentrations of black carbon and the total number of particles (PN) were monitored near
74
two major Dutch waterways, as well as the wind speed and direction. The two sites are
75
shown in Fig. 1, which also shows other important waterways in the Netherlands and
76
Europe. Also shown is a typical inland water vessel in the Netherlands. The average length
77
of these vessels is 110 m with a width of 11 m and a loading capacity of 3,000 tons (IVR,
78
2013).
M AN U
SC
72
79 80
Figure 1: Important inland waterways in Europe and the Netherlands and the two
81
measurement sites: east of the Amsterdam-Rhine Canal (1) and north of the river Waal (2);
82
a typical inland water vessel on the Amsterdam-Rhine Canal.
TE D
83
The first monitoring campaign took place at a site east of the Amsterdam-Rhine Canal,
85
four kilometres north of the town Breukelen. The water is kept at a fixed level and there is
86
hardly any water flow. Consequently, the average speed (“speed over ground”) of ships is
87
similar in both directions. The date of this monitoring campaign was selected because of
88
forecasted favourable meteorological conditions for inverse modelling (see: section 2.3):
89
wind directions which carried exhaust plumes perpendicularly across the canal, wind
90
speeds with limited temporal variability and no rainfall. The monitoring equipment was
91
located directly on the side of the canal with the sampling inlet about 4.5m above the
92
surface of the water. The distance down-wind of the exhaust plume and the height of the
93
exhaust pipe were measured manually on-site with a laser. The speed of each ship was
94
retrieved on-line from a website (www.marinetraffic.com).
95
The second monitoring campaign was at a site north of the river Waal about seven
96
kilometres east of the city of Nijmegen. The Waal is the main of the three distributaries
97
that the river Rhine splits into after crossing the Dutch border. Unlike the Amsterdam-
98
Rhine Canal, the Waal has a current and consequently ships use more power going
AC C
EP
84
ACCEPTED MANUSCRIPT upstream than vice versa. The second monitoring campaign was from 25 to 28 July 2013.
100
The measurements were taken directly on the side of the river with the sampling inlet
101
about 6.5 m above the water surface. In the first monitoring campaign the shipping data
102
were measured on-site manually, which is laborious and costly. Therefore, in the second
103
monitoring campaign, the relevant data from passing ships were retrieved and stored on-
104
line from the Automatic Identification System (AIS). The AIS is compulsory for all ships
105
over 300 gross tonnage and reports the GPS position and speed of a ship every second.
106
Passing ships were detected automatically, which enabled exhaust plumes from a large
107
number of ships to be measured unattended. The meteorological data and shipping data
108
were matched after the monitoring campaign to select passing ships during favourable
109
meteorological conditions. Only on 28 July were the meteorological conditions favourable
110
for downwind plume detection at the monitoring site. Due to the relatively large amount of
111
shipping traffic on the Waal exhaust plumes of passing ships mixed regularly and therefore
112
from the AIS data, only ships were selected where exhaust plumes of could be detected
113
separately. The details of the two monitoring campaigns are summarized in Table 1.
114
M AN U
SC
RI PT
99
2.2 Monitoring equipment
116
The concentration of the total number of particles (PN) were measured with a condensation
117
particle counter – CPC (TSI, type 3775) with a 50% cut-off at 4 nm, a measurement
118
frequency of 1 s and an accuracy of ± 10%. Black carbon (BC) concentrations were
119
measured with a multi-angle absorption photometer - MAAP (Thermo Scientific, model
120
5012) with a temporal resolution of 1 minute and a detection limit of 0.1 µg m-3. The BC
121
data were converted to elemental carbon (EC) with a conversion factor of 0.75. This
122
correction was applied because the MAAP overestimates EC when compared to
123
measurements with the reference thermal-optical method (Petzold and Schönlinner, 2004;
124
Chow et al., 2009; Keuken et al., 2013). The meteorological parameters were measured
125
with a Vaisala, weather transmitter WXT 510 (www.vaisala.com) for wind speed and
126
direction with a response time of 1 s. The distance between the monitoring equipment and
127
passing ships was measured manually with a laser during the first monitoring campaign
128
(Leica, CRF 1200) with an accuracy of ± 1 m up to 350 m. Hence, the uncertainty due to
129
measuring the distances and height of the exhaust pipe is less than 1%. The AIS data on
130
passing ships were collected during the second monitoring campaign with an AIS receiver
131
(COMAR systems). All the data collected were stored on a laptop computer.
AC C
EP
TE D
115
ACCEPTED MANUSCRIPT 132
2.3 Inverse modelling
133
The well-known Gaussian plume model provides the following relationship between the
134
concentration (C) of a pollutant directly downwind at ground level and the emission factor
135
(Q) from a stack emission:
136
Q = (C * π * σy * σz * v)/(exp[-1/2 (H/σz)2])
(1)
RI PT
137 138
with Q in µg s-1, C in µg m-3 and σy and σz the lateral and vertical plume standard
140
deviations in m, v the wind speed at height H in m s-1 and H the effective height of the
141
plume emission in metres.
