Shipping emission forecasts and cost-benefit analysis of China ports and key regions’ control

Environmental Pollution 236 (2018) 49e59

Contents lists available at ScienceDirect

Environmental Pollution journal homepage: www.elsevier.com/locate/envpol

Shipping emission forecasts and cost-benefit analysis of China ports and key regions’ control* Huan Liu a, b, *, Zhi-Hang Meng a, b, Yi Shang a, b, Zhao-Feng Lv a, b, Xin-Xin Jin a, b, Ming-Liang Fu a, b, Ke-Bin He a, b a b

State Key Joint Laboratory of ESPC, School of Environment, Tsinghua University, Beijing 100084, China State Environmental Protection Key Laboratory of Sources and Control of Air Pollution Complex, Beijing 100084, China

a r t i c l e i n f o

a b s t r a c t

Article history: Received 29 April 2017 Received in revised form 25 December 2017 Accepted 7 January 2018

China established Domestic Emission Control Area (DECA) for sulphur since 2015 to constrain the increasing shipping emissions. However, future DECA policy-makings are not supported due to a lack of quantitive evaluations. To investigate the effects of current and possible Chinese DECAs policies, a model is presented for the forecast of shipping emissions and evaluation of potential costs and benefits of an DECA policy package set in 2020. It includes a port-level and regional-level projection accounting for shipping trade volume growth, share of ship types, and fuel consumption. The results show that without control measures, both SO2 and particulate matter (PM) emissions are expected to increase by 15.3 e61.2% in Jing-Jin-Ji, the Yangtze River Delta, and the Pearl River Delta from 2013 to 2020. However, most emissions can be reduced annually by the establishment of a DECA that depends on the size of the control area and the fuel sulphur content limit. Costs range from 0.667 to 1.561 billion dollars (control regional shipping emissions) based on current fuel price. A social cost method shows the regional control scenarios benefit-cost ratios vary from 4.3 to 5.1 with large uncertainty. Chemical transportation model combined with health model method is used to get the monetary health benefits and then compared with the results from social cost method. This study suggests that Chinese DECAs will reduce the projected emissions at a favorable benefit-cost ratio, and furthermore proposes policy combinations that provide high cost-effective benefits as a reference for future policy-making. Crown Copyright © 2018 Published by Elsevier Ltd. All rights reserved.

Keywords: Shipping emissions Sulphur DECA China Costs Benefits

1. Introduction In recent years, the air pollution from Chinese shipping is becoming increasingly prominent. On the one hand, the Chinese share of world seaborne trade is overwhelming. Chinese ports handled 26.5% of the global throughput of containers at ports in 2014 (UNCTAD, 2015). In China, ports are distributed intensively particularly in Jing-Jin-Ji (JJJ, which includes Beijing, Tianjin and Hebei province), the Yangtze River Delta (YRD), and the Pearl River Delta (PRD). In 2016, 6 of the world's top 10 ports as well as 10 of the world's top 20 were located in the above regions (not counting the Hong Kong port as the emission control policy are different) (UNCTAD, 2016). On the other hand, these areas have the highest

*

This paper has been recommended for acceptance by Joerg Rinklebe. * Corresponding author. State Key Joint Laboratory of ESPC, School of Environment, Tsinghua University, Beijing 100084, China. E-mail address: [email protected] (H. Liu). https://doi.org/10.1016/j.envpol.2018.01.018 0269-7491/Crown Copyright © 2018 Published by Elsevier Ltd. All rights reserved.

population density (352, 541 and 502 km
2 for JJJ, YRD and PRD) and the fastest developing rate (7.6e8.6% GDP growth annually). Consequently, shipping emissions cause more severe environmental problems in these areas (Corbett et al., 2007; NRDC, 2014). Considering the expansion of international trade and the important role Chinese marine system plays (19.6% increment for China import and export value), shipping emission will continue to increase in the future. In this background, the Chinese government is paying more attention to the shipping emission issue and has introduced a series of policies. SO2 emissions are directly proportional to the sulphur content of marine fuel; thus, one of the most straightforward and effective methods of reducing them is to switch from bunker fuel to low-sulphur fuel (Wang and Corbett, 2007). IMO enacts regulations for ships through the MARPOL Convention. Annex VI of this convention was revised in 2008 and allows signatory countries to apply for the designation of an Emission Control Area (ECA) (IMO, 2008). Ships are restricted to use low-sulphur fuel or an approved

50

H. Liu et al. / Environmental Pollution 236 (2018) 49e59

equivalent method in sulphur ECA areas. The established ECA areas are as follows: the Baltic Sea area (SOx), the North Sea area (SOx), the North American area (SOx and NOx), the United States Caribbean Sea area (SOx and NOx) (Simon et al., 2013; Viana et al., 2015). China is one of the signatory countries of Annex VI. In December 2015, the Ministry of Transport of China created three sulphur DECAs: in JJJ, the YRD, and the PRD (The Ministry of Transport of China, 2015). Analysis of the costs and benefits of implementing ECAs can be found in the studies of Wang and Corbett for the US West Coast (Wang and Corbett, 2007), EPA for the North American ECA (EPA, 2009), Sieber et al. for Europe (Sieber and Kummer, 2013), Antturi et al. for the Baltic sea ECA (Antturi et al., 2016) and AEA for the Baltic Sea ECA and the North Sea ECA (AEA, 2009). Most of these studies found that the health and environmental benefits far outweighed the costs of meeting ECA requirements except that of Antturi et al. Antturi's study found that the annual cost was roughly V465 M, whereas the benefit was V105 M. However, based on their sensitivity analysis, the benefits yet have a potential to exceed the costs. The costs of DECAs in JJJ, the YRD, and the PRD would be specific to the local conditions in China. Therefore, a locally-based cost-benefit analysis must certainly be developed. Efforts have been made to develop regional-level ship emission inventories and analyze the emission reduction effect in China. Several groups (Li et al., 2016; Liu et al., 2016; Chen et al., 2017; Fu et al., 2017) used an Automatic Identification System (AIS) system to develop the ship emission inventory for China and related regions. A report conducted by two local universities in Hong Kong (Simon et al., 2013), assessed the impacts of emissions from OGVs operating in Hong Kong and the rest of the PRD region. They set and compared four different ship emission control scenarios for Hong Kong and the PRD. Establishing a 100 nautical miles (nm) DECA in the PRD has been revealed to bring a reduction of SO2 by 95% and PM by 85% and the greatest benefits. The emission reduction effects of DECA in the PRD far overweighed those of other control measures, such as mandatory fuel switching at-berth or only in Hong Kong waters, and restricting vessel speeds to 12 knots in Hong Kong waters for OGVs. The impacts of emission controls from OGVs in JJJ and the YRD are still not reported. The negative environmental impacts caused by air pollution can be quantified and monetized as environmental costs, although such calculations can be an inevitable source of uncertainties (Hazilla and Kopp, 1990; Clarkson and Deyes, 2002). According to EPA, the social cost includes changes in net agricultural productivity, human health, property damages from increased flood risk, and the value of ecosystem services (IAWG, 2010). In current studies, the concept of social cost is similar to those of external cost, economic cost, marginal cost, and so forth (Gallagher, 2005; Muller and Mendelsohn, 2007; Tichavska and Tovar, 2015). These concepts are deemed to be identical in this study. Social cost has been used in the calculation of shipping emissions at the port, regional, and national levels. Several European studies addressed the social cost estimation at the port and regional levels, such as those of Tzannatos et al. for the Piraeus port (Tzannatos, 2010), of Kalli et al. for the Gulf of Finland (Kalli and Tapaninen, 2008), and of Tichavska et al. for the Las Palmas Port (Tichavska and Tovar, 2015). There are also investigations at the national and wider regional level such as those for Greece, America, and Europe (Gallagher, 2005; Wang and Corbett, 2007; AEA, 2009; Notteboom et al., 2010; Maragkogianni and Papaefthimiou, 2015). No other previous studies can be found on measuring the social cost of ship emission in China except that on the Yangshan port (Song, 2014). There has been a debate over whether controlling port emissions or controlling regional emissions should be priority. There are

