Impact of port emissions on EU-regulated and non-regulated air quality indicators: The case of Civitavecchia (Italy)

Science Direct


• Meteorology, ship positioning and engine type influence the port role on nearby city air quality.

• The port activities contribute 33% of NO2 43% of PM10 and 60% of SO2 causing no exceedance of the AQ limits

• Low Sulphur fuels do not prevent release of pollutants as ultra-fine particles, and black carbon.

• High loads of black carbon and ultrafine particles coexist with admitted loads of NO2, SO2, and PM.


Current shipping activities employ about 3% of the world-delivered energy. Most of this energy is conveyed by diesel engines. In Europe, release of NOx and particulate matter (PM) from shipping is expected to equal the road-transport one by the year 2020. This paper addresses a typical central Mediterranean city-port condition to evaluate the relative contribution of shipping activities to the local air quality. A 3-year long air quality dataset collected at the boundary between the port of Civitavecchia (the major port in central Italy) and the city itself was analyzed to evaluate the long-term, relative contribution of the port and of the city to determining loads of EU-regulated pollutants (NO2, PM10, and SO2). In addition, black carbon and ultrafine-to-coarse particles data collected along a short-term, intensive campaign were used to assess the port’s role in emitting these unregulated pollutants. Cross-analysis of the measurements, allowed to assess of which shipping-related activities and port’s sectors represent the principal emitters. At the city-port boundary, the annual share of regulated pollutants originating in the port area by shipping and ground movements is of 33% for PM10, 43% for NO2, and 60% for SO2. Analysis of non-regulated pollutants shows the in-port, high polluting potential of some ship categories, in particular those employing low-sulfur but poorly refined oils. These conditions appear to be more often associated with Ro-Ro passenger ships. Piers closest to the Civitavecchia urban settlements are also observed to host the largest emissions. Meteorology and location of the piers with respect to residential areas are confirmed to govern the port’s share at impacting the city air quality. Even though air quality thresholds for regulated pollutants are not exceeded in Civitavecchia, constant consideration of an enlarged set of environmental variables should drive actions implemented to mitigate the port’s impact onto the nearby city’s air quality.

READ ARTICLE -> Science Direct

1. Introduction

The globalization of industrial and agricultural processes makes maritime transport (shipping) a fundamental sector of the world economy (UNCTAD, 2017EEA, 2017a). Currently, over 80% of the world trade is carried by sea (e.g., Cullinane et al., 2014UNCTAD, 2017). Overall, 25% of world delivered energy consumption is employed for transport. About 75% of this energy is employed for road transport, 12% for shipping and 12% for air transport (EIA, 2016).

The majority (95%) of the world’s shipping fleet runs on diesel engines (Deniz et al., 2010), and it is expected that shipping energy use and emissions will keep growing in the near future (Buhaug et al., 2009Eyring et al., 2010UNCTAD, 2017).

International shipping generates between 1.6% and 4.1% of global CO2 emissions (Psaraftis and Kontovas, 2009), a minor contribution with respect to road emissions (21% of global CO2 emissions) or energy production (35%) (Buhaug et al., 2009). Still, about 15% of global anthropogenic NOx, and 5–8% of SOx emissions are attributable to ocean-going ships (Eyring et al., 2005Corbett et al., 2007Maragkogianni et al., 2016).

Since 1990, the European Union (EU) transport-related emissions of air-quality-impacting species have been reducing with the exclusion of the international aviation and shipping sectors, whose particulate matter (PM), NOx and SO2 releases have increased instead (EEA, 2017b). In 2015, the transport sector (including road, aviation and shipping) contributed <20% of the total EU emissions of PM2.5, PM10, and SO2, while emitting 55% of total NOx. Shipping generated over 90% of transport-related SO2 emissions, while its PM2.5, PM10, and NOx emissions represented 45, 28, and 35%, respectively (EEA 2017b). In Europe, despite a four-time lower use of energy (EEA, 2017c), shipping releases of NOx and PM are expected to equal road-transport ones in 2020, (EEA, 2013Aksoyoglu et al., 2016). Such large NOx, SO2 and PM emissions originating from marine transport are caused by the predominant use of scarcely regulated diesel engines, which, on turn, run on low-grade oil (e.g., Cullinane, 2014).

Common oils used in marine engines are rich in sulfur and residuals as the heavy fuel oil (HFO, with sulfur content S > 1.5%). Conversely, the more expensive low-sulfur, partly-refined marine diesel oil (MDO, S ≤ 1.5%), and marine Gas Oil (MGO, S ≤ 0.1%) are only employed to respond to specific constraints (e.g., Buhaug et al., 2009). For instance, low sulfur, unrefined oil is often employed in sulfur emission control areas (SECAs.). Still, this limitation in sulfur content does not prevent the emission of other major pollutants as NOx, ash, heavy metals, and organics. Currently, the sulfur content of oil to be used in EU ports and emission control areas (ECA) is of 0.1%, while it is of 1.5% in non-ECA, EU waters (EU Directive 2005/33/CE), and of 3.5% in all other conditions. Starting 2020, global limits will converge to 0.1% fuels in ECA and port environments and to 0.5% in all other conditions (Buhaug et al., 2009EEA, 2013). As a reference, EU road diesel has a maximum sulfur content of 0.001%. Reducing from 0.5% to 0.1% the oil’s sulfur content results into a cut of about 50% in PM emissions (Buhaug et al., 2009). Reducing sulfur content of marine fuels can reduce PM mass, not necessarily soot, ashes and related health effects (e.g., Oeder et al., 2015). In this respect, after-treatment of diesel exhausts as done by emission gases conditioning systems (EGCS) represents the mandatory step to abate such emissions (e.g., EEA, 2013Barregård et al., 2014Winebrake et al., 2009).

Diesel engines are strong emitters of both primary and secondary PM (e.g., Winnes and Fridell, 2009Lack and Corbett, 2012). Diesels primary PM, mostly emitted as ultra-fine particles (i.e., smaller than 100 nm in diameter), includes soot, ash and a variety of organics as polycyclic aromatic hydrocarbons (PAHs), with strong light absorption properties (e.g., Buhaug et al., 2009Lack and Corbett, 2012). These dark PM emissions are commonly addressed as black and brown-carbon (BC and brC), respectively, e.g., Andreae and Gelencsér (2006). Shipping secondary PM emissions include sulfates, nitrates and organics (e.g, Viana et al., 2014Anderson et al., 2015). In terms of mass, shipping contributes more PM in the form of secondary particles with respect to primary ones (Viana et al., 2014).

Diesel emissions as a whole have been defined as “carcinogenic through genotoxicity” by the International Association of Cancer Research (IARC, Benbrahim-Tallaa et al., 2012). BC, an important component of diesel PM emissions, is also reported as a robust indicator (more than PM10 or PM2.5 metrics) of PM-induced mortality and morbidity (WHO, 2012). In a similar way, NO2, another significant emission of ship diesels, is known to bear adverse health consequences for humans (e.g., WHO, 2013).

In 2007, global mortality caused by ship emissions was estimated at 60,000 per year, with an expected growth of 40% by 2012 (Corbett et al. (2007)). In fact, Sofiev et al. (2018) estimate the 2020 global mortality due to shipping emissions will grow to about 250,000.

The Mediterranean Sea embodies only 0.7% of the world ocean’s surface. Still, 5% of the global shipping transits over this sea (UNCTAD, 2012Eyring et al., 2010). The Mediterranean region, and the Italian coast in particular, are one of the world hotspots in terms of shipping pollution and consequent health effects (Winebrake et al., 2009, Sofiev et al., 2018). In this respect, emissions from ship traffic are expected to have a significant, increasing impact on inland air quality (Viana et al., 2014Aksoyoglu et al., 2016). Overall, between 10 and 30% of PM2.5 in large Mediterranean coastal cities originates from shipping (Thunis et al., 2018).

Due to maneuvering, fueling and hoteling phases, ship-generated air pollution can be rather large in port areas (e.g., Barregård et al., 2014Murena et al., 2018). Involving non-optimal engine loads, maneuvering can generate much more pollution (3–6 times) than cruising and hoteling phases (Petzold et al., 2010Moldanova et al., 2013Lack and Corbett, 2012). When at berth, most ships supply their services by means of auxiliary diesel engines. Depending on ship type, the energy needed during hoteling ranges between 30% and 50% of the one employed at cruising (e.g., Tzannatos, 2010). Overall, in-port emissions of NOx and SO2 represent 5–6% of the total generated by ships in all their navigation phases (Whall et al., 2002).

All these elements point out that in both port areas and port cities important fractions of air pollutants can originate from ships (Tzannatos, 2010Cullinane et al., 2014Viana et al., 2014). To evaluate and reduce risks associated with atmospheric pollution, it is therefore important to know both amount and type of pollutants attributable to shipping, particularly in port-cities. This paper addresses previously unobserved aspects of air pollution in the Mediterranean port city of Civitavecchia, the major passenger-commercial port serving the Rome area, in central Italy. In fact, EU air quality thresholds for PM10, SO2, and NO2 (as well as for all the regulated pollutants) are rarely exceeded in Civitavecchia (Section 3.1.2). Still, epidemiological studies show higher mortality of residents in this city (mainly related to respiratory and cancer pathologies) with respect to the surrounding region (Fano et al., 2004Fano et al., 2006). Recently, a new epidemiological study confirmed a 31% increase in mortality due to lung cancer and a 51% increase due to neurological diseases for people residing within 500 m from the port area (Bauleo et al., 2018). As a consequence of these ongoing health issues, Civitavecchia become one of the most monitored sites in central Italy.

