Characteristics of Trace Metal Elements in Ambient Sub-Micron Particulate Matter in a Coastal Megacity of Northern China Influenced by Shipping Emissions from 2018 to 2022

Abstract

Various shipping emission restrictions have recently been implemented locally and nationally, which might mitigate their impacts on regional air quality, climate change, and human health. In this study, the daily trace metal elements in PM₁ were measured in a coastal megacity in Northern China, from autumn to winter from 2018 to 2022, spanning DECA 1.0 (domestic emission control area), DECA 2.0, IMO 2020, and Pre-OWG Beijing 2022 stages. The trace element changes of V, Ni, Pb, and Zn in PM₁ were analyzed. The concentrations of V declined with shipping emission regulations implemented in 2018–2022 at 3.61 ± 3.01, 1.07 ± 1.04, 0.84 ± 0.62, and 0.68 ± 0.61 ng/m³, respectively, with the V/Ni ratio decreasing at 1.14 ± 0.79, 0.93 ± 1.24, 0.35 ± 0.24, and 0.22 ± 0.18. The V/Ni ratio was dominated by the shipping emissions in the DECA 1.0 stage but has been more affected by the inland sources since DECA 2.0. The V/Ni ratio of local transport air mass was higher than that of long-distance transportation, indicating that some ships were still using high-sulfur fuel oil, especially for the ships 12 nautical miles from the coastline. The multiple linear regression model showed a better fit using V as a tracer for ship emission sources of ambient SO₂ in the DECA 1.0 stage, while the indication effect reduced since DECA 2.0. The V and V/Ni ratios should be carefully used as indicators of ship sources as more vessels will use clean fuels for energy, and the contribution of inland sources to V and Ni will gradually increase.

1. Introduction

Economic development brought about by growing ocean-going trade has negatively impacted urban climate, air quality, and human health [1,2,3,4,5]. SOx, NOx, OC, BC, and heavy metals released from the use of low-grade, high-sulfur fuels by marine vessels pose a serious threat to the atmospheric and marine environments, to the global climate, and to human health [1,2,3]. Particulate emissions from ships affect ocean chemistry and climate change [4], and their metal elements are stored in high concentrations in marine organisms, posing a risk of ingestion by humans and threatening human health [5,6]. In densely populated coastal areas with busy sea transportation, the pollutants emitted from ships impose a heavy burden on the quality of the atmospheric environment and human health [7]. Approximately 70% of ship emissions globally occur within 400 km of the coast, significantly impacting the air quality in and around ports, especially in densely populated urban areas [3,6]. Ship emissions within 12 nautical miles of the coast account for 51.2–56.5% of total ship emissions [7]. Affected by sea and land winds, the concentration of particulate matter in coastal cities increases by 4.7–38.1% on average due to ship emissions [8].

To reduce the impact of air pollution emissions from ships on the air quality of coastal cities, a series of policies have been introduced at home and abroad [9,10,11,12,13,14,15,16]. On 1 January 2016, the Ministry of Transport of China (MOT) determined the Beijing–Tianjin–Hebei region (BTH), the Yangtze River Delta (YRD), and the Pearl River Delta (PRD) as domestic emission control areas (DECA 1.0). It required ships berthed in the control areas to use 0.5% m/m of low-sulfur fuel [17]. On 1 January 2019, the domestic emission control area (DECA 2.0) expanded upon the DECA 1.0 policy, with the emission control area extending about 12 nautical miles along the entire coastline of the country and ships being required to use 0.5% (m/m) sulfur fuel for navigation and berthing [9,10]. On 1 January 2020, the International Maritime Organization (IMO) issued a regulation that the sulfur content of fuel oil used by ships in the global seas shall not exceed 0.5% m/m in the future (IMO 2020) [2] (Figures S1–S3). At the same time, before the 2022 Beijing Winter Olympics (October 2021–January 2022) (Pre-OWG Beijing 2022), the Ministry of Ecology and Environment implemented a series of policies to adjust the industrial, energy, and transportation structure in Tianjin, Hebei, and surrounding areas.

Studies have reported the impact of different ship emission control policies on air quality in coastal cities [17,18,19]. Observations conducted at Jingtang Port in Tangshan of northern China from December 2016 to January 2017 found that ambient SO₂ concentrations decreased rapidly since the DECA 1.0 policy was implemented on 1 January 2017 [20]. Observations at the Gaoqiao Port and Yangshan Port in Shanghai showed that, following the adoption of the DECA policy, the concentration of SO₂ reduction rate in port regions outpaced that in urban areas [21]. Compared with DECA 1.0 (2016–2019), the implementation of DECA 2.0 (2019–now) had a more significant effect on ship-source emission reduction [22,23].

V and Ni are high in emissions from ship fuel combustion and are often used as tracer elements [18,24,25,26,27]. The oceanic Ni and V cycling play essential roles in biogeochemical balance [28,29]. Ship emissions significantly contribute to V and Ni in most East Asian coastal areas, contributing more than 60% to V and 40% to Ni [22]. The observations at Chongming Island in Shanghai showed that the V concentrations in the DECA 2.0 stage significantly reduced between 2015 and 2019 compared with the DECA 1.0 stage. Implementation of DECA 2.0 resulted in a 60–70% reduction in V concentrations [22]. Based on the online monitoring data in port areas of Shanghai for four years (2017–2020), the concentration of V was reduced by 58%, and the concentration of Ni was decreased by 27% after the implementation of DECA 2.0. As the IMO 2020 policy was implemented, the V concentration decreased by 74%, and the Ni concentration decreased by 18% [30]. Liu et al. (2022) [31] observed PM₁ and its metal elements in Qingdao, showing that the V and Ni concentrations decreased by 73.3% and 22.1% in the DECA 2.0 stage compared with the DECA 1.0 stage. In addition to V and Ni, emissions of other heavy traces (Pb, Cd, As, and Zn) would be reduced accordingly [23].

The V/Ni ratio of direct emissions from heavy fuel oil combustion was much higher than those of the thermal power industry and chemical production [31,32,33]. Previous studies have reported the relationship between the dynamic changes in the V/Ni ratio and the impact of ship sources on air quality in inland and coastal sites [30]. Before the implementation of DECA 2.0, the V/Ni ratio in particulate matter in coastal cities such as Shanghai (more than 1.0) and Qingdao (more than 0.7) were higher than those of Zhengzhou (0.34), Xinxiang (0.22) and Taiyuan (0.43), Chengdu (0.41), and other inland cities [31,34,35,36]. However, current research needs more reports on the proportion of metal trace elements in the follow-up policy of DECA 2.0.

