Abstract
Urban environments function as major hotspots for anthropogenic greenhouse gas emissions, yet quantifying the contribution of specific sources is still a challenge due to the complex interplay between emissions, meteorology, and biospheric activity. Continuous high-frequency monitoring of atmospheric CO2 concentration and stable isotope composition provides critical insight into the sources and processes governing urban carbon dynamics. In this study, airborne CO2 concentration, δ13C-CO2, and δ18O-CO2 were measured at 5-min intervals at the ACO-Pa1 station in Palermo (Italy) from November 2023 to September 2025, together with a comprehensive set of meteorological variables. The dataset reveals that CO2 concentrations exceeded the global atmospheric background by approximately 4.5%. Probability plot analysis showed four distinct CO2 populations, highlighting urban CO2 enhancement and revealing some pollution episodes. Carbon isotope data show systematic 13C-depletion during elevated CO2 conditions, and Keeling plot analysis shows that fossil fuel and natural gas combustion represent the dominant forcing sources, with episodic contributions from landfill gas under specific atmospheric conditions. Principal component analysis demonstrated that while δ13C-CO2 and CO2 concentration are strongly coupled to anthropogenic emissions and seasonal house-heating demands (explaining 30.5% of the dataset variance), the δ18O-CO2 signal behaves as a rather independent tracer linked to biospheric exchange and hydrological processes in the urban area of Palermo. Seasonal patterns are marked by higher CO2 concentrations and stronger 13C-depletion during fall and winter, which is consistent with enhanced residential heating and reduced biospheric uptake, while the spring–summer periods show lower concentrations and isotopic signature driven by photosynthetic carbon fixation. Oxygen isotope variability displays a pronounced seasonal cycle that is partly decoupled from δ13C-CO2, reflecting biosphere–atmosphere oxygen exchange. These results prove that integrated monitoring of the concentration and stable isotopes is a powerful tool for resolving urban CO2 sources and processes, providing a robust observational framework for evaluating emission mitigation strategies in Mediterranean urban environments.
1 Introduction
The Earth’s atmosphere experiences a continuous increase in carbon dioxide (CO2), influenced by natural and anthropogenic factors such as volcanic emissions, diffuse degassing from fault zones and soils, and human-induced gas emissions (; ; ; ; ; ). Densely populated urban zones significantly contribute to increasing CO2 levels, thereby profoundly impacting ongoing global warming.
Climate change affects global key sectors such as infrastructure, economies, energetic resources, international relations, biodiversity, and essential resources such as water and food availability (; ; ; ; ; ). It also drives extreme weather events, polar ice melting, and rising sea level (; ; ). In some regions, intense rainfall becomes more frequent, while others are affected by unprecedented droughts and heat waves (; ). The effects of recent floods in Spain (), Italy (), France, and Greece, just to mention some of those that occurred in 2023 and 2024, demonstrate the high vulnerability of cities in the Mediterranean region to extreme weather events ().
Adaptation and mitigation actions such as land-use planning, urbanization management, and enhancement of green infrastructure can partially combat the effects of climate change. In response to the climate emergency, the United Nation Environmental Program supports countries in facing climate change through a variety of actions, including adaptation and building resilience to climate change, mitigation and moving toward low carbon society, reduction of greenhouse gas emissions from deforestation and forest degradation, and supporting new models for green economy (; ; ; Wimbadi et al., 2021). Achieving these targets requires strategic policy realignment, technological transition, and international collaboration. Prominent examples include the EU’s Green Deal and China’s Dual-Carbon Target, which set specific timelines for peaking emissions and achieving carbon neutrality between 2030 and 2060. In contrast, while the United States has introduced comprehensive climate frameworks (e.g., the Green New Deal), it currently lacks a definitive roadmap for a complete renewable energy transition.
Greenhouse gas (GHG) reduction targets require verification tools based on reliable and timely data. Currently, official GHG catalogs provide annual estimates of national CO2 emissions with a lag of 1 year, revealing the long-term variations by an a-posteriori approach (; ; ; ). A decrease in CO2 emissions due to closures during the COVID-19 pandemic (; ; ) was detected statistically from a bottom-up dataset. Some low latency catalogs publish dilated CO2 emission estimates by several months, but near-real time (daily) data by individual country are not available yet. For example, Eurostat estimates annual CO2 emissions for individual EU countries with a lag of approximately 5 months (EEA/PUBL/2024/046), while the Global Carbon Project publishes global CO2 emission projections for the current year ().
Urban areas are prominent sources of CO2 emissions that lead to an increased greenhouse effect. Conurbations serve as focal points where multiple gas sources increase the CO2 concentration above the atmospheric baseline. Studies regarding the urban CO2 dynamics have highlighted the challenges of identifying and monitoring diverse CO2 sources. According to , who combined the measurements of CO2 concentrations and carbon isotope analyses, the urban atmospheric CO2 in Paris can be accurately modeled using a binary mixing approach involving a well-defined unpolluted background and variable pollution sources. The authors distinguished specific inputs, such as vehicle emissions and human respiration, providing a promising framework for air quality and pollutant quantification research (). Exploratory measurements in the Dallas metropolitan area indicated that combined CO2 concentration and carbon isotope analysis can successfully detect seasonal patterns in anthropogenic emissions, specifically from natural gas combustion, even amidst the noise of biological processes (). Spatial and temporal monitoring of CO2 concentration and carbon isotope composition in the urban atmosphere has suitably tracked the source of the pollutant and identified the magnitude of anthropogenic impacts (; ; ) and volcanic emissions (; ). By integrating sporadic historical data with modern continuous sampling, researchers in the Los Angeles basin found that the local urban CO2 increment has decreased relative to the rising global background. However, the isotopic composition of the fossil fuel source mix has remained consistent with changes in fuel consumption (). Anthropogenic CO2 production from economic activities such as transportation, cement manufacturing, residential heating, and landfill emissions creates an urban CO2 dome that affects the air quality and enhances health risks. Comparative analysis between urban and remote areas proves the presence of a distinct atmospheric CO2 dome, where local anthropogenic fluxes induce significant enrichment in CO2 and amplify the diurnal variations compared to regional marine and rural baselines (). Geogenic CO2 emissions naturally occur in volcanic and tectonically active regions, contributing to the global CO2 emissions by approximately 1%. However, under specific conditions of volcanic activity, geogenic CO2 emissions can locally attain magnitudes comparable to those of anthropogenic origin, as observed in areas such Vulcano and Pozzuoli, Italy (; ). In such cases, quantification for the two components is compelling, although concentration measurements solely cannot disentangle the uncertainty about the origin of CO2. In volcanic areas, the application of stable isotope analysis to airborne CO2 has only recently allowed differentiation between volcanic and anthropogenic CO2 sources (; ; ).
This paper concerns the transformations involving the atmospheric CO2 and the stable isotopic composition of its components (i.e., carbon and oxygen) in Palermo, a medium-sized urban zone at sub-tropical latitudes in the Mediterranean region. The integrated monitoring of atmospheric CO2 concentration and its stable isotopes helps to disentangle the fof natural and anthropogenic CO2 on the urban CO2 levels and monitor their variations over time. The study garnered δ13C-CO2 and δ18O-CO2 data using a laser-based isotope analyzer, providing crucial insights into the sources and mechanisms influencing atmospheric CO2 dynamics at Palermo, which can be considered as representative of the medium-sized city in Europe.
2 Methods
2.1 Atmospheric carbon and oxygen laboratory (ACO-Lab)
this paper discusses the dataset collected in Palermo in the laboratory of the National Institute of Geophysics and Volcanology (INGV-Pa) by the Atmospheric Carbon and Oxygen Laboratory (ACO-Lab) Pa1 station (latitude: 13°21′41.2″ E; longitude: 38°6′56.88″ N). ACO-Lab is a research infrastructure that is designed to develop a stable isotope monitoring network at the local and regional scales with the aim of analyzing the causes of spatial and temporal variations in atmospheric CO2 from the north to the south of the Italian peninsula (Figure 1). The Pa1 station enables real-time, high-frequency assessment of atmospheric CO2 (i.e., concentration, δ13C-CO2, and δ18O-CO2) and some environmental parameters (i.e., temperature, pressure, relative humidity, solar radiation, rain, wind speed, and wind direction). In addition to the Pa1 station, which is located at sub-tropical latitudes, further stations will be deployed in urban areas and remote regions across a range of latitudes in the Italian peninsula. Anthropogenic carbon is released in the northern extratropical regions and oxidized to CO2 in the tropical troposphere. Isotopically tracing these lateral variations of CO2 enables reducing the spatial bias and better constrains the top-down estimations of the CO2 emissions through better knowledge of the position of the focal point of the emissions.
FIGURE 1
2.2 Instrument setup
The ACO-Pa1 station is equipped with a continuous flow spectrophotometer that measures the concentration of CO2 isotopologues (i.e., COO, 13COO, and C18OO) using a medium-infrared laser light (). The measurements collect the air at the height of 13.6 m above the ground (i.e., 94.6 m a.s.l.) at a distance of approximately 80 km from the nearest source of volcanic CO2 (). The continuous flow spectrophotometer does not require purification of the gas sample and provides measurements of stable isotope composition at high frequency (i.e., up to 1 Hz). Furthermore, the measurement accuracy is comparable with that of other laboratory techniques, thus offering several advantages over isotopic ratio mass spectrometry (IRMS) methods. Specifically, the Delta Ray Isotope Ratio Infrared Spectrometer from Thermo Fisher Scientific measures concentrations of 44, 45, and 46 CO2 isotopologues based on the absorption of mid-infrared light (λ ≈ 4.3 μm) in the analytical cell. The Lambert–Beer law allows for quantifying the concentration of the three isotopologues, and the Qtegra™ internal software calculates the 13C/12C ratio, the 18O/16O ratio (i.e., δ13C-CO2 and δ18O-CO2 values, respectively), and the CO2 concentration.
