Integrated Surface and Tropospheric Column Analysis of Sulfur Dioxide Variability at the Lamezia Terme WMO/GAW Regional Station in Calabria, Southern Italy

Abstract

Sulfur dioxide (SO₂) can be of natural and anthropogenic origin and is one of the sulfur compounds present in the atmosphere. Among natural sources, volcanoes contribute with relevant annual outputs, and major eruptions lead to spikes in these outputs. In the case of anthropogenic pollution, SO₂ emissions are mostly correlated with the sulfur content of fuels, which has been the focus of specific emission mitigation policies for decades. Following other examples of cyclic and multi-year evaluations, an analysis of SO₂ at the Lamezia Terme (code: LMT) WMO/GAW (World Meteorological Organization—Global Atmosphere Watch) station in Calabria, Southern Italy, was performed. The coastal site is characterized by wind circulation patterns that result in the detection of air masses with low or enhanced anthropic influences. The presence of the Aeolian Arc of active, quiescent, and extinct volcanoes, as well as Mount Etna in Sicily, may influence LMT observations with diffused SO₂ emissions. For the first time in the history of the LMT, a multi-year analysis of a parameter has been integrated with TROPOMI data gathered by Sentinel-5P and used to test total tropospheric column densities at the LMT itself and select coordinates in the Tyrrhenian and Ionian seas. Surface and satellite data indicate that SO₂ peaks at the LMT are generally linked to winds from the western–seaside wind corridor, a pattern that is compatible with active volcanism in the Tyrrhenian Sea and maritime shipping to and from the Gioia Tauro port located in the same region. The findings of this research provide the basis for enhanced source apportionment, which could further differentiate anthropogenic sources in the area from natural outputs.

1. Introduction

Sulfur dioxide (SO₂), sometimes spelled as sulphur dioxide, is one of the sulfur compounds present in the atmosphere and can be natural or anthropogenic in origin [1,2,3,4].

Among natural sources of SO₂, volcanoes contribute a significant emission rate [5,6,7,8]. The direct impact of volcanic eruptions on SO₂ concentrations in the troposphere and the stratosphere has been documented in the past few decades [9,10,11,12]. The Mt. Pinatubo eruption of June 1991 set a milestone in the implementation of plume-tracking technologies, satellite data, airborne measurements, and global collaborative networks used to monitor SO₂ emissions from the volcano and the consequent diffusion [13,14,15,16,17]. The Hunga Tonga–Hunga Haʻapai eruption of 2022, which is considered the most explosive eruption of the past 140 years, resulted in significant SO₂ output that was promptly tracked with multiple methodologies [18,19].

Wildfires, agricultural fires, and other forms of biomass burning also result in SO₂ emissions, as well as the release of other sulfur compounds, which influence the overall budget on a global scale [20,21,22,23,24,25]. Volcanic and wildfire emissions of SO₂ have a direct impact on the atmosphere’s ERF (effective radiative forcing), in conjunction with aerosols; for this reason, SO₂ is also the subject of evaluations meant for assessing the climate change potential [26,27,28].

Anthropogenic emissions are primarily due to fossil fuels and their sulfur content, whose impact on the environment does not affect all areas of the globe via the same mechanisms [29]. Vehicular traffic has been a major source of anthropogenic SO₂, and regulations—in addition to new technologies—have led to the implementation of low-sulfur fuels, which reduce the output from vehicles [30,31,32,33]. The oil and gas sector is also linked to SO₂ emissions at various stages [34]. Emissions from this sector are such that they pose a health hazard for workers operating in the field of oil and gas [35].

Shipping is among the main sources of anthropogenic SO₂ [36,37,38], and air quality parameters in areas where ports are present are known to be affected by maritime traffic [39]. In the Mediterranean, the Suez Canal is a major hotspot for such activities, and a comparative study between standard pollution levels and the levels during periods of exceptionally low traffic, such as during the COVID-19 pandemic and the Ever Given incident, showed relevant reductions in SO₂ levels [40]. In the past few years, new technologies have been implemented in the effort to model [41], and also mitigate [42], SO₂ emissions related to maritime traffic. The use of low-sulfur fuel in maritime shipping can effectively reduce the degrees of pollution [43,44]. Aviation fuel is also affected by similar issues [45,46,47], and the physical–chemical processes that occur at high altitudes in exhaust plumes have been the focus of scientific research meant to assess the extent of aviation fuel-related pollution [48]. Most SO₂ emissions related to aviation occur during the cruise phase and therefore have an impact over wide regions in a short amount of time [49]. Many techniques have been tested to provide accurate information on the impact of aircraft engine exhausts on the total budget of anthropogenic emissions [50,51]. With the rise in SAF (sustainable aviation fuel), sulfur content has been reduced, and the impacts of such emissions on the environment are partially mitigated [52].

As a pollutant, SO₂ is short-lived: depending on a number of environmental factors, its persistence in the atmosphere may be limited to two days [53,54,55,56]. In the atmosphere, chemical reactions between SO₂ and other compounds occur [57]: following the reaction with water droplets, sulfuric acid (H₂SO₄)—which is responsible for the acid rain phenomenon—is released [58].

The effects of SO₂ on human health are well documented [59]. SO₂ affects the respiratory system [60] and also poses mental health concerns [61,62]. Globally, it causes more than one million premature deaths every year [63] and prolonged exposure can negatively affect pregnancies [64,65,66,67].

The adverse effects of SO₂ have led to global mitigation policies that have managed to reduce the annual rate of anthropogenic emissions [68]. In China, which is characterized by significant SO₂ outputs from transportation, industry, and other activities [69], assessments of PCSO₂ (Per Capita SO₂) emissions [70] have demonstrated that, once applied, policies can effectively reduce SO₂ emissions [71,72]. It is worth noting, however, that the reduction in SO₂ emissions may lead to an increase in the release of other pollutants resulting from the burning process of alternate fuels such as coal [73].

