3.1.1 Strongly polluted regions: Arabian Gulf, Gulf of Suez and Suez Canal

Compounds with short lifetimes are generally more relevant for total OH reactivity than long-lived species (see also Sect. 3.3). Since emissions from oil and gas production and cities contain high mixing ratios of reactive alkenes, the Arabian Gulf, the Gulf of Suez and the Suez Canal, where such sources abound, showed the highest average OH reactivities (Fig. 1c, d). In these regions, obvious nearby emission sources were cities, industrial complexes, and especially oil/gas production/processing (Fig. 2). The large alkane mixing ratios associated with oil and gas production (Fig. 1e; Bourtsoukidis et al., 2019) were not reflected in an equally large share of total OH reactivity due to the slow reaction rate coefficient of alkanes with the OH radical (Fig. 1d). The average total OH reactivity (± standard deviation) was 12.9±6.2 s⁻¹ (median: 11.2 s⁻¹) over the Arabian Gulf, 13.2±6.9 s⁻¹ (median: 10.8 s⁻¹) over the Suez Canal and 11.6±4.2 s⁻¹ (median: 10.4 s⁻¹) over the Gulf of Suez. There were short periods with higher total OH reactivities (± total uncertainties) up to 32.8±9.6 s⁻¹ over the Arabian Gulf and 26.6±8.2 s⁻¹ over the Suez Canal (see Sect. 3.2). These ranges are comparable to OH reactivities reported from some urban areas, e.g., Beijing, China (Williams et al., 2016), Houston, Texas (Mao et al., 2010), and New York City (Ren et al., 2006), and to OH reactivities calculated from ship-based VOC observations in Houston–Galveston Bay impacted by fossil fuel production (Gilman et al., 2009). The AQABA observations are, however, more than 3 times as high as VOC-attributed OH reactivity calculated from observations downwind of gas and oil production in south Texas (Schade and Roest, 2016) and twice as high as at the Boulder Atmospheric Observatory when impacted by gas wells (Swarthout et al., 2013).

The largest percentage of attributed OH reactivity over the Arabian Gulf was provided by OVOCs (35 % of total, in contrast to 14 % observed over Houston–Galveston Bay (Gilman et al., 2009). This was surprising, as the Arabian Gulf region was expected to be similarly dominated by direct hydrocarbon emissions from the fossil fuel industry as the Houston–Galveston Bay. Although the alkanes and OVOCs had broadly equivalent mixing ratios over the Arabian Gulf (Fig. 1e), the former have generally lower OH reactivity. Conditions in the Arabian Gulf region during summer favor rapid photooxidation of hydrocarbon emissions to OVOCs, due to high temperatures (on average 34.5 ^∘C), high humidity (up to 92 % RH) and strong irradiation. It should also be noted that as the ship passed through the Arabian Gulf, it was only exposed to primary emissions for relatively short periods of time, when located directly downwind of the sources. The more oxidized regional background air was sampled for longer periods. Furthermore, a bias towards OVOCs in the assessment of speciated OH reactivity is also possible in this study because they were well covered by PTR-ToF-MS measurements, while longer-chain or branched alkenes and alkanes, which are components of oil and other fuels (D'Auria et al., 2009; Gueneron et al., 2015), were limited to C₂–C₈ compounds in the AQABA measurements. Nonetheless, considering the whole campaign, the Arabian Gulf was the region with the highest OH reactivity from alkanes (on average ∼1.2 s⁻¹ or ∼9 % of total OH reactivity). An important influence of oil and gas production on total OH reactivity in the Arabian Gulf is therefore evident in our data.

Petroleum extraction facilities and refineries are located at both shores of the Gulf of Suez and Suez Canal (Fig. 2), and wind from both east and west led to a relatively large contribution of alkanes (5 %–6 %; Fig. 1d, e).