This study provides original insights about the role of the port at influencing the air quality in Civitavecchia. This is done by analyzing measurements of both regulated (NO2, PM10 and SO2) and unregulated atmospheric pollutants (black carbon, and particle size-resolved distributions from the ultrafine to the coarse modes). Principal emitters of these pollutants are inferred through a statistical/graphical analysis coupling observations with wind data to determine their azimuthal provenance and relevant loads. This approach allows to interpret large, multi-variate data-sets without need for modeling or analytical assumptions. Based on this analysis, we separate the port’s (shipping plus ground activities) from the city contributions to the Civitavecchia air quality indicators. Recommendations about improvements to mitigate the port’s impact on the city regulated and “unregulated” air quality are formulated.

2. Methods

2.1. The study area

The port of Civitavecchia (42.1°N, 11.8°E) is located on a flat, hill-edging headland of the Tyrrenian coast of central Italy (Fig. 1a). It extends to the NW of the city for about 3 km (Fig. 1b and 1c). In addition to regular ferry links (Ro-Ro passenger) to Sardinia, north Africa, and Spain, the port hosts an important traffic of cruise ships, cargo ferries, and carrier ships summing-up to some 3000 ship movements per year (about 1500 Ro-Ro passengers, 1000 cruise, 500 cargo/carriers), involving 4 million passengers, 1 million vehicles, and some 16 Mt of goods (Rome Ports Authority, 2014 data). Twenty-six port piers serve mainly Ro-Ro passengers ships in the southern port area, cruise ships in the central-west area, and cargo and goods-carriers in the central-north areas (Fig. 1c). The whole port area is flat, with sparse buildings spanning maximum heights of 10–15 m (Fig. 1b). Port extension works are underway in the NW sector, i.e., north of the cargo area (Fig. 1c). Road transport in the port sums up to about 1.000.000 visiting-vehicles per year. Just inland of the central-northern portion of the port, jet-fuel, gasoline, diesel, and heavy-fuel tanks store oil products unloaded by tankers (about 1Mt/year) through an off-shore pipeline located some 2 km off the coast.

Some 1.5 km NW of the northern, heavy-duty vehicles port’s entrance (Varco Molinari), stands the ENEL generating board, 1980 MW, coal-fired power plant of Torrevaldaliga Nord (TVN, yellow ellipse in Fig. 1c). Here ships unload coal to dome-covered deposits directly at the plant’s pier, detached from the port’s area. Emissions from this power plant are released out of a 250 m high chimney, at a maximum emission allowance of 2100 tons of SO2, 3450 tons of NOx, 160 tons of PM, 2000 tons of CO, and 195 tons of NH3 per year (ENEL, 2016). Some 400 m SE of TVN, a second power plant (1200 MW, gas-fired) named Torrevaldaliga Sud (TVS) is operated by Tirreno Power. TVS emissions are released at 90 m MSL, with maximum allowances of 2500 tons of NOx, and 1500 tons of CO per year. TVN and TVS stand at a distance of 4.5 km from the southern portion of the port, where the entrance named “Varco Vespucci” directly connects to the NW sector of Civitavecchia. Here also stands the air quality station “Arpa Porto” (Fig. 1c, d). The city itself (about 52,000 inhabitants) further extends some 2 km SE along the coast, and some 2 km inland (NE) from this point. The city center is located about 0.7 km SE of the “Varco Vespucci” port entrance (see also Fig. 6).

2.2. Measurements

Our analysis employs “receptor type” observations performed at fixed monitoring stations operated in the port area. The investigation takes into account the main emission sources related to the Civitavecchia port and city activities. Timing of measurements and relevant analyses is given in UTC. Civitavecchia local time is UTC + 1.

2.2.1. Regulated pollutants

Measurements of EU-regulated air pollutants employed in this study, namely NO2, SO2, and PM10, were collected by the regional environmental protection agency (ARPA Lazio) at the Arpa Porto station, during the 3-year period May 2013-April 2016. This monitoring station stands exactly at the boundary between the port and the city (Fig. 1c, d), i.e., it is assumed to intercept the most of the air pollution originated in the port area and directed to the city. The type of instrumentation performing the air quality and meteorological measurements employed in this analysis is listed in Table 1. All variables addressed were measured at 1-h time resolution. All the instrumentation run at Arpa Porto matches the quality assurance and quality control (QA/QC) specifications given in the EC Directive 2008/50/CE (ISPRA, 2014).

Table 1. List of the instrumentation/measurements employed in the analysis. Abbreviations indicate, respectively: PM10 = particulate matter smaller than 10 µm in aerodynamic diameter; NO, NO2 = Nitrogen oxide, and dioxide; SO2 = Sulfur dioxide; AWS = automatic weather station; p = atmospheric pressure; T = temperature; RH = relative humidity, ws, and wd = wind speed, and direction; APS = aerodynamic particle sizer; SMPS = scanning mobility particle sizer; Scattering-Coeff = scattering coefficient; eBC = equivalent black carbon.

SiteInstrumentMeasured VariableTime resol.Begin dateEnd dateData points #
Arpa PortoEnvironment
Arpa PortoTeledyne
NO + NO21-h1–4-201331–3-201623,308
Arpa PortoTeledyne
Arpa PortoVaisala AWSp,T,RH, ws, wd1-h1–4-201331–3-201624,690
Pier 24TSI APSSize Distribution 0.5–30 µm5′05–4-201626–4-20166274
Pier 24Tropos SMPSSize Distribution 0.008–0.7 µm5′05–4-201626–4-20165996
Pier 24Magee AE33 AethalometereBC1′05–4-201626–4-201631,986
Pier 24Ecotech Aurora 3000 NephelometerScattering Coeff.1′05–4-201626–4-201631,127
Pier 24Lufft AWS 700p,T,RH, ws, wd1′05–4-201626–4-201631,375
Pier 24Lufft CHM15kBackscatter profiles 1064 nm5′05–4-201626–4-20166243

2.2.2. Non-regulated pollutants

As a complementary information, high time-resolution measurements of equivalent black carbon (eBC), ultra-fine particles (UFP), and particle size distributions (PSD) have been collected at Pier 24 (Fig. 1c, 42.104 N − 11.778 E, 7 m MSL) by our mobile laboratory AEROLAB, during the month of April 2016. These three metrics are recognized to bear an association with health impairment stronger than PM10 (e.g., Benbrahim-Tallaa et al., 2012WHO, 2013). Pier 24 is located at the core of the port, about 1 km NW of the port Ro-Ro passenger piers and 1.2 km NW of the Arpa Porto station (Fig. 1c). The city center stands about 1 km SE of the Ro-Ro passenger piers. Meteorological variables, and atmospheric backscatter profiles collected by an automated lidar ceilometer (ALC) were also measured at Pier 24 during this campaign. The list of the AEROLAB measurements employed in this analysis is also given in Table 1.

All the in-situ aerosol instrumentation run within the AEROLAB at Pier 24 matches the standard operating procedures and measurement guidelines for aerosol particle variables defined by the ACTRIS (the EU Research Infrastructure for the observation of Aerosol, Clouds and Trace Gases,, and by WMO/GAW (2016). All these instruments were calibrated within the 4 months preceding the campaign.

2.3. Estimated emissions in the Civitavecchia area

An evaluation of the main pollutants emitted in the Civitavecchia area has been performed to provide an “order of magnitude” reference in support to our analysis. The estimates include emissions from: (i) ships, (ii) city road transport and heating; (iii) TVN and TVS power plants. No quantitative use of these emission estimates is made throughout the paper.

Timing and category of ships calling at the port of Civitavecchia in the period May 2013 – April 2016, as provided by the Rome Ports Authority have been employed to estimate relevant emissions. Table 2 reports the number of ship calls occurred at Civitavecchia in that period. Ships are divided into the four main types calling at this port: (1) Ro-Ro passenger, (2) cruise, (3) bulk material carrier, and (4) cargo. The identification number of relevant mooring piers is also specified in Table 2. Quantity of ships at anchor or moored at the TVN power plant’s dock, both stationing being outside the port (Fig. 1c), are reported in the last column. Over this time-span, the port borne some 2700 ship movements per year, 51% made by Ro-Ro-passenger type, 33% by cruise ships, and 8% made by both cargo and bulk carriers. Maximum ship movements occurred in summer and minimized in winter, with a summer to winter decrease of 45% in the case of Ro-Ro Passenger and of 70% for Cruise ships. These data also show the docks closer to the city (the Ro-Ro passenger ones in Fig. 1c, d) to host the largest amount of annual ship traffic.