busier shipping transportation and higher concentrations of emissions in port areas, which makes controlling port emissions potentially a better deal. However, some believed that regional control measures were expected to cut more emissions than that of port control measures, even if it might be less economic. This study filled the gaps of previous studies on evaluation and cost-benefit analysis for current and possible Chinese DECA policies, both on port-level and regional-level for a long time scale. In this study, a shipping emission forecast and cost-benefit analysis model was developed to feature: (1) reliable port and regional shipping emission inventory building on the research for developing the East Asia 2013 OGV inventory by Liu et al. (2016); (2) shipping emission forecast in 2020, associated with based on the growth of port throughput, share of ship type, and fuel consumption reduction; (3) evaluation of current DECA policy by applying effectiveness and a cost-benefit analysis; (4) assessment of a DECA policy package set that are likely to be proposed in the future, involving changing the DECA size and the fuel sulphur content limit. 2. Materials and methods 2.1. Study area The Chinese DECA control measures consist of two phases: control emissions from ships berthed in ports of DECA areas from 2017 to 2018 and control emissions from ships in the whole DECA areas after 2019. To measure and analyze the emission reduction effects, we set up two inventories: the port inventory and the regional inventory, corresponding to the two control phases. The study area of the port inventory consisted berthing area of all ports in three DECA areas, including 10 core ports (shown in the map in Fig. 1) and 15 non-core ports. The core ports included Tianjin port, Qinhuangdao port, Tangshan port and Huanghua port

Fig. 1. Maps of the study areas, scenarios of DECA and locations of the core ports.

H. Liu et al. / Environmental Pollution 236 (2018) 49e59

in JJJ; Shanghai port, Ningbo-Zhoushan port, and Nantong port in the YRD (Suzhou port was dismissed as an inland harbor); and Shenzhen port, Guangzhou port, and Zhuhai port in the PRD. Table S1 gives the details of core ports and non-core ports. Fig. S1 provides an example of how to determine the berthing area of port region for a port with multiple port areas and Table S2 shows an example of longitude and latitude of port areas determining by our method. We dismissed inland ports and river vessels. On one hand, the current DECA is targeted to control the ocean-going vessels. On the other hand, our AIS data have poor coverage of inland ships, especially those small ones. The China DECA allows an exception for ships due to be at berth for less than 2 h. Time period excludes 1 h after berthing and 1 h before departure. According to our field research in Tangshan Port, ships turn off the main engines in 1 h after they were berthed. So we assume this 2 h’ exception allow ships to use high sulphur fuels when the main engines were still on. So we distinguished emissions from main engines from others, and considered all auxiliary engines and boilers changed to low sulphur fuel in port region without exception hours when DECA was implemented. Considering the port regions were selected small enough to represent the berth area only, the errors from dismissing possible high sulphur fuel emissions from auxiliary engines or boilers during maneuvering in selected area can be very small. To evaluate the effect of the expanding size of DECA, we selected study areas of the regional inventory based on several possible scenarios. Fig. 1 shows the study area and different scenarios for current and future DECAs. The study area of JJJ was the same as the current DECA regulation (115036 km2), because it is the area of Bohai sea. The study area of YRD and the PRD regions were expanded to 50 and 100 nm away from the coastline (different scenarios). We considered multiple factors and then chose 100 nm as the potential scenarios of the largest DECA. Firstly, this region can cover almost all the routes along the coast so the control effects are better than current DECA. Secondly, although 200 nm from the baseline of territorial sea (Exclusive Economic Zone) was used for some previous ECAs, the complex ocean area around China adds on the difficulty to include a 200 nm DECA. To avoid possible policy issues, we chose 100 nm to do the research and provide implications. This range does not reflect any government opinions. In YRD and PRD, when the DECA was expanded from 12 nm to 50e100 nm, the area of control increased from 57935 km2 to 107567e163237 km2 in YRD, and from 23594 km2 to 39753e67326 km2 in PRD. 2.2. Methods and uncertainty analysis for baseline emission inventory The bottom-up method described by Liu (Liu et al., 2016) was used in the model to build port and regional inventories (including SO2 and PM emissions) for JJJ, the YRD, and the PRD in 2013. Five vessel types were used in our model (bulk carriers, container ships, general cargo, oil tanker, and others), with a breakdown into four modes (cruise, maneuvering, at anchor, and at berth) and three different engine types (main engine, auxiliary engine, and boiler). Ship activity data were collected from the AIS and the Ship Technical Specification Database (STSD), combined mainly with those of the Lloyd's Register and China Classification Society. A Monte Carlo method (500 simulations) was performed to evaluate the uncertainty for these bottom-up emission inventories. The uncertainty in the average CO2 emission factor is set as ±1%. This estimate is in line with that of Skjølsvik et al. (2000). and Endresen et al. (2003). The uncertainty for other pollutants are not as clear. Cooper et al (Cooper and Gustafsson, 2004). quantified the uncertainty of PM and SO2 roughly which were over ±50% and