By virtue of their double bond, alkenes are more reactive towards OH than alkanes. Indeed, in Saudi Arabian cities, OH reactivity of alkenes has been reported to be twice as high as the reactivity of alkanes, despite higher abundance of the latter (Barletta et al., 2017). Similarly, over the Arabian Gulf, the AQABA campaign results show an equivalent average share of OH reactivity from alkenes and alkanes (both ∼1.2 s⁻¹, which equals ∼9 %), despite an ∼11 times larger average mixing ratio of alkanes. Certainly a fraction of highly reactive alkenes emitted on land will have already been oxidized before the emission-impacted air mass reached the point of observation on the ship. Alkene-attributed OH reactivity was slightly higher in the Suez Canal (1.4 s⁻¹ vs. 1.2 s⁻¹), where the distance to emission sources is generally shorter due to the narrowness of the canal. Additionally, and probably more importantly, alkenes are only minor components in fossil fuel emissions generally. They are emitted from vehicle exhausts (a possible source near the Suez Canal) due to incomplete combustion (Koppmann, 2007). Alkanes, on the other hand, can be so abundant in emissions from oil and gas production that they dominate OH reactivity, as was observed in Colorado, where 60 % of OH reactivity calculated from VOC measurements (not total OH reactivity) was due to alkanes and only 8 % from alkenes (Gilman et al., 2013). In another study at the same site, alkanes contributed 26 %, alkenes 3 %, and aromatics 2 % of total OH reactivity (Swarthout et al., 2013). In contrast to that, alkenes (41 %) dominated over alkanes (23 %) in the OH reactivity calculated from VOC observations over Houston–Galveston Bay (Gilman et al., 2009).

Aromatics are important components of fossil fuels besides alkanes and alkenes (Koppmann, 2007). In Egyptian cities, ambient VOC mixing ratios from traffic have been reported to be dominated by aromatics (Matysik et al., 2010), whereas in Saudi Arabian cities, alkanes were more abundant (Barletta et al., 2017). Consistent with these findings, we observed higher average speciated OH reactivity from aromatics (1.3 s⁻¹) in the Suez Canal (influenced by air from Cairo, roads next to the Suez Canal, a refinery, and a power plant) compared to the Arabian Gulf (0.9 s⁻¹, influenced by oil/gas industry emissions and partly by urban air from Saudi Arabia, Iraq and Kuwait).

A strong influence of maritime traffic in the narrow and highly frequented Suez Canal and Gulf of Suez is evident from the large NO_x contribution to total OH reactivity (∼10 % and ∼12 %, respectively). Here, other vessels and their exhaust plumes passed by closely during waiting periods (before entering the Suez Canal in the northern Gulf of Suez and inside the Suez Canal).

The Arabian Gulf and Suez Canal will be discussed in more detail in case studies in Sect. 3.2.

3.1.2 Seaways influenced by maritime traffic: Red Sea, Gulf of Aden and Gulf of Oman

The Red Sea, the Gulf of Aden, and the Gulf of Oman displayed similar total OH reactivities with medians of 7.9–8.5 s⁻¹ during the AQABA campaign (northern Red Sea: median 8.5 s⁻¹/mean ±  standard deviation 9.1±3.3 s⁻¹, southern Red Sea: 7.9 s${}^{-\mathrm{1}}/\mathrm{8.7}\pm \mathrm{4.2}$ s⁻¹, Gulf of Aden: 8.0 s${}^{-\mathrm{1}}/\mathrm{8.2}\pm \mathrm{3.1}$ s⁻¹, Gulf of Oman: 8.4 s${}^{-\mathrm{1}}/\mathrm{8.4}\pm \mathrm{3.1}$ s⁻¹; also see Table 1 and Fig. 1c, d). These values are comparable to urban observations in the lower range, such as in Helsinki, Finland (Praplan et al., 2017), or Mainz, Germany (Sinha et al., 2008).

The OH reactivity over the Red Sea, Gulf of Aden and Gulf of Oman can largely be attributed to the influence of maritime transport in these busy, relatively constricted seaways. Refineries at the eastern shore of the northern Red Sea (Fig. 2) are not thought to have influenced our measurements due to prevailing wind from the northwest and west.