Qtegra™ software allows for scheduling the isotopic measurements of the airborne CO2 and a gas reference. An isotopically marked CO2 gas reference was used for the calibration of the instrument at a scheduled time. The hourly-based measurement cycle includes 11 measurements of the airborne CO2 and a reference against the working standard to achieve the measurement precision of σ = ± 0.25‰ for both δ13C-CO2 and δ18O-CO2. The adoption of an hourly measurement cycle, including referencing against the working standard, ensures that any instrumental drift is compensated, maintaining long-term stability that is consistent with the analytical precision. Furthermore, the hourly referencing procedure ensures that the Qtegra™ software automatically normalizes the raw measurement to the Vienna Pee Dee Belemnite (VPDB) scales using a single-point anchoring method. The Qtegra™ software implements an automatic calibration on a weekly basis, which includes two distinct procedures. The first procedure assesses the linearity of the absorption signal in the CO2 concentration range of 200 ppm–3,500 ppm vol by diluting pure CO2 by the CO2-free synthetic air. The linearity calibration is crucial to correct the concentration-dependence of the instrument’s isotopic signal, ensuring that the δ-values remain accurate over the range of CO2 concentration in urban environments (e.g., approximately 400 to over 600 ppmv). The second procedure establishes the correspondence of the signal in the analyzer for δ13C-CO2 and δ18O-CO2 values in the expected range of concentration on the analytical samples.
Following 11 measurements of airborne CO2, the instrument reiterates the isotopic measurements on the isotopically marked reference at the average CO2 concentration measured in the air samples (i.e., the so-called smart referencing). The CO2 reference for both the δ13C-CO2 and δ18O-CO2 was commercial CO2 delivered from Naftia emissions (i.e., a productive plant in Palagonia, southern Italy) and has the values of δ13C-CO2 = −1.76‰ vs. VPDB and δ18O = −7.62‰ vs. VPDB (). Both the δ13C-CO2 and δ18O-CO2 values of this internal working standard compared with the VPDB international standard were measured in the INGV laboratory by the Thermo Fisher Scientific Delta V-Plus isotope mass ratio spectrometer with dual inlet system (instrumental precision σ = ± 0.10‰ for both δ13C-CO2 and δ18O-CO2). The traceability to the VPDB international scale was ensured by the cross-calibration of our internal working standard with GasBio™ from Air Liquide (δ13C-CO2 = −25.5‰ ± 0.3‰ vs. VPDB and δ18O-CO2 = −24.4‰ ± 0.5‰ vs. VPDB). The isotopic values of the GasBio™ were anchored to the international VPDB scale using the primary reference material IAEA-603 (Carrara Marble), which replaced the exhausted NBS 19 standard. The traceability was ensured by maintaining the certified values of +1.95‰ for δ13C and −2.37‰ for δ18O during the calibration. Isotopic results provided by the Delta Ray analyzer are reported on the VPDB scale in accordance with minimum requirements for publishing stable-isotope delta results (). The working standard (i.e., the CO2 reference from Naftia) was calibrated via IRMS against the GasBio™ reference standard, ensuring traceability to the VPDB international scale (). During routine operations, the Delta Ray spectrometer performs an hourly single-point referencing (i.e., the so called ‘smart referencing’ procedure) against this calibrated Naftia CO2 reference to compensate for short-term drift. The consistency of the isotopic scale and the absence of scale contraction effects were verified through periodic linearity calibration across the CO2 concentration range observed in the urban environment.
Air temperature (accuracy ±0.3 °C), atmospheric pressure (accuracy ±1 mbar), relative humidity (accuracy ±2%), rain rate (accuracy ±0.2 mm), radiation (accuracy ±5%), wind direction (accuracy ±3), and wind speed (accuracy ±3 km h-1) are pivotal parameters in evaluating the gas dispersion and the pattern of pollutants. These environmental variables were measured using a Davis Vantage Pro2 Plus weather station (https://www.meteoproject.it/davis-vantage-pro2.php) placed on the roof of the INGV laboratory. The positioning of the weather station avoids interference with buildings in the surrounding area and provides information about the air circulation at the air CO2 monitoring site. Both the Delta Ray and the weather station record measurements with a 5-min sampling frequency, allowing synchronous detection of chemical, isotopic, and environmental parameters. The results of a pilot experiment performed in 2021 () demonstrated the suitability of this method for the purpose of analyzing the patterns of airborne CO2 in urban areas.
2.3 Palermo monitoring site
The city of Palermo, located in the northwest sector of Sicily by the Tyrrhenian Sea, is heavily influenced by its seacoast environment, which affects its urban climate through daily wind reversals caused by the differential heating of land and sea. The city’s geography includes rugged mountains, with peaks averaging 780 m, which shield the Palermo plain, where the city spans 160 km2 (Figure 1). Approximately 700,000 residents make Palermo the fifth most populous city in Italy. However, the total community numbers to 1.253 million, including the suburban district, with several daily commuters from the outskirts to downtown. The CO2 levels in the air at the Palermo urban area exceed the global average value due to emissions from industry, agriculture, transportation, and building heating. Variations in the anthropogenic activity in the urban area of Palermo strongly influence the air quality, as observed during the COVID-19 lockdown (). Although renewable energy sources such as wind and small solar plants are increasingly being used, the region remains dependent on fossil fuels for energy production, thus contributing to raising the CO2 levels. A significant CO2 source in the area is the Bellolampo landfill, southwest of Palermo, where decomposing organic waste emits CO2 and CH4 (). The CO2 from the landfill has a distinct isotopic signature compared to that produce from fossil fuels, providing a tool for identifying the source of emissions (). The pilot studies conducted in Palermo revealed the influence of the nearby highways, industries, and shopping centers, with a major green zone to the east (i.e., the Favorita Park) and the Mondello coast to the north (Figure 1). Given the urban size, the location at the boundary between the sub-tropical and intermediate latitudes, and the climatic conditions, Palermo is an ideal site to study changes in CO2 concentration and isotopic composition to evaluate the effects of human activities.
2.4 Data processing and multivariate analysis
The dataset collected at the ACO-Pa1 station (i.e., chemical, isotopic, and environmental data) at 5-min sampling periods were used for time-series and statistical analyses. To reduce the dimensionalities (i.e., the number of variables describing the environmental system well) and preserve a large part of the significant information, the dataset was processed using the statistical technique of principal component analysis (PCA). PCA transforms the original dataset into a smaller set of linearly independent new variables, each one being a linear combination of the original one (). The PCA is a multivariate dimensionality reduction technique that allows disentangling the complex covariance structure characterizing the original dataset of CO2 and environmental variables aimed at identifying the underlying drivers of atmospheric CO2 variability.
Prior to analysis, the dataset variables were normalized to the [0–1] range to eliminate scale discrepancies caused by heterogeneous measurement ranges. The PCA transforms the correlated observations into a smaller set of orthogonal variables (i.e., linearly independent variables), or principal components (PCs). PCs are ordered so that the first few components retain most of the variance of the original dataset (). Therefore, the first principal component (PC1) stands for the greatest proportion of the total variance of the dataset, while the second component (PC2) represents the greatest proportion of the residual variance, after the effect of the PC1 has been removed, and so on. Since a few components explain a significant part of the total information in the dataset, the remaining PCs can be neglected. The number of significant components to retain was decided by applying Kaiser’s criterion, which recommends the conservation of the components with eigenvalue >1, and confirmed its agreement with the distribution of PCs on the scree plot method (). In that case, the amount of information carried by each PC exceeds that of a single standardized variable. A diagnostic tool for the interpretation of the correlation among the original variables and the calculated PCs are biplots. Each vector projected in the biplot represents and depends on both the PCs’ scores of the observed variable. In the biplots, vectors falling near the origin do not show unique features along the PCs, while points falling far from the origin are extreme points, and the direction in which they spread is crucial. The projection of a vector shows the value of each observed variable on the specific PC. Vectors either falling near an axis or aligned with it have particularly high eigenvalue (i.e., saturation) on those PCs. Vectors clustered in a distinct group identify conditions in the dataset, while isolated vectors that are far from other groups represent unique and meaningful conditions.
3 Results
The dataset collected in Palermo for urban CO2 monitoring includes CO2 concentration values, which are reported as parts per million by volume (ppm vol), and isotopic data expressed in δ-notation (δ13C-CO2 and δ18O-CO2, respectively) against VPDB. The dataset includes data collected at the monitoring site for some environmental variables [i.e., temperature (°C), relative humidity (%), pressure (mbar), rain rate (mm h-1), radiation (mW m-2), wind speed (m s-1), and direction (°)].
3.1 Airborne CO2 concentration
Figure 2 shows the time-series of concentration, δ13C-CO2, and δ18O-CO2 of airborne CO2 collected at 5-min intervals. The complex evolution of airborne CO2 concentrations and stable isotope composition at the monitoring site consists of remarkable variations that can be observed throughout the monitoring period. From 17 November 2023 to 30 September 2025, the minimum and maximum concentrations measured were 421 ppm vol and 551 ppm vol, respectively, with an average of 442 ppm vol. These values are 2%, 29%, and 3% higher, respectively, than those observed in 2021 at the same monitoring site, in agreement with data reported in , although those measurements were collected over 4 weeks in winter (January to February 2021).
FIGURE 2
Concentration values show the frequency peaks at 442 ppm vol, with a long tail [skewness (Sk) = 1.60; kurtosis (Ks) = 4.35] extending toward values higher than 460 ppm vol (Figure 3a). This statistical distribution of the concentration values mirrors a combination among multiple normal distributed sources of CO2. A statistical graphical method for classifying the measurements was utilized with the purpose of identifying and speculating about processes originating or involving different populations. This method involves the calculation of the cumulative probability of the measured CO2 concentrations and plotting these values in a probability plot (). This approach is widely used for classifying geochemical data, including measurements of soil CO2 flux and airborne CO2 concentrations in volcanic regions (; ). The probability is based on the graphical evidence that the logarithm of the measured values against cumulative probability provides a straight line for each log-normal data population. An inflection point in the data distribution separates two distinct populations. Three inflection points in the dataset distribution were identified following Sinclair’s method, revealing four distinct populations in the probability plot (Figure 3b).
FIGURE 3
The average value of the background subset (R2 = 0.880) is 424 ppm vol (Table 1), while the average for the high pollution subset (HP-events) is 517 ppm vol (R2 = 0.482). In the probability plot, the average cumulative probability coincides with the 50% values of each population. The first population (background in Figure 3b) consists of CO2 concentrations comparable to those reported by the National Oceanic and Atmospheric Administration-Interactive Atmospheric Data Visualization (NOAA-IADV) as the global reference for airborne CO2 concentration during the monitoring period (https://gml.noaa.gov/dv/iadv/, accessed on 30/09/2025). This population reflects the local background during the observation period. The highest population (HP in Figure 3b) consists of <0.08% of the dataset and corresponds to CO2 concentrations that were occasionally reached in Palermo.