51

±20e50%, respectively. A recent study reports an uncertainty of 21% for SO2 (Beecken et al., 2015). The Second IMO GHG Study (IMO, 2009) estimates a ±20% uncertainty in all emissions and Endresen et al. (2003). report a ±10% uncertainty of SO2. In our research, the uncertainty of PM and SO2 are referenced from the previous study of Cooper et al. (Cooper and Gustafsson, 2004), which were ±50% and ±35% (the median), respectively. The standard deviation of the uncertainty of AIS speeds is 11% compared with the mean speed, which was set as the uncertainty for speed (IMO, 2014a,b). Ship specifications are obtained from multiple databases and improved using our adaptive algorithm. Considering the maximum continuous rated power of the propulsion engine and the maximum design speed obtained for each single ship, these parameters are assumed to be with no uncertainty. 2.3. Forecast of 2020 port and regional inventories The proxies that were used to estimate future total emissions included: trade activity, fuel usage, installed engine power, etc. (Entec, 2007). We made the following assumptions: (1) Port throughput history data and government plan for future are key data for other assumptions. (2) For ship fleets: the new ships would be added into the fleet with a better fuel economy while the average installed engine power was constant for each specific ship type; The share of ship type was forecasted based on global shipping fleet forecasting; (3) By analyze AIS data combined with STSD, it's found the average deadweight and corresponding sharing rate keep constant from 2013 to 2016 in each ship type. So the ship calls were assumed to be proportional to the port throughput; (4) The regional emissions were proportional to the sum of the port emissions in that region. The details are introduced as follows. Port throughput data in different years were obtained: the data for 2013 and 2014 of all ports were collected from China Ports Year Book (China Port Magazine, 2014, 2015); the 2015e2016 data of 11 ports from the Ministry of Transport of China; and 2020 throughput target of 19 ports from the development plans of various cities and provinces, as shown in Table S3. For the 19 ports with 2020 throughput target, we assumed that the port throughputs grew in geometric progression before 2019. For the other six ports without a 2020 throughput target, we assumed an annual growth of 4% based on the analysis of the Ministry of Transport of China (Liu, 2015). The 2013e2020 port throughput projection in JJJ, the YRD, and the PRD are shown in Fig. 2a. The throughputs of all ports were expected to grow quickly from 2013 to 2020. The sum of the port throughputs of JJJ, the YRD, and the PRD would increase by 46%, 20%, and 40%, respectively. There are only five ports in study scope of the YRD DECA, the lowest among the all studied regions. Furthermore, at the time of the research, the Shanghai port and Ningbo-Zhoushan port in the YRD were the top two busiest ports in the world in terms of cargo tonnage. With their large throughput base, the growth rate of port throughput in the YRD would be obviously slower. The throughput proportion of core ports (core ports are defined in Chinese DECA Regulation, see Method part) is shown in Fig. 2b. On the one hand, JJJ has thirteen ports (only four of them are core ports), which is the largest number of ports in the three regions. On the other hand, the throughput of JJJ core ports is not significantly greater than those of the YRD and the PRD. Consequently, the proportion of JJJ core ports is lower than those of the other two regions, only 53.3% in 2013. In contrast, the core ports in the YRD and the PRD have a high contribution of 93.5% and 75.8%, respectively, in 2013. Because all core ports had a large throughput base, their growth was steadily high but slower than that of non-core ports. In 2020, the proportion of core ports throughput will fall to

52

H. Liu et al. / Environmental Pollution 236 (2018) 49e59

Fig. 2. Port throughput in JJJ, the YRD and the PRD: (a) Port throughput projections from 2013 to 2020; (b) the proportion of core ports in 2013 and 2020.

52.2%, 92.2%, and 70.9%, respectively. The newly added ships each year have a direct influence on fuel consumption of fleets. The data on the age distribution of the world vessel fleet in 2013 were obtained from a report by UNCTAD (Fig. S2) (UNCTAD, 2013). Ship age distribution was steady from 2013 to 2016 (UNCTAD, 2013, 2014; 2015, 2016); thus, we assumed the 0e4 year age proportion was shared equally among the four years. Besides, we also speculated that the annual quantity of new ships accounted for one-quarter of the proportion of 0e4 years and assumed the proportion of new ships would remain constant from 2013 to 2020. Proportion of newly added ships each year in this study is shown in Table S4$In 2011, IMO adopted amendments to Annex VI Regulations that required an increase of 10% in the efficiency of new ships from beginning 2015 and of 20% by 2020 (IMO, 2011). Taking the fuel consumption of the new ships after 2015 into account, we developed a fuel consumption projection of the world fleet (Fig. S3). Based on historical data (2013e2016), the average deadweight of each ship category didn't show any increase trend. So we assume deadweight of the fleet should be equal with average deadweight of 2016. Then only throughput growth and fuel consumption change could influence the emissions. Based on 2013 port inventory, we assumed that port emissions were proportional to port throughput and fuel consumption together to obtain the port inventory in 2020. Not only throughput growth and fuel consumption changes, but also the shares of ship types were considered in calculating the regional inventory. The steps of the calculation process of the regional inventory were as follows: (1) The port throughput in the three regions in 2013 and 2020 was summed up to be the regional throughput; (2) The share of the throughput by ship type in 2013 was derived from the UNCTAD report (UNCTAD, 2013); (3) The share of the throughput by ship type in 2020 was obtained from the EPA report (shown in Table S5) (EPA, 2009). In both 2013 and 2020, the sum of the proportion of bulk carriers, container ships, and general cargo and oil tankers accounted for more than 88%. There are two major uncertainties here. Both the shares of throughput in 2013 and 2020 are not based on local data, which is not available for

China now. In addition, the EPA report provides a forecast for 2020, but the projection is not updated since 2009. (4) The throughput growth rate of multiple ship types from 2013 to 2020 was calculated; (5) The emission growth rate of ship types based on fuel consumption projections in 2020 was calculated; (6) SO2 and PM emissions in 2013 were evaluated by ship types in the three regions using our model (see in Table S6); (7) The 2020 regional inventory was performed and evaluated. A linear regression was used to evaluate uncertainty for emission forecasts of 2020 based on the uncertainty of emissions in 2013. A qualitative discussion of notable and likely uncertainties for forecasting shipping emissions is presented in Table S7. The most significant inaccuracies could have been caused by forecasting the regional throughput of ports without taking into account the 2020 throughput target. 2.4. 2020 scenario design Three scenarios for port inventory and seven scenarios (three for JJJ) for regional inventory are designed in this work. The policy options considered in these scenarios are measures that will be proposed or are most likely to be proposed in the future, including two aspects: expanding DECA size (only for regional inventory) and using fuel with lower sulphur content. The sulphur content in the Business as Usual (BAU) scenario in both port and regional inventories were set as 2.43% for heavy fuel oil (HFO) according to an IMO survey (IMO, 2014a,b). BAU scenario assumes that no DECA policy in the future. So the BAU scenario aims to provide a benchmark to evaluate the control effects but not a real emission condition. Detailed scenario design for port and regional inventory is illustrated in Table 1. BAU, P1, and P2 scenarios were included in the port inventory. According to Chinese DECA policy vessels berthing in core ports in DECA must use low-sulphur fuel (0.5%) after 2017 and vessels berthing in all ports in DECA must use low-sulphur fuel (0.5%) after 2018. To assess the effect of Chinese DECA policy, we set P1 and P2 scenarios that assume that vessels use low-sulphur fuel (0.5%) berthing in core ports and all ports in 2020, respectively.

H. Liu et al. / Environmental Pollution 236 (2018) 49e59

53

Table 1 Scenario design for port and regional inventory (2020). Scenario design for port inventory Scenario Description BAU P1

P2

Controlled ports

No control No control All vessels change over to low sulphur fuel berthing in core ports in Core ports in JJJ: Tianjin port, Qinhuangdao port, Tangshan port, a DECA Huanghua port Core ports in YRD: Shanghai port, Ningbo-Zhoushan port and Nantong port Core ports in PRD: Shenzhen port, Guangzhou port and Zhuhai port All vessels change over to low sulphur fuel berthing in all ports in All ports in three areas DECAa

Sulphur, g/g fuel 2.43% 0.5%

0.5%

Scenario design for regional inventory in the YRD and the PRD Scenario Description

DECA area (distance to shore)

Sulphur, g/g fuel

BAU R1 R2 R3 R4 R5 R6

None 12 nm 50 nm 100 nm 12 nm 50 nm 100 nm

2.43% 0.5% 0.5% 0.5% 0.1% 0.1% 0.1%

No control All vessels change over to low sulphur fuel prior to entering DECA

Scenario design for regional inventory in JJJb Scenario Description BAU No control