Among the three regions, the Gulf of Aden was the one with the lowest average total OH reactivity. Likely, this is due to its location in the transition zone to the open northwestern Indian Ocean (see Fig. 1c). Here, air masses were less influenced by pollution than over the Red Sea or Gulf of Oman, as the AQABA cruise took place during Northern Hemisphere summer, when the Intertropical Convergence Zone (ITCZ) is over India and comparably clean air is drawn north over the northwestern Indian Ocean (Ajith Joseph et al., 2018). In accordance with this finding, Bourtsoukidis et al. (2019) showed, by lifetime-variability regression of alkanes, that the Gulf of Aden was the area with the most remote character during AQABA.

In ship exhausts, the highest emission factors (apart from CO₂) have been found for NO_x (Endresen, 2003; Eyring, 2005; Huang et al., 2018). Large ship engines emit an average of 76 kg of NO_x per metric ton of fuel combusted (Eyring, 2005). This is reflected in the OH reactivity share of NO_x, which was in the range of 4 %–9 % in these relatively narrow seaway regions (Table 1). OVOCs accounted for 12 %–21 % of total OH reactivity there, alkenes 4 %–9 %, inorganics (dominated by SO₂ and CO) 4 %–6 %, aromatics 3 %–8 % and alkanes 4 %–5 %. All those classes of compounds can stem from fuel combustion in ships (Endresen, 2003; Eyring, 2005; Huang et al., 2018; Xiao et al., 2018), which emit an average of 43 kg of SO_x, 4.7 kg of CO and 7 kg of hydrocarbons per metric ton of fuel combusted (Eyring, 2005).

The C₃ carbonyls (C₃H₆O) vs. propane (C₃H₈), C₅ oxidation products (C₅H₁₀O) vs. pentanes (C₅H₁₂) and toluene vs. benzene ratios (Table 1) were used here as indicators of the oxidation state (and, thereby, the extent of photochemical aging) of the air. We note that this approach is based on two assumptions: (1) that production of the carbonyls by oxidation dominates over direct emission, although some direct combustion emissions cannot be excluded, and (2) that the only carbonyl precursor is the corresponding C₃ or C₅ compound without significant fragmentation of larger VOCs. The toluene vs. benzene relationship is based on the assumption of simultaneous emission of both compounds. Notably, a distinct difference between the northern (less oxidized, [C₃H₆O]∕[C₃H₈] ≈1.9 and [C₅H₁₀O]∕[C₅H₁₂] ≈ 1.3) and southern Red Sea (oxidized, [C₃H₆O]∕[C₃H₈] ≈9.4 and [C₅H₁₀O]∕[C₅H₁₂] ≈1.9) was observed. Although it did not result in large differences in total OH reactivity values (median of 7.9 s⁻¹ in the photochemically aged air masses in the south Red Sea and 8.5 s⁻¹ in the north Red Sea), it reflects the origin of air masses: urban centers in northeast Africa for the north Red Sea, and the less populated and more distant Sudan and Chad for the south Red Sea.

3.1.3 Cleaner regions: Mediterranean and Arabian Sea

Over the more open seas, notably the Mediterranean and Arabian Sea (northwestern Indian Ocean), the air was aged, cleaner and less reactive to OH. Here, total OH reactivities were mostly close to the detection limit of 5.4 s⁻¹. The median over the Mediterranean was 6.8 s⁻¹ (average ± standard deviation: 7.2±2.9 s⁻¹), which compares to the lower limits of observations from suburban areas such as Whiteface Mountain (Ren et al., 2006), Norfolk, UK (Ingham et al., 2009), central Pennsylvania (Ren et al., 2005), or El Arenosillo at the Spanish coast (Sinha et al., 2012) and more remote locations such as the Rocky Mountains (Nakashima et al., 2014). Total OH reactivity measured over the Arabian Sea during AQABA was, with a median of 4.9 s⁻¹ (average ± standard deviation: 5.6±2.0 s⁻¹), below the detection limit of 5.4 s⁻¹. The marine OH reactivity calculated from trace gas measurements over the central Gulf of Mexico was 1.01 s⁻¹ (Gilman et al., 2009), a value which was slightly exceeded in our observations over the open Arabian Sea with a speciated OH reactivity of 1.5 ±0.2 s⁻¹.