Two intermediate populations encompass most of the dataset (i.e., approximately 98.18%). The Pop-2 population includes approximately 47.49% of the dataset, consisting of CO2 concentrations ranging from 440 to 508 ppm vol, which is equivalent to 6083.67 h of airborne CO2 above 440 ppm vol in the urban area of Palermo. The 50th cumulative percentile of this population (467 ppm vol) is 10.4% higher than the global average CO2 concentration and 5.91% higher than the average local background during the monitoring period. These differences are higher than those typically observed among the reference sites in the global monitoring network (i.e., <1%). According to observations at Mauna Loa, Hawaii (), the annual average CO2 concentration for 2025 was 427.03 ppmv. This represents a year-over-year increase of 2.67 ppmv relative to 2024 and a 3.62 ppmv increase compared to the 2023 average (https://gml.noaa.gov/ccgg/trends/gl_data.html late accessed on 2026, April 6).
The Pop-1 population (i.e., 426 ppm vol < airborne CO2 < 440 ppm vol; R2 = 0.980) encompasses approximately 50.68% of the dataset and has concentration values between the local background CO2 and Pop-2 values (i.e., 440 ppm vol < airborne CO2 < 508 ppm vol; R2 = 0.803). Pop-1 and Pop-2 subsets reflect the daily fluctuations in airborne CO2 concentration. Similar variations have been attributed to the dilution of CO2-polluted air with cleaner air (i.e., air with background CO2 concentration) due to the reversal of sea–land circulation affecting the city of Palermo ().
3.2 Carbon isotope composition of the airborne CO2
The δ13C-CO2 showed significant variations throughout the survey period (Figure 2b). The range of δ13C-CO2 values (Δδ13C-CO2 = 7.16‰) is an order of magnitude greater than the instrumental precision (σ = ±0.25‰). The most 13C-depleted CO2 (e.g., δ13C-CO2 < −12.00‰ vs. VPDB) has been occasionally recorded from December 2023 to April 2024 and from November 2024 to March 2025, with negative peaks coinciding with positive peaks in CO2 concentration (Figure 2a). The carbon isotopic signal seldom exhibited δ13C-CO2 values above −8‰ vs. VPDB, the reference background δ13C in airborne CO2, while the average δ13C-CO2 = −9.21‰ was more 13C-depleted. Decreases in δ13C-CO2 contrasting with increases in airborne CO2 concentration indicate a negative correlation between these variables, which is in agreement with the data reported by and already noted for the urbanized area of Palermo ().
The frequency distribution (Figure 3c) shows that over 25% of the measurements have δ13C-CO2 values in the range of −10‰ to −9‰, indicating a CO2 source that is more 13C-depleted than the global average atmospheric CO2. The distribution is projected toward more negative values of δ13C-CO2 (Sk = −1.08; Ku = 0.14), pointing to a 13C-depleted CO2 source raising the airborne CO2 concentration in the urban area of Palermo. The average δ13C-CO2 in Palermo is currently Δδ13C-CO2 = 1.49‰ more 13C-depleted than the average value measured in 2021 (; ). A closer inspection of the time-series indicates a long-term pattern in the evolution of δ13C-CO2, with less 13C-depleted CO2 measured from April to November 2024 and since May 2025, which is in coincidence with the spring–summer period in the northern hemisphere. In contrast, the 13C-depleted carbon isotope compositions have been observed in airborne CO2 from December 2023 to March 2024 and November 2024 to March 2025, which correspond to the fall–winter period in the north hemisphere.
3.3 Oxygen isotope composition of the airborne CO2
The oxygen signal has the average value δ18O-CO2 = −2.96‰ vs. VPDB (Figure 3d). The range of values is consistent with the oxygen isotope composition of airborne CO2 at latitudes between 25 °N and 40 °N (see NOAA Global Monitoring Laboratory–Earth System Laboratory, https://gml.noaa.gov/dv/iadv/(). Furthermore, the δ18O-CO2 dataset shows a frequency distribution encompassing more 18O-depleted airborne CO2 (Figure 3d). The range of δ18O-CO2 (Δδ18O-CO2 = 5.29‰) was more than an order of magnitude greater than the instrumental precision (σ = ±0.25‰). This finding supports previous observations that the short-term evolution of δ18O-CO2 in the urban area of Palermo is correlated with patterns in airborne CO2 and δ13C-CO2 (). A closer inspection of Figure 2c reveals a less 18O-depleted CO2 recorded from March to June during both 2024 and 2025 (i.e., δ18O-CO2 > −3‰ vs. VPDB) and the most 18O-depleted CO2 observed from November 2023 to March 2024 and from September 2024 to March 2025. A recent remarkable 18O-depletion period started in early September 2025 (Figure 2c).
Although δ18O-CO2 typically exhibits negative values in coastal zones (), this parameter in the airborne CO2 is affected by the CO2–H2O reaction and the consequent oxygen isotope fractionation during both photosynthesis and plant respiration. According to , δ18O-CO2 data measured in Palermo are consistent with δ18O-CO2 values measured by NOAA stations at comparable latitudes in the Mediterranean region (e.g., Lampedusa and the Weizmann Institute of Science at the Arava Institute, Ketura), in the Atlantic region (e.g., Terceira and Tenerife), and in European continental zones (e.g., Centro de Investigación de la Baja Atmósfera).
3.4 Meteorological conditions in Palermo
The weather variables show remarkable variations during the monitoring period from fall 2023 to summer 2025 (Figure 4). The air temperature fluctuated both diurnally and seasonally from 7.7 °C recorded on December 2024 and reached 39.1 °C in July 2025 (Figure 4a). The daily temperature range occasionally was wider than 10 °C. For instance, air temperature reached 30 °C on 31 March 2024 (an unusual occurrence for spring at the latitude of Palermo) and fluctuated at approximately 18 °C on 1 April 2024. This latter value is close to the average temperature measured during the monitoring period (i.e., 20.3 ° C). A clear upward trend in air temperature began in March 2024 (Figure 4a), following several weeks of constant temperatures (i.e., 10 °C–15 °C) in February 2024.
FIGURE 4
The relative humidity time-series mirrored the air thermal pattern (Figure 4b). The average value of the relative humidity was 65.01%, with fluctuations ranging from 15% to 87%. From November 2023 to March 2024, humidity values were higher than those recorded from April to July 2024. A rapid change occurred on 31 March 2024, when a decrease in relative humidity coincided with the onset of rising temperatures and lower atmospheric pressure (Figure 4c).
Figure 4d shows the rain rate (mm h-1) measured by the Davis Vantage Pro2 station. The rain rate represents the instantaneous rainfall intensity, which is expressed as the equivalent depth of rain (mm) that would accumulate in 1 hour if the precipitation rate remained constant. Therefore, this value does not represent the total rainfall but the rainfall intensity at the time of measurement.
To identify the most significant rainfall episodes during the monitoring period, we identified 98 rain events, defined as rain periods with average rain rates >0.2 mm h-1 and separated by at least 1 h without rainfall (Figure 4d). For each event, cumulative rainfall was calculated as:where Ri is the rain rate (mm h-1), Δti is the measurement time interval (5 min), and n is the number of measurements within the accumulation period (e.g., n = 12 in 1 h). In agreement with this calculation, two of these events were classified as extreme rain events. In this study, extreme rain events are defined as those with cumulative rainfall exceeding 15 mm (Supplementary Figure S1).
Solar radiation follows cyclic patterns dependent on daylight duration (Figure 4e), which averaged 12 h, 19 min, and 48 s, ranging from 9 to 15 h in December and May–July, respectively. Solar radiation higher than 1,200 W m-2 has been observed on 19 April and 21 August 2024 and on 28 March and 19 June 2025. A comparison of the time-series for the various parameters revealed that this period also occurred at lower relative humidity and higher temperatures, as typically observed from spring to summer at the latitude of Palermo.
Both the wind speed (Figure 4f) and direction (Figure 4g) can play a crucial role in gas dispersion in the atmosphere. These variables affect the pollutant concentrations by either mixing air from various sources during the turbulent diurnal evolution of the planetary boundary layer (PBL) or by promoting chemical layering when turbulence ceases at night or during periods of high atmospheric pressure (Figure 4c). During the monitoring period, wind speed reached a maximum of 19.67 m s-1 based on 5-min recordings, while the average wind speed (2.08 m s-1) was an order of magnitude lower (Table 1). The maximum wind speed was recorded on 7 January 2024, with a direction of 325° N. Wind direction in the Palermo urban area aligned with the southwest and northeast quadrants, with occasional winds from the northwest quadrant (Figure 4g).
