Silent but deadly: The case of shipping emissions

Earlier this month a group of researchers from the ICCT, The George Washington University Milken Institute School of Public Health, and the University of Colorado Boulder released a new study assessing premature mortality associated with air pollution from transportation. The study found that fine particulate matter (PM2.5) and ozone from on-road vehicles, non-road engines, and oceangoing vessels was linked to an estimated 385,000 * premature deaths in 2015 worldwide. About half of these deaths were attributed to air pollution from diesel cars, trucks and buses. But a surprisingly large fraction of the early mortality—approximately 15%, or 60,000 deaths—were due to air pollution from the 70,000 international ships that ply the world’s oceans. That equates to about 160 billion dollars of health damages annually.

The study assessed health impacts using methods from the Global Burden of Disease (GBD) 2017. The methodology used can be considered conservative* in the number of deaths estimated. As a result, the estimates of air pollution health impacts are lower than other studies, and could be revised upward if any of these assumptions are relaxed.

Still, the study highlights the uneven distribution of premature mortality due to air pollution from international shipping. It provides the raw data, which allows anyone to run their own secondary analysis. We put together a follow-up analysis of shipping impacts using that data, and found some interesting results, namely that many of these deaths occur in places one might not expect.

Top 20 shipping mortality table

The table above summarizes two metrics: total early deaths attributable to shipping air pollution in 2015, and early deaths per 100,000 population. It’s not too much of a surprise that China, which hosts seven of the ten busiest ports by throughput and has many millions living near impacted shores, accounts for more than one third (37%) of the estimated 60,000 odd premature deaths. Likewise, Japan (4,100), India (3,400), the UK (3,200), and Indonesia (1,900) each ranked within the top 5 by total early deaths due to their large populations and exposure to air pollution from major shipping lanes.

What is perhaps more surprising is that the per-capita early death rate, as expressed in deaths per 100,000 population, shows a very different set of countries. On this metric, Singapore is the country most impacted by air pollution from ships. Moreover, six of the 10 most impacted countries are in Europe. Only Japan and the UK appear in the top 10 most impacted countries on both metrics.

Why are these two lists so different? Obviously, countries with larger populations, like China, India, Brazil, and the United States, will have a large number of deaths even if the per-capita damages are relatively low. Per-capita early mortality is a bit different because it’s a function of the magnitude and exposure to emissions, and the sensitivity of the local population (the elderly, for example, have a higher incidence of both baseline disease and incremental impacts). It is also dependent on the baseline air quality, with a given level of emissions having a smaller marginal impact on human health as the air gets dirtier. So, countries like Singapore and European countries with relatively “clean” air and a higher fraction of elderly citizens end up with more per-capita mortality when it’s all tallied up.

The United States fares reasonably well for its population size, in part because it’s protected by the fuel switching and aftertreatment requirement of the North American Emission Control Area (ECA). But even the U.S. figure of 1,200 premature deaths per year is nothing to sneeze at. But where, might you ask, are those deaths taking place?

The study has an answer for that as well. It estimates transportation-attributable health impacts for the 200 most populated urban areas worldwide, of which 21 were located in the United States. The urban area definitions used in the study treat dense contiguous urban areas as one. For example, the urban area labeled ‘San Francisco’ includes much of the San Francisco Bay Area, encompassing a population of 4.6 million—much more than the municipal area of ~880,000 people.

The figure below shows the estimates for these 21 U.S. urban areas in 2015. The number of estimated premature deaths from shipping emissions and applicable population are shown on a log scale to account for the substantial variation in population size among these 21 urban areas. The bubble size indicates the number of shipping deaths per 100,000 population.

premature deaths from shipping emissions

As can be seen in the graphic, within the United States, the areas of Seattle and San Francisco lead in terms of early deaths per 100,000 residents (1.8 and 1.6), or more than double the global average, due to air pollution from the ports of Seattle, Tacoma, Oakland, and San Francisco. Other notable port cities, including Philadelphia, Miami, Houston, and New York, also showed per-capita premature mortality rates at or above the global average. It may come as some surprise to residents of these cities, many of which have reasonably clean air, that their health is at risk due to exposure to shipping air pollution.

So, what can be done? In the near-term, this study strengthens the case for impacted countries to apply for Emission Control Area (ECA) status to the IMO. The ECA system, which establishes tighter regional emission standards for engine emissions and fuel quality along the coast, has been found to be a cost-effective, reliable means of reducing air pollution and improving public health. China in particular is currently assessing the feasibility of applying an ECA around its shores.

A mid-term solution is for IMO to enforce new emission controls for black carbon, which in the form of elemental carbon impacts public health when fine particulates are inhaled deep into the lungs. Longer term, eventually we’ll need completely carbon-free ships powered by electricity, hydrogen, or potentially even ammonia. When that comes to pass, blissful ignorance about air pollution from shipping may be justified once more.


* The estimate of 385,000 premature deaths in 2015 is the central estimate. The 95% confidence interval reflecting uncertainty in the concentration-response function is 274,000–493,000. The methodology used can be considered conservative in the number of deaths estimated for several reasons. For example, the study evaluates the combined impacts of outdoor and indoor air pollution to avoid potential double counting of disease incidence and assumes no increased health risk from particulate matter below the single digit μg/m3 lower thresholds observed in outdoor air pollution cohort studies. The study also assumes that an increase in air pollution at extremely high concentrations has a smaller marginal effect on health than the same increase at lower concentrations (i.e. a non-linear concentration-response), and models increased risk only for the six diseases currently included in the GBD 2017 air pollution methods.