Larger OH reactivity occurred over the Mediterranean Sea during short episodes of polluted air from the mainland in the Strait of Messina (between Sicily and Italy), visible in Fig. 1b. In the Strait of Messina and surroundings, total OH reactivity amounted to between 6.7±3.6 s⁻¹ and 26.7±8.2 s⁻¹ for 5 h, whereas before and after, it was close to the detection limit. OVOCs were with ∼11 %–12 % of total OH reactivity the largest group of identified OH sinks in both the Mediterranean and Arabian seas. Over the Mediterranean, the influence of shipping emissions was larger than over the Arabian Sea, evident from NO_x-attributed OH reactivity, which was on average 0.26 s⁻¹ or ∼4 % of the total in the Mediterranean in contrast to 0.05 s⁻¹ or ∼1 % of the total in the Arabian Sea. Johansson et al. (2017) found that ship traffic over the Arabian Sea causes 3 % of the global NO_x emissions from maritime transport, which explains why even here the conditions were not completely pristine.

Figure 3OH reactivity composition and average ± standard deviation in the Arabian Gulf, separated by origin of the measured air masses: from Iraq and/or Kuwait, from Iran and from southern Arabian Gulf and/or Saudi Arabia. Background: ethane emissions map using data from the EDGAR database (http://edgar.jrc.ec.europa.eu, last access: 31 August 2018, for 2012) as an indicator for the distribution of petroleum extraction and other anthropogenic sources.

3.2.1 Arabian Gulf

Figure 1 shows the average OH reactivity composition for the Arabian Gulf. However, the air measured over the Arabian Gulf originated partly from Kuwait and/or Iraq, partly from Iran and partly from the south. The composition of total OH reactivity between the different regions of origin displayed distinct differences, and can therefore be divided according to origin of the measured air masses, shown in Fig. 3.

Notably, the air from Iraq and/or Kuwait showed obvious influence from oil extraction industries with large fractions of alkanes, alkenes and aromatics (Fig. 3). Here, emission sources were closest to the point of observation, which was reflected in the highest total OH reactivity (median: 18.8 s⁻¹, average ± standard deviation: 18.7±7.9 s⁻¹) and low C₃ carbonyls ∕ propane and high toluene ∕ benzene ratio medians of 0.3 and 0.8, respectively. In contrast, OH reactivity in the air originating from Iran was lower (median: 9.8 s⁻¹, average ± standard deviation: 10.8±3.4 s⁻¹), with a slightly higher C₃ carbonyls ∕ propane and a lower toluene ∕ benzene ratio (both 0.4). The OH reactivity here was similar to that found in air from the south (Saudi Arabia) with a median of 9.8 s⁻¹ and a mean of 9.7±3.3 s⁻¹; however, the C₃ carbonyls ∕ propane ratio of 2.0 points to more oxidized air masses. The high OH reactivity in air originating from Kuwait and Iraq can be attributed to emissions from petrochemical extraction and processing industries that would explain the large contribution of alkanes, alkenes and aromatics. Apart from two notable exceptions, the distribution of OH reactivity did not differ significantly between air from Iran and the south. In air from Iran, the alkene fraction was larger than in air from the south, probably due to closer and different emission sources (gas extraction at the Iranian coast), which would also explain the lower C₃ carbonyls ∕ propane ratio. The NO_x fraction was larger in air from the south, most likely for reasons of ship traffic influence.

Figure 4(a) C₃ carbonyls ∕ propane ratio, C₅ carbonyls ∕ pentane ratio, toluene ∕ benzene ratio, relative humidity, solar irradiation, wind direction, speciated and measured OH reactivity in the Arabian Gulf. Error bars of total OH reactivity display the 1σ total uncertainty of the measurement. (b) Ship track of the campaign in the Arabian Gulf with HYSPLIT air mass back trajectories and numbered labels of case study points. Ethane emissions from the EDGAR database (http://edgar.jrc.ec.europa.eu, last access: 31 August 2018, for 2012) as an indicator for petroleum extraction/processing and other anthropogenic sources are shown on a logarithmic color scale. Emissions out of scale (> 1000 t a⁻¹) are shown with larger circles.