Sulphur content, g/g fuel 2.43%

R1 R2

0.5% 0.1%

All vessels change over to low sulphur fuel prior to entering DECA

a

Excluding the first hour at berth and the last hour before leaving berth. Bohai sea is a semi-enclosed sea (also as territorial sea), all surrounding area are China territory. According to previous ECAs, e.g. the Baltic Sea, the whole sea area is defined as domestic ECA. So, the scenario design for JJJ includes sulphur level difference only, without area difference. b

For regional scenarios, since the size of JJJ DECA was fixed, we developed three scenarios for JJJ regional inventory: BAU, R1, and R2. BAU scenario is used as a benchmark to reflect emissions without any control in the future. Vessels have to use low-sulphur fuel (0.5%) in DECA area after 2019 according to Chinese DECA policy. R1 (JJJ) that has a sulphur limit of fuel of 0.5% is set to assess the effect of Chinese DECA policy after 2019. As IMO limits vessel fuel sulphur content to 0.5% globally after 2020, we set R2 (JJJ) which has a sulphur limits of fuel of 0.1% to assess the effect of possible DECA policies in China after 2020. Seven scenarios were set in the YRD and the PRD regional inventory each: BAU, R1, R2, R3, R4, R5, and R6. We applied a respective sulphur limit of fuel of 0.5% and 0.1%, and DECA sizes of 12 nm, 50 nm, and 100 nm distance to the shore. R1 is in line with the DECA policy in China after 2019 and R2-R6 represent the possible DECA policies in China after 2020. The detailed description for scenarios is in Table 1. IMO has made great efforts on pushing forward the control of global ships. A restriction on fuel sulphur content below 0.5% would be in effect by Jan 2020 globally. Thus, our DECA scenarios P2, R1 for JJJ region, R1, R2 and R3 for YRD and PRD regions are in accordance with the IMO's sulphur limit in 2020. The results from these scenarios can reflect the results in the light of IMO's major recent policy. Emissions from ships using low-sulphur fuel are modified using emission factors in Third IMO Greenhouse Gas study 2014 in Eq. (1):


EFLS ELS ¼ ERO $ EFRO

(1)

where ELS (t) represents emissions using low-sulphur fuel; ERO (t) represents emissions using residual fuel; EFLS and EFRO (g/kWh) are emission factors for residual fuel and low-sulphur fuel, which were derived from Third IMO Greenhouse Gas study 2014 report shown in Table S8 (IMO, 2014b).

2.5. DECA control benefits: social cost method versus air quality and health modeling Establishing DECAs provides a range of benefits, including mainly health and environmental benefits. There are two methods could be used to evaluate the benefits from switching oil. A simplified method is to use social costs, which do not consider the atmospheric process of emissions. A combination of chemical transportation model (CTM) with health model can be a better choice to evaluate the health impacts. In this study, both the social cost and CTM methods are used to determine the benefits of DECA establishment. A range of research were surveyed on estimating social costs of air emissions, mainly aiming at American and European cases but few at Chinese local case (Table S9). The results of the estimated social cost factors were significantly different due to assumptions, cost categories, and methodological variation (Tichavska and Tovar, 2015). Meanwhile, port layout, port function, and population intensity varied in separate regions, resulting in greater distinctions. In this study, the social cost factors which were used to estimate the social costs of shipping emissions in Yangshan port (Song, 2014) are taken for the following three reasons. First, Yangshan is a representative port area of Shanghai port. Therefore, its social cost factors can be close to those of China's reality. Second, the costs of social factors used were slightly higher than those in Kaohsiung's case (Berechman and Tseng, 2012). Considering the higher concentration of ship movements, maritime activities, and higher population density in Yangshan port, their social cost factors were reasonable. Finally, both SO2 and PM social cost factors were centrally located. In this study, the costs established of emission social factors are 12,329 $/ton for SO2 and 76,867 $/ton for PM. Previous studies reported very large variation of social cost factors: 37941,087 $/ton for SO2 and 200e375,888 $/ton for PM, there will be large uncertainty in results of social cost method. In this study, we

54

H. Liu et al. / Environmental Pollution 236 (2018) 49e59

compared the social cost results with the health benefit results using CTM, health models for one scenario. To verify the accuracy of social cost method, CTM method was applied for comparison–in a case study in JJJ. We used the air quality model to estimate the PM2.5 concentration reductions due to the differences between ship emissions in scenario R1 and BAU, then calculated the human health benefit from it. The WRF-CMAQ one-way coupled model was applied to simulate the PM2.5 during August of 2013. In order to reduce the uncertainty of boundary conditions, the simulation of this study was performed over two nested domains. The outer domain covered all of China and part of East Asia with a grid resolution of 36 km  36 km, and the inner domain covered the most part of eastern China such as JJJ, YRD, PRD, at a 12 km  12 km grid resolution. The system consists of two main components: WRF version 3.8.1 developed by US NCAR (National Center for Atmospheric Research) and CMAQ version 5.1 developed by US EPA (Environmental Protection Agency). The detailed configurations in WRF and CMAQ were described in supporting documents. The estimate of model performance is provided in Fig. S4 and Table S10. The method used in this study to estimate the relative risk (RR) for the burden of four kinds of diseases (ischemic heart disease, IHD; cerebrovascular disease, stroke; chronic obstructive pulmonary disease, COPD; lung cancer, LC) related to PM2.5 exposure caused by ship emissions, was introduced in Burnett et al. (2014), shown in Eq. (2):

For Ci < C0 ; RRd;i ¼ 1   d For Ci > C0 ; RRd;i ¼ 1 þ a$ 1
e
g$ðCi
C0 Þ

(2)

where RRd,i was relative risk for the d disease at i grid, Ci was modelling monthly PM2.5 concentrations at i grid, C0 was the counterfactual concentration of d disease below which we assumed there is no additional risk, a, g and d were parameters used to describe the concentration-response of d disease (Table S11). To estimate morbidity caused by PM2.5, we selected cardiovascular and respiratory hospital admissions as health endpoints. The RR for morbidity was calculated as

RRd;i ¼ eb$ðCi
C0 Þ

(3)

where b was the factor describing concentration-response (Table S11). Finally, we calculate the mortalities attributable to PM2.5 pollution in JJJ area in two scenarios respectively based on the method described in Anenberg et al. (2010), as defined in Eq. (4):

DE ¼

XXRRd;i
1 $y0 $Popi RRd;i i

(4)

d

where △E was the mortalities caused by PM2.5 pollution in JJJ, y0 was the baseline mortality or incidence rate (Table S11), Popi was the exposed population at gird i. The 1 km  1 km gridded population used in this study was provided by Data Center for Resources and Environmental Sciences Chinese Academy of Sciences (RESDC) (http://www.resdc.cn). To evaluate the monetary health benefits from reductions of shipping emissions, the value of a statistical life (VSL) was determined based on the prediction for 2020 in China by Wang et al. (Wang and Mauzerall, 2006). The unit value for cardiovascular and respiratory hospital admission in Beijing was introduced in a local research (Guo et al., 2010).