In a work by Altamira-Colado and collaborators [74], a number of patents based on the implementation of UAS (Uncrewed Aerial Systems, commonly referred to as drones)-compatible SO₂ detectors were reviewed, thus highlighting recent advances in environmental monitoring technology and the possibility of pinpointing sources of SO₂ emissions. Overall, monitoring SO₂ concentrations and their possible sources is therefore important in the assessment of anthropogenic and natural pollution in a given area [75,76,77]. This study is aimed at the multi-year characterization of SO₂ variability at the World Meteorological Organization—Global Atmosphere Watch (WMO/GAW) regional station of Lamezia Terme (code: LMT) located in the southern Italian region of Calabria. After methane [78] (2016–2022) and ozone [79] (2015–2023), SO₂ is the third compound subject to a detailed multi-year evaluation at the LMT. Using eight years of continuous surface SO₂ measurements (2016–2023), this study also integrates tropospheric column data obtained by satellite to further assess the sources of emission in the area. This paper is divided as follows: Section 2 described the LMT site, its peculiarities in the context of the Mediterranean basin, and the instruments/methodologies used in this evaluation; Section 3 shows the results of the analysis; and Section 4 and Section 5 discuss and conclude the findings of this paper, respectively.

2. The Lamezia Terme Station and Surface/Tropospheric SO₂ Databases

2.1. The WMO/GAW Regional Coastal Site of Lamezia Terme

Within the WMO/GAW network, the Lamezia Terme (LMT) regional coastal station located in the municipality of Lamezia Terme in Calabria, Southern Italy, gathers continuous data on a number of meteorological and chemical parameters in the atmosphere. Its coordinates are 38.87 °N, 16.23 °E (Figure 1A), and it has been operated by the National Research Council of Italy—Institute of Atmospheric Sciences and Climate (CNR-ISAC)—since 2015. The observation site is located at an elevation of 6 m above sea level, approximately 600 m from the nearby Tyrrhenian coastline of Calabria. The Catanzaro isthmus where the station is located constitutes the narrowest point in the entire Italian peninsula, as the distance between the Tyrrhenian and Ionian coasts is ≈32 km.

In terms of anthropogenic and natural SO₂ emission hotspots, the area where the LMT station is located is characterized by the presence of the Gioia Tauro port (52 km S–SW from the site) [80] and volcanic activities [81] (Figure 1B). The Mount Etna volcano is located in the nearby region of Sicily (160 km S-SW from the LMT) [82,83,84,85]. The Aeolian Arc, which includes the Aeolian Islands and a number of underwater volcanoes [86], many of which have not erupted in historical times, are also close to the LMT. Stromboli (88 km W–SW from LMT) is the nearest active volcano and a well-documented source of SO₂ in the region [87,88]. Vulcano, located in the southernmost part of the Aeolian Islands, has not experienced eruptions in the past few centuries but is subject to SO₂ degassing in the form of fumarole/vents, a phenomenon that also affects Stromboli and Mt. Etna [89,90,91,92].

Previous works based on preliminary data [93] and nearly a decade of LMT observations [78,79] have reported the influence of local wind circulation on the surface concentrations of many parameters. In fact, the site is characterized by two distinct wind axes: a western “seaside” wind corridor normally yielding lower concentrations of pollutants, and a northeastern “continental” corridor that is generally linked to the highest peaks observed at the LMT, although notable exceptions have been reported [78,79,94]. The Lamezia Terme International Airport (IATA: SUF; ICAO: LICA), which is located approximately 3 km north from the WMO/GAW station, has an RWY 10/28 (runway 100–280 °N) orientation, and local air traffic is constantly influenced by the same wind circulation patterns reported at the LMT.

Prior to the LMT’s continuous atmospheric observations, wind patterns were subject to a number of studies. Locally, wind circulation is dominated by breeze regimes, which also regulate the climate across all seasons, and the Marcellinara Gap in the middle of the Catanzaro isthmus is responsible for channeling wind circulation between the Tyrrhenian and Ionian coasts [95,96]. Although seasonal changes in wind circulation occur, the main corridors at low altitudes are oriented on a NE–ENE/W–WSW axis, while NW large-scale circulation is dominant at the 850 hPa layer [95]. Between November and February, daytime circulation is mainly subject to large-scale forcing, and breezes dominate nocturnal flows; in the other periods of the year (spring, summer, and October), daytime breezes are dominated by both local and large-scale flows [96].

Figure 1. (A) Location of Lamezia Terme’s observation site (LMT) in the Mediterranean Basin on NOAA’s ETOPO1 DEM (Digital Elevation Model) [97,98]. (B) EMODnet DEM map [99] showing the location of the LMT in central Calabria and the SO₂ emission hotspots present in the area: the Gioia Tauro Port, Stromboli and Vulcano in the Aeolian Islands, and Mount Etna in Sicily. The main underwater volcanoes in the arc are marked with dark blue labels. “Lametini” refers to the two twin underwater volcanoes (LamN and LamS) named after the municipality of Lamezia Terme. Local maps showing sources of pollution in the Lamezia Terme municipality area with greater detail are available from D’Amico et al. (2024a) [78], (2024b) [100], (2024c) [94].