Figure 4a depicts a time series of measurements along the ship track from entering the Arabian Gulf to Kuwait and back. The upper part of the graph shows indicators of the photochemical age of the air mass. The [C₃H₆O]∕[C₃H₈] ratio decreases and the toluene ∕ benzene ratio increases when getting closer to Kuwait, meaning that the research vessel was approaching the emission sources (refineries, oil platforms). Both these indicators show opposite patterns, as expected. The [C₅H₁₀O]∕[C₅H₁₂] ratio follows the [C₃H₆O]∕[C₃H₈] ratio except for an increase towards Kuwait, which shows that the fresh emissions observed here were richer in propane than in pentanes and indicates a mix of fresh emissions with aged air masses. This observation will be discussed in more detail in Sect. 3.4.2. Changes in relative humidity indicate the varying influence of dry desert air and more humid air masses.

The following five case studies (orange numbered labels in Fig. 4 with corresponding air mass back trajectories) will go into more detail regarding the origin of OH reactivity (reported with ±1σ total uncertainty) in the Arabian Gulf.

1. At the entrance to the Arabian Gulf after passing the Strait of Hormuz, total OH reactivity was comparably low (8.0±4.0 s⁻¹). Sampled air had traversed the open Indian Ocean, but passed over the Musandam Peninsula, where urban areas probably contributed some OH reactivity. OVOCs dominated (36±20 % of OH reactivity), with an insignificant unattributed fraction of 38±76 %.

2. At this location in the northern part of the Arabian Gulf, ca. 100 km from the coast of Saudi Arabia, total OH reactivity amounted to 22.1±7.2 s⁻¹, mainly due to OVOCs (57±29 %), with large fractions of alkanes (10±1 %), alkenes (16±8 %) and aromatics (10±5 %). The back trajectories originate from eastern Iraq, with the urban areas of Amara and Basra. The large fraction of OVOCs is an indicator for chemically aged air. The aromatics, alkenes and alkanes were probably emitted from offshore oil/gas extraction platforms (Fig. 4b) and mixed with the oxidized air contaminants coming from a larger distance. Within the uncertainty of the measurement, there is no unattributed OH reactivity at this location; with values of 19.4±9.9 s⁻¹ for the speciated OH reactivity and 22.1±7.2 s⁻¹ for the measured total OH reactivity.

3. Approximately 30 km from the coast of Kuwait, the back trajectories indicate that the sampled air masses traveled over the Rumaila oil field in Basra (Iraq). Consequently, the total OH reactivity of 26.4±8.1 s⁻¹ was mainly due to alkanes from oil extraction (24±2 %). In contrast to the previous case study, significant unattributed OH reactivity was observed here (40±37 %), which might be due to components of oil higher than C₈ and/or branched hydrocarbons that were not measured on board the Kommandor Iona.

4. Just after departure from the port of Kuwait, the highest observed OH reactivity (32.8±9.6 s⁻¹) of the campaign was measured (except for refueling). HYSPLIT back trajectories show that it likely originated from large pollution sources at the coast: refineries (Saudi Aramco refinery and industrial area with storage tanks in Khafji, Saudi Arabia) and oil fields (Wafra oil field in Kuwait; oil platforms at the sea). Notably, alkenes (23±14 %) were a much larger OH sink here than alkanes (5±0.5 %), and the unattributed OH reactivity was zero within the uncertainty.

5. The largest OH reactivity measured during the whole campaign was 303.6±83.9 s⁻¹ during refueling of the vessel in front of the port of Fujairah (United Arab Emirates). A back trajectory would not be relevant here, as these are direct diesel fuel evaporation emissions, mainly composed of alkenes (27±18 % of speciated OH reactivity) and aromatics (21±10 %), mixed with NO_x (39±4 %) from the bunker ship's own fuel combustion. Wu et al. (2015) measured total OH reactivity from gasoline evaporation with the CRM method. In their study, alkenes contributed 40 % of measured OH reactivity, and they found an unattributed OH reactivity fraction of 43 %. In our study, a larger percentage (75±42 %) of total OH reactivity remains unexplained in the fuel evaporation measurement, which is most likely due to the fact that fewer of the higher and branched hydrocarbons were measured.