2.6. DECA control costs Conversely, the fuel switch represents various and significant control costs, including fuel price premium, capital cost of ships to retrofit for fuel switch, and operating cost (Wang and Corbett, 2007; EPA, 2009; Carr and Corbett, 2015). We excluded the capital and operating costs because they are relatively lower than fuel price premiums (Wang and Corbett, 2007; EPA, 2009) and only estimate price premium for fuel switch. Fuel consumption of each ship was calculated by SFOC (specific fuel oil consumption) method (IMO, 2009; IMO, 2014a,b) using Eq. (5):

R1 ¼

X

Pactual  SFOCload Dt

(5)

R1(t) represents the fuel consumption calculated by SFOC method, Pactual (kW) and SFOCload (g/kWh) represent the actual power of ship and adjusted SFOC based on ship's engine load, Dt(h) is the duration for engine. The actual power, load and the duration for engine of ships are calculated using AIS speed and time, which are the same procedure with the emission calculation. Table S12 provides equations and parameters for SFOC calculation. Fuel consumptions in 2013 are calculated by SFOC method. The fuel consumptions in 2020 are obtained by the same method of forecast port and regional inventories. The regional fuel consumptions in 2020 are shown in Table S13. As there was no LSMGO (0.1% sulphur content) on sale in East Asia, we use the average price of IFO380 (2.43% sulphur content), MGO (0.5% sulphur content) and LSMGO (0.1% sulphur content) from Singapore port during September 8th to September 22, 2017, which are $338.5, $513.6, $519.8 per ton, respectively (derived from https://shipandbunker.com/prices/apac/sea/sg-sinsingapore#IFO380 on Oct. 1st, 2017). Fuel prices are presented in Table S14. 3. Results and discussion 3.1. BAU scenarios In this study, we set up two types of inventories: the port inventories and the regional inventories and designed a set of control scenarios (see Materials and Methods part). Port emissions (which only consist the berthing area) under the BAU scenario are presented in Fig. 3. The resulting total port SO2 emissions in 2013 in JJJ, the YRD, and the PRD are 23,900 t, 37,285t, and 15,260 t, and the total port PM emissions are 2412 t, 3,665t, and 1500 t, respectively. Although the number of ports in the YRD is the lowest, the emissions of the YRD ports exceed those in JJJ and the PRD due to the large throughput of core ports in the YRD. Frequent and intensive shipping activities in the core ports have caused more shipping emission than that in non-core ports. The emissions from the core ports contribute 96% and 70% in the YRD and the PRD, and only 30% in JJJ, values that are close to those of the shares of port throughput. Port emissions are projected to increase rapidly from 2013 to 2020 as depicted in Fig. 3, rising by 61.2%, 15.3%, and 33.9% in JJJ, the YRD, and the PRD. Port emissions will rise as port throughputs continue to grow, except in Qinhuangdao port. When the unchanged target throughput for the period from 2013 to 2020 is treated with fuel consumption reduction, emission in Qinhuangdao port will drop off. More than two-fold augmentation in the emissions in Dongying port, Panjin port, Weifang port, and Weihai port is forecast due to a huge leap forward of throughput. Especially Dongying port is expected to reach an increase of nearly six times. There is a creeping increase in the YRD port emissions owing to the slower port throughput growth in the Shanghai and Ningbo-

H. Liu et al. / Environmental Pollution 236 (2018) 49e59

55

Fig. 3. Port emissions in 2013 and 2020 (berth area only), and their growth rates from 2013 to 2020 without control (Port names in red indicate core ports). (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

Zhoushan ports. We also evaluated emissions, growth rate and under control scenarios (will be discussed in next section) in both the port and the regional inventories, the results of which are shown in Fig. 4 and Fig. S5 for details. In 2013, the shipping emissions in JJJ, the YRD, and the PRD were estimated to be 115,554 ± 7456 t, 318,986 ± 16,870 t, and 126,818 ± 9816 t for SO2 and 13,985 ± 1278 t, 39,732 ± 2867 t, and 15,887 ± 1699 t for PM. In 2020, the regional emissions will grow to 168,349 ± 10,864 t, 392,873 ± 20,771 t, and 191,533 ± 14,793 t for SO2 and 20,477 ± 1862 t, 49,000 ± 3530 t, and 24,121 ± 2560 t for PM. The regional emission in the YRD is higher than those in JJJ and the PRD due to its higher port throughput. The changing share of ship types leads to a slight difference between the growth rates of SO2 and PM. SO2 emissions are expected to increase by 46%, 23%, and 51% in JJJ, the YRD, and the PRD, respectively, while PM emissions will grow by 46%, 23%, and 52%. However, the growth of emissions is faster than that of the throughput except for JJJ area. The following reasons can account for this phenomenon. On the one hand, there is a noticeable increase of the container share of the throughput from 2013 to 2020 (from 12.7% to 17.4%). On the other hand, the

container share of emissions is larger than that of the throughputs in the YRD and the PRD, whereas the container share of the throughput in JJJ is relatively small.

3.2. Scenarios with control measures This section evaluates control scenarios for port and regional inventories in 2020, shown in Fig. 4. The efficiencies of port emission control measures can be observed by comparing the BAU scenario with P1 and P2 scenarios (Fig. 4a). Both P1 scenario and P2 scenario are expected to cut SO2 and PM significantly; the reduction ration in SO2 will be higher than that in PM. The reduction ratios of the P2 scenario are identical in the three areas, but the reduction ratios of the P1 scenario vary with the proportion of core ports throughputs. Under the P1 scenario, the reduction ratios are merely approximately 20% of both SO2 and PM in JJJ, much less than that in the YRD, which is more than 70%. In YRD, most of the ports are key ports, so there is little difference between P1 and P2 scenarios. For JJJ and PRD, only half of the ports are recognized as key ports, so the emission reduction from P1 to P2 is relatively large. The reduction ratios of the P2 scenario are higher than those of the P1 scenario,

56

H. Liu et al. / Environmental Pollution 236 (2018) 49e59

Fig. 4. Port and regional emissions under different scenarios in 2013 and 2020: (a) port inventory; (b) regional inventory. The blue columns indicate emissions in 2020 and the green columns indicate emissions in 2013. The darkness of blue columns indicates emissions percentage reduction of different scenarios. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

demonstrating that the fuel switch in all ports is a comparably better practice than the fuel switch only in core ports. All scenarios in the regional inventory indicate considerable emission reduction effects, as depicted in Fig. 4b. It should be noted that the trends for control become better with the expansion of DECA and the reduction in sulphur content limit. Meanwhile, SO2 also has a higher reduction ratio than PM. In JJJ, the scenarios R1 (0.5% sulphur) and R2 (0.1% sulphur) are expected to cut 81.4% and 96.3% of SO2, as well as 74.8% and 86.7% in PM. In the YRD, the efficiencies of the scenarios in controlling SO2 can be arranged in the following ascending order: R1, R2, R4, R3, R5, and R6. The scenario R6 (100 nm, 0.1% sulphur), scenario R5 (50 nm, 0.1% sulphur), and scenario R3 (100 nm, 0.5% sulphur) exert a relatively better effect of SO2 reduction ratio (more than 80%) and PM reduction ratio more than 70%. In scenario R1 (12 nm, 0.5% sulphur) the restrictive measures of 2019 are maintained, leading to its lowest comparative efficiency. Scenario R4 (12 nm, 0.1% sulphur)

has a better control effect on SO2 emission than scenario R2 (50 nm, 0.5% sulphur), but a lower control effect on PM. The efficiencies of the scenarios in controlling SO2 and PM in the PRD can be arranged in the following ascending order: R1, R2, R3, R4, R5, and R6. By comparing scenarios R1, R2, and R3 (0.5% sulphur) and scenarios R4, R5, and R6 (0.1% sulphur), it is suggested that a limit of fuel sulphur of 0.1% can cut more emissions than enhancing DECA size. In the YRD and PRD, only when a 0.1% fuel sulphur limit is coupled with a 100 nm DECA, can a deepest cut be achieved (96.3% for SO2 and 86.7% for PM), as the comparison of scenario R3, R5, and R6 demonstrates. By comparing the total amounts of emission reduction between port and region control scenarios, the regional control measures were far more effective in reducing emissions than port control. Therefore, the transition from port control to regional control is highly necessary in the long-term. Furthermore, IMO will restrict vessel fuel sulphur content to 0.5% globally after 2020, which is the

