2.1 Emissions

Land-based emissions have been provided by the International Institute for Applied Systems Analysis (IIASA) within European FP7 project ECLIPSE. In this study we use version 5a (hereafter “ECLIPSEv5a”), a global emission data set at 0.5×0.5^∘ resolution, which has been widely used in recent years by the scientific community (http://www.iiasa.ac.at/web/home/research/researchPrograms/air/ECLIPSEv5.html, last access: 27 February 2019). ECLIPSEv5a is available in 5-year intervals from 2005 onwards, and in this study we have chosen data for 2015 and 2030.

The ECLIPSE v5a emissions were re-gridded using the TNO-MACC-III 0.125×0.0625 long–lat emission distribution (Kuenen et al., 2014) for the year 2011. During the re-gridding process only the spatial distribution of the ECLIPSE v5a emissions was modified, while the national and sector totals remained unchanged. Where TNO-MACC-III emissions are not available (such as northern Africa), the gridded ECLIPSE v5a emissions were interpolated to the TNO-MACC grid resolution. Any missing sectors for countries which were included in the TNO-MACC-III emission data were also completed from the interpolated ECLIPSE v5a emissions.

In reality land-based emissions will change between years. Annual emissions from years 2000 to 2016 for the European countries are listed in EMEP Status Report 1/2018 (2018). In the Baltic region reported changes in country emission are small, with the exception of SO_x emissions in Poland dropping by almost 20 % from 2014 to 2016.

In regard to ship emissions in the BAS, we use emission data as provided by the FMI (Finish Meteorological Institute) for the year 2014 (i.e. with 1 % maximum sulfur content in fuels in the SECA) and 2016 (maximum sulfur content reduced to 0.1 % in the SECA). For the remaining sea areas, ship emissions for the year 2015 are used from a previous global data set (Johansson et al., 2017).

The emissions from shipping have been calculated with the Ship Traffic Emission Assessment Model (STEAM) based on ship movements from the AIS which provides real-time information on ship positions. The model requires as input detailed technical specifications of all onboard fuel-consuming systems and other relevant technical details for all ships considered. The data from IHS Global (2017) constituted the most significant source of this information. The STEAM model is described in Jalkanen et al. (2009, 2012, 2016) and Johansson et al. (2013, 2017). Hourly emission grids for Baltic Sea ship emissions were produced based on vessel-specific modelling, considering the changes in fuel sulfur content that occurred between 2014 and 2016.

In STEAM scrubbers can operate in closed- or open-loop mode, depending on the equipment installed. If a hybrid scrubber system is known to be installed, it is assumed to operate in open-loop mode when the vessel operates in an area where open-loop systems are feasible. Closed-loop mode of a hybrid scrubber is assumed in the Bothnian Bay and restricted zones, like German waters. If a vessel has an open-loop scrubber installed and it enters a restriction zone, the model assumes a fuel switch to low-sulfur fuels. Emission modelling uses scrubber equipment type (closed/open/hybrid), vessel identity, and installation date as input to emission modelling. All future scrubber scenarios introduce hybrid scrubbers to the fleet.

Globally, during 2014 there were 77 vessels using a scrubber, of which 30 % were of open-loop, 48 % of closed-loop, and 22 % of hybrid type. By 2016 scrubber installations were doubled globally to 155 units. In the Baltic Sea area during 2016, there were 85 vessels operating a scrubber releasing 73 million tonnes of wash water into the sea. Almost all of this (99.8 %) discharge came from open-loop operation of scrubbers.

Ship-emitted pollutants were modelled using AIS data for the years 2014 and 2016. Any changes in vessel activity, fleet size, and development will have an impact on energy use and all pollutant emissions. However, the sulfur rule was the only significant change which had a large impact on emitted pollutants. Both PM and SO_x were reduced by this change, but only the sulfate fraction of PM was reduced accordingly, whereas other components of PM were less affected.

From 2021 onward, NO_x emissions for new ships have to comply with IMO Tier 3 regulations. These contributions were taken into account in the emission modelling. Future emission projections for the year 2030 also include changes in

• energy efficiency improvements, modelled following the method of Kalli et al. (2013), which goes beyond the Energy Efficiency Defined Index (EEDI) requirements of the IMO;

• fleet size increase;

• vessel size growth, assuming a linear annual growth dependent on ship types.