TABLE 1
| Month | CO2 (ppm vol) | δ13C-CO2 (‰VPDB) | δ13O-CO2 (‰VPDB) | Temp. (°C) | Hum. (%) | Pressure (hPa) | Wind (m s-1) | Rain (mm h-1) | Radiation (W m-2) | U-wind (E-W) | V-wind (N-S) | Et (mm) |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Nov-23 | 437 | −9.19 | −3.23 | 17 | 61 | 1,013 | 2.69 | 0.22 | 93.64 | −0.490 | 0.750 | 0.05 |
| Dec-23 | 442 | −9.48 | −3.36 | 15 | 67 | 1,021 | 2.12 | 0.10 | 74.04 | −0.395 | 0.637 | 0.04 |
| Jan-24 | 441 | −9.34 | −2.97 | 15 | 63 | 1,022 | 2.59 | 0.09 | 85.44 | −0.816 | 0.881 | 0.05 |
| Feb-24 | 444 | −9.44 | −3.05 | 14 | 68 | 1,020 | 2.09 | 0.09 | 120.80 | −0.386 | 0.319 | 0.05 |
| Mar-24 | 441 | −9.30 | −2.97 | 16 | 63 | 1,016 | 2.54 | 0.09 | 183.99 | −0.461 | 0.663 | 0.07 |
| Apr-24 | 444 | −9.33 | −2.58 | 18 | 55 | 1,019 | 2.24 | 0.01 | 223.78 | −0.405 | 0.325 | 0.08 |
| May-24 | 443 | −9.19 | −2.38 | 21 | 65 | 1,016 | 2.02 | 0.11 | 265.55 | 0.065 | 0.349 | 0.08 |
| Jun-24 | 439 | −8.98 | −2.54 | 25 | 64 | 1,017 | 2.07 | 0.01 | 282.15 | −0.031 | −0.404 | 0.09 |
| Jul-24 | 436 | −8.84 | −2.71 | 28 | 63 | 1,015 | 1.98 | 0.00 | 294.74 | 0.307 | 0.265 | 0.10 |
| Aug-24 | 436 | −9.00 | −3.12 | 29 | 64 | 1,014 | 1.97 | 0.09 | 247.82 | 0.407 | 0.272 | 0.10 |
| Sep-24 | 436 | −9.01 | −3.64 | 25 | 64 | 1,016 | 2.19 | 0.07 | 181.41 | 0.139 | 0.910 | 0.07 |
| Oct-24 | 444 | −9.33 | −3.76 | 22 | 69 | 1,020 | 1.66 | 0.03 | 147.30 | −0.298 | 0.259 | 0.06 |
| Nov-24 | 446 | −9.39 | −3.23 | 18 | 69 | 1,023 | 1.75 | 0.06 | 105.75 | −0.214 | 0.814 | 0.05 |
| Dec-24 | 446 | −9.41 | −3.56 | 14 | 69 | 1,022 | 2.91 | 0.13 | 67.97 | 0.123 | 1.673 | 0.04 |
| Jan-25 | 447 | −9.43 | −3.35 | 14 | 69 | 1,022 | 2.01 | 0.16 | 90.02 | −0.408 | 0.243 | 0.04 |
| Feb-25 | 451 | −9.59 | −3.23 | 13 | 71 | 1,024 | 1.75 | 0.08 | 118.53 | −0.309 | 0.223 | 0.04 |
| Mar-25 | 446 | −9.38 | −2.87 | 16 | 67 | 1,017 | 2.34 | 0.11 | 156.76 | −0.292 | 0.289 | 0.06 |
| Apr-25 | 447 | −9.34 | −2.65 | 17 | 67 | 1,017 | 1.96 | 0.07 | 227.86 | −0.003 | 0.206 | 0.07 |
| May-25 | 443 | −9.15 | −2.13 | 20 | 67 | 1,017 | 1.96 | 0.06 | 268.62 | 0.559 | 0.566 | 0.08 |
| Jun-25 | 445 | −9.19 | −2.19 | 26 | 59 | 1,019 | 1.67 | 0.04 | 313.64 | 0.256 | 0.052 | 0.11 |
| Jul-25 | 437 | −8.90 | −2.62 | 28 | 60 | 1,015 | 2.25 | 0.01 | 299.13 | 0.616 | 0.676 | 0.11 |
| Aug-25 | 434 | −8.94 | −3.18 | 27 | 64 | 1,016 | 1.99 | 0.00 | 261.63 | 0.634 | 0.524 | 0.10 |
| Sep-25 | 437 | −8.76 | −3.54 | 25 | 65 | 1,019 | 1.65 | 0.08 | 203.97 | 0.200 | 0.133 | 0.08 |
| Average | 442 | −9.21 | −2.96 | 20 | 65 | 1,018 | 2.10 | 0.07 | 187.59 | −0.052 | 0.462 | 0.07 |
| Background | 424 | −8.49 | −2.44 | | | | | | | | | |
| Pop-1 | 435 | −8.97 | −2.77 | | | | | | | | | |
| Pop-2 | 467 | −9.65 | −2.96 | | | | | | | | | |
| HP-events | 517 | −12.24 | −4.53 | | | | | | | | | |
Monthly average values of the recorded variables.
The average values reported for background, Pop-1, Pop-2, and high pollution events (HP-events) coincides with the 50th percentile value retrieved from the probability plot in agreement with the method proposed by .
4 Discussion
4.1 Origin of the airborne CO2
Stable isotope monitoring allows tracking the variations of the airborne CO2 due to either the patterns in the atmospheric variables or variations in the origin of the emissions produced by the anthropogenic activities.
The isotopic mass balance model suitably identifies the causes of increase in CO2 concentration in the atmosphere produced by a variety of sources of CO2 at the local level using observational data. The Keeling plot method illustrates a correlation between the carbon isotope composition of CO2 and the inverse of airborne CO2 concentration (). Their correlation relies on mass balance principles, wherein a local CO2 source alters the concentration from the atmospheric baseline. Mathematically, this is expressed by the following equations:andwhere C and δ13C denote the CO2 concentration and δ13C-CO2, respectively. In Equations 1, 2, subscripts denote the measured values (m), atmospheric background (a), and relevant local forcing source (fs). The linear combination of these equations generates a straight line in the δ13C vs. 1/C plot (), as delineated by the equation:
Equation 3 provides insight into the carbon isotope composition of the local CO2 source under constant background and CO2 source conditions.
Figure 5 shows the concentration dataset normalized by the global reference for airborne CO2 concentration (i.e., 423 ppm vol). Each straight line in Figure 5 depicts the evolution of the carbon isotope composition of the atmospheric CO2 as the additional CO2 sources increase the concentration above the local background value. The intercept on the isotopic axis provides the carbon isotopic signatures of the endmember, facilitating the identification of the forcing source of CO2 that is relevant at the local scale. Atmospheric CO2 originates from both natural and anthropogenic sources. Natural contributions include soil and plant respiration (; ; ; ; ), chemical weathering (; ), and geogenic degassing from volcanic systems (; ). Although the level of volcanic degassing may change according to the volcanic activity (; ; ; ), natural emissions of CO2 are typically balanced within the global carbon cycle. Since the last centuries, the combustion of fossil fuels for transportation and natural gas for residential heating have become the dominant anthropogenic CO2 sources. These emissions are characterized by distinct carbon isotopic signatures (; ), allowing for source partitioning. Furthermore, urban CO2 levels are significantly influenced by landfill emissions caused by the oxidation of organic matter in such environments (), adding further complexity to the urban carbon footprint. Figure 5 shows theoretical mixing lines among the background air and common geogenic and anthropogenic sources of CO2 in urban environments for a graphical comparison with the recorded data. The adopted δ13C-CO2 values for the endmembers that force CO2 above the background levels were retrieved for CO2 from fossil fuel and natural gas combustion (; ), CO2 from soil/plant respiration (), landfill gas emissions (; ; ; ), and volcanic CO2 (; ).
FIGURE 5
The high-frequency dataset recorded at the ACO-Pa1 station indicates that fossil fuel combustion plays a relevant role in enhancing the CO2 concentrations above the atmospheric background (Figure 5). The intercept of the dataset regression line (δ13C-CO2 = −27.1‰) points toward a13C-depleted source of CO2. However the linear regression model of the dataset can introduce a bias in the retrieved carbon isotopic signature of the forcing source, shifting it toward a less negative value (), particularly when the correlation is <0.92 (). To account for this bias, we followed the approach of , scaling the regression slope by the r-coefficient of the δ13C-CO2 and 1/CO2 correlation. This correction yields an isotopic signature of the emitting source of δ13C-CO2 = −32.20‰ vs. VPDB. This value is less 13C-depleted than the CO2 typically produced by the combustion of the natural gas (e.g., δ13C-CO2 = ∼ −42.00‰ vs. VPDB) but more 13C-depleted than the CO2 produced by the fossil fuel combustion (i.e., δ13C-CO2 = ∼ −28.00‰ vs. VPDB), indicating a mixed anthropogenic contribution. Moreover, the CO2 sources may be relevant at a certain time of the year rather than during the whole period. This occurrence depends on the anthropogenic CO2 production that has a seasonal variability dependent on the relevant activities of the citizens throughout the years (e.g., working days and holiday periods). Therefore, the intermediate values of the carbon isotopic signature retrieved from the continuous measurements of the airborne CO2 indicate that a combination of those anthropogenic sources of CO2 increases the CO2 concentration above the background in the urban area of Palermo. The carbon isotopic signatures shown by the Keeling plot analysis in Palermo align well with those reported for other major urban areas. However, the observed range of atmospheric CO2 concentrations is notably more constrained than in comparable cities (Figure 5). The comparison among different cities shows a general agreement with an increase of the background CO2 concentration in comparison to that in the last decades and a more pronounced 13C-depletion of the airborne CO2 as background CO2 concentration increases. Furthermore, the agreement between the bottom-up estimations of the total CO2, natural gas-derived CO2, and CO2 from oil combustion provided by the Global Carbon Project (; ; ; ; ) and variations of the CO2 observed in the city of Palermo supports the hypothesis of a variable role of the dominant source of CO2 throughout the period of monitoring. The time-series shows a relevant seasonality in the total CO2 emissions composed by an approximately constant level of the CO2 emission of from fossil fuel combustion (i.e., largely derived from vehicle mobility) and a relevant seasonality pattern in the CO2 derived by the combustion of natural gas (Supplementary Figure S2), which is mostly used for heating houses in the city of Palermo.
4.2 Seasonal variability of the airborne CO2
Figure 2a shows that the CO2 concentration rises above the local background value (i.e., 442 ppm vol) and that the baseline CO2 concentration follows smoothed fluctuations, which follow a fifth-order polynomial function of the time. Such smoothed waveforms have their maximum on 31 January 2024 at 445 ppm vol and on 23 February 2025 at 449 ppm vol (red line in Figure 2a). The daily average CO2 concentration recorded at the ACO-Pa1 station shows a seasonal pattern consistent with the daily data collected at the Mauna Loa Station by NOAA (Supplementary Figure S3). This comparison also reveals that CO2 concentrations in Palermo are, on average, approximately 4% higher than that in Mauna Loa, with daily offsets ranging from −0.08% to +8.81%. Consistent with the CO2 concentration trends, both carbon and oxygen isotope compositions showed remarkable variations (Figures 2b,c, respectively). For instance, the baseline of the oxygen isotope composition shows the most 18O-depleted values from August 2024 to March 2025 and after August 2025, while the less 18O-depleted CO2 has been measured in the summer of the north hemisphere. The baseline of the oxygen isotopes at Palermo exhibits the less 18O-depleted values on April 2024 and June 2025 (i.e., δ18O-CO2 from ∼ −2.63 to −2.28‰ vs. VPDB, respectively). Accordingly, the δ13C-CO2 shows the most 13C-depleted values from November 2023 to March 2024 and November 2024 to March 2025, with the minimum values of the smoothed δ13C-CO2 trend (red line in Figure 2b) at approximately −9.49‰ vs. VPDB on 16 January 2024 and 15 February 2025.
