Content uploaded by James J. Winebrake
Author content
All content in this area was uploaded by James J. Winebrake on Jan 08, 2018
Content may be subject to copyright.
Mortality from Ship Emissions: A
Global Assessment
JAMES J. CORBETT,*
,†
JAMES J. WINEBRAKE,
‡
ERIN H. GREEN,
‡
PRASAD KASIBHATLA,
|
VERONIKA EYRING,
⊥
AND AXEL LAUER
⊥
College of Marine and Earth Studies, University of Delaware,
305 Robinson Hall, Newark, Delaware 19716, Department of
STS/Public Policy, Rochester Institute of Technology,
1356 Eastman, Rochester, New York 14623, Nicholas School of
the Environment, Duke University, Box 90328, Durham,
North Carolina 22708, and Deutches Zentrum fuer Luft- und
Raumfahrt (DLR) DLR-Institute fuer Physik der Atmosphaere,
Oberpfaffenhofen, Wessling, Germany
Received July 09, 2007. Revised manuscript received September
28, 2007. Accepted October 04, 2007.
Epidemiological studies consistently link ambient concentrations
of particulate matter (PM) to negative health impacts,
including asthma, heart attacks, hospital admissions, and
premature mortality. We model ambient PM concentrations
from oceangoing ships using two geospatial emissions inventories
and two global aerosol models. We estimate global and
regional mortalities by applying ambient PM increases due to
ships to cardiopulmonary and lung cancer concentration-
risk functions and population models. Our results indicate that
shipping-related PM emissions are responsible for approximately
60,000 cardiopulmonary and lung cancer deaths annually, with
most deaths occurring near coastlines in Europe, East Asia,
and South Asia. Under current regulation and with the expected
growth in shipping activity, we estimate that annual mortalities
could increase by 40% by 2012.
Introduction
The marine transport sector contributes significantly to air
pollution, particularly in coastal areas (1–8). Annually, ocean-
going ships are estimated to emit 1.2–1.6 million metric tons
(Tg) of particulate matter (PM) with aerodynamic diameters
of 10 µm or less (PM10), 4.7–6.5 Tg of sulfur oxides (SOxas
S), and 5–6.9 Tg of nitrogen oxides (NOxas N) (9–12). Recent
studies have estimated around 15% of global NOxand 5–8%
of global SOxemissions are attributable to oceangoing ships
(10, 11). Given nearly 70% of ship emissions occur within
400 km of land (2, 11, 12), ships have the potential to
contribute significant pollution in coastal communities–––
especially for SOx. For instance, Capaldo et al. (1) estimate
that ship emissions contribute between 5 and 20% of non-
sea-salt sulfate concentrations and 5–30% of SO2concentra-
tions in coastal regions.
Numerous studies in recent years have consistently linked
air pollution to negative health effects for exposed popula-
tions (13, 14). Ambient concentrations of PM have been
associated with a wide range of health impacts including
asthma, heart attacks, and hospital admissions. An important
PM-related health effect is premature mortality; in particular,
increases in concentrations of PM with aerodynamic diam-
eters of 2.5 µm or less (PM2.5) have been closely associated
with increases in cardiopulmonary and lung cancer mortali-
ties in exposed populations (15). Cohen et al. estimated
approximately 0.8 million deaths per year worldwide from
outdoor urban PM2.5 air pollution, 1.2% of global premature
mortalities each year (16).
Emissions from international ships are increasingly a focus
for proposed regulation in local, national, and international
arenas (8, 17, 18). Yet, in many ways regulatory deliberations
have not been fully informed, as the extent of shipping
emissions health impacts has been unknown. Previous
assessments of regional shipping-related health impacts
focused on European or Western United States regions, and
ignore long-range and hemispheric pollutant transport (8, 19).
This undercounts international shipping impacts within local
and regional jurisdictions, and does not properly inform
international policy decision making.
Assessing Mortality from Atmospheric Modeling of Ship
Emissions
Our approach is similar to that of other studies (15, 16, 20, 21):
(1) determine pollutant emissions from ships; (2) apply
atmospheric transportation and chemistry models to estimate
the increased concentrations due to ships; (3) estimate
increased risk to exposed population due to these additional
concentrations; and (4) calculate additional mortalities due
to that increased risk.