142
The above equation (1) was used to derive emission factors of EC from the PN and BC
143
concentrations measured in the shipping plumes at the monitoring sites during the two
144
monitoring campaigns. The background concentrations measured before and after a
145
shipping plume passed over the monitoring sites were averaged and subtracted from the
146
measured concentrations. The effective plume height was the sum of the exhaust pipe on
147
the ship and the plume rise. These were estimated manually by a laser for each ship during
148
the first monitoring campaign. The averages were 2.5 m (exhaust pipe) and 4 m (plume
149
rise). These figures were also applied to all ships during the second monitoring campaign
150
since the fleet composition is similar on both waterways. The plumes downwind of the
151
ships may be characterized by the standard deviations of the plume concentrations in the
152
lateral and vertical directions (σy and σz). These are a function of the distance downwind of
153
the ship and atmospheric stability. The latter can be defined by the Pasquill Stability
154
Classes relating to wind speed and cloud cover during daytime (Pasquill, 1971). The
155
meteorological conditions (i.e. daytime, half cloud cover and wind speed over 2 m s-1)
156
during both monitoring campaigns were similar and classified as class B. The related
157
plume standard deviations may be computed from the following equations (2) and (3):
159
M AN U
TE D
EP
AC C
158
SC
139
σy = a * X * exp(b) + c
(2)
σz = α * X * exp(β)
(3)
160 161 162 163
with a: 0.038, b: 1.149, c: 3.3, α: 0.275, β: 0.903 and X the downwind distance from the
164
ship to the monitoring site (Pasquill, 1971). The results from the two monitoring sites are
165
presented in Table 2.
ACCEPTED MANUSCRIPT 166
Because wind directions during the monitoring campaigns were almost perpendicular to
168
the two waterways (75o for the Amsterdam-Rhine Canal and 60o for the Waal), the ship
169
plumes passed over the monitoring sites with the speed of the passing ships. Consequently,
170
the concentrations of PN and BC were measured over a cross-section of the passing plume.
171
In this cross-section, the concentrations follow a normal distribution in accordance to the
172
Gaussian plume model. The width of this cross-section may be estimated from four times
173
the lateral dispersion (σy) of a plume at the monitoring locations. The factor “four” takes
174
into consideration the fact that 95% of the observations in a normal distribution are
175
between the average and plus-and-minus two times the standard deviation. From Table 1, it
176
was derived that the width of the plumes during the first monitoring campaign are in the
177
order of 48 m: four times the average σy of 9 m and 14 m for northbound and southbound
178
ships respectively. For the second monitoring campaign, the width of the plumes was in the
179
order of 104 m: four times the average σy of 23 and 28 m for ships moving at the river
180
Waal downstream and upstream respectively. Considering the speed of the ships during the
181
first and second monitoring campaign – of around 3 and 4 m s-1 respectively – the time for
182
the passage of the plumes over the first and second monitoring sites was in the order of 16
183
and 26 s respectively.
184
The PN measurements were performed at a frequency of once every second, while the BC
185
measurements, which were converted to EC concentrations (see: section 2.2) had a
186
frequency of once every minute. Hence, the average concentration of EC in the plume may
187
be measured but the maximum concentration in the cross-section of the shipping plume
188
passing over the monitoring equipment was missing. Since the latter is required for inverse
189
modelling with equation (1), the maximum concentration per second of EC (EC-max) was
190
inferred from the ratio between the maximum PN peak per second (PN-max) and the
191
average PN concentration (PN-avg) by equation 4:
193
SC
M AN U
TE D
EP
AC C
192
RI PT
167
EC-max = (PN-max/PN-avg) * EC-avg
(4)
194 195
with EC-max and EC-avg in µg m-3, and PN-max and PN-avg in # cm-3. The ratio of the
196
maximum and average concentration in the cross-section of a plume is equal for all
197
pollutants. Because, this ratio is a function of atmospheric dilution during atmospheric
198
transport of the exhaust plume. Dilution (“the size of the plume”) is characterized by sigma
199
y and sigma z. It depends on atmospheric conditions and the downwind distance of the
ACCEPTED MANUSCRIPT 200
plume from the exhaust pipe. The concentration of a pollutant in a plume may also change
201
by other processes such as condensation or deposition. This is relevant for PN but it is
202
assumed that these additional processes both affect PN-max and PN-avg and therefore with
203
a limited effect on their ratio. After determining EC-max in each shipping plume, inverse
204
modelling was applied
205
waterways have been estimated.