A comprehensive analysis of the correlations among the chemical, isotopic, and environmental data reveals increases in the CO2 concentration when the air temperature decreases (Supplementary Table S1). The agreement of an increase in CO2 concentration with the observed 13C-depletion indicates that the combustion of fossil fuels for heating houses during winter and other anthropogenic emissions the elevated levels of CO2 in the atmosphere. On the other hand, an increase in solar radiation (Figure 4) and atmospheric turbulence (e.g., wind speed), which can be considered as proxies of the upcoming productivity season (i.e., from April to August in the northern hemisphere), contributes to reduce the average concentration of the airborne CO2 (r = −0.349, r = −0.198, and r = −0.191 for radiation, U-wind, and V-wind components, respectively). Therefore, a combination of anthropogenic activities and local environmental variables lead to variations in the concentration of the atmospheric CO2 in the air at Palermo. Analyzing smoothed variations of CO2 concentration captures changes in seasonal dynamics of airborne CO2. For instance, smoothing short-term fluctuations reveals information about the impacts of CO2 emissions in the long-term. In this framework, the δ13C-CO2 trends embody the isotopic signature of the main CO2 sources contributing to increase its concentration above the baseline atmospheric levels. On the other hand, δ18O-CO2 carries independent information with respect to the carbon signal. It reveals the fractionation effects involving atmospheric CO2 in plant leaves and offers insights into hydrological cycles and climate forcing (). Similarly, assessing the average values of stable isotope compositions provides insights into the seasonal evolution of CO2 origin and allows an evaluation of the potential influences of environmental factors on urban airborne CO2 patterns, specifically for Palermo city.
4.3 Forcing causes atmospheric CO2 variations
The PCA allows disentangling the physical, anthropogenic, and biological processes behind the evolution of atmospheric CO2 in the urban area of Palermo over approximately 2 years of observations. The results of this analysis provide key information on the spatial and temporal scales over which the evolution of the atmospheric CO2 occurs. In agreement with the Kaiser criterion (), the components that should be retained are PC1–PC3, since only the components with eigenvalues >1 are sufficient to form a reduced number of components to represent the total variance of the dataset (Table 2).
TABLE 2
| Correlation eigenvalues | PC1 | PC2 | PC3 | PC4 | PC5 | PC6 | PC7 | PC8 | PC9 | PC10 | PC11 |
|---|---|---|---|---|---|---|---|---|---|---|---|
| Eigenvalue | 3.350 | 1.395 | 1.204 | 0.966 | 0.874 | 0.805 | 0.766 | 0.716 | 0.441 | 0.382 | 0.101 |
| Variance proportion | 0.305 | 0.127 | 0.109 | 0.088 | 0.079 | 0.073 | 0.070 | 0.065 | 0.040 | 0.035 | 0.009 |
| Cumulative variance | 0.305 | 0.431 | 0.541 | 0.629 | 0.708 | 0.781 | 0.851 | 0.916 | 0.956 | 0.991 | 1.000 |
Results of the principal component analysis.
The eigenvalues of the correlation matrix > 1 indicate that the variance of each PC is greater than the variance of the recorded variable. The variance proportion and the cumulative explained variance are reported.
However, the PC1–PC3 achieve a rather moderate value of the cumulative variance (i.e., 54.1%) because of the complexity and multidimensionality of the dataset. Furthermore, this result also indicates that the evolution of the atmospheric CO2 in the urban areas involves additional variables to the environmental parameters (e.g., the anthropogenic activities). A combination of eigenvalues matrix and the scree plot analysis (Figure 5) show that the elbow in the distribution of PCs occurs on PC4, which has an eigenvalue close to 1 and explains 8.8% of the total variance (Table 2). Therefore, PC1–PC4 have been considered for interpreting the dataset variance, achieving the percentage of the cumulative total variance of 62.9%. The moderate value of cumulative variance is a direct reflection of the urban atmosphere’s complexity, where anthropogenic emissions and local atmospheric turbulence introduce a remarkable stochastic component. In such a complex environment, the PCA serves as a filter to separate structured biogeochemical signals from local noise.
Interpretations of PCA results rely on the matrix of loadings (Supplementary Table S2), which collect the correlations among the original variables and the linearly independent components (PCs). The correlation of the variables with respect to the PCs (i.e., component saturations) allows the construction of the biplots and the identification of some processes driving the evolution of both the concentration and isotope composition of the atmospheric CO2 (Figure 6; Supplementary Figure S4).
FIGURE 6
In this paper, we consider that the saturation (correlation of an observable with respect to a PC) is strong for values of the correlation >0.5 without sign, while moderate saturations (correlations) have values between |0.4| and |0.5|. The PC1 carries most of the variance of the system (i.e., 30.5%) and acts as the dominant driver of the total variance. The PC1 has a strong a correlation with several observed variables (i.e., loadings are δ13C-CO2 = 0.835, CO2 = −0.800, temperature = 0.738, radiation = 0.608, and relative humidity = −0.575, respectively) and moderate correlation with δ18O-CO2, evapotranspiration, and atmospheric pressure (i.e., 0.485, 0.493, and −0.444, respectively). The ensemble of these environmental variables underscores the biogeochemical processes governing the carbon exchange in the urban ecosystem, driven by photosynthesis, respiration, and anthropogenic gas emissions. Positive values of PC1 are associated with high solar radiation and temperature (Figure 6). These environmental variables promote carbon fixation, which is quantified by gross primary productivity (GPP) and has a strong seasonality, being at its high efficiency during spring and summer periods (; Ming-Wei et al., 2025). The enhancement of GPP reduces both the CO2 concentration in the atmosphere and the 13C-depletion in the residual atmospheric CO2. On the contrary, enhancement of the CO2 levels due to the anthropogenic CO2 emissions and GPP decreases during fall and winter promote 13C-depletion due to the specific isotopic carbon signature of the CO2 produced by the combustion of fossil fuel and natural gas. Consistently, PC1 correlations are negative with the relative humidity and positive with evapotranspiration, which represent the dry conditions and high stomatal activity during periods of high insolation (i.e., the hours of high insolation during warm season), when the GPP is high. Therefore, PC1 represents the axis that contrasts photosynthetic carbon fixation with respiration and gas emissions (Figure 6b).
The PC2 has a strong correlation with both the u and v components of the wind (i.e., the zonal and meridional components of the wind pattern saturating PC2 with 0.560 and 0.748, respectively), while the correlation with the relative humidity is moderate (i.e., 0.419). Accordingly, this component accounts for the atmospheric dynamics, the turbulence, and the gas dispersal at the local scale. The PC2 explains approximately 12.6% (0.126) of the total variance and is clearly attributable as a dynamic factor, separating it from both the anthropo-biogenic activity (i.e., PC1) and the physical processes at the wide synoptic scale. The PC3 is strongly saturated by the atmospheric pressure, the u component of the wind, and the solar radiation (i.e., 0.636, 0.549, and 0.532, respectively), explaining approximately 11% of the total variance (Table 3). These correlations indicate a dependence of PC3 on the atmospheric stability (i.e., the high barometer) and dominant wind from the western direction proving clear sky conditions above the urban zone of Palermo (), which is largely controlled by the atmospheric conditions at the synoptic scale.
TABLE 3
| Measured variable | PC1 | PC2 | PC3 | PC4 |
|---|---|---|---|---|
| δ13C-CO2 | 0.834 | 0.219 | −0.190 | −0.090 |
| δ18O-CO2 | 0.483 | −0.232 | −0.006 | 0.098 |
| CO2 | −0.800 | −0.285 | 0.183 | 0.103 |
| T | 0.738 | −0.262 | 0.035 | 0.049 |
| Rh | −0.575 | 0.418 | −0.148 | −0.024 |
| P | −0.444 | 0.002 | 0.636 | 0.125 |
| Rain | −0.032 | 0.234 | −0.340 | 0.901 |
| Rad | 0.608 | −0.006 | 0.532 | 0.178 |
| ET | 0.493 | −0.202 | −0.003 | 0.051 |
| u-wind | 0.286 | 0.560 | 0.549 | 0.130 |
| v-wind | 0.115 | 0.748 | −0.084 | −0.237 |
Saturation matrix (loadings) of the selected PCs based on the Kaiser criterion.
PC4 explains 8.8% of the total variance, separating almost exclusively the variance associated with rain because it has a strong saturation from the rain variable (Table 3). This result is in good agreement with the general event-based patterns of the rain, which are often detached from patterns of other environmental variables. A comparison of PC5 and PC6 with other PCs shows that evapotranspiration correlates with PC5 and oxygen isotope variable correlates with PC6 (Supplementary Table S2).
Therefore, PC4, PC5, and PC6 separate the variance of specific variables, indicating that the dynamics of these variables are largely independent (orthogonal) from the co-variability axis of the photosynthesis/respiration, the atmospheric dynamics, and the synoptic scale structure of the atmosphere (PC1, PC2, and PC3, respectively). Interestingly, the independence of the oxygen isotopic signal (PC6) from that of the carbon isotope (PC1) indicates that drivers of the oxygen fractionation (e.g., the oxygen exchange between CO2 and water in either the plant leaves or the soil water on the synoptic scale) are distinct from those of the carbon fractionation (i.e., diurnal local photosynthesis, carbon fixation, and local CO2 emissions). Notably, the low saturation of δ18O-CO2 in the primary components (PC1–PC3) is a meaningful physical result rather than a statistical limitation. Its emergence in a separate, orthogonal component (PC6) statistically validates the decoupling of the oxygen exchange processes (primarily driven by the water cycle and leaf-water equilibration) from the carbon fixation and combustion processes that dominate the first components. Moreover, the temporal scale of significant variations of oxygen signal diverges from those of the carbon signal, providing an independent tool for analyzing the processes involving CO2 in urban zones.