Table 2. Yearly and seasonal average number of calls at the port of Civitavecchia per ship type and mooring pier in the 3-year period 2013–2016 (grand total 9207 calls). Usual mooring pier numbers are reported under the ship type. Ship calls at the TVN power plant pier, or at anchorage outside the port are reported in the last column.

Average periodRo-Ro passenger 214-1618-2021Cruise 1011-1213-25SBulk carriers 2223-24Cargo 25 N-2627-28Port TotalAnchor & TVN

In accordance with Whall et al. (2002), and EEA (2016), emissions of NOx, SO2 and PM originated from ships maneuvering and hoteling in the port have been estimated by considering single ship type and engine power, as provided by the Port Authority records. Engines have been assumed to be medium speed diesels employing marine gas oil (MGO) in all port maneuvering and hoteling phases. Evaluation of the yearly average emissions per ship type are summarized in Table 3. Seasonal estimates of the specific emission daily rates can be derived by combining the data reported in Table 2Table 3.

Table 3. Estimated yearly emissions (2013–2016 average) per ship category and hoteling pier at the port of Civitavecchia. Estimates of emissions originating outside the port area (off-port anchor and TVN piers) are included in the last row.

Ship typeNOx [t/year]TSP [t/year]SO2 [t/year]Piers
Ro-Ro Pax407.913.842.32–14-16–18-20–21
Cargo85.72.98.925 N-26–27-28
Port Total939.631.997.5All Piers
Port + Anchor + TVN pier1177.740.0122.2All piers + Anchor & TVN pier

Yearly-average emissions attributable to the city’s road transport, heating, and to the TVN and TVS power plants are summarized in Table 4. The city emissions have been estimated on the basis of the emission inventories published by the National environmental agency, ISPRA. Emissions generated by road transport have been extracted from the provincial disaggregation of the national inventory (ISPRA, 2010). Conversely, emissions generated by heating (November 15 to March 15) in the city of Civitavecchia have been estimated on the basis of the provincial disaggregation of the national inventory ISPRA (2009), and of the 1990–2016 emissions national inventory (ISPRA, 2018). The TVN power plant emissions have been extracted from the ENEL generating board publication ENEL, 2016. Emissions of the TVS power station have been obtained from the Ministry of Environment web site (

Table 4. Estimates of annual emissions generated by the TVN and TVS power plants, by road transport, and by heating (November 15-March 15) in Civitavecchia.

EmitterNOx [t/y]TSP [t/y]SO2 [t/y]Reference
TVN2994621943ENEL (2016)
TVS2500n.a.n.a.Ministry of Environment web site
Road330220ISPRA, 2010ISPRA, 2018
Heating8711210ISPRA, 2009ISPRA, 2018

According to these estimates, release of NOx from the two power stations is about 5-times higher than the shipping one, and 15-times higher than the city road transport one. Conversely, PM emission from TVN is 6-times the shipping one and 12-times the city road traffic one. In terms of SO2, TVN emission is 17-times the shipping one. Emissions from the TVN and TVS plants are released at a single point, positioned respectively at 250 m and 90 m MSL, some 4.5 km away from the city, while port emissions occur between ground and some 60 m MSL, along a 2 km-long strip, partly encompassed in the urban area (Fig. 1).

Comparison of emission estimates from shipping (“port total” in Table 3) and the city (“road” plus “heating” in Table 4) indicates shipping releases to be larger in terms of NOx (by a factor ≈ 2-3), PM10 (by a factor ≈ 2 in non-heating seasons), and SO2 (by a factor ≈ 10). Conversely, PM emissions from the city are expected to dominate during the heating season. In the annual average, city emissions of PM are a factor ≈ 3 larger than the port’s ones. The analysis presented in the next sections (Section 3.1) will provide an observation-based evaluation of the actual balance between these sources as measured at the boundary between the port and the city.

2.4. Bivariate polar analysis

To separate the “port” from the “city” components of the AQ variables collected at the port area measurement sites (receptors), data have been analyzed employing the R-package “Openair” bivariate polar analysis (Carslaw et al., 2006Carslaw and Ropkins, 2012). This processing allows to visualize and quantify how the concentration of a variable (commonly a pollutant) at a receptor point changes as a function of both wind direction and a third parameter (usually wind-speed).

Other approaches employed to identify pollution sources in port areas include source apportionment (SA) analysis, and dispersion modelling. SA has been mainly applied to chemical speciation of PM data to define the type of emitters affecting measured PM loads (Cesari et al., 2014Pérez et al., 2016Jeong et al., 2017Saraga et al., 2019). Whereas source apportionment analysis identifies categories of emitters but not their location, dispersion modelling can identify contributions of various emitters to pollutant loads measured at a receptor site (e.g., Saxe and Larsen, 2004Gariazzo et al., 2007). However, the detailed temporal and spatial description of emissions and meteorology needed to run these models in an effective way can lead to large discrepancies between modeled and observed data (e.g., Gibson et al., 2013Casazza et al., 2019). To the end of providing a representative view of the provenance of pollutants in the Civitavecchia port area on the basis of a three-year dataset, the bivariate polar analysis was then preferred to dispersion modelling. This straightforward analysis well suits the Civitavecchia port’s conditions since the whole area is flat and maximum distances between main emitters and receptors are of the order of 1–2 km.

Two Openair functions have been chiefly employed to this goal: Polar-Plot and Pollution-Rose. The first one allows to observe bivariate average, weighted mean, and conditional probability function (CPF) distributions of the addressed variable (e.g., Uria-Tellaetxe, and Carslaw, 2014). The second function permits to evaluate the polar-dependent contribution to the mean value of the investigated variables. In particular, the Polar-Plot function permits some discrimination between local and distant sources by exploiting the fact that the first are evidenced at low wind speeds, while the latters are brought to the receptor by increasing wind speeds (e.g., Carslaw et al., 2006). These conditions are expected to hold particularly in a flat area as the Civitavecchia port’s one. Combined use of the options available within these two functions allowed us to infer important information about the origin and relevant weight of the addressed pollutants.

3. Data analysis

In Section 3.1 hereafter we analyze first the “long-term” (3-year) record of NO2, SO2, and PM10 collected at the Arpa Porto station. In Section 3.2, we then address the April 2016 intensive measurements performed by the AEROLAB mobile station at Pier 24.

3.1. The “long-term” picture

3.1.1. Meteorology

Knowledge of local meteorological patterns is important to understand how pollutants are conveyed and dispersed in the area being studied. Analysis of the Arpa Porto 3-year wind record depicts the typical circulation patterns occurring at the port-city interface. These are summarized in Fig. 2. In particular, Fig. 2a shows the seasonally-resolved weighted mean of the wind speed as a function of wind direction (angle), and time of day (radius, spanning from 0 h UTC at the center to 24 h UTC at the outer circle). This variable represents mean wind speeds weighted by their frequency of occurrence at each wind direction and time of day. Conversely, Fig. 2b illustrates the seasonal polar distribution of the components of the average wind speed. As visible from the maps in Fig. 1, only winds proceeding from the 250° to 340° quadrant can transport air masses from the port to the urban area. Fig. 2b indicates this condition to reach a minimum occurrence (15% of the time) in winter and a maximum (30% of the time) in summer. This is also associated with a sea breeze-type circulation mainly occurring between 10 am and 8 pm (Fig. 2a). Proceeding from angles of 270–300°, these breezes typically convey towards the city the emissions originating at the Ro-Ro passenger and at the cruise-ship piers, that is the port’s largest emission sources.

However, Fig. 2 shows the most frequent wind conditions encountered at Arpa Porto to be those proceeding from SE. These occur at all day times over 25% of the time in summer, and 35% of the time in winter (Fig. 2b). This “port pollution-removal” condition is likely associated with the formation of a coastal low-level jet generated by the interaction of S-N pressure gradients and the hills running from the SE to the N-NW, just inland of Civitavecchia (e.g., Stull, 2017). A similar jet pattern (in this case proceeding from N-NE), appears to be generated by the same hills, at their turning towards NE, some 5 km north of the city. These latter conditions occur about 15% of the time in summer, and 25% of the time in the remaining seasons. Both these jet-like patterns decrease in summer, i.e., when pressure gradients minimize.

Overall, these wind statistics indicate that pollutants from the port (and from the Civitavecchia area in general) preferentially disperse to the NW and SW, i.e., away from the city and off the coast. However, diffusion of port’s pollution towards Civitavecchia is important in the summer months (about 25% of the time), driven by the sea-breeze pattern, with average wind speeds of 3–4 m/s. Overall, the above-described wind patterns account for about 70–75% of the atmospheric circulation conditions encountered at Civitavecchia during this three-year period, the remaining ones pertaining to diffusion to the N-NE sectors.

3.1.2. “Regulated” pollutants

A long-term representation of the NO2, SO2 and PM10 pollutants measured at the Arpa Porto station is given in Fig. 3. Based on the addressed three-year dataset, it depicts, respectively: (1) the seasonal mean values of the NO2, SO2 and PM10 hourly measurements (panels a, b, and c) as a function of wind direction (angle), and time of day (0–24 h UTC, radius), and (2) the polar contribution to the average NO2, SO2 and PM10 loads (panels d, e, and f). On the basis of the wind-rose shown in Fig. 1d, we assume as originating in the port area all the components reaching the Arpa Porto monitoring station from the angular range 180°–360°. Conversely, the city component reaching the Arpa Porto monitoring station is associated with winds from the 0° to 180° quadrant.