206
2.4 Shipping emissions and dispersion modelling
207
The average emission factors at the two waterways (see: Section 2.3) were combined with
208
the number of ships per year to derive emission rates for EC in µg per meter and per
209
second for each of the two waterways. In the next step, available data for the energy
210
consumption at all waterways in the Netherlands were used to determine emission rates of
211
EC for these waterways. Finally, the annual contribution to EC concentrations within 200m
212
of the waterways was estimated with a line source model. Further details on the method
213
and the uncertainty are provided in the Sections 3.2-3.6.
214
3.
M AN U
SC
RI PT
and emission factors of EC for individual ships at the two
RESULTS AND DISCUSSION
215
3.1 PN and BC monitoring data
217
The PN and BC concentrations were measured during the two monitoring campaigns. The
218
average one-minute results are presented in Figure 2 with the results for EC as converted
219
from the BC concentrations, multiplied by ten for presentational reasons. The passage of
220
ships during the monitoring campaigns are also indicated..
EP
221
TE D
216
Figure 2: One minute average number of particles (#/cm3) (“PN”) and elemental carbon
223
(10*ng/m3) (“EC”) during the first (Amsterdam-Rhine Canal) and second (Waal)
224
monitoring campaign and passing ships in both directions.
225
AC C
222
226
Fig. 2 illustrates that the plume concentrations near the Amsterdam-Rhine Canal are higher
227
than near the Waal mainly because the down-wind distances of the exhaust plume to the
228
monitoring site was shorter(see: Table 1). In the first monitoring campaign all passing
229
ships were included in the data analysis but in the second campaign specific ships were
230
selected from the AIS data. This was required as the shipping intensity at the Waal resulted
231
in mixing of exhaust plumes and not all exhaust peaks could be detected due to the larger
232
distances to the ships and variability in the wind direction. The data indicate that emissions
ACCEPTED MANUSCRIPT 233
of black carbon and PN vary considerably between ships, as well as the ratio of PN
234
emissions to BC emissions. These measurements have been used to determine the EC-max
235
per shipping plume as presented in Table 3.
236 237
3.2 EC emission factor per ship
238
The PN and EC measurements were combined with shipping data to link the plume
239
concentrations to specific ships. After subtracting
240
measured before and after a plume passage over the monitoring equipment, the maximum
241
and average PN concentration and the average EC concentrations in each plume were
242
calculated. Subsequently, the maximum EC concentration in each plume was inferred, as
243
described in section 3.1. The maximum EC concentration and the plume characteristics σy
244
and σz (see: Section 2.3) for each plume were used to calculate the emission factor per ship
245
of EC in µg s-1, using equation (1). The average values are presented in Table 4 for each
246
waterway and each shipping direction.
247 248
The average emission factors in both directions at the Amsterdam-Rhine Canal are very
249
similar. This was expected because there is hardly any flow in the canal. The average
250
emission factor for ships going downstream on the Waal was in line with ships on the
251
Amsterdam-Rhine Canal, but by contrast ships travelling upstream on the Waal have
252
higher emission factors which is attributed to the use of more engine power.
253
The data in Table 4 show a large range in emission factors between ships. For both
254
directions on the two waterways, it was found that around 30% of all ships emit over 80%
255
of the total emissions. This may be related to parameters such as engine power, the age of
256
the engine, maintenance, and loaded versus unloaded vessels. In the Netherlands, the
257
BIVAS database has been established which combines traffic and freight data and includes
258
shipping activities on all inland waterways in the Netherlands related to the number of
259
ships and engine power (http://bivas.chartasoftware.com). Therefore, it is known that the
260
number of unloaded ships is considerably less than two-thirds of all vessels, which would
261
indicate that “loaded versus unloaded” ships is unlikely to explain the variation in emission
262
factors completely. Engine type and maintenance are other likely important factors in
263
explaining the variation in emission factors. This indicates that one out of three ships is a
264
“gross” polluter. This conclusion has implications for measures to reduce shipping
265
emissions by targeting a limited number of ships rather than the whole fleet.
266
AC C
EP
TE D
M AN U
SC
RI PT
the background concentrations
ACCEPTED MANUSCRIPT 3.3 EC emission rates for each waterway
268
The impact of shipping emissions on air quality near waterways was modelled with a line-
269
source model analogous to the dispersion of traffic emissions on air quality near
270
motorways (Wesseling et al., 2003; Beelen et al., 2010; Keuken et al., 2012a). As well as
271
meteorological parameters, the emission rate for air pollutants is required as an input for
272
the model. The emission rate was derived from the product of the emission factor for each
273
ship and the number of ships on the waterway during a specific time period. The emission
274
rate (E) may be computed from equation 5:
RI PT
267
275
E = (N/3600 * EF)/V
(5)
SC
276 277
with E in µg m-1 s-1, N the number of ships per hour, EF the emission factor per ship in µg
279
s-1 and V the speed over ground of each ship in m s-1. From the BIVAS data base, it was
280
known that in 2011 the average numbers of ships on the Amsterdam-Rhine Canal and the
281
Waal were 105,000 and 175,000 ships, respectively or 12 and 20 ships per hour.