Particulate matter is still a major issue in North China in autumn/winter, while the sources of V and Ni in particulate matter in coastal cities are still unclear [31,37]. Qingdao is a crucial coastal megacity with a world-leading port in Northern China [31,38]. Even while facing the challenge of the COVID-19 pandemic, the foreign trade cargo throughput in the Port of Qingdao still steadily increased [39]. In this study, the trace metal elements in sub-micron particulate matter (PM₁) were measured in Qingdao in Northern China in the autumn/winter seasons for four consecutive years (2018–2022), spanning different shipping policy implementation stages (DECA 1.0, DECA 2.0, IMO 2020, and Pre-OWG Beijing 2022). The potential transport and spatial distribution of trace metal elements were analyzed using the airflow backward trajectory method. This study also provides further insights into the future air quality in coastal cities in eastern China under the increasingly stringent emission control from land and marine sources.

2. Materials and Methods

2.1. Sample Collection

2.1.1. Sampling Site and Time

The sampling site (36°05′ N, 120°21′ E) is located on the 7th-floor platform of Qingdao University of Technology (QUT) in Shibei District (71 m above sea level) (Figure 1). The improved six-channel ambient air particle sampler (ZR-3930 multi-channel type, ZhongRui, Inc., Qingdao, China) was used to collect PM₁ samples. One channel was for a 47 mm Teflon filter (Whatman, Marlborough, MA, USA), and another five channels were for quartz filters. The sampling time was in autumn and winter (November to January of the following year) (

Table 1. The observation period under different shipping policies.

Shipping Policy Stage	Implementation Time	Observation Time	Number of Samples
DECA 1.0	1 January 2016–31 December 2018	1 November 2018–31 December 2018	61
DECA 2.0	Since 1 January 2019	1 January 2019–20 January 2019 1 November 2019–31 December 2019	80
IMO 2020	Since 1 January 2020	1 January 2020–20 January 2020 1 November 2020–31 December 2020 1 January 2021–20 January 2021	101
Pre-OWG Beijing 2022	October 2021–January 2022	1 November 2021–31 December 2021 1 January 2022–20 January 2022	81

). The flow rate was 16.7 L/min. The daily sampling period was 09:30 to 08:30 of the next day. After sampling, the filter was stored at −18 °C.

2.1.2. Sample Analysis

The Teflon filter was equilibrated for 24 h (temperature: 20 ± 2 °C; relative humidity: 50 ± 5%) before and after sampling and then weighed (XPE-105, Mettler Toledo, Columbus, OH, USA). The detection and analysis process of metal elements in PM₁ includes Teflon filter pretreatment (digestion, acid removal, and constant volume) and instrument detection. The Teflon filter was cut and placed in the digestion tank, and 6 mL HNO₃ (GR) and 2 mL HF (GR) were added. A microwave digestion apparatus was used to heat to 190 °C for 25 min. Excess acid in the canister was evaporated in a fume hood and brought to a certain volume and fixed to 30 mL with ultrapure water. The concentrations of metal elements (Zn, Pb, Ni, V, and other trace metals) in PM₁ samples were determined by an inductively coupled plasma mass spectrometer (ICP-MS, PerkinElmer, Waltham, MA, USA).

2.1.3. Quality Assurance/Quality Control (QA/QC)

Before sampling, the quartz filter was baked in a muffle furnace at a temperature of 500 °C for 5.5 h. When analyzing the metal elements in the digestion solution, two blank filters were added to the same batch of filter samples to determine whether the samples were contaminated during the digestion and detection. Rhodium was used as the internal standard solution for ICP-MS analysis, and 1% HNO₃ was used to flush the pipeline before detection. In this study, the concentrations of the above metal elements were all within the detectable range, and the correlation coefficients of the standard curves were above 0.999. At the same time, every ten samples were tested and analyzed for quality control samples, and the standard error was required to be less than 5%. The background value of the blank sample was deducted.

2.2. Pollutant and Meteorological Data

The location of the national automatic air-monitoring substation was 618 m away from the sampling site. The hourly pollutant concentrations (PM₁₀, PM_2.5, CO, NO₂, SO₂, and O₃) in the site were obtained from the Qingdao Eco-environment Monitoring Center of Shandong Province. The hourly meteorological data (temperature (T), relative humidity (RH), wind direction (WD), and wind speed (WS)) were obtained from Fulongshan Meteorological Station (120°19′43″ E, 36°04′20″ N), 4.6 km away from the sampling site. The visibility (VIS) came from the Qingdao Liuting Airport Station (https://rp5.ru, accessed on 20 December 2023). The height of the planetary boundary layer (PBL) data were extracted from MeteoInfo 2.2.3.

2.3. Analysis Method

2.3.1. K-Means Clustering

The method assumes that the dataset X contains n data points with d-dimensional attributes, $X=\left\{x_1, x_2, \dots x_n\right\}$, where $x_1\in {\mathrm{R}}^{\mathrm{d}}$. K-means clustering aims to divide the n sample points into the specified K clusters according to the similarity of the samples between the datasets, and each sample only belongs to one of the clusters with the smallest distance from the center point. First, the K-means clustering algorithm randomly generates K cluster center points $C=\{c_1,c_2,\dots c_k\}$, and then calculates the Euclidean distance from each data point to all cluster centers. The data points are assigned to the cluster center points with the smallest distance. Finally, the clustering effect is evaluated by calculating the sum of the squares of the distances between each data point and its cluster center points, even if the following function values are minimized:

$$\sum _{\begin{array}{c}i=1\\ j\in \{\mathrm{1,2},\dots k\}\end{array}}^{n}min {‖x_i-c_j‖}^2$$

(1)

This study performed K-means clustering based on V and Ni concentrations in the DECA 2.0 stage by software SPSS 26.

2.3.2. Enrichment Factor

The enrichment factor (EF) method was used to calculate the enrichment of trace elements in PM₁, which was used to determine whether the PM₁ fraction was more influenced by natural sources or anthropogenic sources [40,41]:

$$\mathrm{EF}={(\mathrm{C}}_{\mathrm{i}}/{\mathrm{C}}_{\mathrm{r}})/{(\mathrm{X}}_{\mathrm{i}}/{\mathrm{X}}_{\mathrm{r}})$$

(2)

where EF is the enrichment factor, C_i is the concentration of trace elements in PM₁, C_r is the concentration of reference trace elements in PM₁, X_i is the crustal abundance value of the measured trace elements, and X_r is the crustal abundance value of the reference trace elements. Al was selected as the reference factor for this study [42]. When the EF is less than 10, it indicates that the trace element is mainly influenced by natural sources from the earth’s crust. When the EF is greater than 10, it suggests that the trace element is more influenced by anthropogenic sources [43,44].