We use two different geospatial ship data sets to help us
construct geospatial emission inventories: the International
Comprehensive Ocean-Atmosphere Data Set (ICOADS) by
Corbett et al. (10), and the Automated Mutual-assistance
Vessel Rescue system (AMVER) by Endresen et al. (12). These
two data sets combine detailed information about vessel
characteristics with vessel traffic densities to determine
emissions geospatially. However, each data set allocates ship-
traffic intensities differently. For example, while all oceango-
ing commercial ship types are included in these data sets,
ICOADS oversamples container ship traffic and refrigerated
cargo ship (i.e., reefer) traffic, and AMVER oversamples bulk
carrier and tanker traffic. Ship inventory differences affect
regional atmospheric pollution concentrations, potentially
influencing health effects estimates. Both inventories provide
emissions data on a monthly time-resolution; for atmospheric
modeling, we assume emissions occur uniformly throughout
each month.
We generated three emissions inventory data sets for
comparison. First, we used monthly resolved ICOADS 2002
emissions estimates of NOx,SO
x, black carbon (BC), and
particulate organic matter (POM) at a 0.1°×0.1°global grid
resolution (Inventory A). Second, we used AMVER 2001
emissions estimates of NOx,SO
x, BC, and POM at a 1°×1°
global grid resolution from Eyring et al. (Inventory B)(11).
Because of recent attention on the growth in commercial
shipping activity, we also produced ICOADS-based ship
inventories for 2012 (Inventory C) forecast using a uniform
global average growth rate of 4.1% (3, 10). Both inventories
represent shipping routes for most cargo shipping, and some
oceangoing passenger shipping activity, but neither ad-
equately represents typical fishing fleets and passenger ferry
service; therefore, we adjust global inventories to represent
only cargo and passenger ships. Table 1 shows total annual
shipping-attributable emissions for each inventory.
* Corresponding author phone: (302) 831-0768; e-mail:
jcorbett@udel.edu.
†
University of Delaware.
‡
Rochester Institute of Technology.
|
Duke University.
⊥
DLR-Institute fuer Physik der Atmosphaere.
Environ. Sci. Technol. 2007, 41, 8512–8518
8512 9ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 41, NO. 24, 2007 10.1021/es071686z CCC: $37.00 2007 American Chemical Society
Published on Web 11/05/2007
Global-scale models may model differently the PM2.5
concentrations used in health-effects estimates. We compare
increased ambient PM2.5 concentrations from marine ship-
ping using two atmospheric models. The first, GEOS-Chem
(22), is a global 3-D atmospheric composition model driven
by assimilated meteorological observations from the Goddard
Earth Observing System (GEOS). GEOS-Chem output pro-
vided us with ambient dry concentrations of BC, POM, and
sulfates from ocean-going ships separately from total PM2.5
attributed to all other sources. The second model, ECHAM5/
MESSy1-MADE (referred to as E5/M1-MADE), is an aerosol
microphysics module (MADE) coupled to a general circula-
tion model (ECHAM5), within the framework of the Modular
Earth Submodel System MESSy (23). Along with global PM2.5
concentrations attributed to nonship sources, the E5/M1-
MADE model provided ambient concentrations of BC, POM,
and sulfates for direct comparison with GEOS-Chem results;
separately the model produced concentrations of total PM2.5
constituents related to shipping (including nitrates and
ammonium ions). The Supporting Information includes
additional detail for both models.
Comparing results of each model with and without ship
inventories of PM2.5 components, we quantify ambient
concentrations of PM2.5 due to marine shipping. Worldwide
concerns about SOxemissions from ships are motivating the
replacement of marine residual oil (RO) with cleaner fuels,
such as marine gas oil (MGO) and marine diesel oil (MDO),
which will directly impact BC, POM, and sulfates attributed
to ships; therefore, we model total PM and the subset of PM
from ships most commonly associated with current marine
fuels. We defined the following cases to investigate robustness
of mortality estimates under different inventory and modeling
choices:
Case 1 compares PM2.5 concentrations with and without
ship emissions from model simulations with Inventory A.
This was done three times: Case 1a examines BC, POM, and
sulfates only, using the GEOS-Chem model; Case 1b uses the
E5/M1-MADE model to examine BC, POM, and sulfates for
direct comparison with GEOS-CHEM; Case 1c uses the E5/
M1-MADE model to examine total PM from ships.
Case 2 compares PM2.5 concentrations with and without
ship emissions from model simulations with Inventory B in
the E5/M1-MADE model. This was done twice: Case 2a for
BC, POM, and sulfates only; and Case 2b for all PM
constituents.
Case 3 compares PM2.5 concentrations with and without
ship emissions from model simulations with Inventory C
representing estimated 2012 emissions from increased ship-
ping activity. The case examines BC, POM, and sulfates only,
using the GEOS-Chem model. Note that Case 3 estimates
ignore potential emissions growth (or reduction) from other
sources between 2002 to 2012; however, we use Case 3 only
to estimate the additional mortality from oceangoing trade
growth, not to estimate total change in mortality due to all
sources of PM2.5.