As expected, the NO2 time-resolved plots (Fig. 3a) show the city-bound contribution to reach a maximum at traffic rush hours (7 am and 7 pm, approximately). This is particularly evident in autumn and winter. Conversely, the port’s component spreads over the full 5 am–8 pm time range, peaking at about 4 pm (i.e., when the sea-breeze reaches its maximum, e. g., Fig. 2). The port signal is clearly visible in all the SO2 records, with a summertime maximum (Fig. 3b). Conversely, the PM10 plots do not show dominant fingerprints originating from the port (Fig. 3c). Similarly to what was observed for NO2, the traffic daily cycle is visible in the city-oriented half-circle in autumn and winter. PM10 from the marine quadrants is found to peak mainly before 10 am and at night, not necessarily proceeding from the port hotspots identified by the NO2 and SO2 records. As hypothesized by means of the emission estimates discussed in Section 2.3, this point confirms the minor role of port’s emissions at determining the PM10 load, particularly in winter.

Table 5 reports the Arpa Porto seasonal averages of the three variables addressed here, i.e., NO2, SO2 and PM10, as obtained from the analysis of Fig. 3d, 3e and 3f, respectively. Wind speed occurrences obtained from data in Fig. 2b are also included in this table. Components from the 180°–360° quadrant (i.e., originating from the port area and reaching the station), and from the angular range 250°–340° (that is the portion then proceeding from the station towards the city center) are also included in this table as ‘‘from port’’ and ‘‘to-the-city’’ components, respectively. Table 5 shows that in terms of contribution to the mean, 55% of the 29.8 µg/m3 NO2 averaged in summer originates from the port area. This reduces to 20% in winter, when the mean NO2 load is of 26.8 µg/m3. The NO2 average yearly contribution from the port is of 43%.

Table 5. The three-year (2013–2016) statistics of regulated pollutants and wind speeds at the Civitavecchia Arpa Porto station. Wind speed averages “from port” (angular range 180°-360°), and “from port and to the city” (angular range 250°-340°) are given as % occurrences.

Site Arpa PortoNO2 [µg/m3]SO2 [µg/m3]PM10 [µg/m3]Wind Speed [m/s]
Year average ± ± 22.01.21 ± 2.722.4 ± 18.53.4 ± 2.1
Year average from port (% of year average)11.4 (43%)0.7 (60%)7.4 (33%)40% of wind data from port
Year average from port to Civitavecchia6.6 (25%)0.5 (43%)4.2 (19%)20% of wind data to Civitavecchia
Winter average26.80.824.53.7
Winter average from port (% of winter average)5.4 (20%)0.2 (30%)4.9 (20%)28% of wind data from port,
13% to Civitavecchia
Summer average29.81.823.52.9
Summer average from port (% of summer average)16.4 (55%)1.4 (80%)11.8 (50%)50% of wind data from port,
30% to Civitavecchia

In the case of SO2, 80% of the summer mean value of 1.8 µg/m3 is contributed by the port area. This reduces to 30% in winter, when the mean SO2 descends to 0.8 µg/m3. Highest SO2 concentrations are generated in the port area, with a yearly average contribution of 60% of the total load. Both SO2 and NO2 port components peak at wind directions of 270°–285°, i.e., the direction corresponding to Ro-Ro passenger and cruise ship piers (Fig. 1). In this respect, the intensive observations made at Pier-24 and presented in the following Section 3.2, will help identifying the actual origins of these pollutants.

In terms of contribution to mean PM10 values, 50% of the 23.5 µg/m3 averaged in summer originates from the port area. This reduces to 20% in winter, when the mean PM10 rises to 24.5 µg/m3. In this case, average yearly contribution from the port is of 33%. Both the PM10 and NO2 components from the city show a principal peak at 45°–60°, and a secondary one at 105°–135°, that is, the directions pointing to the closest road traffic source, and to the city center, respectively.

Overall, the analysis of the 3-year record collected at the port-city border indicates that concentrations of regulated pollutants (NO2, SO2 and PM10) always kept below EU annual limits, and that the port area (including ship and road movements) represents a minority contribution (33, 43%) to PM10 and NO2, respectively, while contributing 60% of the SO2 loads. These contributions descend to some 19% for PM10, 25% for NO2, and 43% for SO2 if considering the airmasses proceeding to influence the city center. Minimum concentrations were recorded in spring and fall. Summer maxima with respect to winter were observed for SO2 and NO2 (attributable to the larger ship traffic and increased sea-breeze condition). A winter maximum of PM10 is possibly due to city heating and low-level-jet, south easterly circulation. SO2 first, and NO2 act as the clearest tracers of the port’s emissions. Finally, neither NO2 nor SO2 signals appear to proceed from the power-plants sector (320–330°) towards the city (e. g., Fig. 3).

3.2. Regulated and non-regulated pollutants during the intensive campaign of April 2016

To further investigate the pollution associated with particulate matter in the port area, an intensive observational period was performed in April 2016, in the framework of the CNR’s “AirSeaLab” project. The data addressed here were collected by the ISAC-CNR AEROLAB at Pier 24 (high temporal resolution, e.g., Table 1), and by ARPA Lazio at the Arpa Porto station (1-h resolution) in the period 5–26 April 2016. During this intensive campaign, Ro-Ro passenger, cruise, and carrier ships stopped at the Civitavecchia piers 860, 523 and 378 hours, respectively. Median age of these ships was 14, 8 and 9 years. Median mooring times were 5, 12 and 9 hours, respectively. In terms of ship mass-weighed mooring time, this corresponds to approximately 61, 86, and 20 kt*hour per day, respectively. That is, cruise ships embodied the largest mass*permanence-time factor (a proxy for the required amount of energy production) in the port. At the same time, some 56% of the cruise ships calling at the port was running EGCSs (emission gases conditioning systems) for the reduction of both SO2 and PM emissions.

Polar plots of the regulated pollutants load and contribution to the mean as measured at Arpa Porto during this campaign are presented in Fig. 4, using same plotting formats as in the long-term analysis of Section 3.1.2. Relevant average values and port origins are summarized in Table 6. This shows that measured average loads are all close (well within one standard deviation) to the long-term ones presented in Table 5. During the April 2016 campaign, this station at the port-city boundary registered a harbor contribution to the NO2, SO2 and PM10 loads of 43%, 73% and 38%, respectively (Table 6 and Fig. 4). As found in the long-term analysis, polar plots of NO2 and SO2 presented in Fig. 4 display a clear fingerprint of the port sector. In fact, maxima in NO2 and SO2 proceed from 285° (see maps in Fig. 1), with median values of 60 and 6 µg/m3, respectively (Fig. 4a, b, d, and e). Conversely, a good deal of the PM10 originating from the port sector points also to ground mobility corridors (Fig. 4c), that is, it is likely associated with both shipping and road traffic within the port. In analogy with the 3-year statistics, the city emissions originate mainly from the SE, providing important contributions to both PM10, and NO2 loads (e.g., Fig. 4d and 4f).

Table 6. April 2016 campaign: statistics of NO2, SO2 and PM10 at the Arpa Porto station. Pollutants are defined as arriving from the port area for winds proceeding from the angular range 180°-360°.

Empty CellNO2 [µg/m3]SO2 [µg/m3]PM10 [µg/m3]Wind Speed [m/s]
Campaign average24. (21% to CV)
Average from Port area (% of campaign average)10.5 (43%)0.8 (73%)10.2 (38%)(40% from port)
Mean value (% contribution) from Ro-Ro passenger piers (285°)3 (13%)0.2 (22%)1.6 (6%)(7% from 285°)

Average values of equivalent black carbon (eBC), total particle number concentration (N, sum of particles with diameter 0.008 < D < 10 μm), and total particle volume, V10 (that is a proxy for PM10, obtained by summing the volumes of all the particles with D < 10 μm), measured by the AEROLAB at Pier 24 are reported in Table 7. Polar-plots of these variables are presented in Fig. 5. All maximum values in these plots originate from the S-SE area of the port. Only eBC, and to some extent V10, present secondary maxima at low wind speeds, i.e., originating nearby the AEROLAB site. These are possibly generated by local circulation of cargo trains, heavy and light road vehicles reaching nearby piers.

Table 7. Statistics of equivalent black carbon (eBC), particulate number concentrations (N, size range 0.008 < D < 10 µm), and volume V10 (D < 10 µm), as measured by the AEROLAB at Pier 24 during the April 2016 campaign and average contribution from the angular range 150°-185° (Ro-Ro passenger piers).