Figure 1. (A) Location of Lamezia Terme’s observation site (LMT) in the Mediterranean Basin on NOAA’s ETOPO1 DEM (Digital Elevation Model) [97,98]. (B) EMODnet DEM map [99] showing the location of the LMT in central Calabria and the SO₂ emission hotspots present in the area: the Gioia Tauro Port, Stromboli and Vulcano in the Aeolian Islands, and Mount Etna in Sicily. The main underwater volcanoes in the arc are marked with dark blue labels. “Lametini” refers to the two twin underwater volcanoes (LamN and LamS) named after the municipality of Lamezia Terme. Local maps showing sources of pollution in the Lamezia Terme municipality area with greater detail are available from D’Amico et al. (2024a) [78], (2024b) [100], (2024c) [94].

Following the introduction of surface measurements of greenhouse gases (GHGs) and other parameters at the LMT in 2015, additional campaigns targeted the enhanced characterization of local wind circulation at low altitudes. Using a Zephir lidar 300, several wind profiles have been observed at the site at various altitude thresholds, from 10 to 300 m above the ground level [101]. Additionally, based on the results of a short 2009 summertime campaign performed via the implementation of multiple techniques and instruments, two evaluations of PBL (planetary boundary layer) variability at the site were conducted [102,103]. These studies, however, lacked information on additional parameters, which were, however, covered in a longer 2024 campaign on integrated PBL/GHG/aerosol variability that demonstrated the influence of PBLH (PBL height) and wind regime categories on LMT observations [104].

Prior to this research study and the assessment of possible volcanic activity influence on the LMT, multiple local sources of pollution have been mentioned in the past as responsible for some of the peaks observed at the site. Landfills, agricultural and livestock farming, the A2 highway, the SUF/LICA airport, and urban traffic in the town center have been reported in various studies as possible sources of certain peaks [78,93,100]. Specifically, many of the hypotheses on these sources were confirmed in an analysis of the first 2020 COVID-19 lockdown period, during which the restrictions introduced by the Italian government were strict and allowed researchers to better assess a number of pollution sources in a period of exceptionally low anthropic activities [94]. Furthermore, in considering the location of the site in the context of the Mediterranean basin, summer open-fire emissions [105,106] and Saharan dust events [107] have also characterized LMT observations.

2.2. Measurements of Surface Sulfur Dioxide and Meteorological Parameters

SO₂ concentrations at the LMT observation site have been gathered with a Thermo Scientific 32i (Franklin, MA, USA). The 43i model operates under the principle by which SO₂ molecules in ambient air absorb ultraviolet (UV) light, consequently becoming excited, at a specific wavelength and emit additional UV light at a given wavelength during their decay to a lower energy state. Specifically, sample air in the 43i model flows through a hydrocarbon (HC) “kicker” that removes HC from gathered ambient air by forcing these compounds to permeate through the kicker tube wall, while SO₂ molecules are unaffected by the filter. The standard flow rate is 0.5 L per minute. Ambient air is then channeled toward a fluorescence chamber where UV light pulses excite SO₂ molecules. A condensing lens, following these pulses, focuses UV light on the mirror assembly, which contains four selective mirrors that reflect only the wavelength capable of exciting SO₂ molecules. As described above, following the absorption of UV light, SO₂ molecules decay and emit UV at an intensity that is proportional to SO₂ concentration within the chamber. Thanks to a bandpass filter, only the wavelengths emitted by excited SO₂ in sampled air can reach the PMT (photomultiplier tube), which detects UV light emissions attributable only to these decaying molecules. A photodetector located at the back of the fluorescence chamber is configured to monitor UV light pulses and is connected to a compensation circuit that counterbalances fluctuations in UV light. The sampled ambient air moves from the chamber to a flow sensor, followed by a capillary and the exhaust bulkhead. The instrument provides, at a rate of ten measurements per minute (one every six seconds), observed SO₂ concentrations and stores them in a record for further data analysis. The detection limit is <0.5 ppb of SO₂. In this research study, data were aggregated on an hourly, daily, monthly, seasonal, or yearly basis depending on each evaluation.

A Vaisala WXT520 (Vantaa, Finland) instrument was used to gather data on key meteorological parameters (wind speed, wind direction, relative humidity, and air temperature) at 10 m above sea level, with no obstacles nearby at the same elevation. The WXT520 measures temperature in Celsius degrees with a precision of 0.3 °C via an RC oscillator and two reference capacitors. Wind data were obtained using ultrasonic transducers placed on a horizontal plane. Wind speed was measured with a precision of 0.3 m per second, while wind direction had a precision of 3 degrees. Additional details on WXT520 measurements at the LMT are available from D’Amico et al. (2024c) [94]. Data aggregation algorithms were applied to meteorological data depending on the evaluation type (hourly, daily, etc.).

As described in Section 2.1, the LMT observation site is characterized by a daily cycle that results in a clear differentiation between western–seaside and northeastern–continental winds, which in turn result in differences in a number of parameters [78]. Figure 2A shows a wind rose of hourly data gathered between 2016 and 2023, and the presence of two distinct wind corridors is noticeable.

The western wind corridor, which has virtually no obstacles in that direction for hundreds of kilometers, has been used as a baseline to define an offshore cone aimed at satellite data evaluation. The points TYR1–3 located 60 km west from the coast, point TYR4 located 3 km west from the observatory, and point ION1 in the Ionian coast of the Catanzaro isthmus are shown in Figure 2B. These points are considered representative of the airmasses passing at the LMT observatory from both wind corridors. In particular, due to local wind circulation in the isthmus being channeled through the Marcellinara Gap, point ION1 is considered representative of eastern air masses prior to the influence of anthropogenic emissions in continental Italy.

Following other studies based on multi-year evaluations on LMT data [78,79], the four seasons were divided as follows: summer (JJA—June, July, August); fall (SON—September, October, November); winter (JFD—January, February, December); spring (MAM—March, April, May).