3.2.2 Suez Canal

The Suez Canal transit on 24 August 2017 was among the episodes of highest total OH reactivities during the whole campaign and will therefore be discussed in more detail here. In August 2017, 49 ships transited the Suez Canal per day (Suez Canal Authority, 2018). It is one of the world's most heavily used shipping lanes and there are no regulations concerning the fuel used or emission limitations. The Suez Canal is 193 km long and at the water surface 313 m wide (Suez Canal Authority, 2017). Given its narrowness, other vessels and their exhausts as well as urban areas or streets were relatively close to the Kommandor Iona while it was traveling along the Suez Canal. Consequently, variations in OH reactivity can mainly be attributed to local emissions. Therefore, back trajectories are usually not helpful for explaining OH reactivity over the Suez Canal.

Figure 5(a) C₃ carbonyls ∕ propane ratio, C₅ carbonyls ∕ pentane ratio, toluene ∕ benzene ratio, relative humidity, solar irradiation, wind direction, speciated and measured OH reactivity in the Suez Canal. Error bars of total OH reactivity display the total uncertainty of the measurement. (b) Ship track of the cruise in the Suez Canal with HYSPLIT air mass back trajectories and numbered labels of case study points. Benzene emissions as an indicator for anthropogenic sources from the EDGAR database (http://edgar.jrc.ec.europa.eu, last access: 31 August 2018, for 2012) are shown on a logarithmic color scale.

Four case studies (orange numbered labels in Fig. 5) will elucidate the origin of OH reactivity (reported with ±1σ total uncertainty) measured over the Suez Canal.

1. This 4 h period of higher OH reactivity (15.2±5.8 to 19.1±6.6 s⁻¹) occurred during a waiting period at the southern entrance to the Suez Canal, in the port area of Suez City. Interestingly, OH reactivity was dominated by alkanes here (∼10 %–15 %). This can be explained by the wind coming from north where directly at the shore a petrol storage area is located (Nasr Petroleum Company, Suez, Egypt). The toluene ∕ benzene ratio was between 1 and 3, indicating fuel combustion emissions as a source in Suez City and/or from ships, rather than biomass burning, which tends to be richer in benzene.

2. The Great Bitter Lake is part of the Suez Canal and is used by ships as a “passing lane”. During the 4 h waiting time in the lake, the highest OH reactivity of the transit was observed (20.6±6.9 to 26.6±8.2 s⁻¹). Alkenes, aromatics and OVOCs contributed almost equally to this value (∼16 %, ∼17 %, and ∼17 %, respectively). In particular, the large amount of aromatics is notable. With 7±1 % contribution, alkanes were less relevant here than in the Suez City port (case study 1). A visible source of nearby emissions at the northern lakeshore (with northerly winds) was a power plant running on natural gas and/or heavy fuel oil (Abu Sultan Power Station, Ismailia, Egypt (Global Energy Observatory, 2015). Its emissions may have been mixed with those from the lakeshore settlements and from bypassing cargo ships, with a toluene ∕ benzene ratio between 1 and 3, characteristic for fuel combustion emissions.

3. This point of observation was low in OH reactivity (5.5±3.2 s⁻¹) as well as in relative humidity (29.2 %). Combined with the back trajectory, this indicates that the air mass sampled here had passed over the Sinai Desert. In this location, the C₃ carbonyls ∕ propane ratio was 20.7, indicating a photochemically aged air mass. For comparison, the ratio was never above 10 anywhere else in the area, and in case studies 1 and 2 even between 0 and 1. Similarly, the C₅ carbonyls ∕ pentanes ratio was elevated (2.6) and the toluene ∕ benzene ratio was low (0.5).