Annual growth rates in fleet size are implemented as percentage increase per type of ship: for example, if the annual percentage growth is n % for container ships, we duplicate n % of the container ships in the current fleet in the following year. The following growth rates are assumed for vessel DWT: vehicle carriers and RoRo: 1.25 % per annum; dry cargo: 0.4 % per annum; container carriers: 1.2 % per annum; liquid cargo: 2.0 % per annum; passenger vessels, ferries, and high-speed craft: 0.3 % per annum; cruise ships: 0.3 % per annum; fishing vessels: 0.3 % per annum. Vessel size growth for other types was set to zero. For those vessels, the vessel size remains at the 2014 level.

As the ship emission data are used for multiple meteorological years (see the next section), we did not retain the high (hourly) temporal resolution in the data, but rather aggregated them to monthly resolution before use in the chemistry transport model.

2.2 Model calculations of air pollutants and depositions

Concentrations of air pollutants and depositions of sulfur and nitrogen have been calculated with the EMEP MSC-W model (hereafter “EMEP model”), version rv4.14, at 0.1×0.1^∘ resolution for the domain between 30^∘ W and 45^∘ E and between 30 and 75^∘ N. A detailed description of the EMEP model can be found in Simpson et al. (2012), with later model updates being described in Simpson et al. (2018), and references therein. The EMEP model is available as open source (see https://github.com/metno/emep-ctm, last access: 27 February 2019) and is regularly evaluated against measurements as part of the EMEP status reports. See Gauss et al. (2016, 2017, 2018) for evaluations of the meteorological years 2014, 2015, and 2016, respectively. In addition, the EMEP model has successfully participated in model intercomparisons and model evaluations presented in a number of peer-reviewed publications: Colette et al. (2011, 2012), Angelbratt et al. (2011), Dore et al. (2015), Karl et al. (2019b), Stjern et al. (2016), and Jonson et al. (2018). Vivanco et al. (2018) evaluated depositions of sulfur and nitrogen species in Europe calculated by 14 regional models, showing good results for the EMEP model.

In the present study the model is driven by meteorological data from the European Centre for Medium-Range Weather Forecasts (ECMWF) based on the CY40R1 version of their IFS (Integrated Forecast System) model. All simulations for this paper have been run for the 3 meteorological years 2014, 2015, and 2016 and then averaged, in order to cancel out meteorological variability. The simulations are the following.

• Present_Base: base case with ship emissions of 2016. Land-based emissions for 2015 (from ECLIPSEv5).

• Present_NoShip: as Present_Base, but without ship emissions in the BAS.

• Present_HiSulphur: as Present_Base, but with ship emissions of 2014 (i.e. high sulfur content) in the BAS.

• Future_Base: ship emissions of 2030 (assuming NECA and business-as-usual development) and land-based emissions of 2030 (from ECLIPSEv5).

• Future_NoShip: as Future_Base, but without ship emissions in the BAS.

The emissions are also summarised in Table 1. In the future scenarios it is assumed that ships that are in compliance with the NECA regulations will operate the equipment (i.e. be compliant) also when sailing outside the NECA.

Table 1All model scenarios have been calculated for the 3 meteorological years 2014, 2015, and 2016. In the comparisons to measurements in Table 2 only year 2016 model calculations are shown. The land-based ECLIPSE emissions for 2016 have been interpolated between 2015 and 2020. SECA regulations for the North Sea are included in the remaining sea ship emissions. The 2020 sulfur cap is included in the 2030 ship emissions outside the SECAS.