282
Subsequently, the average EC emission rates (EEC) at the monitoring sites near the
283
Amsterdam-Rhine Canal and the Waal in each shipping direction have been calculated
284
using equation 5. The emission factor for each ship (EF in equation 5) was computed as
285
described in Section 3.2. The average EC emission rates (EEC) are presented in Table 5.
TE D
286
M AN U
278
The next step was to generalize these derived emission rates at the monitoring sites near
288
the Amsterdam-Rhine Canal and the Waal in Table 5 to all Dutch waterways. The BIVAS
289
data base provided the annual energy consumption for a specific length of a waterway: at
290
the monitoring sites near the Amsterdam-Rhine Canal 0.024 TJ m-1 and near the Waal 0.05
291
TJ m-1 respectively. Considering the average EC emission rates at both locations in Table
292
5, a conversion factor may be derived between EC emission rates and energy consumption.
293
At the monitoring sites near the Amsterdam-Rhine Canal and the Waal, the conversion
294
factors are 0.0074 and 0.0058 µg EC per TJ energy consumption per s. This provides an
295
average conversion factor of 0.0066 between annual EC emission rates and energy
296
consumption on Dutch inland waterways.
AC C
EP
287
297 298
3.4 Uncertainty of the established emission factors and rates
299
The emission factors have been derived from measurements during the two monitoring
300
campaigns of about 2% of the total number of Dutch inland vessels. Whether the size of
ACCEPTED MANUSCRIPT 301
this sample was representative for the fleet has been assessed as follows. During the first
302
monitoring campaign, twenty seven and twenty eight
303
southbound, respectively at the Amsterdam-Rhine Canal were measured. The ships going
304
in both directions are assumed to be similar in fleet composition. Also, it is expected that
305
the engine load in both directions is similar as there is no current in the canal and the wind
306
direction was almost perpendicular to the shipping direction. The similarity in established
307
emission factors for both directions at the Amsterdam-Rhine Canal was tested with two
308
nonparametric statistical tests: the Wilcoxon’s T-test and Spearman’s correlation
309
coefficient. These nonparametric tests have been applied, as the large range in emission
310
factors presented in Table 4 suggests that the emission factors may not be normally
311
distributed. The result of the Wilcoxon test with a T value of 0.939 at a 95 percentage
312
confidence interval, indicates that the emission factors for both shipping directions at the
313
Amsterdam-Rhine Canal are similar. Also, the Spearman’s correlation coefficient of 0.999
314
shows that the established emission factors in both shipping directions are highly
315
correlated. Both statistical tests confirm that there is no difference in emission factors
316
derived from two samples of twenty seven and twenty eight
317
concluded that the number of fifty five
318
adequate to estimate shipping emissions at the Amsterdam-Rhine Canal.
319
The uncertainty to apply the established emission factors at other waterways than the
320
Amsterdam-Rhine Canal has been investigated as follows.
321
Canal, the annual energy consumption is 0.024 TJ m-1 (http://bivas.chartasoftware.com)
322
and the average emission rate is 3.2 µg EC m-1.s-1 (see: Table 5). This provides a ratio of
323
0.0074 between the energy consumption and the emission rate of EC. This ratio has been
324
applied to derive an emission rate of EC for the Waal from its energy consumption of 0.05
325
TJ m-1 near the monitoring location. This results in an emission rate of 6.8 µg EC m-1 s-1
326
which is 78% of the established emission rate of 8.7 µg EC m-1.s-1 (see: Table 5). From the
327
conversion ratios at the Amsterdam-Rhine Canal and the Waal, an average conversion ratio
328
was established of 0.0066 with a relative standard deviation of 17%. The latter provides an
329
estimate of the uncertainty to apply the average conversion factor to all Dutch waterways.
ships. It is therefore
ships in the first monitoring campaign was
At the Amsterdam-Rhine
AC C
EP
TE D
M AN U
SC
RI PT
ships going northbound and
330 331
3.5 Contribution from inland shipping emissions to EC concentrations near waterways
332
The conversion factor was applied to the BIVAS data base on energy consumption to
333
derive the annual average EC emission rates from inland waterways in the Netherlands.
334
These emission rates were projected at two line sources for each waterway as an input for
ACCEPTED MANUSCRIPT the line-source model (Wesseling et al., 2003; Beelen et al., 2010; Keuken et al., 2012a). It
336
is based on a Gaussian plume model which takes into account vehicle-induced turbulence,
337
the upwind roughness of the terrain and atmospheric stability. This model for road traffic
338
was applied to model the dispersion of black carbon shipping emissions but the vehicle-
339
induced turbulence was not included in the model. For canals, the emission rates of the two
340
line sources are similar while for rivers the emission rates differ for upstream and
341
downstream ships in accordance with the results for the Waal. The annual contribution to
342
EC concentrations were modelled for the Amsterdam-Rhine Canal and the Waal in 2011.