Additional details on the data gathered at the LMT during the observation period 2016–2023 are available in

Table 1. Coverage rates, shown as percentages compared to the actual number of hours, of the three datasets used in this study. Both 2016 and 2020 are leap years, with 24 extra hours each.

Year	Hours	SO2 (%)	Meteo (%)	Combined (%)
2016	8784	95.53%	96.34%	93.54%
2017	8760	97%	93.8%	92.42%
2018	8760	80.59%	77.05%	58.21%
2019	8760	40.23%	98.59%	40.21%
2020	8784	93.47%	99.98%	93.46%
2021	8760	70.78%	99.74%	70.77%
2022	8760	66.65%	89.85%	65.34%
2023	8760	51.31%	96.3%	49.92%
Total	70,128 1	74.44% 2	93.95% 2	70.48% 2

¹ Sum of all hours. ² Average rate.

, which reports the coverage rate of Thermo Scientific 43i (“SO₂”) and Vaisala WXT520 (“Meteo”) instruments and their percentages of the total number of hours in each year. The “Combined” dataset, used for data evaluations accounting for both SO₂ concentrations and wind speed/direction, refers to instances of both instruments fully operating at the same time, therefore resulting in overall lower coverage rates. Due to maintenance, the SO₂ dataset was affected by low rates in 2019, 2022, and 2023; however, the high coverage rate of the Meteo dataset throughout the entire study period ensured a minimum additional loss of coverage in the Combined set. Furthermore, wind direction filters were applied to differentiate the western–seaside corridor (240–300 °N) at the observation site of the LMT from the northeastern–continental corridor (0–90 °N).

Data aggregation algorithms were used in R 4.4.0 using the ggplot2, ggpubr, tidyverse, and dplyr libraries and their respective packages. The aggregations were based on the data analyses performed on multi-year methane [78] and ozone [79] concentrations at the same observation site. In this research, evaluations showing SO₂ data alone were based on the SO₂ dataset from Table 1, while those combining SO₂ concentrations with meteorological data were based on the Combined dataset. All hours are in UTC.

2.3. Tropospheric Column Measurements of SO₂

On 13 October 2017, the TROPOspheric Monitoring Instrument (TROPOMI) installed onboard the Sentinel-5P satellite, under the ESA’s Copernicus program, was launched [108]. The satellite and instrument carried out detailed analyses on the SO₂, CH₄ (methane), CO (carbon monoxide), O₃ (ozone), NO₂ (nitrogen dioxide), and HCHO (formaldehyde) concentrations in the atmosphere. The TROPOMI is a remote sensing instrument of the “passive” type, equipped with a hyperspectral nadir sensor of several UV-Vis-NIR spectrometers. Specifically, the advanced instrument relies on four spectrometers set at two bands: SWIR (2305–2385 nanometers, nm); UV and Vis (270–495 nm); and NIR (675–775 nm). The spatial resolution is in the 5–15 km range, with an option for 50 km at wavelengths lower than 300 nanometers [109].

The instrument images a swath for about one second, yielding a spatial resolution at the center of the swath of 7 × 7 km². The results of the TROPOMI scans were processed with an algorithm providing accurate vertical column densities, or VCDs, in molecules per square centimeter (molecules/cm²) of the SO₂, CO, and NO₂ [110]. Study areas were processed with one full scan, with an option for two partial scans; available data were used in this research using level 2 TROPOMI products from 1 January 2020 to 31 December 2023, in accordance with the recommendations and technical documents provided with these products [111]. Column density data in molecules per square meter (molecules/m²) were converted into molecules per square centimeter (molecules/cm²) using the multiplication factor of 6.02214 × 10¹⁹ recommended by the ESA [112].

Downloaded products were processed with an algorithm programmed in MATLAB-R2016a, based on a number of steps that have also been used in a summer campaign on formaldehyde tropospheric column data at the LMT [113]. Specifically, the steps were the following: TROPOMI products were downloaded in netCDF format and parsed through; coordinates and time were extracted from the products and converted; data on the tropospheric gas density were extracted from the matrix and stored in a georeferenced array; the entire dataset was filtered; and all data with a Quality Assurance Value (Q_a) lower than 0.5 were excluded. Q_a, and its related qa_value flag in the dataset, is a data quality value ranging from 0 to 1 that accounts for several factors. As per the recommendations, only data with Q_a >0.5 were used for further evaluation in this study, representative of the following conditions: Solar Zenith Angle (SZA) ≤ 70°; air mass factor > 0.1; cloud radiance fraction at 340 nm <0.5; surface albedo ≤ 0.2; no error flag; no snow/ice warning. In this work, tropospheric column data from the LMT’s coordinates, as well as the coordinates of the TYR1–4 and ION1 points shown in Figure 2B, were selected, and their coverage rate is reported in

Table 2. Coverage rate, shown as a percentage of the total number of days in a given year, of TROPOMI satellite data on SO₂ concentrations; 2020 was a leap year, with one extra day.

Year	Days	Sat. SO2 (%)
2020	366	38.25%
2021	365	34.52%
2022	365	42.73%
2023	365	38.08%
Total	1461 1	38.39% 2

¹ Sum of days. ² Average rate.

. Additional details on the algorithm and its evaluations are available from Barrese et al. (2024) [113].

In order to show the large-scale influence of natural sources such as volcanoes in the area, Level 3 (L3) products were used to generate yearly maps of SO₂ [114]. Specifically, maps referring to the period 2020–2023 were generated based on the COvariance-Based Retrieval Algorithm (COBRA) developed to compensate for a number of artifacts in Differential Optical Absorption Spectroscopy (DOAS) of SO₂ via the retrieval of one specific parameter, which was the SO₂ Slant Column Density (SCD) [115,116].