2.3.3. Backward Trajectory and Clustering Analysis

The airflow backward trajectory is the air mass trajectory with a similar spatial distribution that arrives at the research site, and the TrajStat model is used for cluster analysis to explore the long-distance transport of atmospheric particles [31,45]. The MeteoInfo software was adopted with a start time of 8:00 am and corresponding daily PM₁ concentration. The backward pollution trajectory for 48 h was tracked.

2.3.4. Potential Source Contribution Factor Analysis

Potential source contribution function analysis (PSCF) is a conditional probability expressing the grid cell (divided into 0.5° × 0.5° latitude and longitude grids) that the pollutant concentration at the target location exceeds the standard threshold. This study set the 75th percentile of the sample as the potential contribution of a standard threshold [37,46], defined as

$${PSCF}_{ij}=x_{ij}/y_{ij}$$

(3)

y_ij represents the total number of trajectories passing through a grid unit; x_ij is the number of trajectories exceeding the threshold standard in the same grid unit.

As PSCF is a probability value, the smaller the pollution trajectory value (y_ij), the easier it is to cause the PSCF value to fluctuate and increase the uncertainty of the PSCF value. Therefore, a weighting function W_ij is introduced to obtain the revised potential source contribution factor (WPSCF) to reduce the uncertainty of this value [45]:

$${WPSCF}_{ij}={PSCF}_{ij}·W_{ij}$$

(4)

$$W_{ij}=\left\{\begin{array}{c}1.00 \left(80<y_{ij}\right) \\ 0.72 \left(20<y_{ij}<80\right)\\ 0.42 \left(10<y_{ij}<20\right)\\ 0.05 \left(y_{ij}<10\right)\end{array}\right.$$

(5)

3. Results

3.1. Meteorological Parameters and Pollutant Concentrations

Under different shipping policies (DECA 1.0, DECA 2.0, IMO 2020, and Pre-OWG Beijing 2022), the prevailing wind direction was northwest but the area was also affected by easterly and southeasterly winds (Figure 2). The wind speeds under different policies were 3.29 ± 1.68, 3.60 ± 1.53, 3.52 ± 1.54, and 3.44 ± 1.58 m/s, respectively (

Table 2. Meteorological conditions and concentrations of gaseous pollutants and particulate matter during observation period.

Types	DECA 1.0 (n = 61)		DECA 2.0 (n = 80)		IMO 2020 (n = 101)		Pre-OWG Beijing 2022 (n = 81)
	Range	Average	Range	Average	Range	Average	Range	Average
T (°C)	−6.22–16.1	6.50 ± 6.04	−3.79–17.5	5.60 ± 5.60	−10.9–17.3	3.98 ± 5.96	−7.10–17.8	6.24 ± 5.55
VIS 1 (km)	1.03–29.6	13.2 ± 8.08	0.74–28.6	13.8 ± 8.65	1.88–30.0	15.2 ± 8.49	1.50–30.0	15.9 ± 8.98
WS (m/s)	1.13–7.93	3.29 ± 1.68	1.39–7.96	3.60 ± 1.53	1.17–8.77	3.52 ± 1.54	1.28–8.43	3.44 ± 1.58
RH (%)	39.8–93.9	63.7 ± 11.5	29.3–97.3	62.9 ± 14.5	36.4–91.8	60.1 ± 14.6	27.5–91.2	59.2 ± 14.6
SO2 (μg/m3)	2.96–34.9	12.4 ± 7.03	4.00–31.2	13.4 ± 6.08	5.00–39.4	15.1 ± 6.71	4.23–22.0	9.68 ± 3.41
NO2 (μg/m3)	21.8–118	55.8 ± 21.6	14.2–106	58.2 ± 21.8	5.62–115	47.4 ± 20.6	11.8–84.5	45.8 ± 18.2
PM10 (μg/m3)	30.3–493	114 ± 78.7	18.6–375	109 ± 64.4	25.0–325	96.0 ± 59.4	14.5–205	77.5 ± 44.3
PM2.5 (μg/m3)	12.5–186	59.4 ± 41.4	10.6–286	67.1 ± 54.8	7.86–257	60.8 ± 49.2	8.04–161	47.3 ± 32.5

¹ VIS represents visibility.

; Figure S4). The relative humidity differences were minor from DECA 1.0 to Pre-OWG Beijing 2022 stages, which were 63.7 ± 11.5%, 62.9 ± 14.5%, 60.1 ± 14.6%, and 59.2 ± 14.6%, respectively. Compared with the DECA 1.0 stage, the VIS in the Pre-OWG Beijing 2022 stage was improved by 15.2% (DECA 1.0: 13.2 ± 8.08 km, DECA 2.0: 13.83 ± 8.65 km, IMO 2020: 15.2 ± 8.49 km, Pre-OWG Beijing 2022: 15.9 ± 9.0 km).

The PM₁ concentrations during DECA 1.0, DECA 2.0, IMO 2020, and Pre-OWG Beijing 2022 were 37.9 ± 22.2, 42.2 ± 29.3, 41.9 ± 28.8, and 34.3 ± 22.4 μg/m³, respectively (

Table 3. V, Ni, Zn, and Pb in PM₁ and their ratios under different policies.

Types	DECA 1.0		DECA 2.0		IMO 2020		Pre-OWG Beijing 2022
	Range	Mean ± SD	Range	Mean ± SD	Range	Mean ± SD	Range	Mean ± SD
PM1 (μg/m³)	9.36–108	37.9 ± 22.2	7.74–123	42.2 ± 29.3	6.29–154	41.9 ± 28.8	4.99–95.8	34.3 ± 22.4
V (ng/m³)	0.03–11.1	3.45 ± 3.01	0.19–5.39	1.07 ± 1.04	0.00–3.02	0.84 ± 0.62	0.00–3.27	0.68 ± 0.61
Ni (ng/m³)	0.25–10.9	2.86 ± 2.27	0.11–5.83	1.90 ± 1.20	0.71–15.6	2.90 ± 2.41	0.21–13.8	3.57 ± 2.43
Zn (ng/m³)	12.32–298	64.4 ± 50.8	3.75–262	59.6 ± 49.9	3.26–193	64.6 ± 38.4	6.31–323	65.3 ± 46.8
Pb (ng/m³)	3.92–60.7	19.8 ± 13.4	0.96–54.8	16.0 ± 12.3	2.38–68.0	21.0 ± 11.7	2.15–34.8	12.7 ± 6.54
V/Ni	0.05–2.98	1.14 ± 0.79	0.10–7.65	0.93 ± 1.24	0.00–1.43	0.35 ± 0.24	0.00–0.95	0.22 ± 0.18
V/Zn	0.00–0.48	0.07 ± 0.09	0.00–0.48	0.05 ± 0.10	0.00–0.70	0.03 ± 0.08	0.00–0.17	0.02 ± 0.03
V/Pb	0.00–0.90	0.20 ± 0.20	0.01–1.77	0.17 ± 0.35	0.00–0.96	0.07 ± 0.12	0.00–0.50	0.07 ± 0.08