Empty CelleBC [µg/m3]N [1/cm3]V10 [µcm3/m3]
Campaign average ± st. dev.1.3 ± 1.620,400 ± 22,50011.7 ± 6.5
Mean from Ro-Ro piers (% contribution to total)2.0 (45%)50,000 (58%)15 (43%)

Compared to wintertime measurements performed in the city center of Rome (e.g., Costabile et al., 2017), the average eBC concentration recorded at Pier 24 is about halved (1.3 µg/m3 vs. 2.6 µg/m3, while particle number concentrations are almost doubled (20,400 vs. 12,000 cm−3). Some 45%–58% of these average loads originate in the sector 150°–185°, that is, the Ro-Ro passenger piers area (Fig. 1d). In fact, mean loads of 2 µg/m3 for eBC, and 50,000 cm−3 for N are observed to originate in this sector (Fig. 5d and e, and Table 7). These loads are rather high with respect to polluted urban sites, particularly in what concerns particle number concentrations. Conversely, even assuming a high aerosol density as of 2 g/cm3, the corresponding V10 would translate into an average PM10 < 30 µg/m3, that is well below the EU daily limit of 50 µg/m3Fig. 5 shows that eBC concentrations maximize when proceeding from the Ro-Ro passenger piers segment. The V10 plot behaves similarly to eBC, except for some enhancements in the W-SW region, likely originating in the cargo and the cruise-ship sectors. In a similar way, high particle number concentrations (about 90% made of ultrafine particles with D < 100 nm) mainly originate in the port’s Ro-Ro passenger area.

The triangulation of sectors originating maximum eBC, and N concentrations at the AEROLAB site (that is the 150°–185° arch), and those originating maximum SO2 at Arpa Porto (i.e., the 210°–300° arch) places the maximum of particulate emissions in the Ro-Ro passenger zone evidenced in red in Fig. 6. In sea-breeze conditions, this sector is upwind, and less than 1 km apart from the city center.

To better identify properties and origins of pollutants in the addressed port areas, we examined polar-plots of the following variables (Fig. 7): i) the ratio between 40 nm and 100 nm particle number concentrations (dN40/dN100Fig. 7a), and its reversal, (dN100/dN40, Fig. 7b); ii) the ratio between black carbon concentration (eBC) and particle’s volume in the “ultra-fine” size fraction D < 100 nm (V100Fig. 7c); and iii) the particle’s absorption Angstrom exponents (AAE, Fig. 7d), respectively. The dN40 vs. dN100 relationship provides information about the relevant weight of nucleation-mode particles with respect to accumulation-mode ones. The eBC/V0.1 ratio is an indicator of the relative importance of BC within ultra-fine particles. The particle’s absorption Angstrom exponent provides information about the relative importance of BC (low AAE) with respect to other light absorbing material as organic carbon or dust (high AAE) in the sampled particles (e.g., Andreae, and Gelencsér, 2006). A further description of these variables will be also given in the following paragraphs.

Examination of Fig. 7 indicates four main combustion sources are likely to drive the average properties of particulate matter in the port:i)

medium/low speed diesel engines running on low-sulfur (still with high contents of HFO) marine diesel oil (MDO), mostly operating in the Ro-Ro passenger and carrier sectors of the port;ii)

medium–high speed diesel engines fitted with EGCS, mainly operating in the cruise-ship and cargo sectors of the port;iii)

diesel, heavy-duty vehicles operating on road; andiv)

light duty vehicles operating either on diesel or gasoline engines.

This interpretation is based on the following arguments:

Ultrafine modes. In plumes emitted by medium/slow speed marine diesels running on MDO, the ratio dN40/dN100 (that is between concentrations of “nucleation type” (D ≈ 40 nm) and “accumulation mode” (D ≈ 100 nm particles) is commonly>1 (≫1 for Heavy Fuel Oil). Conversely, this ratio is expected to be < 1 in the case of engines running on MGO (e.g, Anderson et al., 2015) or automotive diesels (e.g., Harris and Maricq, 2001). This is mainly due to the high concentration of residuals (heavy metals, ash and sulfur) present in marine unrefined oils (as MDO), a condition leading to an enhanced generation of “nucleation mode” nanoparticles (e.g., Kasper et al., 2007Anderson et al., 2015Streibel, 2016ICCT, 2016 ), and black carbon (Lack and Corbett, 2012). In addition to the use of HFO and MDO, presence of “nucleation mode” particles in marine emissions is also associated with low engine load operations (e.g., ICCT, 2016). As opposite, and similarly to road diesels, marine diesel engines running at optimal loads tend to emit particles centered at about 100 nm (Harris and Maricq, 2001). Recent cruise ships often run on battery-fed electric motors, whose energy is generated by diesel engines operating at optimal speed. Emissions from engines using less-refined oils should then present high dN40/dN100 ratios. In this respect, the SMPS-derived dN40/dN100 ratios presented in Fig. 7a maximize in the Ro-Ro passengers sector first, with secondary maxima in the carriers (South) sector. Conversely, the inverse ratio dN100/dN40 (Fig. 7b) presents maxima in the cruise ship sector (SW and W of Pier 24), and in the road traffic areas of the port (NW to SW half circle), particularly in low-wind conditions, i.e., for advected air associated with nearby yards, where cars and containers were handled before shipping.

Specific eBC content. Even though most of the eBC emissions originate from the Ro-Ro passengers sector (Fig. 5d), black carbon contents at specific particles volume (the eBC/ V100 ratio of Fig. 7c) are highest along the road traffic areas of the port. These areas are characterized by heavy-duty vehicle operations in the NW (port construction, cargo loading, and vehicle maneuvering areas) and eastern (heavy duty vehicles and trains) sectors, and by light vehicle traffic in the SE, and NE sectors. In fact, Kasper et al. (2007) points out that the ratio between particle number and BC concentrations is higher in low-speed marine diesels with respect to light vehicle diesels. Interestingly enough, cruise ships are found to emit less BC (Fig. 5a) and less BC per aerosol unit volume than all other ships and ground vehicles operating in the port (Fig. 7c). In the comparison of ship emissions, this is possibly attributable to cruise ships employing: i) a better fuel quality (as inferred by the previous point); ii) EGCS devices on 56% of the calling vessels; iii) engines operating at optimal regimes (e.g., ICCT, 2016). In the ship vs. road-vehicle comparison, the unbalance possibly depends on to the higher BC emission factors characterizing road vehicles with respect to ship engines (e. g., Ježek et al., 2015Zavala et al., 2017ICCT, 2016). In fact, the ICCT, 2016 tests report distillate fuels (MGO) to generate the lowest BC emissions, followed by conventional HFO. The highest BC emission factors were found to be associated with the low sulfur, residual fuel (MDO) tested.

Light absorption coefficients. The absorption Angstrom exponent (AAE) is a measure of the spectral variation of the aerosol absorption coefficient. Amongst light absorbing aerosols, black carbon is characterized by low spectral variation of the absorption coefficient, i.e., AAE ≤ 1. Conversely, light absorption of organic aerosols and mineral dust shows a marked spectral dependence (AAE > 1), e.g., Andreae and Gelencsér, 2006Bond et al., 2013. Analysis of the AAE plot in Fig. 7d indicates that both cruise ships (W-SW sector), and light-vehicle road traffic (NE sector) are mainly characterized by organic carbon type emissions (AAE > 1). Such finding can be explained by the higher proportion of organics gasoline cars emit with respect to both diesel cars (a factor of about 10 higher) and ship diesels (about two-times higher), e.g., EEA (2016). Conversely, prevailing BC conditions (AAE ≤ 1) are confirmed to dominate aerosol properties in the whole Ro-Ro passenger berthing areas, in the port channel leading to the open sea (associated to medium wind speeds), and in the new port construction sectors to the NW.

4. Discussion

Principal goal of this paper was the separation of emissions originated in the port area from the ones pertaining to the city of Civitavecchia. This was achieved by the bivariate data analysis of AQ measurements collected over three years at the exact border between the city and the port, that is, a city area where the impact of port emissions is expected to be the highest. This analysis identified main points of emission and their contribution to the measured loads of regulated pollutants as NO2, PM10, and SO2. The overall contribution from the port area vs. the one originating from the city area could then be evaluated for these three markers. Combination of intensive observations from two receptor stations made in April 2016, also allowed to infer piers and ship categories emitting the highest amounts of UFP, eBC, NO2, PM10, and SO2. Main strength of this analysis resides in its direct, statistical representation of real data (e.g., Carslaw and Beevers, 2013). Main limitations of the approach are related to the attribution of pollutant sources based essentially on provenance and speed of winds during measurements. In this respect, both the flatness of the investigated area and the short distances (ranging between 50 and 2000 m) separating emitters from observing stations support the reliability of the results presented in the paper. Considering the type of data employed (a 3-year record of hourly NO2, PM10, and SO2) and a 1-month record of aerosol size distributions and eBC concentrations at minute time resolution, analogous results could only be obtained by high resolution dispersion modeling (e.g., Gibson et al., 2013). In addition to the general port vs. city pollutants share obtained in our analysis, dispersion modeling might also isolate contributions of single emitters. At the same time, such an approach would have required a huge effort to define time evolution and properties of the emissions, and high resolution meteorological dispersion parameters along a 3-year timeframe. This could possibly lead to important errors in the case of rapidly varying pollutants as eBC, PM10, and particle number concentration (e.g., Gibson et al., 2013). We therefore believe the results provided in this paper represent a reasonable estimate of the whole port’s area contribution to air pollution at the port-city border and a good quantification of sources of unregulated, still important pollutants as UFP and BC.