The PCA provides an integrated analysis of the variables affecting the evolution of atmospheric CO2 in the city of Palermo and allows disentangling the effects of photosynthesis, anthropogenic emissions, and biogeochemical variations (PC1); the atmospheric dynamics at the district scale (PC2); and the synoptic scale circulation effect (PC3) as the dominant processes affecting the evolution of the atmospheric CO2. Moreover, some environmental variables, such as rain and evapotranspiration, have self-explained variances and are independent from the relevant PCs. The rain variable saturates PC4 and its eigenvalue approximately to the threshold of acceptance (i.e., the Kaiser criterion), which indicates that it is relevant at the temporal scale of this study (i.e., the year temporal scale). The evapotranspiration dominates PC5 (loading = 0.874), although it is moderately correlated with PC1, indicating that it relies on a combination of several factors (e.g., soil humidity and vapor pressure deficit) that are not completely captured by PC1. However, the most relevant result of the PCA over the dataset collected in Palermo is the wide independence of δ18O-CO2 (i.e., PC6 loading = 0.730), in contrast to the moderate dependence on PC1, which is saturated by δ13C-CO2. This result supports the hypothesis that oxygen isotopic signal carries additional information about the origin of CO2. Arguably, the independent component of the oxygen isotope signal carries information about the carbon amount fixed in the ecosystem after removing the carbon amount released in the environment by either autotrophic or heterotrophic and autotrophic respiration. This is due to several biogeochemical reactions involving CO2 with water vapor (i.e., i.e., CO2 that enters the plant leaves and the water in the stomata and equilibration with soil water that depends on the hydrology of the region). It has been established that these processes leave a distinct isotopic signal on the oxygen isotopes, which is wider than that on the carbon isotopes (). This phenomenon involves extensive oxygen isotopic exchanges that occur during the process of establishing isotopic equilibration between carbon dioxide and water vapor. The CO2–H2O isotopic equilibration in the atmosphere requires approximately 30 s of CO2 exposure to H2O to reach equilibrium (). In contrast, CO2 achieves the oxygen isotopic equilibrium with water in the plant leaves approximately 104–105 times faster because of the enhancement of the photosynthetic reactions, providing early and complete equilibration. Consequently, variations in airborne δ18O-CO2 reflect a shift in the interaction between the carbon and water cycles mediated by the biosphere. At the global scale, variations of δ18O-CO2 depend on the land–water distribution (i.e., the distribution of continental plates on the Earth surface). At the regional scale, the pattern observed in the yearly recordings of δ18O-CO2 carries the signal of the growing season (; ; ; ; ; ; ; ). The reduced range of CO2 concentrations in the city of Palermo compared to that in other cities is consistent with the PCA results, which highlight two dominant mitigation mechanisms that are specific to the Palermo urban area. First, the strong influence of wind dynamics (PC2) indicates that coastal ventilation and regular sea-breeze circulation effectively disperse local emissions, preventing the accumulation and formation of an extreme CO2 dome. Second, the coupling of CO2 and δ13C-CO2 with solar radiation (PC1) indicates that the region’s lower latitude may support a vigorous and extended photosynthetic season. Under these conditions, enhanced biospheric uptake could potentially mitigate the anthropogenic impact. Consequently, both advective mixing and biological carbon fixation act to dampen the amplitude of the urban CO2 signal, although a contribution from lower absolute emission rates in the city of Palermo cannot be excluded. Overall, the PCA highlights that variability of atmospheric CO2 in Palermo is primarily controlled by the interaction between biospheric carbon exchange, anthropogenic emissions, and atmospheric transport processes operating across the local and synoptic scales. The independence of the δ18O-CO2 signal further indicates that oxygen isotopes provide complementary information on ecosystem–atmosphere interactions beyond that captured by carbon isotopes alone.
5 Conclusion
The integrated monitoring of stable isotopes and the concentration of CO2 conducted at the ACO-Pa1 station in Palermo establishes a robust observational framework for disentangling the complex drivers of urban atmospheric carbon dynamics. By integrating continuous, high-frequency measurements of CO2 concentration with stable isotope ratios (δ13C-CO2 and δ18O-CO2), in this study, we successfully characterized a sustained urban CO2 enhancement that persistently exceeds global background levels by approximately 5%, modulated by diurnal planetary boundary layer dynamics and seasonal variability. The application of Keeling plot analysis provided critical constraints on source attribution, reliably identifying fossil fuel and natural gas combustion as the primary agents driving urban high-pollution episodes, particularly during the fall and winter months when the anthropogenic heating demands peak.
Probability plot analysis resolves four distinct CO2 populations, underscoring the capability of this approach to detect localized emission events against broader synoptic variability. Furthermore, the statistical identification of four distinct CO2 populations by probability plot analysis highlights the high sensitivity of this method in detecting specific local pollution events against the backdrop of broader synoptic atmospheric circulation.
A pivotal finding of this study is the distinct behavior of isotopic tracers revealed through PCA. While the δ13C-CO2 and bulk concentration showed strong coupling with anthropogenic emissions and seasonal biospheric carbon fixation, the δ18O-CO2 signal emerged as a largely independent tracer. This decoupling underscores the unique utility of oxygen isotopes in tracking biosphere–atmosphere exchanges and local hydrological processes, such as evapotranspiration and equilibration with soil water, offering a complementary layer of information that is often obscured in single-tracer studies. Together, these results show that multi-species isotopic monitoring substantially enhances top-down emission estimates, improves the understanding of coupled carbon–water biogeochemical cycles, and provides a powerful diagnostic tool for urban air-quality assessment, where CO2 variability co-occurs with other harmful pollutants.
Ultimately, these results confirm that multi-species isotopic monitoring is indispensable for validating top-down emission estimates in complex urban terrains. As Mediterranean cities face increasing vulnerability to extreme climate events, such high-resolution diagnostic tools are essential for guiding effective decarbonization strategies and verifying compliance with international mitigation targets. This study, therefore, provides a scalable template for resolving the specific contributions of human activity versus natural variability in medium-sized urban environments.
Statements
Data availability statement
The datasets presented in this study can be found in online repositories. The names of the repository is ZENODO (https://doi.org/10.5281/zenodo.18921912, Years 2023-2024).
Author contributions
RD: Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Software, Supervision, Validation, Visualization, Writing – original draft, Writing – review and editing. SG: Conceptualization, Investigation, Writing – review and editing. ML: Conceptualization, Investigation, Writing – review and editing. GC: Writing – review and editing. MC: Writing – review and editing. CD: Writing – review and editing.
Funding
The author(s) declared that financial support was not received for this work and/or its publication.
Conflict of interest
The author(s) declared that this work was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
Generative AI statement
The author(s) declared that generative AI was not used in the creation of this manuscript.
Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.
Publisher’s note
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.
Supplementary material
The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fenvs.2026.1829977/full#supplementary-material
References
1
AdamO.Shourky WolffM.GarfinkelC. I.ByrneM. P. (2023). Increased uncertainty in projections of precipitation and evaporation due to wet-get-wetter/dry-get-drier biases. Geophys. Res. Lett.50, e2023GL106365. 10.1029/2023GL106365
2
AhmadiP.DincerI.RosenM. A. (2013). “Environmental impact assessment of integrated multigeneration energy system,” in Causes, impacts and solutions to global warming. Editors DincerI.ColpanC. O.KadiogluF. (Springer), 751–777.
3
AndreoniV. (2021). Estimating the european CO2 emissions change due to COVID-19 restrictions. Sci. Tot. Env.769, 145115. 10.1016/j.scitotenv.2021.145115
4
AndrewR. M. (2021). Towards near real-time, monthly fossil CO2 emissions estimates for the european Union with current-year projections. Atmos. Pollut. Res.12 (12), 101229. 10.1016/j.apr.2021.101229
5
ArrighiC.DomeneghettiA. (2025). Brief communication: on the environmental impacts of the 2023 floods in Emilia-Romagna (italy).
6
BallingR. C.Jr.CervenyR. S.IdsoC. D. (2001). Does the urban CO2 dome of phoenix, Arizona contribute to its heat island?Geophys. Res. Lett.28, 4599–4601. 10.1029/2000GL012632
7
BarichivichJ.BriffaK. R.MyneniR. B.OsbornT. J.MelvinT. M.CiaisP.et al (2013). Large-scale variations in the vegetation growing season and annual cycle of atmospheric CO2 at high northern latitudes from 1950 to 2011. Glob. Change Biol.19, 3167–3183. 10.1111/gcb.12283
8
BognerJ. R. E.SweeneyD.ColemanR.HuitricR.RirieG. T. (1996). Using isotopic and molecular data to model landfill gas processes. Waste Manag. Res.14, 367–376. 10.1006/wmre.1996.0037
9
CalidonnaC. R.DuttaA.D’AmicoF.MalacariaL.SinopoliS.De BenedettoG.et al (2025). Ten-year analysis of mediterranean coastal wind profiles using remote sensing and in situ measurements. Wind5 (2), 9. 10.3390/wind5020009
10
CalvertJ. G.HeywoodJ. B.SawyerR. F.SeinfeldJ. H. (1993). Achieving acceptable air quality: some reflections on controlling vehicle emissions. Science261 (5117), 37–45. 10.1126/science.261.5117.37
11
CamardaM.De GregorioS.CapassoG.Di MartinoR. M. R.GurrieriS.PranoV. (2019). The monitoring of natural soil CO2 emissions: issue and perspectives. Earth Sci. Rev.198, 102928. 10.1016/j.earscirev.2019.102928
12
CaminF.BesicD.BrewerP.AllisonC.CoplenT.DunnP.et al (2025). Stable isotope reference materials and scale definitions—outcomes of the 2024 IAEA experts meeting. Rapid Commun. Mass Spectrom.39, e10018. 10.1002/rcm.10018
13
CanadellJ. G.Le QuéréC.RaupachM. R.FieldC. B.BuitenhuisE. T.CiaisP.et al (2007). Contributions to accelerating atmospheric CO2 growth from economic activity, carbon intensity, and efficiency of natural sinks. Proc. Natl. Acad. Sci. U. S. A.104 (47), 18866–18870. 10.1073/pnas.0702737104
14
CarapezzaM. L.Di MartinoR. M. R.CapassoG.Di GangiF.RanaldiM.TarchiniL. (2024). “Stable isotope composition and airborne concentration of CO2 in rome capital city (italy),”, 2024. Vienna, Austria, EGU24–EGU11294. 10.5194/egusphere-egu24-11294EGU General Assem.