Download Print Version | Download XLSX

3.1 Air pollution due to Baltic Sea shipping

Concentrations of NO₂ for Present_Base are shown in Fig. 1a. The lifetime of NO₂ is relatively short, and as a result the concentrations largely reflect the locations of the main source areas. Concentrations of NO₂ are high in central Europe and in and around the English Channel, with markedly lower concentrations north and east of the BAS. In the NOS and the BAS the major ship tracks are clearly visible. Figure 1c shows the difference between the Present_Base and Present_NoShip scenarios. The calculations show that ship emissions account for more than 50 % of NO₂ in central parts of the BAS and for a substantial percentage also in coastal zones, in particular in Denmark, southern parts of Sweden and Finland, and the Baltic states (Estonia, Latvia, and Lithuania). This is also illustrated in Table 2, where measured NO₂ at sites located in the BAS coastal regions are compared to the Present_Base, Present_NoShip, and Present_HiSulphur model calculations calculated with 2016 meteorology. The position of the measurement sites and the corresponding time-series plots for NO₂ are shown in Appendix A. In the Present_NoShip case NO₂ levels are clearly underestimated and correlations and rms errors deteriorated compared to the Present_Base calculation, in particular for those sites located very close to major shipping routes. The comparisons with measurements convincingly show that these measurements can only be reproduced when BAS ship emissions are included. The contributions to individual countries will be further discussed in a later section.

As shown in Table 2, measured SO₂ levels for 2016 are relatively well reproduced by the model for the Present_Base calculation. The position of the measurement sites and the corresponding time-series plots for SO₂ are shown in Appendix A. The effects of excluding the BAS ship emissions in the Present_NoShip scenario have only minor effects on the SO₂ levels. Replacing 2016 BAS emissions with 2014 ones (Present_HiSulphur) has much larger effects, resulting in an overestimation of SO₂ levels at most of the sites listed in Table 2, and in particular for Anholt and Råö, located very close to the shipping routes. This clearly illustrates the effects of the stricter SECA regulations. With the high ship emissions of 2014, the measurements for 2016 can not be reproduced. This is also a strong indication that the ships are largely in compliance with the SECA regulations. As for NO₂, the contributions to individual countries are discussed further in a later section.

PM_2.5 in the atmosphere is a mixture of many chemical species of both natural and anthropogenic origins. It is emitted as a primary pollutant and formed as a secondary pollutant in the atmosphere. As a result PM_2.5 concentrations are more spread out compared to NO₂. Concentrations decrease from south to north from a maximum in central Europe. As shown in Fig. 1d the percentage contributions from BAS shipping, calculated as Present_Base–Present_NoShip, are much smaller for PM_2.5 than for NO₂, but with noticeable contributions in coastal zones, in particular in parts of Denmark, Sweden, and Finland. Figure 1e shows higher contributions when assuming BAS shipping at 2014 levels (Present_HiSulphur), prior to the implementation of the stricter SECA regulations. These results are also illustrated in the comparisons of model scenario calculations at the measurement sites located in BAS coastal regions as listed in Table 2. The positions of the measurement sites and the corresponding time-series plots for PM_2.5 are shown in Appendix A. For PM_2.5 differences between the Present_Base and Present_NoShip cases are much smaller than for NO₂. Likewise, differences are smaller than for SO₂ between Present_Base and Present_HiSulphur. In Table 2 we also show measured and model-calculated concentrations of SO₄.

Continuing a downward trend from the late 1980s, land-based sulfur emissions have decreased by more than 50 %, i.e. more than for any other of the major air pollutants (Tista et al., 2018), and thus the importance of sulfur in particle formation has decreased relative to other anthropogenic emitted species and natural sources. In the SECAs the sulfur content in marine fuels has decreased from the global average of about 2.5 % to 1 % in 2011 and finally to 0.1 % in 2015. As a result of these large emission reductions the fraction of SO₄ in PM_2.5 in the BAS region has decreased even further here. At the sites in Table 2 both the measured and model-calculated fractions of SO₄ in PM_2.5 are about 0.15. As SO₄ makes up a moderate portion of the PM_2.5 composition, this fraction increases only by a small amount with the Present_HiSulphur scenario.

The model results underestimate the measurements at most of the sites listed. Based only on the comparisons between measurements and the different model scenarios for PM_2.5, one can not conclude that the Present_Base scenario is more realistic than the other two. As for NO₂ and SO₂, the contributions to individual countries are discussed further in a later section.

Figure 2(a, b) Calculated depositions of oxidised nitrogen and sulfur. (c, d) Present percentage contributions from BAS ship emissions to depositions of oxidised nitrogen and oxidised sulfur with reference to Base 2016. (e) Percentage contribution to depositions of oxidised sulfur with reference to 2014 BAS emissions.