; Figure S5). Among different policies, the observed results in Qingdao were higher than those in coastal cities at home and abroad (Tianjin: 32.4 μg/m³, Venice: 26.2 μg/m³, Bologna: 15.5 μg/m³ and Salento Peninsula: 15 μg/m³) [47,48,49,50]. But they were lower than that of Xi’an (178.2 μg/m³) [51], Wuhan (117.2 μg/m³) [52], and Beijing (78.2 μg/m³) [45].

Compared with the IMO 2020 stage, the PM₁, PM_2.5, and PM₁₀ decreased by 18.1%, 22.2%, and 19.5% in the Pre-OWG Beijing 2022 stage, respectively. The results showed that the emission reduction policy before the Winter Olympics also played a crucial role in reducing the concentration of atmospheric particulate matter in Qingdao. Under the DECA 1.0, DECA 2.0, and IMO 2020 stages, the SO₂ concentration (12.4 ± 7.03, 13.4 ± 6.08, and 15.1 ± 6.71 μg/m³) slightly increased. During the Pre-OWG Beijing 2022 stage, the SO₂ concentration decreased significantly, with a decrease of 35.9% compared with the IMO 2020 stage.

Emissions from burning crude oil as energy for ships contain V, Ni, Zn, and Pb, which were used as tracers of ship sources [37,53,54,55,56]. The V concentrations during DECA 1.0, DECA 2.0, IMO 2020, and Pre-OWG Beijing 2022 were 3.61 ± 3.01, 1.07 ± 1.04, 0.84 ± 0.62, and 0.68 ± 0.61 ng/m³, respectively (Figure 3). In both the DECA 1.0 and DECA 2.0 stages, the V concentrations in PM₁ were lower than the observation results in Shanghai (10.98 and 9.44 ng/m³) [30], which was related to the higher number of ships in Shanghai Port than in Qingdao Port. Compared with the DECA 1.0 stage, V dropped by 70.3% in the DECA 2.0 stage, 73.4% in the IMO 2020 stage, and 80.3% in the Pre-OWG Beijing 2022. The Ni, Zn, and Pb in the DECA 2.0 stage decreased by 33.6%, 7.45%, and 19.2%, respectively. Compared with the DECA 2.0 stage, the Ni, Zn, and Pb in the IMO 2020 stage had increased by 52.6%, 8.39%, and 31.2%, respectively. With the changes in the policy, the concentration of V had decreased significantly. Ni, Pb, and Zn had no significant changes, indicating other potential sources.

Previous studies have shown that using low-sulfur fuel in ships reduces the concentration of V-containing pollutants [20,57]. Changes from heavy fuel to low-sulfur fuel reduce the concentration of V in the atmosphere [20,31]. The DECA 2.0 policy significantly reduced ship emissions in Qingdao and had similar effect compared with the IMO 2020 stage. On the other hand, the fuel oil used by ships in 2019 and 2020 was mainly heavy oil, light oil, or diesel/gasoline, accounting for 95.03% and 94.01%, respectively [58,59]. Compared with 2019, the use of LNG in 2020 increased by 14.23%. From January 2018 to May 2022, the cargo throughput of Qingdao Port showed an overall growth trend, indicating that the number of ships under different shipping policies increased slowly (Figure S6). From 2016 to 2022, the operation of LNG-fueled vessels and the number of planned equipment increased dramatically on a yearly basis (Figure S7). Consistent with the conclusions obtained from the field experiment in the Pearl River Delta [60], V should be carefully considered as a tracer element for ship sources as more vessels will use clean fuels for energy in the future.

3.2. Source Analysis Based on the Ratio Method

The V/Ni, V/Zn, and V/Pb ratios can be used to identify ship emission sources [31,56,61]. When the V/Ni, V/Zn, and V/Pb were less than 0.7, 0.11, and 0.09, the regional ship emission sources had less impact on the regional ambient air [56]. The V/Ni ratios in this research study were 1.14 ± 0.79, 0.93 ± 1.24, 0.35 ± 0.24, and 0.22 ± 0.18 during DECA 1.0, DECA 2.0, IMO 2020 and Pre-OWG Beijing 2022. Compared with the DECA 1.0 stage, the V/Ni ratio dropped by 18.4%, 69.3%, and 80.7% in the DECA 2.0, IMO 2020, and Pre-OWG Beijing 2022 stages. The V/Ni ratio in the DECA 2.0 stage ranged from 0.10 to 7.65, and there is still a high value (Figure 3). The V/Zn and V/Pb ratios were 0.07 ± 0.09 and 0.20 ± 0.20 in the DECA 1.0 stage, higher than those in Shijiazhuang (0.02 and 0.05) [62], Beijing (0.01 and 0.06) [63], Baoding (0.02 and 0.05) [62], and Langfang (0.02 and 0.02) [64]. The V/Zn and V/Pb ratios decreased by 28.6% and 5% in DECA 2.0, 57.1% and 65.0% in IMO 2020, and 71.4% and 65.0% in the Pre-OWG Beijing 2022 stage. After implementing the IMO 2020 policy, the V/Ni, V/Zn, and V/Pb ratios stabilized at a smaller value.

The V/Pb ratios of DECA 1.0, DECA 2.0, IMO 2020, and Pre-OWG Beijing 2022 were 0.20 ± 0.20, 0.19 ± 0.37, 0.07 ± 0.12, and 0.02 ± 0.03 ng/m³, respectively. The V/Pb ratios in the DECA 1.0 and DECA 2.0 stages were higher than those observed by [32] (0.11–0.27 ng/m³, PM_2.5). Compared with the IMO 2020 stage, the Pre-OWG Beijing 2022 stage dropped by 71.4% due to the stringent control and emission reduction in industrial enterprises in BTH and surrounding cities during the preparatory stage for the Winter Olympics.