H. Liu et al. / Environmental Pollution 236 (2018) 49e59

same sulphur limits with our scenarios P2, R1 for JJJ region, R1, R2 and R3 for YRD and PRD regions. The results indicate that the IMO policy will push a big step forward on emission reductions. For YRD cases, expanding the control area with the same cap of sulphur contents (0.5%) can be very effective, e.g., a 50% reduction on SO2 emissions can be achieved from 12 nm scenario R1 to 100 nm scenario R3. Thus the globally IMO cap will be very effective for YRD region. However, for PRD cases, the effects from enlarge the control zone is not as effective as reducing the sulphur contents. Thus, local efforts beyond the IMO regulation, e.g. with a 0.1% sulphur content limit, would be important to achieve better goals. 3.3. Cost-benefit analysis The previous section demonstrates the mechanism of shipping emission reduction by a DECA. This section provides a vital link in the evidence chain, making a cost-benefit analysis of DECA establishment to evaluate monetary impacts. The costs and benefits of all control scenarios are calculated by social cost method. The costs and benefits of the port control scenarios have been assessed for the year 2020 (Fig. 5). The costs, benefits, and benefit-cost ratios of port scenarios are summarized so that the monetary outcomes can be clearly understood. In the port control scenarios, the total costs can be arranged in ascending order: JJJ scenario P1 (0.05 billion $), PRD scenario P1 (0.06 billion $), PRD scenario P2 (0.09 billion $), JJJ scenario P2 (0.16 billion $), YRD scenario P1 (0.16 billion $), YRD scenario P2 (0.17 billion $), with total benefits of 0.17, 0.21, 0.32, 0.61, 0.65, 0.68 billion dollars and net benefits of 0.13, 0.16, 0.24, 0.45, 0.49, 0.50 billion dollars, respectively. Net benefits increase 262.4%, 4.5%, 51.7% from P1 to P2 in JJJ, YRD and PRD, which demonstrates that more economical returns can be ensured with more ports taken under emission control. The net benefit increment is related to the ratio of core ports emission to non-core ports emission. The high increment of net benefit from P1 to P2 indicts more attention should be paid to non-core ports in JJJ. In our study, the monetary efficiency of the control measures was assessed through benefit-cost ratios, depicted on the right axis of Fig. 5. B/C ratios of ports control scenarios range between 3.7 and 3.9, which means both P1 and P2 control scenarios in the three areas lead to a substantial emission reduction comparing to the cost. The costs and benefits of the control scenarios in the regional inventory are also evaluated for the year 2020 (Fig. 6). The costs of the regional control scenarios range from 0.667 to 1.561 billion $. The costs of the YRD regional control scenarios are higher than that of JJJ and PRD because of high throughput of YRD. Based on social

Fig. 5. Effects and efficiencies of the control scenarios (2020 port inventory).

57

cost method, the benefits of regional control scenarios can be arranged in ascending order: JJJ R1, PRD R1 to R3, JJJ R2, PRD R4 to R6, YRD R1, YRD R4, YRD R2, YRD R3, YRD R5, YRD R6. The highest net benefits are generated by the YRD scenario R6 (6.37 billion $) and the lowest by the JJJ scenario R1 (2.20 billion $). The benefits of the YRD scenarios far exceed those of PRD and JJJ, owing to its significantly larger throughputs, whereas there is little difference in the benefits between JJJ and PRD. With sulphur contents decrease from 0.5% to 0.1%, net benefits increase 21.5%, 21.3%, 21.2% in JJJ, YRD, PRD region. With 0.5% sulphur contents, when DECA area expands from 12 nm to 50 nm and 100 nm, net benefits increase 18.1%, 28.5% and 5.3%, 9.8% in YRD and PRD region. The same increments are obtained with 0.1% sulphur contents. The higher net benefit increment with DECA area expanding in YRD means more emission may occur away from the shore in YRD than in other regions. Fuels with sulphur contents of 0.5% and 0.1% generate B/C ratios of 4.3e4.5 and 4.9 to 5.1, respectively. These results are comparable to AEA (AEA, 2009) and Wang's (Wang and Corbett, 2007) findings (Table S15). The comparatively higher ratios of the 0.1% sulphur fuel can be explained by the little costs increment based on current oil price and great improvement of the emission reduction obtained by the decrease of sulphur in the fuel from 0.5% to 0.1%. The expansion of DECA size contributed to the increase in the benefits and costs, but the B/C ratio remained constant. Furthermore, switching from 0.5% sulphur fuel to 0.1% sulphur fuel generated more benefits and higher B/C ratios, which means it is more effective. The benefit of scenario R1 in JJJ is calculated by CTM method and social cost method for comparison. Fig. 7. (a) and (b) showed the PM2.5 concentration reduction and monetary benefits of human health between scenario R1 and BAU of JJJ based on CTM method. Land areas near the coastline in this region have an obvious PM2.5 reductions about 0.5e4 mg/m3, especially in the coastal areas of Tianjin, Hebei and Liaoning province where the human health benefit reached 100e500 thousand dollars at each grid, indicating shipping emissions had a significant influence on air pollution. The air quality improvement would prevent more than 600 premature deaths and 500 hospital admission. The total Human health benefit was about 0.11 billion $ when implementing R1 control policy in JJJ region. The cost for JJJ R1 scenario is about 0.67 billion $. If more benefits from agricultural productivity, property damages from

Fig. 6. Effects and efficiencies of the control scenarios (2020 regional inventory).

58

H. Liu et al. / Environmental Pollution 236 (2018) 49e59

Fig. 7. Differences between scenario R1 and BAU of JJJ. (a) PM2.5 concentration reduction (b) Benefits of human health.

increased flood risk, and the value of ecosystem services could be considered, there are large chance to get the benefit overtake the costs. Compared with the social cost method, which shows a benefit of JJJ R1 scenario is 2.87 billion $, the differences between these two methods are very large. Such difference was mainly due to the concentration-response method only considering the impacts on human health, without estimating the ecological and environmental benefits from shipping emissions control. Furthermore, as there is few research on estimating social costs based on Chinese local case and the social cost factors differs due to different conditions, the social cost factors that we used from existing researches will lead to a high uncertainty in benefit results based on social cost method. On the other hand, both the CTM and health models have their own uncertainty. The strong non-linear responses in CTM model show lower effects than reality from shipping emission control, because the emission reduction from other sources are not included.