Figure 1 depicts an annual aggregation of one of our two
midrange estimated contributions of PM2.5 concentrations
due to shipping in 2002 (Case 2a). Concentration increases
from ships range up to 2 µg per cubic meter (µg/m3) and
occur primarily over oceans and coastal regions.
TABLE 1. Annual Emission Totals of Particulate Matter and Trace Gases from Shipping in Tg/yr for the Three Different
Inventories Considered in This Study
Inventory A for 2002
(Corbett et al., 2007 (4))
Inventory B for 2001
(Eyring et al., 2005 (11))
Inventory C for 2012
(this study)
spatial ship traffic proxy ICOADS AMVER ICOADS
fuel consumption in million tonnes 200 (cargo and
passengers only)
280 (world fleet
including auxiliary engines)
299 (cargo and
passengers only)
NOx16.4 21.3 24.5
SOx9.2 11.7 13.7
primary SO40.35 0.77 0.50
CO 1.08 1.28 1.61
BC 0.07 0.05 0.10
POM 0.71 0.13 1.06
FIGURE 1. Annual average contribution of shipping to PM2.5 concentrations for Case 2b (in µg/m3)
VOL. 41, NO. 24, 2007 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 98513
Annual PM2.5 concentrations were used to assess annual
mortality due to long-term PM exposure, consistent with
Pope et al. (15). This requires an estimate of exposed
population. We used 2005 global population estimates
(obtained in a 1°×1°format) from the Socioeconomic Data
and Applications Center (SEDAC) at Columbia University
(24). To conform to the population data resolution, we
interpolated to a 1°×1°resolution the atmospheric
concentration output for each of our cases (provided at 2°
latitude ×2.5°longitude in GEOS-Chem and at 2.8°×2.8°
longitude by latitude in E5/M1-MADE). We note that for
most areas (with population growth) the use of 2005
population estimates will slightly overestimate our 2002
mortalities and slightly underestimate our 2012 mortalities.
Our mortality estimates are based on cardiopulmonary
and lung cancer causes of death for adults over 30 years of
age. Therefore, we applied U.S. Census Bureau International
Database estimates to derive, by continent, the percentage
of each grid cell’s population over 30 years old (25).
We also required background incidence rates of mortality
due to the health effects under study. Incidence rates were
estimated using World Health Organization (WHO) 2002 data
aggregated to the WHO region level (26). WHO cause of death
by age estimates were used to derive incidence rates for the
30–99 age group for each of the six WHO regions. Similar to
another assessment of global mortality from outdoor pol-
lution, lung, tracheal, and bronchial cancers were considered
“lung cancers” for our purposes (20); these cancers are
aggregated and nondistinguishable in WHO burden of disease
estimates. United States cardiopulmonary incidence values
obtained from the U.S. EPA (27) were used for North
America.
In calculating mortality effects we used C-R functions
derived from an American Cancer Society cohort study that
examined the relationship between PM2.5 and lung cancer
and cardiopulmonary mortality in the United States (15).
We apply these U.S.-derived C-R functions to our entire
spatial data set, recognizing that transferring U.S.-derived
functions to the global population introduces uncertainty to
the analysis, because socioeconomic factors have been
associated with effects of PM exposure on mortality and
relative risks (28, 29). However, other researchers have
demonstrated that the relationship between short-term PM
exposure and mortality is relatively consistent across several
countries and continents (21, 30, 31). We employ a log-linear
exposure function using Pope (15) to estimate long-term
mortality effects of PM2.5, as recommended and described
by Ostro (21). These equations reduce to an effects equation
as follows:
E)
[
1-
(
X0+1
)
(
X1+1
)
]
β
·B·P(1)
where Erepresents total effects (deaths/year); X1is the
pollutant concentration for the case under study in µg/m3;
X0is the pollutant background concentration in µg/m3;βis
an estimated parameter based on the health effect under
study; Brepresents the general incidence of the given health
effect (e.g., cardiopulmonary deaths/person/year), and P
represents the relevant exposed population (detailed equa-
tions are derived in the Supporting Information).
Ship PM-Induced Global and Regional Premature
Mortality
Exposure to shipping-related PM2.5 emissions in 2002 resulted
in 19,000 (Case 1a) to 64,000 (Case 1c) cardiopulmonary and
lung cancer mortalities globally, depending on the emission
inventory and on the particles considered. Approximately
92% of the estimated premature mortalities are from car-
diopulmonary illnesses. Mortalities increase by approxi-
mately 40% in 2012 due to trade-driven growth in shipping
emissions.