343
The results are presented as a function of the distance to the middle of the waterway in
344
Figure 3. For the Amsterdam-Rhine Canal, the line-sources are situated at 35 m and 60 m
345
from the eastern bank of the waterway and for the Waal at 165 m (downstream) and 195 m
346
(upstream) from the northern bank of the waterway. The reason that the down-wind
347
distances of the exhaust pipe to the monitoring stations in Table 2 are longer than the
348
distances of the line sources to the board of the waterway, is the angle between the wind
349
direction and the orientation of the waterway was not exactly perpendicular during the two
350
campaigns (see: Table 1) but 75o (Amsterdam-Rhine Canal) and 60o (Waal),
M AN U
SC
RI PT
335
351
Figure 3: The annual average contribution of shipping emissions to EC concentrations in
353
µg m-3 on both sides of the Waal (left = north; right = south) and the Amsterdam-Rhine
354
Canal (left = west; right = east) in 2011; straight (Waal) and dotted (Amsterdam-Rhine
355
Canal) lines indicate the banks of both waterways.
EP
356
TE D
352
Fig. 3 illustrates that the contribution of shipping emissions to air quality is not identical on
358
both sides of the two waterways as a result of prevailing wind directions and wind speeds
359
in the Netherlands. Accordingly, the impact on air quality is different north and south of
360
the Waal, and east and west of the Amsterdam-Rhine Canal. Fig. 3 also shows that the
361
concentration of shipping emissions beyond 200 m of the banks of the waterways is less
362
than 5% of the concentrations in the middle of the waterway. Hence, the contribution of
363
shipping emissions beyond 200 m from the waterway to the annual average concentrations
364
of EC may be regarded as insignificant. Using the line-source model, the contribution of
365
shipping emissions to annual average EC at each residential address within 200 m of the
366
inland waterways was calculated for the year 2011. The results at a spatial resolution of
367
25*25 m for each household are shown in Figure 4 for busy Dutch waterways. Busy
368
waterways in this study being defined as having - according to the BIVAS database - 78%
AC C
357
ACCEPTED MANUSCRIPT 369
of the total emissions from inland shipping in the Netherlands. These waterways represent
370
8% of the total length of waterways and 140 000 people live within 200 m of these
371
waterways. The black carbon emissions at these busy waterways are over 3 µg EC per s
372
per km, as derived from this study.
373
Figure 4: Additional annual average EC concentrations in µg m-3 for the population living
375
within 200 m of busy Dutch waterways. and the measurement sites near the Amsterdam-
376
Rhine Canal (1) and the Waal (2).
RI PT
374
377
Fig. 4 illustrates that near the Waal the additional on annual average EC concentrations is
379
in the range of 0.2 to 0.3 µg m-3 while further downstream and near the Amsterdam-Rhine
380
Canal the contribution is in the range of 0.01 to 0.1 µg m-3 EC. In Figure 5, the percentages
381
of the population are presented in accordance with different levels of additional EC
382
concentrations due to shipping emissions.
M AN U
SC
378
383 384
Figure 5: The percentage of households (%) within 200 m from busy waterways in the
385
Netherlands exposed to additional annual average EC (µg m-3) due to shipping emissions.
386
4.
387
A method has been presented to derive EC emission factors for individual ships and EC
388
emission ratesfrom inland waterways. The method is based on downwind measurements
389
of PN and black carbon which were converted to EC concentrations in shipping plumes
390
near waterways followed by inverse modelling. In this approach, it is assumed that
391
atmospheric dilution of PN and BC is similar in the few minutes of transport time from the
392
ship’s exhaust pipe to the monitoring location. Cooling of the exhaust plume will result in
393
particle formation (Fernández-Camacho et al., 2010) and change in PN concentration other
394
than dilution. However, it is assumed that this process does not affect significantly the
395
ratio of the maximum and average PN concentrations which enables to infer the maximum
396
EC concentration in the plume from PN and BC measurements.
397
The method was applied near the Amsterdam-Rhine Canal and the Waal in the Netherlands
398
during two monitoring periods under specific meteorological conditions (e.g. wind speeds
399
of over 2 m s-1 and a wind direction almost perpendicular to the waterway). Despite the
400
more labour-intensive strategy in the first campaign to collect on-site shipping data, it is
401
preferred over the automatic AIS system in the second campaign near a waterway with
AC C
EP
TE D
CONCLUSION
ACCEPTED MANUSCRIPT intense shipping traffic. The emission factors varied from 0.3 to 23.5 mg EC per s between
403
ships which is mainly attributed to differences in engine age and maintenance. The average
404
emission factors were 4.1 and 6.6 mg EC per s for ships going downstream and upstream
405
on the Waal and 3.6 and 3.8 mg EC per s for southbound and northbound ships on the
406
Amsterdam-Rhine Canal. The reproducibility of the method was better than 96% as
407
estimated from the emission factors derived separately for southbound and northbound
408
ships on the Amsterdam-Rhine Canal. The accuracy was 78% based on comparison of the
409
emission rates for the Waal by the method presented and by energy consumption.