appropriate control measures. China established DECAs since 2015, but the limited regional inventory and the lack of cost-benefit analysis makes difficult the precise quantification of DECA impacts by Chinese policy-makers. In this study, a shipping emission forecast and a cost-benefit analysis model were developed to prognosticate 2020 shipping emissions and assess DECA efficiency. Under the BAU scenario, without further control measures, the shipping emissions will continue to grow from 2013 to 2020. The port shipping emissions will grow 61.2%, 15.3%, 33.9% and the regional shipping emissions will grow 46%, 23%, and 51% in JJJ, YRD and PRD, respectively. To assess the efficiency of DECA, we designed an appropriate policy package set, including a fuel sulphur limit and DECA size. The most aggressive control scenario (100 nm and 0.1% sulphur) can reduce SO2 and PM by 96.3% and 86.7%. The maximum emission reduction in 2020 for the three regions could reach 725 kt for SO2 and 81 kt for PM. If only port control is continued, even with all the ports under control, the total emission reductions are only 83 kt for SO2 and 8 kt for PM. This study also compared the cost and emission reduction benefits for multiple port scenarios and regional scenarios. Switching to 0.1% sulphur content fuels reduces more pollutants than switching to 0.5% sulphur content fuels under both regional and port control scenarios which can lead to higher net benefits based on social cost method. Switching to 0.5% sulphur content fuels leads to costs ranging from 0.045 to 0.172 billion $ under port control scenarios and 0.667 to 1.509 billion $ under regional control scenarios, respectively. Under regional control scenarios, switching to 0.1% sulphur content fuels leads to costs ranging from 0.690 to 1.561 billion $. Under both port and regional control scenarios, YRD has higher costs because of its higher throughput. Furthermore, results based on social cost method shows regional control scenarios provides higher B/C ratios (4.3e4.5) than port control scenarios (3.7e3.9) under the 0.5% sulphur content fuels condition. Higher B/C ratios (4.9e5.1) are obtained by switching to 0.1% sulphur content fuels than switching to 0.5% sulphur content fuels (4.3e4.5). The B/C ratios results indict regional control and using 0.1% sulphur content fuels will control emission more efficiently. However, as there is few research on local social cost factors in China, it's difficult to quantify the uncertainty of social cost factors that we used which leads to high uncertainty in the results based on social cost method. A following study is necessary to provide a thorough evaluation of health and ecological benefits with atmospheric and marine dispersion models, dose-response and exposure-response functions. Currently, we only evaluated the health benefits for R1 scenario in JJJ region using WRF-CMAQ models combined with IER functions. A substantial impact on human health from controlling shipping emissions was found. Switching to 0.5% sulphur fuel in core ports in JJJ region could avoid more than 600 premature deaths and 500 hospital admission annually. When converted to monetary benefits, the health benefits of implementing DECA would be 0.11 billion $, with cost of 0.67 billion $. The big difference between results of these two methods shows large uncertainty in both methods. Further researches on generating local social cost factors and evaluate a comprehensive benefit in China are necessary. Overall, this work provides evidence for policymakers in China to facilitate the evaluation of the extent to which policies should be strengthened by 2020. In further studies, air quality and health analysis could be integrated into these cost-benefit estimations to present more comprehensive evaluations.

4. Conclusion Acknowledgement Driven by the significant growth of the shipping trade volume, especially in JJJ, the YRD, and the PRD areas, the shipping emissions in China will continue to rise without the implementation of

This work was supported by the Training Program of the Major Research Plan of the National Natural Science Foundation of China

H. Liu et al. / Environmental Pollution 236 (2018) 49e59

(91544110), the National Key R&D Program (2016YFC0201504), the Beijing NOVA Program (2018051), the special fund of State Key Joint Laboratory of Environment Simulation and Pollution Control (16Y02ESPCT), the National Natural Science Foundation of China (41571447), and the National Program on Key Basic Research Project (2014CB441301). Appendix A. Supplementary data Supplementary data related to this article can be found at https://doi.org/10.1016/j.envpol.2018.01.018. References AEA, 2009. Cost Benefit Analysis to Support the Impact Assessment Accompanying the Revision of Directive 1999/32/EC on the Sulphur Content of Certain Liquid Fuels. https://www.intertanko.com/upload/CBA%20sulphur%20in%20fuels% 20Report%20%20Task%20B2%2025%2006%2009.pdf. Anenberg, S.C., Horowitz, L.W., Tong, D.Q., West, J.J., 2010. An estimate of the global burden of anthropogenic ozone and fine particulate matter on premature human mortality using atmospheric modeling. Environ. Health Perspect. 118, 1189e1195. Antturi, J., H€ anninen, O., Jalkanen, J., Johansson, L., Prank, M., Sofiev, M., Ollikainen, M., 2016. Costs and benefits of low-sulphur fuel standard for Baltic Sea shipping. J. Environ. Manag. 184, 431e440. Beecken, J., Mellqvist, J., Salo, K., Ekholm, J., Jalkanen, J.P., Johansson, L., Litvinenko, V., Volodin, K., Frank-Kamenetsky, D.A., 2015. Emission factors of SO2, NOx and particles from ships in Neva Bay from ground-based and helicopter-borne measurements and AIS-based modeling. Atmos. Chem. Phys. 15, 5229e5241. Berechman, J., Tseng, P., 2012. Estimating the environmental costs of port related emissions: the case of Kaohsiung. Transp. Res. Part D: Transp. Environ. 17, 35e38. Burnett, R.T., Pope, C.A., Ezzati, M., Olives, C., Lim, S.S., Mehta, S., Shin, H.H., Singh, G., Hubbell, B., Brauer, M., Anderson, H.R., Smith, K.R., Balmes, J.R., Bruce, N.G., Kan, H.D., Laden, F., Pruss-Ustun, A., Michelle, C.T., Gapstur, S.M., Diver, W.R., Cohen, A., 2014. An integrated risk function for estimating the global burden of disease attributable to ambient fine particulate matter exposure. Environ. Health Perspect. 122, 397e403. Carr, E.W., Corbett, J.J., 2015. Ship compliance in emission control areas: technology costs and policy instruments. Environ. Sci. & Tech. 49, 9584e9591. Chen, D.S., Wang, X.T., Li, Y., Lang, J.L., Zhou, Y., Guo, X.R., Zhao, Y.H., 2017. Highspatiotemporal-resolution ship emission inventory of China based on AIS data in 2014. Sci. Total Environ. 609, 776e787. China Port Magazine, 2014. China Ports Year Book 2014. China Port Magazine, 2015. China Ports Year Book 2015. Clarkson, R., Deyes, K., 2002. Estimating the Social Cost of Carbon Emissions. HM Treasury, London. http://www.civil.uwaterloo.ca/maknight/courses/CIVE24005/week3/carbon%20social%20cost.pdf. Cooper, D., Gustafsson, T., 2004. Methodology for Calculating Emissions from Ships: 1. Update of Emission Factors. SMHI Swedish Meteorological and Hydrological Institute. Corbett, J.J., Winebrake, J.J., Green, E.H., Kasibhatla, P., Eyring, V., Lauer, A., 2007. Mortality from ship emissions: a global assessment. Environ. Sci. & Tech. 41, 8512. Endresen, O., Sorgard, E., Sundet, J.K., Dalsoren, S.B., Isaksen, I.S.A., Berglen, T.F., Gravir, G., 2003. Emission from international sea transportation and environmental impact. J. Geophys. Res.-Atmos. 108. Entec, 2007. UK Ship Emissions Inventory Final Report. https://uk-air.defra.gov.uk/ assets/documents/reports/cat07/0810291231_7_Entec_Ships_Presentation_ NAEI_workshop_AndrianaStavrakaki.pdf. EPA, 2009. Proposal to Designate an Emission Control Area for Nitrogen Oxides, Sulfur Oxides and Particulate Matter. Environmental Protection Agency. https:// www.epa.gov/sites/production/files/2016-09/documents/420r09007.pdf. Fu, M., Liu, H., Jin, X., He, K., 2017. National- to port-level inventories of shipping emissions in China. Environ. Res. Lett. 12. http://iopscience.iop.org/article/10. 1088/1748-9326/aa897a. Gallagher, K.P., 2005. International trade and air pollution: estimating the economic costs of air emissions from waterborne commerce vessels in the United States. J. Environ. Manag. 77, 99e103. Guo, X.R., Cheng, S.Y., Chen, D.S., Zhou, Y., Wang, H.Y., 2010. Estimation of economic costs of particulate air pollution from road transport in China. Atmos. Environ. 44, 3369e3377. Hazilla, M., Kopp, R.J., 1990. Social cost of environmental quality regulations: a general equilibrium analysis. J. Polit. Econ. 98, 853e873. IAWG, U.S., 2010. Technical Support Document: Social Cost of Carbon for Regulatory Impact Analysis under Executive Order 12866. Interagency Working Group on Social Cost of Carbon, United States Government, Washington, DC. https:// www.epa.gov/sites/production/files/2016-12/documents/scc_tsd_2010.pdf.