3.2 Depositions of sulfur and nitrogen from Baltic Sea shipping

Total depositions (wet and dry) of oxidised sulfur and nitrogen for Present_Base are shown in Fig. 2a, b. The highest depositions of both sulfur and nitrogen are seen over central Europe. For nitrogen, high levels of depositions also extend into northern Germany and Denmark. Based on the difference between Present_Base and Present_NoShip, a significant amount of the nitrogen depositions can be attributed to BAS shipping (Fig. 2c), contributing more than 15 % of the total nitrogen depositions in major parts of the BAS and also in parts of Sweden, Finland, and the Baltic states (Estonia, Latvia, and Lithuania). Dry deposition is parameterised as a function of subgrid-scale ecosystems and is typically higher than the grid average for forest ecosystems (both coniferous and deciduous). This will affect the calculations of critical loads for acidification and eutrophication as the subgrid-scale ecosystem depositions are used in the critical load calculations. Critical loads will be discussed in a companion paper (Repka et al., 2019). Figure 2d shows that the calculated contributions from BAS shipping in 2016 to depositions of sulfur are very low (Present_Base–Present_NoShip) and much lower than what has been calculated assuming 2014 emissions (Present_HiSulphur–Present_Base) as shown in Fig. 2e, with percentage contributions exceeding 10 % in many coastal zones.

Table 3Annual average measured (Obs) and model-calculated concentrations (Calc.) in precipitation of oxidised nitrogen in mg N L−1 and oxidised sulfur in mg (inS) L−1 in 2016 for the present Base, NoShip, and HiSulphur scenarios. Also listed are the normalised mean bias (NMB), the daily correlations (Corr.), and rms errors (rms) between model and measurements. Km Balt. is a classification of the distance in kilometres between the stations and the Baltic Sea coast. The distance is equal to or smaller than the distance listed. The position of the measurement sites and the time-series plots are shown in Appendix A.

Download Print Version | Download XLSX

These findings for the depositions of oxidised nitrogen and sulfur are also illustrated in Table 3, where measured concentrations in precipitation at sites located in the BAS coastal regions are compared to the Present_Base, Present_NoShip, and Present_HiSulphur model calculations. Compared to Present_Base, averaged concentrations in precipitation are about 14 % lower for oxidised nitrogen when BAS ship emissions are excluded (Present_Base–Present_NoShip). The effects of the stricter SECA regulations are demonstrated by an increase of about 9 % in the calculated concentrations of oxidised sulfur in precipitation in the Present_HiSulphur scenario compared to the Present_Base calculation.

Figure 3For each country, the upper bar shows the future (2030) case and the lower bar the present case country average concentration. (a) SO₂, (b) NO₂, (c) PM_2.5, and depositions of oxidised sulfur (d) and oxidised nitrogen (e). The black and green bars represent the Present_NoShip and Future_NoShip calculations, respectively. The additional contributions from BAS (Add Baltic) are shown in blue and the additional effect assuming high sulfur fuel emissions (Add Baltic 2014) in red. (These are also given as numbers. Numerical values for NO₂ Add Baltic 2014 and for SO₂ Add Baltic are not given as they are very small.)

Download

3.3 Contributions to individual countries from BAS shipping.

Figure 3 shows the concentrations of NO₂, SO₂, and PM_2.5 and the depositions of oxidised sulfur and oxidised nitrogen averaged over the individual countries bordering the BAS. The black (Present) and green (Future) bars represent contributions from all other sources (both anthropogenic and natural) than BAS shipping. The blue part of the bars represents the (present and future) contributions from BAS shipping calculated as Base–NoShip, where Base can be either Present_Base or Future_Base and NoShip can be either Present_NoShip or Future_NoShip. The sum of the black or blue and green parts of the bars then adds up to the total concentrations and depositions averaged over the individual countries bordering the BAS for the Present_Base and Future_Base scenarios. The red part is the additional BAS contributions assuming BAS ship emissions at 2014 levels calculated as Present_HiSulphur–Present_Base. The calculations are made assuming linearity. Previous calculations, adding up contributions from different sources, have shown that this assumption is reasonable (Jonson et al., 2017, 2018). Irrespective of species and depositions, the largest contributions are seen for smaller countries with long coastlines exposed to the BAS such as Denmark and the Baltic states, and the least for large countries such as Germany and Poland with major parts of their areas located far from the shipping routes.