The correlation coefficients R2 of V and Ni were 0.48, 0.01, 0.18, and 0.32 in DECA 1.0, DECA 2.0, IMO 2020, and Pre-OWG Beijing 2022, respectively. Since the implementation of DECA 2.0, the primary sources of V and Ni have changed considerably [23,31]. Based on domestic and foreign research, the V/Ni ratios of domestic and foreign coastal sites are summarized in

Table 4. Trace elements and their ratios in particulate matter in coastal cities of China (Unit: ng/m³).

Observation Time		Location	Particle Size	V	Ni	Zn	Pb	V/Ni	V/Zn	V/Pb	References
Before DECA 1.0	October 2014–September 2005	Hong Kong	PM1	16	5.2	210	47	3.08	0.08	0.34	[65]
	March 2006–February 2007	Qingdao	PM2.5	10	6	271	128	1.67	0.04	0.08	[44]
	January 2012–December 2012	Yantai	PM2.5	6	4	94.3	91.7	1.50	0.06	0.07	[56]
	September 2013–August 2014	Ningbo	PM2.5	7.36	8.25	190	56.3	0.89	0.04	0.13	[34]
	September 2013–August 2014	Shanghai	PM2.5	16.5	14.9	215	69.7	1.11	0.08	0.24	[34]
	January 2015–December 2015	Tianjin	PM2.5	17.1	10.8	80.5	19.8	1.58	0.21	0.86	[66]
DECA 1.0	January 2016–December 2016	Tangshan	PM2.5	5.6	7.1	186	183	0.79	0.03	0.03	[67]
	January 2017–December 2017	Shanghai	PM2.5	10.98	5.33	/	/	2.06	/	/	[30]
	January 2018–December 2018	Shanghai	PM2.5	9.44	5	/	/	1.89	/	/	[30]
	January 2017–December 2017	Tangshan	PM2.5	3.2	4.6	198	146	0.70	0.02	0.02	[67]
	January 2018–December 2018	Shenzhen	PM2.5	11.6	4.76	/	31	2.44		0.37	[35]
	May 2018–July 2018	Tianjin	PM2.5	2.2	12.1	86	19.4	0.18	0.03	0.11	[47]
	August 2018–May 2019	Qingdao	PM2.5	10.68	22.88	130	355.2	0.47	0.08	0.03	[37]
DECA 2.0	January 2019–December 2019	Shanghai	PM2.5	4.6	3.87	/	/	1.19	/	/	[30]
	January 2018–December 2019	Shanghai	PM2.5	9.57	9.57	31.2	8.64	1	0.31	1.11	[68]
	August 2019	Tangshan	PM2.5	2.3	1.53	41.6	14.2	1.5	0.06	0.16	[62]
IMO 2020	January 2020–December 2020	Shanghai	PM2.5	1.19	3.18	/	/	0.37	/	/	[30]

Observation time of this research spanned from DECA 1.0 to DECA 2.0.

,

Table 5. Trace elements and their ratios in particulate matter in Chinese inland cities (Unit: ng/m³).

Observation Time		Location	Particle Size	V	Ni	Zn	Pb	V/Ni	V/Zn	V/Pb	References
Before DECA 1.0	March 2006–February 2007	Zibo	PM2.5	8.23	10.55	1070	1630	0.78	0.01	0.01	[69]
	November 2011	Chengdu	PM2.5	1.7	2.5	350	172	0.5	0	0.01	[70]
	September 2013–August 2014	Nanjing	PM2.5	9.88	9.3	247	90.9	1.06	0.04	0.11	[34]
	September 2013–August 2014	Hangzhou	PM2.5	15.3	10.4	495	122	1.47	0.03	0.13	[34]
	April 2013–June 2013	Beijing	PM2.5	3.1	2.6	117	90	1.19	0.03	0.03	[51]
	December 2014–February 2015	Baoding	PM2.5	10	40	1170	460	0.25	0.01	0.02	[71]
	March 2014–June 2015	Langfang	PM2.5	10	40	330	140	0.25	0.03	0.07	[71]
DECA 1.0	October 2016–November 2016	Shijiazhuang	PM2.5	8.85	17.7	398	185	0.5	0.02	0.05	[62]
	January 2016–December 2016	Beijing	PM2.5	3.22	3.01	291	53.6	1.07	0.01	0.06	[63]
	September 2016–November 2016	Baoding	PM2.5	8.85	17.7	398	185	0.5	0.02	0.05	[62]
	September 2016–January 2017	Baoding	PM2.5	1.8	6.2	340	192	0.29	0.01	0.01	[64]
	September 2016–January 2017	Shijiazhuang	PM2.5	2.8	6.1	288	124	0.46	0.01	0.02	[64]
	September 2016–January 2017	Langfang	PM2.5	3.6	5.8	228	150	0.62	0.02	0.02	[64]
	January 2016–December 2017	Chengdu	PM2.5	1.24	3.06	357	87.2	0.41	0	0.01	[36]
	January 2016–December 2017	Nanjing	PM2.5	6.05	3.62	199	50.8	1.67	0.03	0.12	[72]
	January 2017	Shijiazhuang	PM2.5	1.1	14.2	212	51.2	0.08	0.01	0.02	[73]
	September 2017–January 2018	Baoding	PM2.5	1.2	4.4	188	80	0.27	0.01	0.02	[64]
	September 2017–January 2018	Langfang	PM2.5	2.1	4.8	123	56	0.44	0.02	0.04	[64]
	October 2017–January 2018	Langfang	PM2.5	4	6	198	68	0.67	0.02	0.06	[74]
	September 2017–January 2018	Shijiazhuang	PM2.5	1.5	4.9	161	59	0.31	0.01	0.03	[64]
	June 2018	Shijiazhuang	PM2.5	1.7	1.26	114	63.8	1.35	0.01	0.03	[75]
	July 2018	Shijiazhuang	PM2.5	0.25	1.1	115	32.6	0.23	0	0.01	[75]
	November 2018–December 2018	Baoding	PM2.5	0.51	3.12	979	279	0.16	0	0	[75]
	January 2018–December 2018	Taiyuan	PM2.5	4.46	10.4	/	63.3	0.43	/	0.07	[35]
DECA 2.0	October 2018–January 2019	Shijiazhuang	PM2.5	1.2	8	183	51	0.15	0.01	0.02	[76]
	December 2018–January 2019	Langfang	PM2.5	0.16	2.2	184	67.5	0.07	0	0	[75]
	January 2019	Baoding	PM2.5	0.7	1.44	123	38	0.49	0.01	0.02	[75]
	March 2019	Baoding	PM2.5	4.66	2.96	698	156	1.57	0.01	0.03	[75]

and

Table 6. Trace elements and their ratio in particulate matter in inland and coastal cities abroad (Unit: ng/m³).