3. Results

3.1. Daily Cycles of Sulfur Dioxide at LMT

As described in Section 2, the LMT observation site is affected by local wind circulation patterns, which results in peculiar daily cycle for gases and aerosols at the site. The characterization of daily cycles based on multi-year data on methane [78] and ozone [79] showed two distinct behaviors. Figure 3 shows the results of SO₂ daily cycle analysis at the LMT.

Overall, the daily cycle of SO₂ is unlike the cycles observed for methane and ozone, although ozone shows a similar behavior in diurnal peaks [79]. In this case, the peaks are narrowed down in a specific time window (11:00–15:00 UTC). The behavior of SO₂ with respect to the daily cycle at the LMT demonstrates the importance of compound-specific evaluations using multi-year datasets, as the variability of each observed compound at the site possessed certain distinctive properties.

3.2. Bivariate Analysis with Wind Direction and Speed

In integrating the data shown in the wind rose based on 2016–2023 hourly averages (Figure 2), which is thus representative of eight years of local wind circulation at the LMT, a total of four seasonal wind rose scatter plots of SO₂ are shown in Figure 4. These plots, which integrate the daily cycle evaluations seen in 3.1, show SO₂ concentrations from the western–seaside wind corridor at wind speeds higher than their northeastern–continental counterparts. Supplementary Materials S1A–H cover seasonal patterns in each year between 2016 and 2023.

D’Amico et al. (2024a) [78] showed that methane concentrations at the LMT from the northeastern–continental sector follow a hyperbola branch pattern (HBP), with low wind speeds correlated with high concentrations and, vice versa, high wind speeds yielding very low mole fractions. The same pattern was not observed from the western–seaside air corridor. An identical analysis was performed by D’Amico et al. (2024d) [79] on surface ozone and did not show any evidence of a similar pattern. Eight years of continuous surface SO₂ records at the LMT were tested, as shown in Figure 5, with respect to wind speeds and wind corridor. Figure 5A,B refer to the W and NE sectors, respectively, while Figure 5C considers all data, including those falling outside the W and NE filters. Supplementary Materials S2A1–H3 cover the entire 2016–2023 period.

As already reported for the daily cycle, the observed behavior of SO₂ at the LMT observation site is different from that of methane and surface ozone. Specifically, the variability of SO₂ with respect to wind speed at the site from all sectors is not as well defined as that observed for methane [78].

3.3. Evaluation of Weekly Cycles of SO₂ at LMT

Multi-year evaluations of compounds at the LMT have also been conducted on the weekly cycles as possible indicators of natural vs. anthropogenic sources [78,79,100]. In the case of SO₂, considering the implementation of satellite data that were not used in similar research studies on the LMT, weekly cycles were assessed and are shown in Figure 6. Supplementary Materials S3A1–H3 cover surface data at the LMT from each year within the entire 2016–2023 observation period.

The analysis of surface data does not shown specific weekly patterns, with the exception of a number of spikes falling outside weekends. Tropospheric column data are more heterogeneous and show various fluctuations mostly affecting the LMT and ION data. At the LMT, weekly trends and their respective seasonal changes have been observed frequently, especially in the case of greenhouse gases and aerosols; the absence of a well-defined weekly variability pattern is linked to natural sources, which tend to be spread equally over the course of a standard week [100].

3.4. Variability During the Observation Period

After specific evaluations aimed at daily, weekly, and seasonal trends, in conjunction with the cross-analysis with wind data, the multi-year variability of SO₂ at the LMT was assessed, and the results of the surface data are shown in Figure 7. The plots are based on past research studies at the site, which demonstrate the presence of wind sector-dependent concentrations of gases and aerosols, which in turn indicates the presence of multiple sources of pollution [104].

Tropospheric column densities gathered between 2020 and 2023 at the designated LMT, TYR, and ION locations were also analyzed to monitor multi-year variability, and the results are shown in Figure 8. This methodology was not applied to studies on the multi-year variability of other compounds previously assessed at the LMT. Specifically, in Figure 8A, the monthly variability of the total SO₂ column measured by satellite is reported.

To highlight the differences between the two Calabrian coasts and broadly identify sources of SO₂ emissions, in Figure 8B, the monthly means of TCSO₂ (Total Column SO₂) at points TYR4 and ION1 are reported. A comparison between the in situ (surface) SO₂ concentration data and satellite tropospheric column data is shown in Figure 8C. In Figure 8D, points TYR1–3 are analyzed, which are further away from the Tyrrhenian coast (Figure 2), in order to highlight possible SO₂ outputs from maritime shipping and volcanic sources. These points are representative of the western–seaside sector of the LMT, while ION1 is deemed representative of easterly air masses prior to their contact with the continent, which results in increased anthropogenic influence. The plots indicate that TYR1, whose distance to the Stromboli volcano is similar to that of TYR2, is less subject to the transport of air masses enriched in SO₂.

3.5. Statistical Evaluations

The findings of this study were subjected to a statistical evaluation performed using Jamovi v. 2.3. Key surface SO₂ and meteorological data gathered during the entire observation period (2016–2023) were evaluated to determine the mean, median, variance, standard deviation (SD), and quantiles. Pearson’s Correlation Coefficient (PCC) and Spearman’s Rank Correlation Coefficient (SR) were also calculated to test a number of correlations between the chemical and meteorological data [117,118,119].

Regarding the SR, particularly for a sample of size n, the observed variables x and y are converted into ranks (Rx and Ry, respectively); the SR itself ρ is then calculated as described in Equation (1):

ρ = (cov(R_x,R_y))/(σR_x σR_y)

(1)

where “cov” refers to the covariance of both rank variables, and the two values σRx-σRy are the respective standard deviations. The SR ranges between −1 and 1, and its statistical interpretation is similar to that of PCC. The results of the primary statistical evaluation are shown in

Table 3. Main statistical analysis of hourly surface SO₂ and meteorological parameters (RH, relative humidity; T, air temperature; WS, wind speed) in the study period 2016–2023; 1st Q. and 3rd Q. refer to the first and third quartiles, respectively.