410
The energy consumption data for busy Dutch waterways was converted into annual
411
average emission rates for EC. These emission rates were used as an input for a line source
412
model to calculate the contribution
413
concentrations up to 200 m from busy waterways. The shipping volume and the distance of
414
ships to the bank of a waterway are important parameters for the air quality near
415
waterways. The Waal, with about 175,000 ships per year, is the busiest waterway in the
416
Netherlands with an EC emission rate similar to that of a motorway with more than
417
120,000 vehicles per day. However, the distance of the ships from the bank of the river is
418
in the order of 135 m and therefore shipping emissions are considerably dispersed before
419
reaching the side of the waterway. The contribution to air quality near a waterway ranged
420
from 0.2 to 0.3 µg m-3 EC for a busy but wide waterway (360 m) such as the river Waal,
421
and from 0.01 to 0.1 µg m-3 EC for a less busy but relatively narrow waterway (110 m)
422
such as the Amsterdam-Rhine Canal.
423
The emissions factors measured in this study were compared to the emission factors used
424
in the official Netherlands emission inventory where an average emission rate of 36 kg
425
PM10 per TJ is applied for inland shipping (BIVAS). With an average contribution of 40%
426
EC in exhaust PM in diesel engines for road traffic, this would result in an emission rate of
427
9 kg EC per TJ or 14.4 µg EC m-1.s-1 at the Waal. This is about 50% higher than in this
428
study. PM emissions are known to be uncertain and a 50% discrepancy is not extreme.
429
However, it suggests that our exposure calculations will be a lower estimate, as the
430
emission factors used were lower than currently in-use emission factors. Therefore,
431
monitoring EC concentrations near waterways, as well as additional emission factor
432
measurements are recommended to verify the modelling results in this study.
433
Emission standards for shipping still lag behind the strict standards in place for road
434
transport. Such measures could be the introduction of LNG and low(er) sulphur content
435
fuels in this sector. In view of the relatively limited number of ships that likely cause more
SC
RI PT
402
AC C
EP
TE D
M AN U
of shipping emissions on annual average EC
ACCEPTED MANUSCRIPT 436
than 80% of the EC emissions, identifying and targeting “gross” polluters may be another
437
successful mitigation option. Follow-up research is recommended to detect these gross
438
polluters and understand the cause for their relatively high emissions. The long-term
439
monitoring of EC concentrations in cities near busy waterways with docking locations for
440
inland ships is also recommended to determine the actual exposure of people.
RI PT
441
5.
ACKNOWLEDGEMENTS
443
This work was supported by The Netherlands Ministry of Infrastructure and Environment
444
and the 7th European Framework project: TRANSPHORM directed to emissions,
445
dispersion and health impact of transport-related PM (Grant Agreement No. 243406).
446
We gratefully thank Ernst Bolt of the Ministry for Infrastructure and Environment
447
(Rijkswaterstaat Water, Verkeer en Leefomgeving) for access to the BIVAS data.
SC
442
M AN U
448
AC C
EP
TE D
449
ACCEPTED MANUSCRIPT 450
6.
REFERENCES
451 452
Bond, T.C., Doherty, S.J., Fahey, D.W., Forster, P.M., et al., 2013. Bounding the role of
453
black carbon in the climate system: a scientific assessment. Journal of Geophysical
454
Research 10.1002/jgrd.50171
RI PT
455
Beelen, R., Voogt, M., Duyzer, J., Zandveld, P., Hoek, G., 2010. Comparison of the
457
performances of land-use regression modelling and dispersion modelling in estimating
458
small-scale variations in long-term air pollution concentrations in a Dutch urban area.
459
Atmospheric Environment 44, 4614-4621
SC
456
460
Chow, J.C., Watson, J.G., Doraiswamy, P., Antony Chen, L-W, Sodeman, D.A.,
462
Lowenthal, D.H., Park, K., Arnott, W.P., Motallebi, N., 2009. Aerosol light
463
absorption, black carbon, and elemental carbon at the Fresno Supersite, California.
464
Atmospheric Environment 93, 874-887
M AN U
461
465
Eyring, V., Köhler, H.W., van Aardenne, J., Lauer, A., 2005. Emissions from international
467
shipping. The last 50 years. Journal of Geophysical Research 110, D17305.