4. The wind direction at case study point 4 was north-northwest. Total OH reactivity was in a 4 h peak of 11.8±5.0 to 18.2±6.4 s⁻¹. Circa 15 % of it was caused by NO_x and ∼9 % by alkenes. The toluene ∕ benzene ratio was ∼1, which indicates a fuel combustion source. There is a highway located west of this northern part of the Suez Canal (Ismailia-Port Said Road), but given the wind direction, exhaust plumes of other vessels sailing in front of the research ship were probably relevant for the OH reactivity as well as traffic on land.

3.4.1 Ozone production regimes

Ozone and OH reactivity are connected via a chemical cycle that involves NO_x and VOCs (Sillman, 1999; Ren et al., 2013): OH oxidizes VOCs to form peroxy radicals, which, in turn, oxidize NO to NO₂. NO₂ thus formed is subsequently photolyzed to result in ozone production. Ozone formation is, however, neither linearly dependent on NO_x concentration nor VOC reactivity, and a decrease in one parameter does not necessarily lead to a decrease in ozone formation (Pusede and Cohen, 2012).

Defining ozone production regimes in terms of the OH reactivities of VOCs and NO_x is a way of assessing the sensitivity of ozone production to the prevailing conditions (Kirchner et al., 2001; Sinha et al., 2012). The method and underlying chemistry have been extensively evaluated for different conditions and compared to other methods in Kirchner et al. (2001). The amount of peroxy radicals produced by VOC oxidation is linked to the OH reactivity of VOCs. The OH reactivity of VOCs was calculated here as the difference of measured total OH reactivity (in 5 min time resolution) and OH reactivity of measured NO_x. Only data in the daytime (06:00–18:00 local time) were considered because the method is inherently limited to daytime chemistry. Figure 9 shows ozone production regime plot examples, where “s” denotes the relative reactivity of OH towards NO_x and VOCs. Based on the analysis of Kirchner et al. (2001), which was developed for urban VOC mixtures, in between the lines labeled s=0.2 and s=0.01, ozone production is favored by a suitable ratio of NO_x and OH reactivity of VOCs, whereas s > 0.2 indicates VOC limitation, and s < 0.01 NO_x limitation. The Arabian Sea with its comparably clean air was a region with partially NO_x-limited ozone production (Fig. 9a). Here, 20 % of the data points do not fall within the regime of ozone production, mostly due to NO_x limitation. This is reflected in relatively low ozone mixing ratios. The Gulf of Suez (Fig. 9b) is characterized in part by a strong VOC limitation associated with high NO_x (69 % of data points indicate an ozone formation regime). This NO_x can be attributed to emissions from other vessels at close proximity while waiting at the entrance of the Suez Canal. The low ozone mixing ratios associated with high NO_x in the fresh combustion emissions (Fig. 9b, c) point to ozone titration, which will be discussed in greater detail in Sect. 3.4.2. In the Suez Canal (Fig. 9c), the data points are also grouped more towards VOC sensitivity whereas the opposite is true for the Arabian Gulf (Fig. 9d). Nevertheless, in both the Suez Canal and the Arabian Gulf, almost all data points (97 % in the Arabian Gulf and 91 % in the Suez Canal) fall into the regime of ozone formation, meaning that a suitable ratio of VOC and NO_x molecules for ozone formation was present nearly all the time. The same is true for the other environments (not shown in graph) with favored ozone formation 97 % of the time in the Mediterranean, 93 % in the northern Red Sea, 88 % in the southern Red Sea, 87 % in the Gulf of Aden and 82 % in the Gulf of Oman.

Figure 9Ozone production regimes for the (a) Arabian Sea, (b) Gulf of Suez (c) Suez Canal and (d) Arabian Gulf. “s” denotes the relative reactivity of OH towards NO_x and VOCs. s > 0.2: VOC limitation; s < 0.01: NO_x limitation of the ozone formation.

Download

The ozone production favoring conditions with both high NO_x- and VOC-attributed OH reactivities may be contributing factors to the high ozone levels observed and modeled above the Arabian Gulf (Lelieveld et al., 2009; Fountoukis et al., 2018). Conditions favorable for ozone formation resulting in ozone mixing ratios up to 14 ppb have been reported from other sites influenced by oil and gas production (Edwards et al., 2014; Wei et al., 2014). Moradzadeh et al. (2019) found that a hydrocarbon processing plant at the shore of the Arabian Gulf impacted ozone levels even at a 300 km distance. Generally, emissions in lower latitudes (e.g., below 30^∘ N as in the Arabian Gulf) lead to more efficient formation of tropospheric ozone than in higher latitudes because the meteorological conditions lead to faster reaction rates and strong convection (Zhang et al., 2016).