Figure 2 reveals that mortalities are concentrated in
distinct regions. We estimate regional impacts separately in
Table 2 for North America (NA); Europe/Mediterranean
(EUM); East Asia (EA), including China and Japan; South
Asia (SA), including India and Indonesia; and Eastern South
America (ESA). Regional burden of mortality varies, with the
greatest effects seen in the EUM (20–40% of global mortali-
ties), EA (20–30%), and SA (15–30%) regions.
Figures 2, 3, and 4 depict our cardiopulmonary mortality
estimates by grid cell for Case 2a for the entire globe, the
EUM region, and the EA/SA regions, respectively. Mortality
estimates of less than 1 per grid cell are excluded to facilitate
readability.
As expected, regions with the greatest mortality effects
are also those where shipping-related PM2.5 concentrations
FIGURE 2. Cardiopulmonary mortality attributable to ship PM2.5 emissions worldwide, Case 2b.
8514 9ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 41, NO. 24, 2007
TABLE 2. Annual Cardiopulmonary and Lung Cancer Mortality Attributable to Ship PM2.5 Emissions by Region and by Case (Best Estimate from C-R function
a
(95% confidence interval
b
))
Region
Case 1a Case 1b Case 1c Case 2a Case 2b Case 3 (2012 Forecast)
Inventory A Inventory A Inventory A Inventory B Inventory B Inventory C
Model: GEOS-Chem Model: E5/M1-MADE Model: E5/M1-MADE Model: E5/M1-MADE Model: E5/M1-MADE Model: GEOS-Chem
PM: BC, POM, SO4PM: BC, POM, SO4PM: All PM: BC, POM, SO4PM: All PM: BC, POM, SO4
North America (NA) Region
cardiopulmonary 1,860 (680–3,050) 2,820 (1,020 – 4,610) 4,590 (1,660 – 7,510) >5,470 (1,980 – 8,950) 7,910 (2,870 – 12,940) 2,770 (1,010 – 4,540)
lung cancer 210 (80 – 350) 320 (120 – 520) 520 (190 – 850) 620 (230 – 1,020) 900 (330 – 1,470) 320 (120 – 520)
NA Total 2,070 (760 – 3,400) 3,140 (1,140 – 5,130) 5,110 (1,850 – 8,360) 6,090 (2,210 – 9,970) 8,810 (3,200 – 14,410) 3,090 (1,130 – 5,060)
Europe/Mediterranean (EUM) Region
cardiopulmonary 6,770 (2,450 – 11,070) 11,830 (4,290 – 19,350) 24,350 (8,840 – 39,810) 7,250 (2,630 – 11,860) 15,100 (5,480 – 24,690) 8,990 (3,260 – 14,700)
lung cancer 670 (250 – 1,090) 1,100 (410 – 1,800) 2,360 (870 – 3,840) 650 (240 – 1,060) 1,430 (530 – 2,320) 880 (330 – 1,440)
EUM Total 7,440 (2,700 – 12,160) 12,930 (4,700 – 21,150) 26,710 (9,710 – 43,650) 7,900 (2,870 – 12,920) 16,530 (6,010 – 27,010) 9,870 (3,590 – 16,140)
East Asia (EA) Region
cardiopulmonary 3,490 (1,270 – 5,710) 11,970 (4,340 – 19,590) 17,920 (6,500 – 29,300) 9,640 (3,500 – 15,780) 13,800 (5,010 – 22,570) 5,170 (1,880 – 8,460)
lung cancer 370 (140 – 610) 1,300 (480 – 2,110) 1,950 (720 – 3,170) 1,030 (380 – 1,680) 1,480 (550 – 2,410) 550 (200 – 900)
EA Total 3,860 (1,410 – 6,320) 13,270 (4,820 – 21,700) 19,870 (7,220 – 32,470) 10,670 (3,880 – 17,460) 15,280 (5,560 – 24,980) 5,720 (2,080 – 9,360)
South Asia (SA) Region
cardiopulmonary 4,050 (1,470 – 6,630) 7,250 (2,630 – 11,870) 9,440 (3,420 – 15,450) 11,240 (4,080 – 18,390) 15,460 (5,610 – 25,260) 6,090 (2,210 – 9,970)
lung cancer 230 (90 – 380) 390 (150 – 640) 510 (190 – 830) 600 (220 – 970) 820 (300 – 1,340) 350 (130 – 570)
SA Total 4,280 (1,560 – 7,010) 7,640 (2,780 – 12,510) 9,950 (3,610 – 16,280) 11,840 (4,300 – 19,360) 16,280 (5,910 – 26,600) 6,440 (2,340 – 10,540)
East South America (ESA) Region
cardiopulmonary 380 (140 – 620) 520 (190 – 850) 690 (250 – 1,130) 1,120 (410 – 1,840) 1,540 (560 – 2,520) 570 (210 – 930)
lung cancer 50 (20 – 90) 70 (30 – 120) 100 (40 – 160) 160 (60 – 260) 220 (80 – 350) 80 (30 – 130)
ESA Total 430 (160 – 710) 590 (220 – 970) 790 (290 – 1,290) 1,280 (470 – 2,100) 1,760 (640 – 2,870) 650 (240 – 1,060)
Global