468
doi:10.1029/2004jD005619
TE D
466
469
Eyring, V., Isaksen, I.S.A., Berntsen, T., Collins, W.J., Corbett, J.J., Endresen, O.,
471
Grainger, R.G., Moldanova, J., Schlager, H., Stevenson D.S., 2010. Transport impacts
472
on atmosphere and climate; shipping. Atmospheric Environment 44, 4735-4771
474 475 476
AC C
473
EP
470
EC, 2012. Energy, transport and environment indicators. Eurostat Pocketbooks ISSN 17254566. (http://epp.eurostat.ec.europa.eu/portal/page/portal/publications/collections/pocketbooks)
477
Fernández-Camacho, R., Rodríguez, S., de la Rosa, J., Sánchez de la Campa, A.M., Viana,
478
M., Alastuey, A., Querol, X., 2010. Ultrafine particle formation in the inland sea
479
breeze airflow in Southwest Europe. Atmopsheric Chemistry and Physics, 10, 9615-
480
9630
481
ACCEPTED MANUSCRIPT 482
Hulskotte, J., Denier van der Gon, H.A.C., 2010a. Fuel consumption and associated
483
emissions from seagoing ships at berth derived from an on-board survey. Atmospheric
484
Environment 44, 1229-1236
485 486
Hulskotte, J., Denier van der Gon, H.A.C., 2010b. Methodologies for estimating shipping emissions in the Netherlands. RIVM, Bilthoven, the Netherlands. BOP-report
488
http://www.rivm.nl/bibliotheek/rapporten/500099012.pdf
489
491
IVR,
2013.
IVR
–
Total
Western
Europe
Fleet
(April
2013).
http://www.ivr.nl/fileupload/statistieken/2013/Total_fleet_2012-new.pdf
SC
490
RI PT
487
492
Keuken, M.P., Henzing, J.S., Zandveld, P., Elshout van den, S., Karl, 2012a. Dispersion of
494
particle number and elemental carbon from road traffic, a harbour and an airstrip in
495
the Netherlands. Atmospheric Environment 54, 320-327
M AN U
493
496
Keuken, M.P., Jonkers, S., Zandveld, P., Voogt, M., Elshout van den, S., 2012b. Elemental
498
carbon as an indicator for evaluating the impact of traffic measures on air quality and
499
health. Atmospheric Environment 61, 1-8
TE D
497
500
Keuken, M.P., Zandveld, P., Jonkers, S., Moerman, M., Jedynska, A.D., Verbeek, R.,
502
Visschedijk, A., Elshout van den, S., Panteliadis, P., Velders, G.J.M., 2013. Modelling
503
elemental carbon at regional, urban and traffic locations in the Netherlands.
504
Atmospheric Environment 73, 73-80
EP
501
505
Klein, J., Hoen, A., Hulskotte, J., Duynhoven van, N., Smit, R., Hensema, A.,
507
Broekhuizen, D., 2007. Methods for calculating the emissions of transport in the
508 509
AC C
506
Netherlands. Task Force Traffic and Transport of the National Emission Inventory, Central Bureau of Statistics, Voorburg, the Netherlands (in Dutch)
510 511
Kasper, A., Aufdenblatten, S., Forss, A., Mohr, M. and Burtscher, H., 2007. Particulate
512
emissions from a low-speed marine diesel engine. Aersol Science and Technology 41,
513
24-32
514 515
ACCEPTED MANUSCRIPT 516
Moldanová, J., Fridell, E., Popovicheva, O., Demirdjian, B., Tishkova, V., Faccinetto, A.,
517
Focsa, C., 2009. Characterisation of particulate matter and gaseous emissions from a
518
large ship diesel engine. Atmospheric Environment 43, 2632-2641
519
521
Pasquill, F., 1971. Atmospheric dispersion of pollution. Journal of the Royal Meteorological Society, 97, 369-395
RI PT
520
522
Petzold, A., Schönlinner, M., 2004. Multi-angle absorption photometry – a new method for
524
the measurement of aerosol absorption and atmospheric black carbon. Aerosol Science
525
35, 421-441
SC
523
526
Petzold, A., Hasselbach, J., Lauer, P., Baumann, R., Franke, K., Gurk, C., Schlager, H.,
528
Weingartner, E., 2008. Experimental studies on particle emissions from cruising ship,
529
their characteristic properties, transformation and atmospheric lifetime in the marine
530
boundary layer. Atmospheric Chemistry and Physics, 8, 2387-2403
M AN U
527
531
535 536 537 538 539 540 541
TE D
534
fishing and military (shipping), EMEP/EEA Guidebook Version 2013
Tzannatos, E., 2010. Ship emissions and their externalities for the port of Piraeus – Greece. Atmospheric Environment 44, 400-407
EP
533
Trozzi, C. and De Lauretis, R., 2013. International navigation, national navigation, national
Wesseling, J.P., Visser, G.Th., 2003. An inter-comparison of the TNO Traffic Model, field data and wind tunnel measurements. TNO, Utrecht, The Netherlands. Report 2003/207
AC C
532
ACCEPTED MANUSCRIPT 542 543
Table 1: Details of the two monitoring campaigns near the Amsterdam-Rhine Canal and
544
the river Waal.