Location	Country	Observation Time	Particle Size	V	Ni	Zn	Pb	V/Ni	V/Zn	V/Pb	References
Salento Island	Italy	July 2008–May 2010	PM1	3	3	/	4	1	0.75	3.00	[50]
Venice	Italy	November 2010–July 2011	PM1	8.0	1.7	22.1	/	4.7	/	1.70	[77]
Venice	Italy	December 2013–February 2014	PM1	2.4	2.5	28	7	0.96	0.34	2.50	[48]
Salento Island	Italy	July 2008–May 2010	PM2.5	4	5	/	7	0.80	0.57	5.00	[50]
Venice	Italy	January 2011–December 2011	PM2.5	7	4	81	12	1.75	0.58	4.00	[78]
Genoa	Italy	January 2011–December 2011	PM2.5	14	7	19	6	2.00	2.33	7.00	[78]
Barcelona	Spain	January 2011–December 2011	PM2.5	6	3	42	6	2.00	1.00	3.00	[68]
Marseille	France	January 2011–December 2012	PM2.5	6	4	24	8	1.50	0.75	4.00	[78]
Busan	South Korea	January 2013–December 2013	PM2.5	8.3	4.4	92	30	1.89	0.28	4.39	[53]
Sao Paulo	Brazil	January 2014–December 2015	PM2.5	2	17	320	44	0.12	0.05	16.67	[79]
Hayashizaki	Japan	January 2016–December 2017	PM2.5	8.96	3.12	25.3	6.98	2.89	1.28	/	[80]
Tarumi	Japan	January 2016–December 2017	PM2.5	10.2	3.35	22.5	7.61	3.04	1.34	3.36	[80]
Suma	Japan	January 2016–December 2017	PM2.5	8.34	2.7	14.3	4.77	3.09	1.75	2.70	[80]
Estarreja	Portugal	September 2019–November 2019	PM2.5	2.13	0.534	3.99	21.2	3.78	0.10	0.56	[81]
Frankfurt	Germany	June 2009–November 2010	PM1	1.2	2.2	/	4.8	0.54	0.25	2.22	[82]
Tehran	Iran	May 2012–May 2013	PM1	3.04	4.1	84.4	54.46	0.74	0.06	4.11	[83]
Frankfurt	Germany	June 2009–November 2010	PM2.5	2.5	5.0	/	13	0.5	0.19	5.00	[82]
London	UK	January 2012–December 2012	PM2.5	1.3	0.5	8.9	2.3	2.60	0.57	0.50	[84]
Tehran	Iran	May 2012–May 2013	PM2.5	4.19	4.91	133.97	72.65	0.85	0.06	4.93	[83]
Makkah	Saudi Arabia	January 2012–December 2014	PM2.5	3.2	20	19	15	0.16	0.21	20.00	[85]

. The V/Ni ratios of particulate matter in coastal cities were similar to the slope of the regression line in the DECA 1.0 stage (Figure 4). After the implementation of DECA 2.0, the V/Ni ratios in inland sites were similar to the slopes of DECA 2.0, IMO 2020, and Pre-OWG Beijing 2022. Since the implementation of DECA 2.0, the V/Ni ratio has decreased due to the steady decline in ship emission sources and it was gradually influenced by inland coal-burning and motor vehicle sources.

K-means clustering of V and Ni concentrations could classify the data in the DECA 2.0 stage into three classes (Type I, Type II, and Type III) (Figure 5). The V/Ni ratios of different linearity were 0.49 ± 0.27, 1.28 ± 0.94, and 5.52 ± 1.90, respectively. This indicated that the DECA 2.0 stage belonged to the transition stage of the DECA 1.0 and the IMO 2020. The linear slope of Type I was 0.11, which was similar to the linear slope of the IMO 2020 and Pre-OWG Beijing 2022 stages. The dominant wind directions of Type I were northerly (wind direction frequency: 36%) and northwesterly (wind direction frequency: 30%), which were less affected by the discharge from ships (Figure 6). The linear slope of Type II (2.06) was much higher than that in DECA 1.0, indicating dominance by ship emissions. The dominant wind direction in Type II was south (frequency of 35%), further suggesting the impact of ship emissions. This further verified that a number of ships still used heavy fuel oil outside the emission control area in the DECA 2.0 stage. The oceanic transport air mass arrived in Qingdao, increasing the V/Ni ratio in the region [61]. The dominant wind direction in Type III was northwesterly, with wind speed of 3.89 ± 2.49 m/s and PBL of 520 ± 293 m. V and Ni concentrations ranged from 0.83 to 1.69 ng/m3 and 0.11 to 0.42 ng/m³, respectively. The V and Ni concentrations were lower than DECA 2.0, but the Ni concentration decreased more. The inland emission reduction policy also impacted the V/Ni ratio and resulted in a high V/Ni ratio [31]. After the implementation of DECA 2.0, the V/Ni ratio should be carefully considered as an indicator for identifying regional pollution from ship sources.

Based on the results of EFs, the uncertainty of the V, Ni, and V/Ni ratio in ship source traceability after implementing the DECA 2.0 policy was further confirmed as the ship source traceability standard. Pb and Zn had EFs above 500 under different policies, indicating the influence of anthropogenic sources (Figure 7). In the DECA 1.0, DECA 2.0, IMO 2020, and Pre-OWG Beijing 2022 stages, the EFs of V were 30.0, 32.6, 8.46, and 4.18; the EFs of Ni were 45.0, 71.2, 41.2, and 32.7, respectively. Before the IMO 2020 stage, V and Ni were more affected by anthropogenic sources, consistent with the results in Jingtang Port in Tangshan [20]. Compared with the DECA 1.0 stage, the V concentration in the Pre-OWG Beijing 2022 stage decreased by 86.1%. After the implementation of IMO 2020, the source of V was affected by both anthropogenic and natural sources.