cardiopulmonary 17,340 (6,290 – 28,390) 35,610 (12,910 – 58,260) 58,640 (21,270 – 95,900) 36,970 (13,410 – 60,490) 56,790 (20,600 – 92,870) 24,780 (8,980 – 40,540)
lung cancer 1,580 (580 – 2,570) 3,260 (1,200 – 5,310) 5,540 (2,050 – 9,020 3,220 (1,190 – 5,240) 5,050 (1,870 – 8230) 2,240 (830 – 3,650)
Global Total 18,920 (6,870 – 30,960) 38,870 (14,110 – 63,570) 64,180 (23,320 – 104,920) 40,190 (14,600 – 65,730) 61,840 (22,470 – 101,100) 27,020 (9,810 – 44,190)
a
Values are rounded to the nearest 10.
b
Confidence interval range is based on uncertainty in the concentration–response function coefficients.
VOL. 41, NO. 24, 2007 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 98515
are high (compare Figures 1 and 2)—near coastal regions,
major waterways, and in highly populated areas. For Case
2a we estimate annual cardiopulmonary mortalities from
shipping reaching densities greater than 300 per grid cell in
FIGURE 3. Case 2b annual cardiopulmonary mortality attributable to ship PM2.5 emissions for Asia.
FIGURE 4. Case 2b annual cardiopulmonary mortality attributable to ship PM2.5 emissions for Europe/Mediterranean.
8516 9ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 41, NO. 24, 2007
regions of Asia, and between 100 and 200 in the EUM region,
as shown in Figures 3 and 4; coastal health-impact densities
are thousands of times greater than those seen in inland
regions.
Multiscale Cross-Comparisons
We compare our findings with other studies of PM2.5 related
mortality that employed alternative modeling or inventories
to estimate PM2.5 concentrations and health effects on three
scales: global, national/continental, and state/regional.
Concentration–response functions are used to estimate
global mortality for PM2.5 from anthropogenic sources
including shipping. These are compared to an analysis of
global mortality associated with long-term exposure to PM2.5
pollution (16, 20). Cohen et al. estimated that approximately
712,000 cardiopulmonary deaths are attributable to urban
outdoor PM2.5 pollution annually. With adjusting assump-
tions, our Case 1a estimate of 737,000 is within 4% of Cohen’s
(20) findings, and our Case 2b estimate is within 25% (see
table in Supporting Information).
We evaluate potential bias of using WHO region-level
incidence rates and continent-level age demographic esti-
mates in predicting mortalities at the national scale (24–26)].
We compare Case 1a mortality results over the United States
with mortality estimates from a similar analysis using the
U.S. Environmental Protection Agency’s Benefit Mapping
and Analysis Program (BenMAP). BenMAP is a geographic
information systems program which combines U.S. Census-
level population and incidence data at county-level resolution
with user-supplied air quality data to estimate heath effects.
We input our 1°×1°PM2.5 concentration data in BenMAP
for the United States, and applied the C-R functions within
BenMAP. We obtain Case 1a mortality estimates within 6%
of BenMAP estimates, as detailed in the Supporting Infor-
mation. The close agreement indicates that our population
demographics and incidence rate approximations produce
suitably accurate results when examining large regions,
recognizing that our confidence in this statement is based
on a U.S.-based analysis.
Direct comparison of our mortality estimates with recent
work estimating PM health effects in Europe by Cofala et al.
(8) is not possible because that study used an approach that
estimates loss of life expectancy in months rather than total
number of premature deaths. However, our patterns of health
impacts for Europe among our cases appear consistent with
patterns reported for their health-effects analysis (see
Figure 6.1 of Cofala et al.).