5. Conclusions

In spite of air pollution records well below the limits established by the EU Air Quality Directive (2008/50), the city of Civitavecchia suffers from some health issues with respect to the regional average. Main emissions in this area are generated by two collocated power stations emitting out of high stacks some 4 km north of the city, by the port, and by the city’s road traffic and heating. These four sources (considering the two power stations as one) were estimated to contribute some 5500, 1000, 300, and 100 t/y NO2; 60, 40, 20, and 100 t/y PM10, and 2000, 30, 0, and 10 t/y SO2, respectively. In this context, measurements of atmospheric pollutants in the Civitavecchia port area have been studied in relation to wind parameters both on a long-term (3-year monitoring) and on a short-term (one-month intensive observations) basis. The study employed a bivariate polar analysis of “receptor type” observations performed at fixed monitoring stations operated at the port-city boundary, and within the port itself.

Analysis of the 3-year dataset indicated that pollutants from the port (and from the Civitavecchia area in general) preferentially diffuse to the NW and SW directions, i.e., away from the city. This happens thanks to the frequent formation of low-level jets by the nearby hills. Diffusion of port’s pollution to the city of Civitavecchia maximizes in spring and summer, mainly between 10 and 20 UTC, following a sea-breeze pattern with average wind speeds of 3–4 m/s. The long-term measurements collected at the port-city boundary (Arpa Porto station) confirmed that concentrations of regulated pollutants (NO2, SO2 and PM10) always kept below EU limits. The port area (including ship and ground operations) contributed respectively 33% (7.4 µg/m3), 43% (11.4 µg/m3), and 60% (0.7 µg/m3) of the PM10, NO2, and SO2 loads observed at Arpa Porto. Some 4.2 µg/m3 PM10, 6.6 µg/m3 NO2, and 0.5 µg/m3 SO2 were estimated to then proceed to influence the city center annual loads. Out of these three species, SO2 first and NO2 were found to be the clearest tracers of the port emissions. In particular, SO2 best correlates with port origins and with the seasonal variation of the port’s marine traffic. Neither the NO2 nor the SO2 records evidenced clear footprints from the two power plants.

Specifically addressing BC contents, ultrafine particles concentration and size distribution, the intensive campaign carried-out within the port, in April 2016, allowed to unveil some important features of the particulate matter origins and properties in the port area. In this spring period, cruise ships first, and Ro-Ro passenger ships represented the largest hotelling masses, i.e., energy production in the port. Some 56% of the cruise ships calling at the port was employing emission gases conditioning systems to reduce both SO2 and PM emissions. During this campaign, the Arpa Porto station registered a harbor contribution to the PM10, NO2, and SO2 loads of 38%, 43%, and 73%, respectively. These values are in good agreement with the 3-year averages observed at the same station. As found by the long-term analysis. NO2 and SO2 represented the clearest fingerprint of the port sector. Both these components originated mainly from the Ro-Ro passenger and cruise ship piers.

Measurements of equivalent black carbon (eBC), size resolved particle number concentration, and particle volume (V10) collected at Pier 24 by our mobile AEROLAB, indicated maximum number concentrations to originate in the S-SE area of the port. Black carbon, and to some extent V10, presented secondary maxima from sources nearby the AEROLAB site (located at the port’s center). These are possibly due to local traffic of cargo trains, heavy and light-duty vehicles reaching nearby piers. Particle’s number concentrations were dominated by emissions from the Ro-Ro ferry sectors. The average eBC load recorded at Pier 24 was about half the one measured in downtown Rome, in winter (1.3 µg/m3 vs. 2.6 µg/m3). Conversely, particle number concentrations were almost twice as high (20,400 cm−3 vs. 12,000 cm−3). Some 45%–58% of these average loads originated from the sector 135°–185°. Triangulation with SO2 maxima observed at Arpa Porto indicates this to coincide with the Ro-Ro passenger piers area. Monthly median loads of 2 µg/m3 for eBC, and 50,000 cm−3 for N were observed to proceed from this “hot spot” to Pier 24, while PM10 did not exceed 30 µg/m3.

The combined analysis of particle’s size distributions and eBC allowed allocating some different engine and fuel sources. In particular, four main combustion sources were suggested to determine the average properties of particulate matter emitted in the port: i) medium/low speed diesel engines operating on low-sulfur (still with high contents of HFO elements) marine diesel oil, mostly operating in the Ro-Ro passenger and carriers sectors of the port; ii) medium–high speed diesel engines fitted with EGCS, mainly operating in the cruise ship and cargo sectors of the port; iii) diesel, heavy-duty vehicles operating on road and/or rail; and iv) light duty vehicles operating either on diesel or gasoline engines.

Cruise ships were observed to emit less BC, and less BC per aerosol unit volume than all other ships and ground vehicles operating in the port. In the case of ship emissions, this is possibly attributable to cruise ships employing: i) better fuel quality; ii) EGCS devices; iii) engines operating at optimal regimes, or to a combination of these factors. In the ship vs. road-vehicle comparison, the differences could depend on the higher BC emission factors characterizing road vehicles with respect to ship engines.

All these evidences substantiate the picture of the Civitavecchia port-city boundary (that is, the city area most affected by the port’s activities) as standing below EU limits in terms of regulated pollutants (NOx, SO2, and PM10). The port itself is estimated to contribute some 33% and 43% of the PM10 and NO2 loads at Arpa Porto, with diminishing impact at the city center. Chief origins of port’s NO2 coincide with Ro-Ro passenger and cruise ship sectors. When focusing on currently unregulated pollutants, the port is shown to contribute very high concentrations of ultra-fine particles and high concentrations of black carbon, manly generated in the Ro-Ro passenger and carrier sectors. During sea-breeze conditions (more frequent in summer, when ship traffic is at its maximum) this “hot spot” sits upwind and less than 1 km apart from the city center. An impact onto the city air quality is therefore expected out of these unregulated emissions. This could be mitigated by moving north the relevant piers, and/or by reducing/improving the quality of the emissions. To reduce citizen’s exposure to these pollutants, specific alerts could be issued when meteorological conditions favor the diffusion of port emissions to the city.


We wish to thank the Rome Ports Authority for providing data and information about the Civitavecchia ship traffic. We thank ARPA Lazio for sharing the Civitavecchia air quality data employed in this paper. The April 2016 intensive field campaign was run in the framework of the “Air-Sea-Lab”, CNR Twin Laboratory Project, 2015, and with the logistical support of the Rome Ports Authority.