3.3. Airflow Backward Trajectory

The airflow backward trajectory analysis was carried out for different shipping policy stages, and three trajectories were clustered (Figure 8). Cluster 1 originated in the central part of Shandong Province, a typical local transport, which was also affected by the atmosphere over Bohai Bay during local transmission. Cluster 2 had a moderate-distance transport. The contribution of cluster 2 ranged from 22.8% to 29.4%, mainly originating from Mongolia, central Inner Mongolia, China. The air mass arrived in Qingdao from Hebei Province, Beijing, Tianjin, and other heavily polluted regions and passed through the Bohai Bay. Cluster 3 had long-distance transport, originating from Russia and Mongolia, with relative contribution of air mass from 28.5% to 38.0%.

The V concentrations of Cluster 1 in DECA 1.0, DECA 2.0, IMO 2020, and Pre-OWG Beijing 2022 were 4.01 ± 2.80, 1.50 ± 1.40, 1.07 ± 0.75, and 0.85 ± 0.72 ng/m³, respectively (

Table 7. V, Ni, Zn, and Pb and their ratios in PM₁ under different air mass trajectories and policies.

Cluster Types	Policy Stages	Samples	V (ng/m³)	Ni (ng/m³)	Zn (ng/m³)	Pb (ng/m³)	V/Ni	V/Zn	V/Pb
Cluster 1	DECA 1.0	29	4.01 ± 2.80	3.03 ± 1.78	77.5 ± 60.6	23.3 ± 14.2	1.29 ± 0.73	0.08 ± 0.10	0.21 ± 0.18
	DECA 2.0	22	1.50 ± 1.40	2.13 ± 1.26	63.7 ± 56.2	18.4 ± 15.6	1.17 ± 1.38	0.10 ± 0.16	0.38 ± 0.61
	IMO 2020	46	1.07 ± 0.75	3.55 ± 2.70	71.0 ± 47.2	22.8 ± 14.6	0.36 ± 0.25	0.04 ± 0.11	0.10 ± 0.17
	Pre-OWG Beijing 2022	34	0.85 ± 0.72	4.21 ± 2.42	78.7 ± 58.2	15.0 ± 6.85	0.23 ± 0.21	0.02 ± 0.10	0.08 ± 0.04
Cluster 2	DECA 1.0	17	2.96 ± 3.00	3.70 ± 3.30	58.8 ± 33.5	18.1 ± 9.95	0.86 ± 0.66	0.20 ± 0.24	0.06 ± 0.08
	DECA 2.0	22	1.11 ± 1.04	2.35 ± 1.15	68.4 ± 40.7	20.0 ± 11.9	0.66 ± 0.58	0.02 ± 0.03	0.09 ± 0.11
	IMO 2020	26	0.71 ± 0.43	2.53 ± 1.42	68.2 ± 29.6	22.4 ± 8.82	0.31 ± 0.16	0.01 ± 0.01	0.03 ± 0.02
	Pre-OWG Beijing 2022	14	0.53 ± 0.54	3.73 ± 3.18	64.5 ± 27.4	12.9 ± 5.12	0.17 ± 0.17	0.01 ± 0.06	0.05 ± 0.01
Cluster 3	DECA 1.0	15	2.92 ± 3.35	1.89 ± 1.62	48.2 ± 41.4	15.6 ± 13.8	1.15 ± 0.96	0.18 ± 0.19	0.06 ± 0.08
	DECA 2.0	26	0.68 ± 0.31	1.37 ± 1.00	48.6 ± 51.2	10.6 ± 6.65	0.91 ± 1.45	0.02 ± 0.03	0.10 ± 0.11
	IMO 2020	29	0.61 ± 0.41	2.23 ± 2.41	51.7 ± 26.3	16.9 ± 7.45	0.37 ± 0.27	0.02 ± 0.03	0.05 ± 0.07
	Pre-OWG Beijing 2022	33	0.55 ± 0.47	2.83 ± 1.90	51.8 ± 36.2	10.3 ± 6.04	0.22 ± 0.15	0.01 ± 0.05	0.06 ± 0.01

). Compared with Cluster 1, the V concentrations of Cluster 2 under different shipping policies decreased by 26.2%, 26.0%, 33.6%, and 37.6%, respectively. The V concentrations of Cluster 3 under different shipping policies decreased by 27.2%, 54.7%, 43.0%, and 36.1%, respectively. The V concentrations of Cluster 3 decreased significantly in the DECA 2.0 and IMO 2020 stage compared with the DECA 1.0 stage. The results showed that local ship emission sources contributed more to V.

The long-distance air mass transport in the DECA 2.0 and IMO 2020 stages contributes less to V concentrations in ports such as Tianjin Port and Tangshan Port [20]. The characteristics of V concentration under different airflows (oceanic airflow, inland airflow, and mixed airflow) in Jingtang Port were investigated. The results showed that oceanic airflow significantly impacted coastal areas. The vessel switched fuels, and low-sulfur fuel reduced the vessel-related V by 97.1% [20,60]. The transmission distance was positively correlated with PM₁ concentration. The closer the air mass transport distance, the greater the PM₁ concentration [31]. The Ni, Pb, and Zn concentrations had similar characteristics during the observation period. The changes in the concentration of V in the DECA 1.0 and DECA 2.0 stages did not have the change rules, but the same changes occurred in the IMO 2020 and Pre-OWG Beijing 2022 stages. The transmission of inland air masses, which brought V from industrial sources, might affect the sampling site. The findings by Bi et al. (2019) [86] also showed that V was primarily derived from coal combustion and motor vehicle emissions.

The V/Ni ratios of Cluster 1 to Cluster 3 in the DECA 1.0 stage were 1.29 ± 0.73, 0.86 ± 0.66, and 1.15 ± 0.96, respectively. The V/Ni ratios of Cluster 1 to Cluster 3 in the DECA 2.0 stage were 1.17 ± 1.38, 0.66 ± 0.58, and 0.91 ± 1.45, respectively. Compared with the DECA 2.0 stage, the V/Ni ratio of Cluster 1 to Cluster 3 in IMO 2020 decreased by 69.2%, 53.0%, and 59.3%, respectively. Compared with the DECA 1.0 stage, the V/Ni ratio in Cluster 3 slightly dropped in the DECA 2.0 stage and dramatically dropped in the IMO 2020 stage. In the DECA 2.0 stage, the V/Pb ratio of different air mass trajectories (Cluster 1 to Cluster 3) were 0.38 ± 0.61, 0.09 ± 0.11, and 0.10 ± 0.11, respectively, which were higher than the V/Pb ratio of other shipping policy stages. The V/Ni ratio of Cluster 1 (1.17 ± 1.38) was higher than that of Cluster 3 (0.91 ± 1.45) in the DECA 2.0 stage (Table 7), indicating that the global low-sulfur fuel oil replacement target had not been fully implemented, especially for the ships 12 nautical miles away from the coastline. There were still a number of ships using high-sulfur fuel oil. In addition, there might be regulatory loopholes in the implementation stage of DECA 2.0, and a number of ships used heavy fuel oil and low-sulfur fuel oil alternately offshore. Ships’ evasion tactics have become another obstacle to achieving the desired emissions reductions [87]. Alongside this, the contribution of inland transportation to V and Ni gradually increased, but the influence of ship sources weakened [31].