3.4.2 OH reactivity/ozone correlations

The relationship between ozone and total OH reactivity during the AQABA campaign is depicted in Fig. 10. Data from most regions display a negative correlation between ozone mixing ratio and total OH reactivity. This is exemplified for the Mediterranean, Suez Canal and Gulf of Suez in Fig. 10a (correlation for Suez Canal: slope $=-\mathrm{0.2}$; r²=0.61). Additionally, there is a large group of data points located at lower OH reactivity (below 10 s⁻¹) and ozone mixing ratios of 50 to 70 ppb.

The reason for the negative correlation seen in parts of the data is displayed in Fig. 10b. Higher OH reactivities in this part of the plot coincide with higher NO_x, stemming from the exhaust plumes of vessels passing close by. In these ship plumes, high levels of VOCs are co-emitted with NO_x. This combination leads to elevated total OH reactivity. At the same time, ozone is depleted by the NO emissions in the fresh ship plumes ($\mathrm{NO}+{\mathrm{O}}_{\mathrm{3}}\to {\mathrm{NO}}_{\mathrm{2}}+{\mathrm{O}}_{\mathrm{2}}$, Sillman, 1999). This effect is seen when an emission source co-emitting NO_x and VOCs is very close (Akimoto, 2016), as ozone values recover rapidly downwind through NO₂ photolysis and mixing. The ozone titration effect has similarly been observed in negative ozone–CO correlations (Parrish et al., 1998).

Figure 10(a) Ozone vs. total OH reactivity for four regions. (b) Same as (a), but with OH reactivity of NO_x as color scale (no data points when NO_x values are missing). (c) Ozone vs. OH reactivity for the Arabian Gulf with the fraction of OVOC in total OH reactivity as color scale. (d) Ozone vs. OH reactivity for the Arabian Gulf with OH reactivity of alkenes as color scale.

Download

The bulk of data points between 50 and 70 ppb of ozone associated with lower OH reactivity and no trend are, on the other hand, more representative of a campaign background/average, uninfluenced by immediate emission sources and with lower NO_x (Fig. 10b). Here, a possible longer-term effect of the co-emission of VOC and NO_x from marine transport and other anthropogenic sources can be seen. As a secondary pollutant and being suppressed by NO, ozone can often be elevated at a distance from the pollution sources (Sudo and Akimoto, 2007). Large quantities of ozone can be produced during long-range transport under sunlight influence (Parrish et al., 1998; Akimoto, 2016; Moradzadeh et al., 2019). Consequently, in the seaways around the Arabian Peninsula, ozone mixing ratios were mostly between 50 and 80 ppb, which is the range of daytime values in polluted Beijing (Williams et al., 2016; Yang et al., 2017).

Over the Arabian Gulf, the same negative relationship between OH reactivity and ozone as in the Suez Canal, Gulf of Suez and Mediterranean existed in those data points where ozone was below 70 ppb. In contrast, at ozone mixing ratios above ca. 80 ppb, no trend or a slightly positive correlation between ozone mixing ratio and OH reactivity can be ascertained (Fig. 10a). Ozone mixing ratios above 70 ppb over the Arabian Gulf were associated with a high fraction of OVOCs (Fig. 10c), indicating more photochemically processed air. Additionally, when OH reactivity was high at the same time as ozone, alkene OH reactivity was also elevated (Fig. 10d). This points to a mixing of fresh emissions (associated with short-lived alkenes) into photochemically aged, oxidized air contaminants (high OVOCs and high ozone) over the Arabian Gulf. Polluted air masses from urban or industrial areas on land were probably oxidized during their transport to the coast/sea, where they were mixed with emissions from oil and gas extraction facilities, as has been discussed in detailed case studies of such occasions in Sect. 3.2.