	N	Missing	Mean	Median	SD	1st Q.	3rd Q.
RH (%)	66,033	6660	67.442	68.950	11.68	60.890	76.250
T (°C)	66,033	6660	17.784	17.210	6.22	13.380	22.660
SO2 (ppb)	52,267	20,426	0.217	0.056	2.83	0.028	0.142
WS (m/s)	65,895	6798	3.413	2.97	2.07	1.79	4.55

. These refer specifically to surface measurements at the LMT site.

The results of PCC and SR evaluations on surface measurements are shown in

Table 4. Correlation matrix combining surface SO₂ data and key meteorological parameters measured during the observation period 2016–2023. The evaluation shows Pearson’s Correlation Coefficient (PCC) and Spearman’s Rank Correlation Coefficient (SR).

		WS (m/s)	T (°C)	RH (%)
T (°C)	PCC	0.062 ***	–	–
	p-value	<0.001	–	–
	SR	0.091 ***	–	–
	N	65,893	–	–
RH (%)	PCC	−0.282 ***	−0.175 ***	–
	p-value	<0.001	<0.001	–
	SR	−0.371 ***	−0.200 ***	–
	N	65,893	66,031	–
SO2 (ppb)	PCC	0.020 ***	0.001	−0.038 ***
	p-value	<0.001	0.89	<0.001
	SR	0.056 ***	0.096 ***	−0.260 ***
	N	49,485	49,623	49,623

*** p < 0.001.

, which also reports the p-values of each pair. The null hypothesis is that the reported variables do not have a direct correlation; statistically relevant p-values (<0.05) show the presence of a correlation between the variables in the pair. Most of the computed p-values are lower than 0.001; however, a peak of 0.89 is observed for SO₂ and air temperature.

A similar approach was applied to satellite data, which are limited in coverage to the 2020–2023 period. The primary statistical information of the satellite (Tropospheric Vertical Column, TVC) data of SO₂ and surface data limited to 2020–2023 are shown in

Table 5. Main statistical evaluations of surface and tropospheric column SO₂ between 2020 and 2023.

SO2 Type	Mean	Median	SD	1st Q.	3rd Q.
Surface (ppb)	0.19	0.11	0.23	0.082	0.19
Trop. Column (molecules/cm2)	1.24 × 1016	9 × 1015	1.13 × 1016	5.15 × 1015	1.63 × 1016

.

The correlation between the surface and tropospheric column data of SO₂ was tested, and the results are shown in

Table 6. Correlation matrix of surface and tropospheric column SO₂, including Pearson’s Correlation Coefficient (PCC) and Spearman’s Rank Correlation Coefficient (SR).

		Trop. Column (Molecules/cm2)
Surface (ppb)	PCC	0.042
	p-value	0.324
	SR	0.085 *

* p < 0.05.

.

3.6. Multi-Year COBRA Maps of Volcanic SO₂

In addition to punctual data analysis, maps computed using COBRA [115,116] are shown in Figure 9, referring to each year between 2020 and 2023. These maps rely on Level 3 (L3) products by Sentinel-5P to highlight dominant SO₂ tropospheric circulation in the area; volcanic activity, mostly attributable to Mount Etna, yields the most relevant results.

COBRA-derived SO₂ column data for volcanic eruptions were calculated for three different assumed plume height thresholds: 15, 7, and 2.5 km. The advantage of COBRA is its capacity to reveal and pinpoint weak sources of SO₂ that are not normally detected with TROPOMIs; plotted data are reported to be more susceptible to the 7 km altitude threshold. The maps clearly indicate the presence of relevant natural sources of SO₂ in southern Italy, i.e., volcanoes.

4. Discussion

At the Lamezia Terme (LMT) WMO/GAW station, a multi-year and cyclic characterization of sulfur dioxide (SO₂) was performed based on 2016–2023 surface data gathered at the observation site. SO₂ is the third parameter subjected to such a data evaluation, following methane [78] and ozone [79]. Due to SO₂’s heterogeneous sources and the presence of a volcanic arc less than 100 km from the observatory (Figure 1), this study integrated the analysis of TROPOMI data on tropospheric column densities gathered by Sentinel-5P between 2020 and 2023. Specifically, a number of target coordinates were designated as representative of the Tyrrhenian and Ionian coasts of central Calabria, with three points (TYR1–3) specifically selected in proximity of the nearest active volcano, Stromboli (Figure 2). This is the first occurrence of an integrated analysis of surface and satellite data at the LMT based on a medium-term data series.

In the analysis of daily cycles of SO₂ at the site, an early morning peak (greater than 1.2 ppb) is observed between 09:00 and 16:00. Two minor peaks in the 0.3 ppb range match rush hour traffic hours. These findings are in accordance with those of Cristofanelli et al. (2017) [93], who first assessed the influence of vehicular traffic at the site based on the analysis of other pollutants, and those of D’Amico et al. (2024c) [94], who compared 2020 COVID-19 lockdown period data with the pre- and post-lockdown peaks to show the differences caused by exceptionally low anthropic activities. The daily cycle of SO₂ at the LMT is characterized by a 12:00 h peak that matches the peak in sea breezes and is compatible with the transport of air masses from the west. This pattern is similar to that observed for surface ozone at the LMT [79] but is narrowed down to a shorter time window. The influence of solar radiation and, in particular, local wind circulation is well documented in Figure 3C, where the observed peaks are differentiated by wind corridor.