Aksoyoglu et al., 2016S. Aksoyoglu, U. Baltensperger, A.S.H. PrévôtContribution of ship emissions to the concentration and deposition of air pollutants in EuropeAtmos. Chem. Phys., 16, 1895–1906 (2016), 10.5194/acp-16-1895-2016 View PDFGoogle ScholarAnderson et al., 2015M. Anderson, K. Salo, Å.M. Hallquist, E. FridellCharacterization of particles from a marine engine operating at low loadsAtmos. Env., 101 (2015), pp. 65-71, 10.1016/j.atmosenv.2014.11.009ArticleDownload PDFView Record in ScopusGoogle ScholarAndreae and Gelencsér, 2006M.O. Andreae, A. GelencsérBlack carbon or brown carbon? The nature of light-absorbing carbonaceous aerosolsAtmos. Chem. Phys., 6 (2006), pp. 3131-3148, 10.5194/acp-6-3131-2006 View PDFGoogle ScholarBarregård et al., 2014L. Barregård, E. Fridell, H. WinnesParticle emissions from ships, targeting the environmental sustainability of European shipping: the need for innovation in policy and technologyEPSD Report, 6 (2014), pp. 20-30View Record in ScopusGoogle ScholarBauleo et al., 2018L. Bauleo, S. Bucci, C. Antonucci, R. Sozzi, M. Davoli, F. Forastiere, C. AnconaLong-term exposure to air pollutants from multiple sources and mortality in an industrial area: a cohort studyOccup. Environ. Med. Published Online First, 14 (2018), p. 2018, 10.1136/oemed-2018-105059 View PDFGoogle ScholarBenbrahim-Tallaa et al., 2012L. Benbrahim-Tallaa, R.A. Baan, Y. Grosse, B. Lauby-Secretan, F. El Ghissassi, V. Bouvard, N. Guha, D. Loomis, K. StraifCarcinogenicity of diesel-engine and gasoline-engine exhausts and some nitroarenesLancet, 13 (7) (2012), pp. 663-664ArticleDownload PDFView Record in ScopusGoogle ScholarBond et al., 2013T.C. Bond, S.J. Doherty, D.W. Fahey, P.M. Forster, T. Berntsen, B.J. De AngeloBounding the role of black carbon in the climate system: A scientific assessmentRetrieved fromJ. Geophys. Res. Atmospheres, 118 (11) (2013), pp. 5380-5552, 10.1002/jgrd.50171 View PDFView Record in ScopusGoogle ScholarBuhaug et al., 2009Ø. Buhaug, J.J. Corbett, Ø. Endresen, V. Eyring, J. Faber, S. Hanayama, D.S. Lee, D. Lee, H. Lindstad, A.Z. Markowska, A. Mjelde, D. Nelissen, J. Nilsen, C. Pålsson, J.J. Winebrake, W. Wu, K. Yoshida(2009), Second IMO GHG study 2009Prevention of air pollution from ships, International Maritime Organization (IMO), London (2009)Google ScholarCarslaw et al., 2006D.C. Carslaw, S.D. Beevers, K. Ropkins, M.C. BellDetecting and quantifying aircraft and other on-airport contributions to ambient nitrogen oxides in the vicinity of a large international airport”Atmos. Environ., 40 (28) (2006), pp. 5424-5434ArticleDownload PDFView Record in ScopusGoogle ScholarCarslaw and Ropkins, 2012D.C. Carslaw, K. RopkinsOpenair – an R package for air quality data analysisEnviron. Modell. Software (2012)27–28, pp. 52–61Google ScholarCarslaw and Beevers, 2013D.C. Carslaw, S.D. BeeversCharacterising and understanding emission sources using bivariate polar plots and k-means clusteringEnviron. Modell. Software, 40 (2013), pp. 325-329, 10.1016/j.envsoft.2012.09.005ArticleDownload PDFView Record in ScopusGoogle ScholarCorbett et al., 2007J.J. Corbett, J.J. Winebrake, E.H. Green, P. Kasibhatla, V. Eyring, A. LauerMortality from ship emissions: A global assessmentEnviron. Sci. Technol., 41 (24) (2007), pp. 8512-8518, 10.1021/es071686z View PDFView Record in ScopusGoogle ScholarCasazza et al., 2019M. Casazza, M. Lega, E. Jannelli, M. Minutillo, D. Jaffe, V. Severino, S. Ulgiatic3D monitoring and modelling of air quality for sustainable urban port planning: Review and perspectivesJ. Cleaner Prod., 231 (2019), pp. 1342-1352, 10.1016/j.jclepro.2019.05.257ArticleDownload PDFView Record in ScopusGoogle ScholarCesari et al., 2014D. Cesari, A. Genga, P. Ielpo, M. Siciliano, G. Mascolo, F.M. Grasso, D. ContiniSource apportionment of PM2.5 in the harbor-industrial area of Brindisi (Italy): Identification and estimation of the contribution of in-port ship emissionsSci. Total Environ. (2014), 10.1016/j.scitotenv.2014.08.007 View PDFGoogle ScholarCostabile et al., 2017F. Costabile, F.H. Alas, M. Aufderheide, P. Avino, F. Amato, F. Barnaba, M. Berico, V. Bernardoni, G. Calzolai, S. Canepari, G. Casasanta, S. Ciampichetti, E. Cordelli, A. Di Ianni, L. Di Liberto, M.C. Facchini, D. Frasca, S. Gilardoni, M.G. Grollino, M. Gualtieri, F. Lucarelli, A. Malaguti, M. Manigrasso, M. Montagnoli, S. Nava, E. Padoan, C. Perrino, E. Petralia, I. Petenko, G. Simonetti, G. Tranfo, G. Valli, S. Valentini, R. Vecchi, F. Volpi, G.P. GobbiFirst Results of the “Carbonaceous Aerosol in Rome and Environs (CARE)” Experiment: Beyond Current Standards for PM10Atmosphere, 8 (249) (2017), 10.3390/atmos8120249 View PDFGoogle ScholarCullinane et al., 2014K. CullinaneIntroduction and overview to: Targeting the environmental sustainability of european shipping: The need for innovation in policy and technologyEPSD Report, 6 (2014), pp. 5-16View Record in ScopusGoogle ScholarDeniz et al., 2010C. Deniz, A. Kilic, G. Civkaroglu(2010) Estimation of shipping emissions in Candarli Gulf, TurkeyEnviron. Monitor. Assess, 271 (2010), pp. 219-228 View PDFCrossRefView Record in ScopusGoogle ScholarEEA, 2013EEA, 2013, European Environment Agency, The impact of international shipping on European air quality and climate forcing, Technical report No 4, 2013.Google ScholarEEA, 2016EEA, 2016, EMEP/EEA air pollutant emission inventory guidebook 2016 Technical guidance to prepare national emission inventories, Report No 21/2016, ISSN 1977-8449, 2016.Google ScholarEEA, 2017aEEA, 2017a, Aviation and shipping – impacts on Europe’s environment: Transport and Environment Reporting Mechanism (TERM) report, EEA Report No 22/2017a.Google ScholarEEA, 2017bEEA, 2017b, ‘Emissions of air pollutants from transport’, TERM indicator 003, European Environment Agency (, 2017b.Google ScholarEEA, 2017cEEA, 2017c, Final energy consumption by mode of transport, Indicator Assessment, Prod-ID: IND-113-en, Also known as: TERM 001,, 2017c.Google ScholarEia, 2016EIA, U.S. Energy Information Administration, International Energy Outlook, DOE/EIA-0484 (2016),, 2016.Google ScholarENEL, 2016ENEL, 2016, Dichiarazione ambientale, Impianto termoelettrico Torrevaldaliga Nord Civitavecchia, 2012.Google ScholarEyring et al., 2010V. Eyring, I. Isaksen S.A., T. Berntsen, W. Collins J., J. Corbett J., O. Endresen, R. Grainger G., J. Moldanova, H. Schlager, D. Stevenson S.Transport impacts on atmosphere and climate: ShippingAtmospheric Environment, 44 (2010), pp. 4735-4771, 10.1016/j.atmosenv.2009.04.059ArticleDownload PDFView Record in ScopusGoogle ScholarEyring et al., 2005V. Eyring, H.W. Köhler, J. van Aardenne, A. LauerEmissions from international shippingJ. Geophys Res., 110 (2005), p. D17305Google ScholarFano et al., 2004V. Fano, P. Michelozzi, C. Ancona, A. Capon, F. Forastiere, C.A. PerucciOccupational and environmental exposures and lung cancer in an industrialized area in ItalyOccup. Envir. Med., 61 (2004), pp. 757-763View Record in ScopusGoogle ScholarFano et al., 2006V. Fano, F. Forastiere, P. Papini, V. Tancioni, A. Di Napoli, C.A. PerucciMortalità e ricoveri ospedalieri nell’area industriale di Civitavecchia, anni 1997–2004Epidemiologia e Prevenzione, 30 (2006), p. 221.226Google ScholarGariazzo et al., 2007C. Gariazzo, V. Papaleo, A. Pelliccioni, G. Calori, P. Radice, G. TinarelliApplication of a Lagrangian particle model to assess the impact of harbour, industrial and urban activities on air quality in the Taranto area, ItalyAtmos. Environ., 41 (30) (2007), pp. 6432-6444, 10.1016/j.atmosenv.2007.06.005ArticleDownload PDFView Record in ScopusGoogle ScholarGibson et al., 2013M.D. Gibson, S. Kundu, M. SatishDispersion model evaluation of PM2.5, NOx and SO2 from point and major line sources in Nova Scotia, Canada using AERMOD Gaussian plume air dispersion modelAtmos. Pollut. Res., 4 (2) (2013), pp. 157-167, 10.5094/APR.2013.016ArticleDownload PDFView Record in ScopusGoogle ScholarHarris and Maricq, 2001Stephen J. Harris, M.Matti MaricqSignature size distributions for diesel and gasoline engine exhaust particulate matterJ. Aerosol. Sci., 32 (6) (2001), pp. 749-764, 10.1016/S0021-8502(00)00111-7ArticleDownload PDFView Record in ScopusGoogle ScholarISPRA, 2009ISPRA, 2009, La disaggregazione a livello provinciale dell’inventario nazionale delle emissioni, Report 92, ISBN 978-88-448-0392-6, 2009.Google ScholarISPRA, 2010ISPRA, 2010, Trasporto su strada, Inventario nazionale delle emissioni e disaggregazione