Lastly, we compare our California global grid results for
Case 1a and Case 2c with results from a report by the
California Air Resources Board (18). As described in the
Supporting Information, our Case 1a estimate is about 186%
of the ARB estimate, and our Case 2b estimate is about 242%
of the ARB estimate. In addition to differences in population
and incidence at local scale, reasons to expect larger California
mortality estimates in our assessment include the following.
First, ARB excluded sulfates from its source-specific analyses.
We include sulfates in our PM2.5 concentrations, which on
average comprise 24% of ambient PM concentrations; ARB
includes nitrates, which on average may comprise some 13%
of ambient PM concentrations (32). Second, ARB only
included PM2.5 emissions from ocean-going ships within 24
nautical miles from shore in its analysis; all other emissions
were allocated to the outer continental shelf air basin (19).
ARB also assumed that between 10% and 25% of ship
emissions reached populated areas. In contrast, our modeling
directly estimates land-exposure from worldwide ocean-
going ship inventories, considering atmospheric transport
of ship emissions to California from unbounded distances
as attributed by atmospheric chemical transport functions
in GEOS-Chem and E5/M1-MADE. Third, our “California”
case is made up of 1°×1°grid cells that overlap small parts
of Nevada, Utah, and Mexico and could lead to slightly higher
estimates than a strict California-only comparison. On the
other hand, ARB used smaller (more resolved) grid cells; all
else equal, we would have expected this to yield larger not
smaller health impacts in the CARB report because CARB
would more accurately capture near-source population
density.
Discussion
Our results indicate that shipping-related PM emissions from
marine shipping contribute approximately 60,000 deaths
annually at a global scale, with impacts concentrated in
coastal regions on major trade routes. Most mortality effects
are seen in Asia and Europe where high populations and
high shipping-related PM concentrations coincide. Based
on previous estimates of global PM2.5-related mortalities (16),
our estimates indicate that 3% to 8% of these mortalities are
attributable to marine shipping. We identify three categories
of uncertainty, ranked by their importance to estimates in
this work: (i) ship inventory and PM constituent uncertainties
most influence our best estimates across all Cases; (ii) the
95% confidence intervals on the health effects C-R functions
represent significant uncertainty (capturing toxicity and
response effects) that similarly affects each case; (iii) atmo-
spheric modeling uncertainties vary where emissions offshore
expose coastal and inland populations. Uncertainties are
discussed in the Supporting Information; results may be more
uncertain at local scales, given the lack of detailed localized
data pertaining to incidence, demographics, PM2.5 concen-
trations, and other factors.
The absence of localized C-R functions and incidence
rates prevents precise quantification of all anticipated PM-
related health effects, such as asthma and hospital admis-
sions, etc. Though we only examine cardiopulmonary and
lung cancer mortalities, we expect that regions where ships
contribute most to mortality effects (concentrated population
areas with high shipping-related PM levels) will also suffer
other related health impacts. We anticipate future work to
investigate variation and uncertainty in these inputs further.
Higher resolved atmospheric models could provide more
accurate or precise results on a regional level by targeting
regions of interest where better localized data for ship
emissions, incidence rates, and population demographics
are available.
Our work demonstrates that mortality and health benefits
in multiple regions globally could be realized from policy
action to mitigate ship emissions of primary PM2.5 formed
during engine combustion and secondary PM2.5 aerosols
formed from gaseous exhaust pollutants. These results
support regional assessments of health impacts from ship
PM2.5 emissions, and identify other regions where similar
impacts may be expected. Current policy discussions aimed
at reducing ship emissions are focused on two concerns: (i)
the geospatial aspects of policy implementation and compli-
ance (e.g., uniform global standards versus requirements for
designated control areas); and (ii) the benefits and costs of
various emission-reduction strategies (e.g., fuel switching
versus aftertreatment technologies or operational changes).
Our work quantifies the baseline estimates of mortality due
to ship emissions from which future work would estimate
mitigation benefits.
Acknowledgments
This work was partly supported by the Oak Foundation (J.J.C.,
J.J.W., E.H.G., P.K.), and the German Helmholtz-Gemeinschaft
Deutscher Forschungszentren (HGF) and by the German
Aerospace Center (DLR) within the Young Investigators Group
VOL. 41, NO. 24, 2007 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 98517
SeaKLIM (V.E. and A.L.). We acknowledge Chengfeng Wang,
currently with the California Air Resources Board, for his
efforts in constructing some of the emissions inventory data.
Supporting Information Available
Description of atmospheric aerosol model parameters,
calculations for cardiopulmonary mortality estimates, dis-
cussion of uncertainty in our analysis, and additional
discussion of our results. This material is available free of
charge via the Internet at http://pubs.acs.org.