15
CattelR. B. (1966). The scree test for the number of factors. Multivar. Behav. Res.1 (2), 245–276. 10.1207/s15327906mbr0102_10
16
CiaisP.YaoY.GasserT.BacciniA.WangY.LauerwaldR.et al (2021). Empirical estimates of regional carbon budgets imply reduced global soil heterotrophic respiration. Nat. Sci. Rev.8 (2), nwaa145. 10.1093/nsr/nwaa145
17
ClarkI.FritzP. (1997). Environmental isotopes in hydrogeology. New York: CRC Press, 328.
18
Clark-ThorneS. T.YappC. J. (2003). Stable carbon isotope constraints on mixing and mass balance of CO2 in an urban atmosphere: dallas metropolitan area, Texas, USA. Appl. Geochem.18 (1), 75–95. 10.1016/S0883-2927(02)00054-9
19
De DatoG. D.De AngelisP.SircaC.beierC. (2010). Impact of drought and increasing temperatures on soil CO2 emissions in a mediterranean shrubland (Gariga). Plant Soil327, 153–166. 10.1007/s11104-009-0041-y
20
Di BellaG.Di TrapaniD.VivianiG. (2011). Evaluation of methane emissions from Palermo municipal landfill: comparison between field measurements and models. Waste Manage.31, 1820–1826. 10.1016/j.wasman.2011.03.013
21
Di MartinoR. M. R.GurrieriS.LiottaM.CapassoG.ChiappiniM.DoglioniC. (2025). Atmospheric carbon and oxygen laboratory - PA1-station_2023-2024. Zenodo10.5281/zenodo.15044199
22
Di MartinoR. M. R.CapassoG. (2021). On the complexity of anthropogenic and geological sources of carbon dioxide: onsite differentiation using isotope surveying. Atm. Env.2556, 118446. 10.1016/j.atmosenv.2021.118446
23
Di MartinoR. M. R.GurrieriS. (2022). Theoretical principles and application to measure the flux of carbon dioxide in the air of urban zones. Atm. Env.288 (4), 119302. 10.1016/j.atmosenv.2022.119302
24
Di MartinoR. M. R.GurrieriS. (2023). Quantification of the volcanic carbon dioxide in the air of vulcano Porto by stable isotope surveys. J. Geophys. Res. Atmos.128, 2022JD037706. 10.1029/2022jd037706
25
Di MartinoR.GurrieriS. (2026). Theoretical principles and application to measure the flux of carbon dioxide in the air of urban zones - dataset [data set]. Atmos. Environ.288, 119302.
26
Di MartinoR. M. R.CamardaM.GurrieriS.ValenzaM. (2016). Asynchronous changes of CO2, H2 and he concentrations in soil gases: a theoretical model and experimental results. J. Geophys. Res. Solid Earth.121, 1565–1583. 10.1002/2015JB012600
27
Di MartinoR. M. R.CapassoG.CamardaM.De GregorioS.PranoV. (2020). Deep CO2 release revealed by stable isotope and diffuse degassing surveys at vulcano (aeolian islands) in 2015–2018. J. Volcanol. Geoth. Res.401, 106972. 10.1016/j.jvolgeores.2020.106972
28
Di MartinoR. M. R.CamardaM.GurrieriS. (2021). Continuous monitoring of hydrogen and carbon dioxide at Stromboli volcano (aeolian islands, Italy). Ital. J. Geosci.141, 79–94. 10.3301/IJG.2020.26
29
Di MartinoR. M. R.GurrieriS.CamardaM.CapassoG.PranoV. (2022). Hazardous changes in soil CO2 emissions at vulcano, Italy, in 2021. J.G.R. Solid Earth127, e2022JB024516. 10.1029/2022JB024516
30
Di MartinoR. M. R.GurrieriS.PaonitaA.CaliroS.SantiA. (2024). Unveiling spatial variations in atmospheric CO2 sources: a case study of metropolitan area of naples, Italy. Sci. Rep.14, 20483. 10.1038/s41598-024-71348-9
31
FarquharG. D.LloydJ.TaylorJ. A.FlanaganL. B.SyvertsenJ. P.HubickK. T.et al (1993). Vegetation effects on the isotope composition of oxygen in atmospheric CO2. Nature363, 439–443. 10.1038/363439a0
32
FeketeA.EstranyJ.RamírezM. Á. A. (2025). Cascading impact chains and recovery challenges of the 2024 Valencia catastrophic floods. Discov. Sustain6, 586. 10.1007/s43621-025-01483-4
33
FlanaganL. B.BrooksJ. R.VaneryG. T.EhleringrJ. R. (1997). Discrimination against C18O16O during photosynthesis and the oxygen isotope ratio of respired CO2 in boreal forest ecosystems. Glob. Biogeochem. Cy.11 (1), 83–98. 10.1029/96GB03941
34
FranceyR. J.TansP. P. (1987). Latitudinal variation in oxygen-18 of atmospheric CO2. Nature327 (6122), 495–497. 10.1038/327495a0
35
FriedlingsteinP.JonesM. W.O'SullivanM.AndrewR. M.HauckJ.PetersG. P.et al (2019). Global carbon budget 2019. Earth Syst. Sci. Data11, 1783–1838. 10.5194/essd-11-1783-2019
36
FriedlingsteinP.HoughtonR. A.MarlandG.HacklerJ.BodenT. A.ConwayT. J.et al (2010). Update on CO2 emissions. Nat. Geosci.3, 811–812. 10.1038/ngeo1022
37
FriedlingsteinP.O’SullivanM.JonesM. W.AndrewR. M.HauckJ.LandschützerP.et alGlobal carbon budget 2024 (2024). Earth Syst. Sci. Data Discuss., [preprint]10.5194/essd-2024-519
38
FriedlyH.SiegenthalerU.RauberD.OeschgerH. (1987). Measurements of concentration 13C/12C and 18O/16O ratios of tropospheric carbon dioxide over Switzerland. Tellus39B, 80–88. 10.1111/j.1600-0889.1987.tb00272.x
39
FujitaD.IshizawaM.MaksyutovS.ThorntonP. E.SaekiT.NakazawaT. (2003). Inter-annual variability of the atmospheric carbon dioxide concentrations as simulated with global terrestrial biosphere models and an atmospheric transport model. Tellus B Chem. Phys. Meteorology55 (2), 530–546. 10.3402/tellusb.v55i2.16721
40
GemeryP. A.TrolierM.WhiteJ. W. C. (1996). Oxygen isotope exchange between carbon dioxide and water following atmospheric sampling using glass flasks. J. Geophys. Res. Atmos.101 (D9), 14415–14420. 10.1029/96jd00053
41
GórkaM.Lewicka-SzczebakD. (2013). One-year spatial and temporal monitoring of concentration and carbon isotopic composition of atmospheric CO2 in a wrocław (SW Poland) city area. Appl. Geochem.35, 7–13. 10.1016/j.apgeochem.2013.05.010
42
GurrieriS.Di MartinoR. M. R.CamardaM.FrancofonteV. (2023). Monitoring CO2 hazard of volcanic origin: a case study at the island of vulcano (italy) during 2021–2022. Geosciences13, 266. 10.3390/geosciences13090266
43
GuyR. D.FogelM. L.BerryJ. A. (1993). Photosynthetic fractionation of the stable isotopes of oxygen and carbon. Plant Physiol.101 (1), 37–47. 10.1104/pp.101.1.37
44
HackleyK. C.LiuC. L.ColemanD. D. (1996). Environmental isotope characteristics of landfill leachates and gases. Ground Water34, 827–836. 10.1111/j.1745-6584.1996.tb02077.x
45
HasanbeigiA.PriceL.LinE. (2012). Emerging energy–efficiency and CO2 emission–reduction technologies for cement and concrete production: a technical review. Renew. Sustain. Energy Rev.16 (8), 6220–6238. 10.1016/j.rser.2012.07.019
46
HeroldA.SchefflerM.EmeleL.AndersonG.Oeko-Institut e.V. - Institut fuer Angewandte Oekologie, Berlin (Germany) (2018). Early CO2 emission estimates for 2016 based on eurostat monthly energy data. Annual project report.