provinciale, Report #124, ISBN 978-88-448-0466-4, 2010.Google ScholarISPRA, 2014ISPRA, 2014, Linee guida per le attività di assicurazione/controllo qualità (QA/QC) per le reti di monitoraggio per la qualità dell’aria ambiente, ai sensi del D.Lgs. 155/2010 come modificato dal D.Lgs. 250/2012, Report #108, ISBN: 978-88-448-0647-7, 2014.Google ScholarISPRA, 2018ISPRA, 2018, Italian Emission Inventory 1990-2016, Informative Inventory Report 284, ISBN 978-88-448-0891-4, 2018.Google ScholarICCT, 2016ICCTBlack Carbon Measurement Methods and Emission Factors from Ships, International Council on Clean Transportation, Washington D.C., USA (2016), pp. 1-182View Record in ScopusGoogle ScholarJeong et al., 2017Ju-Hee Jeong, Zang-Ho Shon, Minsung Kang, Sang-Keun Song, Yoo-Keun Kim, Jinsoo Park, Hyunjae KimComparison of source apportionment of PM 2.5 using receptor models in the main hub port city of East Asia: BusanAtmos. Environ., 148 (2017), pp. 115-127, 10.1016/j.atmosenv.2016.10.055ArticleDownload PDFView Record in ScopusGoogle ScholarJežek et al., 2015I. Ježek, T. Katrašnik, D. Westerdahl, G. MočnikBlack carbon, particle number concentration and nitrogen oxide emission factors of random in-use vehicles measured with the on-road chasing methodAtmos. Chem. Phys., 15 (2015), pp. 11011-11026, 10.5194/acp-15-11011-2015 View PDFView Record in ScopusGoogle ScholarKasper et al., 2007A. Kasper, S. Aufdenblatten, A. Forss, M. Mohr, H. BurtscherParticulate emissions from a low-speed marine diesel engineAerosol Sci. Technol., 41 (1) (2007), pp. 24-32, 10.1080/02786820601055392) View PDFView Record in ScopusGoogle ScholarLack and Corbett, 2012D.A. Lack, J.J. CorbettBlack carbon from ships: a review of the effects of ship speed, fuel quality and exhaust gas scrubbingAtmos. Chem. Phys., 12 (9) (2012), pp. 3985-4000 View PDFCrossRefView Record in ScopusGoogle ScholarMaragkogianni et al., 2016A. Maragkogianni, S. Papaefthimiou, C. ZopounidisMitigating shipping emissions in European portsSpringer Briefs Appl. Sci. Technol. (2016), 10.1007/978-3-319-40150-8_1 View PDFGoogle ScholarMoldanova et al., 2013J. Moldanova, E. Fridell, H. Winness, S. Holmin-Fridell, J. Boman, A. Jedynska, V. Tishkova, B. Demirdjian, S. Joulie, H. Bladt, N. Ivleva P., R. NiessnerPhysical and chemical characterisation of PM emissions from two ships operating in European Emission Control AreasAtmos. Meas. Tech., 6 (2013), pp. 3577-3596, 10.5194/amt-6-3577-2013 View PDFView Record in ScopusGoogle ScholarMurena et al., 2018F. Murena, L. Mocerino, F. Quaranta, D. ToscanoImpact on air quality of cruise ship emissions in Naples, ItalyAtmos. Environ., 187 (2018), pp. 70-83, 10.1016/j.atmosenv.2018.05.056ArticleDownload PDFView Record in ScopusGoogle ScholarOeder et al., 2015S. Oeder, T. Kanashova, O. Sippula, S.C. Sapcariu, T. Streibel, J.M. Arteaga-SalasParticulate matter from both heavy fuel oil and diesel fuel shipping emissions show strong biological effects on human lung cells at realistic and comparable in vitro exposure conditionsPLoS ONE, 10 (6) (2015), Article e0126536, 10.1371/journal.pone.0126536 View PDFView Record in ScopusGoogle ScholarPérez et al., 2016Noemí Pérez, Jorge Pey, Cristina Reche, Joaquim Cortés, Andrés Alastuey, Xavier QuerolImpact of harbour emissions on ambient PM 10 and PM 2.5 in Barcelona (Spain): Evidences of secondary aerosol formation within the urban areaSci. Total Environ., 571 (2016), pp. 237-250, 10.1016/j.scitotenv.2016.07.025ArticleDownload PDFView Record in ScopusGoogle ScholarPetzold et al., 2010A. Petzold, E. Weingarner, J. Hasselbach, P. Lauer, C. Kurok, F. FleisherPhysical Properties, Chemical Composition, and Cloud Forming Potential of Particulate Emissions from a Marine Diesel Engine at Various Load ConditionsEnvironmental Science and Technology, 44 (10) (2010), pp. 3800-3805, 10.1021/es903681z View PDFView Record in ScopusGoogle ScholarPsaraftis and Kontovas, 2009H.N. Psaraftis, C.A. KontovasCO2 emission statistics for the world commercial fleetWMU, J. Maritime Affairs, 8 (2009), pp. 1-25View Record in ScopusGoogle ScholarSaraga et al., 2019D.E. Saraga, E.I. Tolis, T. Maggos, C. Vasilakos, J.G. BartzisPM2.5 source apportionment for the port city of Thessaloniki, GreeceSci. Total Environ., 650 (2) (2019), pp. 2337-2354, 10.1016/j.scitotenv.2018.09.25010ArticleDownload PDFView Record in ScopusGoogle ScholarSaxe and Larsen, 2004H. Saxe, T. LarsenAir pollution from ships in three Danish ports, August 2004Atmos. Environ., 38 (24) (2004), pp. 4057-4067, 10.1016/j.atmosenv.2004.03.055ArticleDownload PDFView Record in ScopusGoogle ScholarSofiev et al., 2018Mikhail Sofiev, James J. Winebrake, Lasse Johansson, Edward W. Carr, Marje Prank, Joana Soares, Julius Vira, Rostislav Kouznetsov, Jukka-Pekka Jalkanen, James J. CorbettCleaner fuels for ships provide public health benefits with climate tradeoffsNat Commun, 9 (1) (2018), 10.1038/s41467-017-02774-9 View PDFGoogle ScholarStull, 2017Stull, R., Practical Meteorology: An Algebra-based Survey of Atmospheric Science, Univ. of British Columbia, 940 pages, ISBN 978-0-88865-283-6, 2017.Google ScholarStreibel, 2016T. Streibel, et al.Aerosol emissions of a ship diesel engine operated with diesel fuel or heavy fuel oilRECENTADVANCES IN CHEMISTRYAND THE ENVIRONMENT, Springer (2016), 10.1007/s11356-016-6724-z View PDFGoogle ScholarThunis et al., 2018P. Thunis, B. Degraeuwe, E. Pisoni, M. Trombetti, E. Peduzzi, C.A. Belis, J. Wilson, A. Clappier, E. VignatiPM2.5 source allocation in European cities: A SHERPA modelling studyAtmos. Env., 187 (9) (2018), pp. 3-106, 10.1016/j.atmosenv.2018.05.062 View PDFGoogle ScholarTzannatos, 2010Tzannatos, E., 2010. Ship emissions and their externalities for the port of Piraeus – Greece. Atmospheric Environment, 44, 400-407, doi:10.1016/j.atmosenv.2009.10.024, 2010.Google ScholarUNCTAD, 2017UNCTAD, 2017, United Nations Conference on Trade and Development, Review of Maritime transport, ISBN 978-92-1-112922-9, UNCTAD/RMT/2017, 2017.Google ScholarUNCTAD, 2012UNCTADUnited Nations Conference on Trade and Development, Review of Maritime Transport 2012, UNCTAD/RMT/2012., United Nations (2012)ISBN 978-92-1-112860-4Google ScholarUria-Tellaetxe and Carslaw, 2014I. Uria-Tellaetxe, D.C. CarslawConditional bivariate probability function for source identification”Environ. Modell. Software, 59 (2014), pp. 1-9, 10.1016/j.envsoft.2014.05.002ArticleDownload PDFView Record in ScopusGoogle ScholarViana et al., 2014M. Viana, P. Hammingh, A. Colette, X. Querol, B. Degraeuwe, I. de Vlieger, J. van AardenneImpact of maritime transport emissions on coastal air quality in EuropeAtmos. Environ., 90 (2014), pp. 96-105ArticleDownload PDFView Record in ScopusGoogle ScholarWhall et al., 2002Whall, C., Cooper, D., Archer, K., Twigger, L., Thurston, N., Ockwell, D., McIntyre, A., Ritchie, A., 2002. Quantification of Emissions from Ships Associated with Ship Movements between Ports in the European Community. Final Report 0617702121. Entec, UK Limited, Norwich, UK, 2002.Google ScholarWHO, 2012WHO, Health effects of black carbon, WHO Regional Office for Europe, ISBN: 978 92 890 0265 3, 2012.Google ScholarWHO, 2013WHO, Health risks of air pollution in Europe – HRAPIE project, Recommendations for concentration–response functions for cost–benefit analysis of particulate matter, ozone and nitrogen dioxide, 2013.Google ScholarWinebrake et al., 2009J. Winebrake J., J. Corbett J., E. Green H., A. Lauer, V. EyringMitigating the Health Impacts of Pollution from Oceangoing Shipping: An Assessment of Low-Sulfur Fuel MandatesEnvironmental Science and Technology, 43 (13) (2009), pp. 4776-4782, 10.1021/es803224q View PDFView Record in ScopusGoogle ScholarWinnes and Fridell, 2009H. Winnes, E. FridellParticle emissions from ships: dependence on fuel typeJ. Air Waste Manag. Assoc., 59 (12) (2009), pp. 1391-1398, 10.3155/1047-3289.59.12.1391 View PDFView Record in ScopusGoogle ScholarWMO/GAW, 2016WMO/GAW, Aerosol Measurement Procedures, Guidelines and Recommendations, GAW Report 227, 2016.Google ScholarZavala et al., 2017M. Zavala, L. Molina T., T. Yacovitch I., E. Fortner C., J. Roscioli R., C. Floerchinger, S. Herndon C., C. Kolb E., W. Knighton B., V.H. Paramo, S. Zirath, J.A. Mejia, A. JazcilevichEmission factors of black carbon and co-pollutants from diesel vehicles in Mexico CityAtmos. Chem. Phys.,, 17 (2017), pp. 15293-15305, 10.5194/acp-17-15293-2017 View PDFView Record in ScopusGoogle Scholar

Cited by (23)