59

IMO, 2008. Prevention of Air Pollution from Ships. International Maritime Organization. http://www.imo.org/en/OurWork/environment/pollutionprevention/ airpollution/pages/air-pollution.aspx. IMO, 2009. Second IMO GHG Study 2009. International Maritime Organization. http://www.imo.org/en/OurWork/Environment/PollutionPrevention/ AirPollution/Documents/SecondIMOGHGStudy2009.pdf. IMO, 2011. Resolution MEPC.203(62). Amendments to the Annex of the Protocol of 1997 to Amend the International Convention for the Prevention of Pollution from Ships, 1973, as Modified by the Protocol of 1978 Relating Thereto (Inclusion of Regulations on Energy Efficiency for Ships in MARPOL Annex VI). International Maritime Organization. https://iea.uoregon.edu/treaty-text/2012amendment-1997-protocoladdingannexvi-1973-pollutionfromshipsentxt. IMO, M., 2014a. Report of the Marine Environment Protection Committee on its Sixty-seven Session. International Maritime Organization. IMO, 2014b. Third IMO Greenhouse Gas Study 2014. International Maritime Organization. http://www.imo.org/en/OurWork/Environment/PollutionPrevention/ AirPollution/Documents/Third%20Greenhouse%20Gas%20Study/GHG3% 20Executive%20Summary%20and%20Report.pdf. Kalli, J., Tapaninen, U., 2008. Externalities of Shipping in the Gulf of Finland until 2015. Publications from the Centre of Maritime Studies, University of Turku. A47, Finland. Li, C., Yuan, Z., Ou, J., Fan, X., Ye, S., Xiao, T., Shi, Y., Huang, Z., Ng, S.K., Zhong, Z., 2016. An AIS-based high-resolution ship emission inventory and its uncertainty in Pearl River Delta region, China. Sci. Total Environ. 573, 1e10. Liu, C.J., 2015. The increment of ports throughput has come into a 'New Normal'. China Ports 07, 10e12 (in Chinese). Liu, H., Fu, M., Jin, X., Shang, Y., Shindell, D., Faluvegi, G., Shindell, C., He, K., 2016. Health and climate impacts of ocean-going vessels in East Asia. Nat. Clim. Change 18, 1e5. Maragkogianni, A., Papaefthimiou, S., 2015. Evaluating the social cost of cruise ships air emissions in major ports of Greece. Transportation Research Part D: Transport and Environment 36, 10e17. Muller, N.Z., Mendelsohn, R., 2007. Measuring the damages of air pollution in the United States. J. Environ. Econ. Manag. 54, 1e14. Notteboom, T., Delhaye, E., Vanherle, K., 2010. Analysis of the Consequences of Low Sulphur Fuel Requirements. Study Commissioned by the European Community Shipowners' Associations (ECSA). University of Antwerp, Antwerp. http:// schonescheepvaart.nl/downloads/rapporten/doc_1361790123.pdf. NRDC, 2014. Prevention and Control of Shipping and Port Air Emissions in China. Natural Resources Defense Council. https://www.nrdc.org/sites/default/files/ china-controlling-port-air-emissions-report.pdf. Sieber, N., Kummer, U., 2013. Environmental Costs of Maritime Shipping in Europe. Publications from the Institute for Energy and Rational Use of Energy, University of Stuttgart, Germany. http://niklas-sieber.de/Publications/Env_Cost_ Ship09_9.pdf. Simon, Ng, Booth, Veronica, Fung, Freda, 2013. Working towards a Quality Living Region e a Pearl River Delta Emission Control Area. Civic Exchange. http://civicexchange.org/cex_reports/201311AIR_PRDEmissionControlArea_en.pdf. Skjølsvik, K.O., Andersen, A.B., Corbett, J.J.S., Magne, John, 2000. Study of Greenhouse Gas Emissions from Ships. International Maritime Organization. Song, S., 2014. Ship emissions inventory, social cost and eco-efficiency in Shanghai Yangshan port. Atmos. Environ. 82, 288e297. The Ministry of Transport of China, 2015. The Implementation Plan of Ship Emission Control Areas (ECAs) in the Pearl River Delta, the Yangtze River Delta and the Bohai Bay Rim Area. Tichavska, M., Tovar, B., 2015. Environmental cost and eco-efficiency from vessel emissions in Las Palmas Port. Transport. Res. E Logist. Transport. Rev. 83, 126e140. Tzannatos, E., 2010. Cost assessment of ship emission reduction methods at berth: the case of the Port of Piraeus, Greece. Marit. Pol. Manag. 37, 427e445. UNCTAD, 2013. Review of Marinetime Transport 2013. United Nations Conference on Trade and Development. http://unctad.org/en/PublicationChapters/ rmt2013flyer_en.pdf. UNCTAD, 2014. Review of Maritime Transport 2014. United Nations Conference on Trade and Development. http://unctad.org/en/PublicationsLibrary/rmt2014_en. pdf. UNCTAD, 2015. Review of Maritime Transport 2015. United Nations Conference on Trade and Development. http://unctad.org/en/PublicationsLibrary/rmt2015_en. pdf. UNCTAD, 2016. Review of Maritime Transport 2016. United Nations Conference on Trade and Development. http://unctad.org/en/PublicationsLibrary/rmt2016_en. pdf. Viana, M., Fann, N., Tobías, A., Querol, X., Rojas-Rueda, D., Plaza, A., Aynos, G., Conde, J.A., Fernandez, L., Fernandez, C., 2015. Environmental and health benefits from designating the marmara sea and the Turkish straits as an emission control area (ECA). Environ. Sci. & Tech. 49, 3304e3313. Wang, C.F., Corbett, J.J., 2007. The costs and benefits of reducing SO(2) emissions from ships in the US West Coastal waters. Transport. Res. Transport Environ. 12, 577e588. Wang, X.P., Mauzerall, D.L., 2006. Evaluating impacts of air pollution in China on public health: implications for future air pollution and energy policies. Atmos. Environ. 40, 1706e1721.