47
IPCC (2022). in Climate change 2022: impacts, adaptation, and vulnerability. Contribution of working group II to the sixth assessment report of the intergovernmental panel on climate change. Editors PörtnerH.-O.RobertsD. C.TignorM.PoloczanskaE. S.MintenbeckK.AlegríaA.et al (Cambridge, UK and New York, NY, USA: Cambridge University Press. Cambridge University Press), 3056. 10.1017/9781009325844
48
JolliffeI. T.CadimaJ. (2016). Principal component analysis: a review and recent developments. Phil. Trans. R. Soc. A374, 20150202. 10.1098/rsta.2015.0202
49
JonesM. W.AndrewR. M.PetersG. P.Janssens-MaenhoutG.De-GolA. J.DouX.et al (2024). Gridded fossil CO2 emissions and related O2 combustion consistent with national inventories (GCP-GridFEDv2024.0). Zenodo. 10.5281/zenodo.13909046
50
JyotiJ.SwapnaP.KrishnanR. (2023). North Indian Ocean sea level rise in the past and future: the role of climate change and variability. Glob. Planet. Change228, 104205. 10.1016/j.gloplacha.2023.104205
51
KaiserH. F. (1960). The application of electronic computers to factor analysis. Educ. Psychol. Meas.20, 141–151. 10.1177/001316446002000116
52
KeelingC. D. (1958). The concentration and isotopic abundance of carbon dioxide in rural areas. Geochim. Cosmochim. Acta24, 277–298. 10.1016/0016-7037(61)90023-0
53
LanX.TansP.ThoningK. W. (2026). Trends in globally-averaged CO2 determined from NOAA global monitoring laboratory measurements. Version 2026-03. 10.15138/9N0H-ZH07
54
LawrenceJ.BlackettP.Cradock-HenryN. A. (2020). Cascading climate change impacts and implications. Clim. Risk Manag.29, 100234. 10.1016/j.crm.2020.100234
55
Le QuéréC.AndresR.BodenT.ConwayT.HoughtonR.HouseJ. I.et al (2013). The global carbon budget 1959-2011. Earth Syst. Sci. Data5, 165–185. 10.5194/essd-5-165-2013
56
Le QuéréC.JacksonR. B.JonesM. W.SmithA. J. P.AbernethyS.AndrewR. M.et al (2020). Temporary reduction in daily global CO2 emissions during the COVID-19 forced confinement. Nat. Clim. Chang.10, 647–653. 10.1038/s41558-020-0797-x
57
Le QuéréC.PetersG. P.FriedlingsteinP.AndrewR. M.CanadellJ. G.DavisS. J.et al (2021). Fossil CO2 emissions in the post-COVID-19 era. Nat. Clim. Chang.11, 197–199. 10.1038/s41558-021-01001-0
58
LiM.-W.ZhangJ.WuZ.-F.FuY.-S. (2025). Effect of growing season length on gross primary productivity increased in the jinsha river watershed. J. Plant Ecol.18, 1. 10.1093/jpe/rtae108
59
LianX.PiaoS.ChenA.HuntingfordC.FuB.LiL. Z. X.et al (2021). Multifaceted characteristics of dryland aridity changes in a warming world. Nat. Rev. Earth Environ.2, 232–250. 10.1038/s43017-021-00144-0
60
LiaoZ.ZhouB.ZhuJ.JiaH.FeiX. (2023). A critical review of methods, principles and progress for estimating the gross primary productivity of terrestrial ecosystems. Front. Environ. Sci.11, 1093095. 10.3389/fenvs.2023.1093095
61
LiptayK.ChantonJ.CzepielP.MosherB. (1998). Use of stable isotopes to determine methane oxidation in landfill cover soils. J. Geophys. Res.103, 8243–8250. 10.1029/97JD02630
62
LiuZ.MacphersonG. L.GrovesC.MartinJ. B.YuanD.ZengS. (2018). Large and active CO2 uptake by coupled carbonate weathering. Earth-Sci. Rev.182, 42–49. 10.1016/j.earscirev.2018.05.007
63
LiuZ.DengZ.DavisS. J.GironC.CiaisP. (2022). Monitoring global carbon emissions in 2021. Nat. Rev. Earth Environ.3, 217–219. 10.1038/s43017-022-00285-w
64
MulunehM. G. (2021). Impact of climate change on biodiversity and food security: a global Perspective—A review article. Agric and Food Secur10, 36. 10.1186/s40066-021-00318-5
65
NastosP.SaaroniH. (2024). Living in mediterranean cities in the context of climate change: a review. Int. J. Climatol.44 (10), 3169–3190. 10.1002/joc.8546
66
NewmanS.XuX.AffekH. P.StolperE.EpsteinS. (2008). Changes in mixing ratio and isotopic composition of CO2 in urban air from the Los Angeles basin, California, between 1972 and 2003. J. Geophys. Res.113, D23304. 10.1029/2008JD009999
67
ParaschivS.ParaschivL. S. (2020). Trends of carbon dioxide (CO2) emissions from fossil fuels combustion (coal, gas and oil) in the EU member states from 1960 to 2018. Energy Rep.6 (8), 237–242. 10.1016/j.egyr.2020.11.116
68
ParkR.EpsteinS. (1960). Carbon isotope fractionation during photosynthesis. Geochim. Cosmochim. Acta44, 5–15. 10.1016/S0016-7037(60)80006
69
PatakiD. E.EhleringerJ. R.FlanaganL. B.YakirD.BowlingD. R.StillC. J.et al (2003). The application and interpretation of keeling plots in terrestrial carbon cycle research. Glob. Biogeochem. Cycles17 (1), 1022. 10.1029/2001gb001850
70
PoungparnS.KomiyamaA.TanakaA.SangtieanT.MaknualC.KatoS.et al (2009). Carbon dioxide emission through soil respiration in a secondary mangrove forest of eastern Thailand. J. Trop. Ecol.25 (4), 393–400. 10.1017/S0266467409006154
71
RacoB.BattagliniR.DotsikaE. (2014). New isotopic (δ13CCO2–δ13CCH4) fractionation factor limits and chemical characterization of landfill gas. J. Geochem. Explor.145, 40–50. 10.1016/j.gexplo.2014.05.003
72
RaichJ. W.SchlesingerW. H. (1992). The global carbon dioxide flux in soil respiration and its relationship to vegetation and climate. Tellus44B, 81–99. 10.1034/j.1600-0889.1992.t01-1-00001.x
73
RaisA. (2020). “Extreme weather events and human health,” in International case studies. Nature Switzerland AG: Springer.
74
RamnarineR.Wagner-RiddleC.DunfieldK. E.VoroneyR. P. (2012). Contribution of carbonates to soil CO2 emissions. Can. J. Soil Sci.92, 599–607. 10.4141/CJSS2011-025
75
RobinsonC.DilkinaB.Moreno-CruzJ. (2020). Modeling migration patterns in the USA under sea level rise. PLoS ONE15 (1), e0227436. 10.1371/journal.pone.0227436
76
ScarpelliT. R.PalmerP. I.LuntM.SuperI.DrosteA. (2024). Verifying national inventory-based combustion emissions of CO2 across the UK and mainland Europe using satellite observations of atmospheric CO and CO2. Atmos. Chem. Phys.24, 10773–10791. 10.5194/acp-24-10773-2024
77
SherwoodO. A.SchwietzkeS.ArlingV. A.EtiopeG. (2017). Global inventory of gas geochemistry data from fossil fuel, microbial and burning sources, version 2017. Earth Syst. Sci. Data9, 639–656. 10.5194/essd-9-639-2017
78
ShumacherM.WernerR. A.MeijerH. A. J.JansenH. G.BrandW. A.GeilmannH.et al (2011). Oxygen isotopic signature of CO2 from combustion processes. Atmos. Chem. Phys.11, 1473–1490. 10.5194/acp-11-1473-2011
79
SinclairA. J. (1974). Selection of threshold values in geochemical data using probability graphs. J. Geochem. Explor.3, 129–149. 10.1016/0375-6742(74)90030-2
80
SkrzypekG.AllisonC. E.BöhlkeJ. K.BontempoL.BrewerP.CaminF.et al (2022). Minimum requirements for publishing hydrogen, carbon, nitrogen, oxygen and sulfur stable-isotope Delta results (IUPAC technical report), Pure Appl. Chem., vol. 94, no. 11-12, pp. 1249–1255. 10.1515/pac-2021-1108
81
TangX.PeiX.LeiN.LuoX.LiuL.ShiL.et al (2020). Global patterns of soil autotrophic respiration and its relation to climate, soil and vegetation characteristics. Geoderma369, 114339. 10.1016/j.geoderma.2020.114339
82
TarquiniS.IsolaI.FavalliM.MazzariniF.BissonM.PareschiM. T.et al (2007). TINITALY/01: a new triangular irregular network of italy, annals of geophysics. 50, 407–425. 10.4401/ag-4424
83
UmairM.KimD.ChoiM. (2020). Impact of climate, rising atmospheric carbon dioxide, and other environmental factors on water-use efficiency at multiple land cover types. Sci. Rep.10, 11644. 10.1038/s41598-020-68472-7
84
United Nations Environment Programme (2025). We are all in this together - annual report 2024.
85
ViethA.WilkesH. (2010). “Stable isotopes in understanding origin and degradation processes of petroleum,” in Handbook of hydrocarbon and lipid microbiology. Editor TimmisK. N. (Berlin: Springer), 97–111.
86
VultaggioM.VarricaD.AlaimoM. G. (2020). Impact on air quality of the COVID-19 lockdown in the urban area of Palermo (italy). Int. J. Environ. Res. Public Health17, 7375. 10.3390/ijerph17207375
87
WatanabeM.ObaA.SaitoY.PurevjavG.GankhuyagB.Byamba-OchirM.et al (2023). Enhancing scientific transparency in national CO2 emissions reports via satellite-based a posteriori estimates. Sci. Rep.13, 15427. 10.1038/s41598-023-42664-3
88
WidoryD.JavoyM. (2003). The carbon isotope composition of atmospheric CO2 in paris. Earth Planet. Sci. Lett.215, 289–298. 10.1016/S0012-821X(03)00397-2
89
WimbadiR. W.DjalanteR.MoriA. (2021). Urban experiments with public transport for low carbon mobility transitions in cities: a systematic literature review (1990–2020). Sustain. Cities Soc.72, 103023. 10.1016/j.scs.2021.103023
90
YakirD. (2003). “The stable isotopic composition of atmospheric CO2,” in The atmosphere, treatise on geochemistry. Editors HollandH. D.TurekianK. K. (Oxford: Elsevier-Pergamon), 4, 175–212. 10.1016/b0-08-043751-6/04038-x
91
ZhuF.ZhuM.YangW.WangZ.GuoY.YaoT. (2024). Drivers of the extreme early spring glacier melt of 2022 on the central Tibetan Plateau. Earth Space Science11, e2023EA003297. 10.1029/2023EA003297
Summary
Keywords
airborne CO2, anthropogenic CO2 emissions, fossil fuel combustion, principal component analysis, stable isotopes, urban CO2 monitoring, δ13C-CO2, δ18O-CO2
Citation
Di Martino RMR, Gurrieri S, Liotta M, Capasso G, Chiappini M and Doglioni C (2026) Deciphering carbon dioxide signals in urban environments: integrated stable isotope and concentration monitoring for evaluating the impact of human activities on the atmospheric composition. Front. Environ. Sci. 14:1829977. doi: 10.3389/fenvs.2026.1829977
Received
13 March 2026
Revised
19 April 2026
Accepted
21 April 2026
Published
02 June 2026
Volume
14 - 2026
Edited by
Greg Skrzypek, The University of Western Australia, Australia
Reviewed by
Maria De Fatima Andrade, University of São Paulo, Brazil
Martina Ferrari, University of Florence, Italy
Updates
Copyright
© 2026 Di Martino, Gurrieri, Liotta, Capasso, Chiappini and Doglioni.
This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
*Correspondence: Roberto M. R. Di Martino, roberto.dimartino@ingv.it
Disclaimer
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article or claim that may be made by its manufacturer is not guaranteed or endorsed by the publisher.