The PSCF results showed that V might distributed in Qingdao and the surrounding coastal areas in the DECA 1.0 stage. Coastal cities such as Tianjin, Yantai, and Rizhao had higher V and similar potential emission source areas to Ni (Figure 9A,E). The potential emission source area of V reduced in the DECA 2.0 stage, mainly distributed in Qingdao and its coastal regions (Figure 9B). The potential emission areas of V and Ni in the IMO 2020 stage were widely distributed in inland areas (Figure 9C,G). The potential emission area of the Pre-OWG Beijing 2022 stage was significantly reduced, mainly concentrated in Qingdao (Figure 9D,H), though V and Ni were still high in several inland cities. A series of stringent control measures were implemented in BTH and surrounding areas to ensure air quality improvement before and during the Beijing Winter Olympics, which might have significantly reduced the potential emission areas of SO₂, V, and Ni (Figure S9).

3.4. Impact of Ship Emissions on Coastal SO₂

An important effect of the emission reductions in SOx and NOx is the resulting reduction in atmospheric concentrations of PM, especially secondary particulate sulfate and nitrate [88]. High-sulfur heavy sulfur oil burned by ships emits 9.5 million tons > of sulfur oxides into the atmosphere each year [89]. Sofiev et al. (2018) have shown that the global limit on sulfur content in ship fuels decreases concentrations of particulate sulfate by 2–4 µg/m³ in the vicinity of busy ship lanes on a worldwide scale, leading to significant reductions in PM_2.5 [90]. Different trace elements in PM₁ can be used as tracers of pollution sources [91]. Previous studies have shown that As was a tracer element for coal combustion sources, Cr was a tracer element for industrial sources, and V was a tracer element for ship sources [9,37,53,70,91,92]. The relative contribution of each pollution source to SO₂ during the observation period was explored, and the multi-linear regression model was as follows:

$$\left[{\mathrm{S}\mathrm{O}}_2\right]={\beta }_1\times \left[\mathrm{A}\mathrm{s}\right]+{\beta }_2\times \left[\mathrm{C}\mathrm{r}\right]+{\beta }_3\times \left[\mathrm{V}\right]+{\beta }_4$$

(6)

$$\left[{\mathrm{S}\mathrm{O}}_2\right]=761.0\left[\mathrm{A}\mathrm{s}\right]+865\left[\mathrm{C}\mathrm{r}\right]+162\left[\mathrm{V}\right]+5732$$

(7)

$$\left[{\mathrm{S}\mathrm{O}}_2\right]=2529\left[\mathrm{A}\mathrm{s}\right]-1.40\times {10}^{-13}\left[\mathrm{C}\mathrm{r}\right]-1.46\times {10}^{-6}\left[\mathrm{V}\right]+7255$$

(8)

$$\left[{\mathrm{S}\mathrm{O}}_2\right]=1496\left[\mathrm{A}\mathrm{s}\right]+1181\left[\mathrm{C}\mathrm{r}\right]+7.82\times {10}^{-14}\left[\mathrm{V}\right]+5839$$

(9)

$$\left[{\mathrm{S}\mathrm{O}}_2\right]=1101\left[\mathrm{A}\mathrm{s}\right]-8.09\times {10}^{-11}\left[\mathrm{C}\mathrm{r}\right]-7.27\times {10}^{-4}\left[\mathrm{V}\right]+5996$$

(10)

In the equation, SO₂ comes from four sources: coal combustion ([As]), industrial emission ([Cr]), ship emission ([V]), and background concentration (β₄). β₁, β₂, and β₃ were correlation coefficients of linear regression. The equations for the multiple linear regression curves for the different policies are shown, where DECA 1.0 is Equation (7), DECA 2.0 is Equation (8), IMO2020 is Equation (9), and Pre-OWG Beijing 2022 is Equation (10).

The multivariate linear curve in the DECA 1.0 stage showed a better fitting result (Figure 10). The coal-fired source had the most significant influence on SO₂ concentration during the observation period. The contribution of ship sources in the DECA 1.0 stage is 1.8%, much higher than other stages. After the implementation of DECA 2.0, the indication effect of V as a shipping source was reduced. Previous observations conducted in Qingdao showed that coal-burning sources contribute more to urban pollution than mobile and industrial sources in winter [31,37]. SO₂ dropped significantly in the Pre-OWG Beijing 2022 stage, which is related to the strict coal combustion restriction policy for BTH areas before the Beijing Winter Olympics.

4. Conclusions

A four-year autumn and winter observation was carried out in Qingdao, a coastal megacity of northern China, from 2018 to 2022. V, Ni, Pb, and Zn characteristics and their ratios in PM₁ under different shipping policies were investigated. The ship emission source dominated the V/Ni ratio in the DECA 1.0 stage, and the V/Ni ratio was gradually affected by the inland source after implementing DECA 2.0. The indication effect of V as a shipping source was reduced and has been affected by other emission sources since 2019.

The V/Ni ratio of Type I in the DECA 2.0 stage was lower than that of Type II, which was related to the influence of meteorological factors. The high V/Ni ratios were mainly due to transmission from the southeastern marine sources to Qingdao. The high V/Ni ratios due to low Ni concentrations from inland transport can be misleading when using V/Ni ratios to identify ship emission sources.

The V/Ni ratio of the local air mass transmission was higher than that of the long-distance transmission. This indicated that some ships still used high-sulfur fuel oil, especially vessels 12 nautical miles from the coastline during the DECA 2.0 stage. The global low-sulfur fuel oil replacement target had yet to be fully implemented in the DECA 2.0 stage.

Alongside this, the contribution of inland transportation to V and Ni gradually increased. Based on the multiple linear regression model, the DECA 1.0 phase showed a better fit using V as a tracer for ship emission sources to ambient SO₂. After the DECA 2.0 stage, the fitting coefficients for V as a shipping source could be ignored due to the influence of other emission sources.