In the context of the Mediterranean basin, the relationship between gases present in the atmosphere, temperatures, and solar radiation has been the subject of research. In the case of ozone [120,121], research studies have found evidence of different patterns between the eastern and western Mediterranean sectors. At the LMT, the complexities in the distribution of gases and aerosols are amplified by the contribution of anthropogenic sources and, with respect to SO₂, by the presence of a volcanic arc and a major volcano in a 160 km radius from the observation site. The Mediterranean is recognized as a hotspot for air quality [122] and an intersection for several distinct air mass transport processes [123,124,125]. For this reason, a precise understanding of the mechanisms driving changes in the observed cycles and trends is necessary.

The correlations between SO₂ concentrations and wind data shown in Figure 4 and Figure 5 also indicate the presence of westerly peaks compatible with possible volcanic and shipping emissions. Many of the SO₂ peaks occur at high wind speeds, which indicate air mass transport from the Tyrrhenian Sea and is likely not related to local emissions in the context of the Calabrian peninsula.

In detail, from the western sectors, SO₂ concentrations reach peaks above 3.4 ppb, both in the case of winds below 5 m/s (breeze regimes) and in synoptic conditions with speeds that can exceed the 8 m/s threshold. From the northeastern sector, winter and fall concentrations are linked to values that exceed the 5 ppb threshold under both breeze regimes and synoptic conditions where wind speeds exceed 7–8 m/s. These patterns affecting cold seasons are compatible with anthropogenic emissions from fossil fuels in nearby urban areas. Wind regimes were found in a 2024 campaign integrating greenhouse gas, aerosol, and planetary boundary layer (PBL) data to influence several parameters at the LMT [104]. The differentiation between two corridors remains one of the main characteristics of the LMT.

Under the assumption that only anthropogenic emissions are subject to weekly cycles, unlike natural emissions, which are not affected [100], weekly assessments of SO₂ concentrations were performed. For the first time in the history of LMT observations, a weekly analysis was also applied to tropospheric column data. In Figure 6, surface fluctuations are linked to winter peaks falling outside weekends, a pattern that may be compatible with anthropogenic emissions, although further investigations would be required. Tropospheric column data reported in Figure 6 are also subject to weekly oscillations, but the TYR points are less affected; this is compatible with natural emissions and could therefore point to volcanic activity (in the form of fumarole and constant gaseous outputs) as responsible for these regular patterns that are not seen at the LMT and ION due to increased anthropogenic influence. Maritime shipping due to the Gioia Tauro port located on the Tyrrhenian coast of Calabria could also be responsible for the TYR and LMT baseline values higher than those of ION.

The multi-year variability of surface SO₂ was evaluated, and the results are shown in Figure 7. For the first time in the history of the LMT, a multi-year evaluation (2020–2023) was also applied to tropospheric column data, and the results are shown in Figure 8.

By differentiating surface SO₂ averages by wind corridor (Figure 7), higher concentrations linked to westerly winds are prominent and are compatible with maritime shipping and volcanic activity. The implementation and evaluation of TROPOMI data (Figure 8), aimed at the TYR1–4 points in the Tyrrhenian Sea and the ION1 point in the Ionian Sea, in conjunction with tropospheric column data on the LMT site itself, constitute a completely new methodology for the assessment of local observations. The LMT’s surface and column data (Figure 8C) are mostly in accordance, peaks included.

A comparison between TYR4 and ION1 was performed to assess coast-to-coast differences in tropospheric column data in the narrowest point of the entire Italian peninsula (≈32 km between the two coasts) (Figure 2), in an area characterized by the presence of the Marcellinara Gap as one of the orographic factors channeling wind circulation between the two seas [95,96]. The graph shows that TYR4, representative of near-shore Tyrrhenian conditions, yields generally higher values compared to ION1, which is its Ionian counterpart. In Figure 8D, the TYR1–3 points—which are further away from the Tyrrhenian coast—are analyzed in detail to highlight contributions from volcanic activity. The results indicate that the three points are not affected in the same way, with the southernmost location (TYR1) generally yielding much lower density values compared to the central (TYR2) and northern (TYR3) points. It is worth mentioning that TYR2 is closer to Stromboli than TYR1 by approximately 1.5 km, especially when considering the volcanic cone and Sciara del Fuoco, which is the southwestern slope of the volcano along which most lava flows occur [87,88]. Due to the characteristics of prevailing large-scale circulation [95] from the west, a Stromboli–TYR2-TYR4-LMT path is compatible with reported observations and TYR2’s higher SO₂ values, while the Stromboli–TYR1-TYR4-LMT path would be less dependent on that. With Vulcano’s activity being mostly limited to fumarole, TYR1’s shorter distance to the second volcano is likely not a key factor.

In addition to the plots and graphs of surface and satellite data, statistical evaluations were performed in Section 3.5 to test possible correlations between a number of parameters. As shown in Table 3 (2016–2023), the mean temperature (17.78 °C) is higher than the median (17.21 °C). This suggests a slight positive skew in the data, with more temperature values falling below the mean. However, the difference still indicates a roughly symmetrical distribution with a slight tendency toward warmer temperatures. The first quartile is 13.38 °C, and the third is 22.66 °C. This indicates that half of the data fall between these temperatures, reflecting the range of conditions experienced during this period. Similarly, SO₂’s mean value is 0.217 ppb with a third quartile threshold of 0.142 ppb, indicating a right skew in the data. This means that the majority of SO₂ measurements are in the low range, but there are sporadic high concentrations driving the mean upward. These conditions are representative of a “multisource” observation site where low background values of a parameter are affected by perturbances.