Literature Cited
(1) Capaldo,K. P.; Corbett, J. J.; Kasibhatla, P.; Fischbeck, P.; Pandis,
S. N. Effects of Ship Emissions on Sulphur Cycling and Radiative
Climate Forcing Over the Ocean. Nature 1999,400, 743–746.
(2) Corbett, J. J.; Fischbeck, P. S.; Pandis, S. N. Global Nitrogen and
Sulfur Emissions Inventories for Oceangoing Ships. J. Geophys.
Res. 1999,104 (D3), 3457–3470.
(3) Corbett, J. J.; Fischbeck, P. S. Emissions from Waterborne
Commerce in United States Continental and Inland Waters.
Environ. Sci. Technol. 2000,34 (15), 3254–3260.
(4) Wang, C.; Corbett, J. J.; Firestone, J. Modeling Energy Use and
Emissions from North American Shipping: Application of the
Ship Traffic, Energy, and Environment Model. Environ. Sci.
Technol. 2007,41 (9), 3226–3232.
(5) Streets, D. G.; Guttikunda, S. K.; Carmichael, G. R. The Growing
Contribution of Sulfur Emissions from Ships in Asian Waters
1988–1995. Atmos. Environ. 2000,34 (26), 4425–4439.
(6) Streets, D. G.; Bond, T. C.; Carmichael, G. R.; Fernandes, S. D.;
Fu, Q.; He, D.; Klimont, Z.; Nelson, S. M.; Tsai, N. Y.; Wang,
M. Q.; Woo, J. H.; Yarber, K. F., An inventory of gaseous and
primary aerosol emissions in Asia in the year 2000. J. Geophys.
Res. 2003,108, (D21).
(7) European Commission; ENTEC UK Limited. Quantification of
emissions from ships associated with ship movements between
ports in the European Community; FS 13881; European Com-
mission: Brussels, Belgium, 2002.
(8) Cofala, J.; Amann, M.; Chris Heyes; Klimont, Z.; Posch, M.;
Schöpp, W.; Tarasson, L.; Jonson, J. E.; Whall, C.; Stavrakaki, A.
Final Report: Analysis of Policy Measures to Reduce Ship
Emissions in the Context of the Revision of the National Emissions
Ceilings Directive; International Institute for Applied Systems
Analysis: Laxenburg, Austria, 2007; p 74.
(9) Corbett, J. J.; Koehler, H. W. Updated Emissions from Ocean
Shipping. J. Geophys. Res., D: Atmos. 2003,108 (D20), 4650–
4666.
(10) Corbett,J. J.; Wang, C.; Winebrake, J. J.; Green, E. Allocation and
Forecasting of Global Ship Emissions; Clean Air Task Force and
Friends of the Earth International: Boston, MA, January, 11,
2007; 26.
(11) Eyring, V.; Köhler, H. W.; van Aardenne, J.; Lauer, A. Emissions
from international shipping: 1. The last 50 years. J. Geophys.
Res., D: Atmos. 2005,110 (D17), D17305.
(12) Endresen, O.; Soergaard, E.; Sundet, J. K.; Dalsoeren, S. B.;
Isaksen, I. S. A.; Berglen, T. F.; Gravir, G., Emission from
international sea transportation and environmental impact. J.
Geophys. Res., D: Atmos. 2003,108, (D17.)
(13) Nel,A. ATMOSPHERE: Enhanced: Air Pollution-Related Illness:
Effects of Particles. Science 2005,308 (5723), 804–806.
(14) Kaiser, J. EPIDEMIOLOGY: Mounting Evidence Indicts Fine-
Particle Pollution. Science 2005,307 (5717), 1858a–1861.
(15) Pope, C. A.; Burnett, R. T.; Thun, M. J.; Calle, E. E.; Krewski, D.;
Ito, K.; Thurston, G. D. Lung cancer, cardiopulmonary mortality,
and long-term exposure to fine particulate air pollution. JAMA
2002,287 (9), 1132–1141.
(16) Cohen, A. J.; Anderson, H. R.; Ostro, B.; Pandey, K. D.;
Krzyzanowski, M.; Künzli, N.; Gutschmidt, K.; Pope, A.; Romieu,
I.; Samet, J. M.; Smith, K. The global burden of disease due to
outdoor air pollution. J. Toxicol. Environ. Health, Part A 2005,
68, 1301–1307.
(17) Bailey, D.; Solomo, G. Pollution prevention at ports: clearing
the air. Environ. Impact Assess. Rev. 2004,24, 749–774.