Following the expected reductions between 2016 and 2030 in both land-based and ship emissions, calculated concentrations and depositions are reduced over the 2016 to 2030 time span. For SO₂ and depositions of sulfur, BAS shipping is already an insignificant source in 2016 and the differences between 2030 and 2016 are almost entirely caused by changes in land-based emissions. For NO₂ concentrations and depositions of oxidised nitrogen, reductions in land-based and BAS ship emissions both contribute to the improvements in pollution levels. In the BAS region the fractional reductions in future concentrations attributed to (mainly) land-based and BAS ship emissions are roughly in the same range.

The largest contributions from BAS shipping are seen for NO₂ (Fig. 3b), depositions of oxidised nitrogen (Fig. 3c), and partially also for SO₂ (Fig. 3a) when assuming 2014 emissions (Present_HISulphur). However, for SO₂ calculated contributions are insignificant following the implementation of the stricter SECA in 2015. The same conclusion also holds for sulfur depositions (Fig. 3d). PM_2.5 contributions from BAS shipping are markedly smaller than for NO₂. Contributions are higher when assuming Present_HiSulphur emissions. After the implementation of stricter SECA regulations in 2015, PM_2.5 from shipping mainly originates from NO₂ and, in part, primary PM emissions. As shown in Fig. 1d, e, elevated PM_2.5 concentrations from BAS shipping are mainly seen in coastal zones close to shipping lanes. Much of these coastal zones is densely populated. When assessing the health effects of PM in a forthcoming companion paper (Barregård et al., 2019), population-weighted PM_2.5 concentrations are used.

Figure 4(a) SOMO35 in ppb days where black bars represent Present_Base levels. (b) Changes in annual ozone in ppb (annual average ozone is in the 30–35 ppb range in all countries). For both SOMO35 and annual ozone green bars represent changes in levels from 2016 to 2030 (Future_Base–Present_Base), red bars contributions from BAS (Present_Base–Present_NoShip), and blue bars contributions from BAS in 2030 (Future_Base–Future_NoShip).

Download

Figure 4 (left) shows calculated SOMO35¹ as an average for countries around the Baltic Sea and the effect of BAS shipping. The effects on annually averaged ozone are shown in the same figure (right). For all countries annually averaged ozone is in the 33–37 ppb range. For most countries both SOMO35 and annually averaged ozone increase only slightly as a result of BAS shipping, and relatively more so for SOMO35 than for annually averaged ozone. However, in Denmark emissions from BAS shipping result in a decrease in annually averaged ozone with present emissions.

Changes in ozone are caused by a combination of ozone production, mainly in the summer months, and ozone titration by NO, mainly in winter. In winter reductions in NO_x emissions (including reductions in emissions from ships) result in a decrease in ozone titration and subsequently higher ozone levels. This is illustrated in Fig. A5a with ozone winter levels in 2030 higher than in 2016 throughout northern and central Europe. Ozone production dominates in the summer months (Fig. A5b) and, with the exception of a region around the English Channel, the expected reductions in the emissions of ozone precursors result in lower ozone levels. For SOMO35 (Fig. A5c, d) the relative increase in winter is much smaller as ozone is largely below the 35 ppb threshold. In summer the increase in ozone from present to future caused by less titration around the English Channel is confined to a much smaller area. As a result annually averaged ozone production and titration in the BAS region partially cancel out, and for some regions and countries titration dominates the annual values. As shown in Fig. 4 (green bars), the expected emission reductions (land-based and from ships) from the years 2016 to 2030 result in overall reductions in ozone levels (both annually averaged ozone and SOMO35) for all countries except Germany and Denmark, where calculated average ozone levels are higher in 2030 (but SOMO35 is reduced). In 2030 the additional emissions from BAS shipping result in increased SOMO35 and annually averaged ozone in all countries. (blue bars in Fig. 4). These results are in good agreement with detailed model calculations with projected emission changes, demonstrating a future transition from NMVOC²-limited to NO_x-limited regimes in large parts of Europe north of the Alps (Beekmann and Vautard, 2010).

It has to be noted that in our model calculations the ship emissions are instantly diluted throughout the model grid cell where the emissions occur. Previous studies (Vinken et al., 2011; Huszar et al., 2010) have shown that this could lead to an overestimation of ozone formation. However, Vinken et al. (2011) found that the overestimation caused by instant dilution was small in polluted regions, such as the central parts of the BAS.