545 First campaign
Second campaign Waal river
547
Waterway
Amsterdam-Rhine Canal
548
Width/Orientation
110 m/north-south
549
Current
550
Monitoring period
07-09-2012; 09-15 h
551
Wind direction/speed
240-270o/3 m s-1
552
Collection shipping data
553
Number of ships/direction
554
Shipping speed
RI PT
546
360 m/east-west 0.5-1.5 m s-1
none
27 (N-S)/28 (S-N) 3.2 m s-1
555 556
SC
185-250o/2.7 m s-1 automatic (AIS)
M AN U
manual (on-site)
28-07-2013; 09-18 h
24 (down-stream)/15 (up-stream) 4.4 m s-1 (down)/ 3.7 m s-1 (up)
Table 2: Distances of the exhaust plume downwind from the passing ships to the
558
monitoring equipment (X) in m and the lateral (σy) and vertical (σz) dispersion in m of
559
shipping plumes at the monitoring sites during the first and second monitoring campaigns.
561
X
σy
σz
(m)
(m)
(m)
563
first campaign (northbound)
51
9
7
564
first campaign (southbound)
76
14
9
565
second campaign (downstream)
233
38
23
566
AC C
562
EP
560
TE D
557
second campaign (upstream)
277
44
28
567 568 569 570
ACCEPTED MANUSCRIPT 571
Table 3: The measured maximum PN, average PN and average EC concentrations (PN-
572
max, PN-avg and EC-avg) and the inferred maximum EC (EC-max) in the shipping plumes
573
(n) near the Amsterdam-Rhine Canal (ARC) and the Waal.
574 575
average and 10th-90th percentile
576
ARC
577
(n=55)
RI PT
Waal (n=39)
578
PN-max (# cm-3)
580
PN-avg (# cm-3)
581
Ratio
582
EC-avg (µg m-3)
22,044 (8,555-41,879)
18,483 (10,431-31,525)
14,812 (5,803-27,855)
2.9 (1.2-5.4) 1.6 (0.6-2.8)
-3
EC-max (µg m )
4.2 (1.4-7.6)
584 585
1.5 (1.1-2.1)
0.7 (0.3-1.0)
M AN U
583
53,712 (14,887-133,199)
SC
579
1.0 (0.4-1.7)
586
Table 4: The average and range of emission factors for EC of ships on the Amsterdam-
587
Rhine Canal (ARC) and the Waal.
TE D
588
EC (µg s-1) per ship
589
number of ships
590
(n)
average
27
3,800
300 – 11,000
28
4,000
400 – 11,000
10th - 90th percentile
ARC (northbound)
592
ARC (southbound)
593
Waal (downstream)
24
4,100
500 – 16,400
594
Waal (upstream)
15
6,600
500 – 23,500
AC C
595
EP
591
ACCEPTED MANUSCRIPT 596
Table 5: The average number of ships N in # h-1, speed V in m s-1 and emission rates EEC in
597
µg m-1 s-1 for the Amsterdam-Rhine Canal (ARC) and the Waal in 2011.
598
N
V
(# h-1)
600
EEC
(m s-1)
601
ARC (northbound)
6
3.8
602
ARC (southbound)
6
3.6
ARC (total)
12
3.7
604
Waal (downstream)
10
4.4
605
Waal (upstream)
10
3.7
606
Waal (total)
20
4.1
1.5
AC C
EP
TE D
M AN U
SC
603
(µg m-1 s-1)
RI PT
599
1.7
3.2
2.7
6.0
8.7
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
AC C
EP
TE D
Figure 5: The percentage of households (%) within 200 m from busy waterways in the Netherlands exposed to additional annual average EC (µg m-3) due to shipping emissions.
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
AC C
EP
TE D
Figure 1: Important inland waterways in Europe and the Netherlands and the two measurement sites: east of the Amsterdam-Rhine Canal (1) and north of the river Waal (2); a typical inland water vessel on the Amsterdam-Rhine Canal.
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
Figure 2: One minute average number of particles (#/cm3) (“PN”) and elemental carbon (10*ng/m3) (“EC”) during the first (Amsterdam-Rhine Canal) and second (Waal) monitoring campaign and passing ships in both directions.
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
Figure 3: The annual average contribution of shipping emissions to EC concentrations in µg m-3 on both sides of the Waal (left = north; right = south) and the Amsterdam-Rhine Canal (left = west; right = east) in 2011; straight (Waal) and dotted (Amsterdam-Rhine Canal)
AC C
EP
lines indicate the banks of both waterways.
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
Figure 4: The impact of shipping emissions to EC concentrations in µg m-3 on the population living within 200 m of Dutch waterways with more than 3 µg EC emissions per m and per s. The measurement sites are also indicated near the Amsterdam-Rhine Canal (1) and the Waal (2).
ACCEPTED MANUSCRIPT
Inland shipping important source of black carbon emissions
•
Additional annual average black carbon near busy waterways up to 0.5µg EC per m3
•
30% of the ships contribute to 80% of the total shipping emissions
AC C
EP
TE D
M AN U
SC
RI PT
•