RH also shows evidence of skewness: the mean is 67.44%, which is slightly lower than the median (68.9%). Overall, there is considerable variability in the data of each parameter: air temperature and RH have moderate standard deviations, while SO₂ and wind speed (WS) show wider spreads, suggesting that these variables fluctuate more.

With respect to statistical correlations, surface data (2016–2023) were evaluated, with the results shown in Table 4. From the analysis, which also relied on Pearson’s Correlation Coefficient (PCC) and Spearman’s Rank Correlation Coefficient (SR), it is possible to determine that there is virtually no linear correlation between air temperature and SO₂ mole fractions measured at the site; in particular, the PCC is close to zero. All p-values in Table 4 are lower than 0.001, except, in fact, for SO₂/T at 0.89. The SR of the relationship is 0.096 with a significance of <0.001, which is still insufficient to indicate a solid linear correlation. A weak but statistically relevant correlation is observed between SO₂ and RH, especially for the SR (−0.26), while the PPC is very low at 0.038. Overall, no clear correlations emerge in the evaluation; the absence of a correlation between SO₂ and air temperature has environmental implications due to the temperature’s variability between boreal warm and cold seasons.

Possible correlations between surface and tropospheric column SO₂ (Table 5), limited to the 2020–2023 period, are evaluated in Table 6. In the correlation matrix, both PCC (0.042) and the SR (0.085) indicate a weak correlation between surface and TVC data, with the SR showing a statistically significant p-value below the 0.5 threshold. These results have a number of possible explanations: Bias errors in satellite data processing algorithms may have had a role; however, it is possible to justify this with the coexistence of near-surface anthropogenic pollution such as maritime shipping and volcanic outputs, which are subject to different transport mechanisms. The outputs of volcanic eruptions from the Aeolian Arc, as well as Mount Etna, reach high altitudes and are therefore subject to transport over a large area: if these outputs pass above the LMT at sufficiently high altitudes, peaks in the TVC data will lack a surface counterpart.

The findings based on the implementation of the TYR and ION locations were further evaluated via satellite products on SO₂ emissions over a large area. TROPOMI products are used in the literature to provide estimates on SO₂ emissions from a range of sources [126]; however, they may have been subject to artifacts. With the COvariance-Based Retrieval Algorithm (COBRA), Level 3 products can be used to generate maps that partially compensate for some of the issues affecting regular evaluations [115,116]. In Figure 9, maps showing the multi-year variability of SO₂ column number densities provide a large-scale picture of natural SO₂ emissions in the area. In fact, large-scale wind circulation has a dominant westerly nature [95], and this leads to prevailing transport toward the east of volcanic outputs such as those attributable to Mount Etna in Sicily.

Overall, the findings described in this study set up new fundaments and methodologies upon which future assessments of SO₂ in the area could be performed to discriminate natural and anthropogenic sources. Additional methodologies such as the carbon isotope fractionation of carbon dioxide (δ¹³C-CO₂) [127], which is also a product of volcanic activity, could be used in source apportionment: the characteristic isotopic fingerprint of volcanic carbon dioxide (CO₂), paired with surface observations of SO₂, could differentiate volcanic and anthropogenic sources [128,129,130,131] and thus provide a clear understanding of volcanic influence over LMT observations. With the PSCF (Potential Source Contribution Function) [132,133,134] and/or CWT (Concentration Weighted Trajectory) [135,136,137,138], case studies could also be subject to further research to provide additional details and insights on the contribution of natural and anthropogenic SO₂ emissions to LMT measurements.

5. Conclusions

Through the implementation of new methodologies that have never been applied to similar studies at the same observation site, integrated surface and tropospheric column data on sulfur dioxide (SO₂) were evaluated at the Lamezia Terme (code: LMT) WMO/GAW station in Calabria, Southern Italy.

Surface (2016–2023) and satellite (2020–2023) data provided new evidence of possible volcanic activity influence over LMT measurements due to the presence of the Aeolian Arc and Mount Etna in a 160 km radius from the station. Surface measurements, in conjunction with satellite data on select coordinates within the Tyrrhenian and Ionian seas, indicate that SO₂ peaks are linked to westerly winds compatible with maritime shipping and volcanic pollution. The target coordinates selected for satellite data evaluations are based on local wind circulation, which is known to be characterized by a western–seaside sector and a northeastern–continental sector. Local wind circulation and air mass transport is subject to winds alternating between these sectors. The findings seem to indicate that volcanic outputs from the Aeolian Arc are subject to westerly winds, which may drive air mass transport toward the LMT observation site. Maps generated from cumulative yearly Level 3 satellite observations—processed by COBRA and meant to compensate for TROPOMI artifacts—also show the presence of multiple sources of the SO₂ in the area and a dominating westerly flow that contributes to the transport of volcanic outputs, especially those attributable to Mount Etna in Sicily.

SO₂ peaks linked to westerly winds from the Tyrrhenian Sea are in contrast with other observed trends at the same site, where northeastern–continental winds are generally enriched in pollutants, while western–seaside winds generally yield low concentrations of compounds attributable to anthropic activity. SO₂ variability at the site is also distinct from that of surface ozone, a compound that shares similar patterns but shows different seasonal and diurnal trends. At the LMT, higher SO₂ concentrations are also observed in the case of synoptic conditions where wind speeds exceed the thresholds of 7–8 m/s, thus indicating the influence of air mass transport from other locations.

This research study sets new fundaments upon which future source apportionment efforts could be based on. Through the implementation of additional methodologies, the influence of volcanic activity on LMT measurements could be assessed with greater detail, thus providing new insights on the role played by active volcanoes in air mass transport within the context of a climatic hotspot such as the Mediterranean basin. Furthermore, this study shows the importance of performing multi-year evaluations of specific compounds to assess their cycles, variability, and trends with enhanced detail.