(18) California Air Resources Board. Proposed Emission Reduction
Plan for Ports and Goods Movement in CA; CA Air Resources
Board: Sacramento, CA, March 22, 2006.
(19) California Air Resources Board. Appendix A: Quantification of
the Health Impacts and Economic Valuation of Air Pollution
from Ports and Goods Movement in CA; CA Air Resources Board:
Sacramento, CA, March 22, 2006.
(20) Cohen, A. J.; Anderson, H. R.; Ostro, B.; Pandey, K. D.;
Krzyzanowski, M.; Kunzli, N.; Gutschmidt, K.; Pope, C. A.;
Romieu, I.; Samet, J. M.; Smith, K. R., Mortality impacts of urban
air pollution. In Comparative Quantification of Health Risks:
Global and Regional Burden of Disease Due To Selected Major
Risk Factors; Ezzati, M., Lopez, A. D., Rodgers, A., Murray, C. J. L.,
Eds.; World Health Organization: Geneva, 2004; Vol. 2, pp 1353–
1394.
(21) Ostro, B. Outdoor air pollution: Assessing the environmental
burden of disease at national and local levels; World Health
Organization: Geneva, 2004.
(22) Bey, I.; Jacob, D. J.; Yantosca, R. M.; Logan, J. A.; Field, B. D.;
Fiore, A. M.; Li, Q.; Liu, H. Y.; Mickley, L. J.; Schultz, M. G. Global
modeling of tropospheric chemistry with assimilated meteorol-
ogy: Model description and evaluation. J. Geophys. Res. 2001,
106 (D19), 23073–23096.
(23) Lauer,A.; Eyring, V.; Hendricks, J.; Jöckel, P.; Lohmann, U. Global
model simulations of the impact of ocean-going ships on
aerosols, clouds, and the radiation budget. Atmos. Chem. Phys.
2007,7(19), 5061–5079.
(24) SEDAC(Socioeconomic Data and Applications Center). Gridded
Population of the World; Columbia University, 2007.
(25) U.S. Census Bureau. International Data Base, IDB Data -IDB
Aggregation - Table 94 Midyear Population, by Age and Sex;
Washington, DC, 2006.
(26) World Health Organization (WHO). Revised Global Burden of
Disease (GBD) 2002 Estimates: Mortality Data, GBD 2002: Deaths
by age, sex and cause for the year 2002; Geneva, 2004.
(27) Abt Associates. BenMap: Environmental Benefits Mapping and
Analysis Program, Technical Appendices; Office of Air Quality
Planning and Standards, U.S. Environmental Protection Agency:
Research Triangle Park, NC, May, 2005; p 275.
(28) O’Neill, M.; Jerrett, M.; Kawachi, I. Health, wealth, and air
pollution. Environ. Health Perspect. 2003,111, 1861–1870.
(29) Krewski, D.; Burnett, R. T.; Goldberg, M. S. Reanalysis of the
Harvard Six Cities Study and the American Cancer Society Study
of particulate air pollution and mortality. A special report of the
institute’s particle epidemiology reanalysis project; Health Effects
Institute: Cambridge, MA, 2000.
(30) Davis, D. L.; Kjellstrom, T.; Sloof, R.; McGartland, A.; Atkinson,
D.; Barbour, W.; Hohenstein, W.; Nalgelhout, P.; Woodruff, T.;
Divita, F.; Wilson, J.; Deck, L.; Schwartz, J. Short term improve-
ments in public health from global-climate policies on fossil-
fuel combustion: an interim report. The Lancet 1997,350, 1341–
1349.
(31) Anderson, H. R.; Atkinson, R. W.; Peacock, J. L.; Marston, L.;
Konstantinou, K. Meta-analysis of time-series studies and panel
studies of Particulate Matter (PM) and Ozone (O3); 5042688;
World Health Organization: Copenhagen, 2004.
(32) Hodan, W. M.; Barnard, W. R. In Evaluating the Contribution
of PM2.5 Precursor Gases and Re-entrained Road Emissions to
Mobile Source PM2.5 Particulate Matter Emissions; 13th annual
emission inventory conference, Clearwater, FL, 8–10 June, 2004;
Administration, M. F. P. U. C. t. t. F. H., Ed.; Emission Factors
and Inventory Group Emission Inventory Improvement Pro-
gram, Emissions, Monitoring and Analysis Division, Office of
Air Quality Planning & Standards, U.S. Environmental Protection
Agency: Clearwater, FL, 2004.
ES071686Z
8518 9ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 41, NO. 24, 2007
