Preliminary results from the direct comparison of four atmospheric background detection methods at a central Mediterranean WMO/GAW station: potential implications for long-term CO, CO2, and CH4 monitoring

Long term trends in greenhouse and reactive gases highlight changes in the atmospheric background (AB) and help determining the impact of anthropogenic emissions. The presence of pollution events calls for the implementation of background detection methods, capable of differentiating the AB from fresh anthropogenic emissions. At the Mauna Loa observatory (MLO) operated by NOAA, ad hoc procedures are applied to measurements in order to filter out pollution events and sinks from the AB of key gases such as carbon dioxide (CO₂) and methane (CH₄). These procedures account for MLO’s known sources and sinks in the area, and its location.

At the Lamezia Terme (LMT) observation site in Calabria, Southern Italy, part of the WMO/GAW network, several methods have been implemented to filter out non-AB measurements of carbon monoxide (CO), CO₂, and CH₄. Namely, the BaDS (Background Data Selection), SM (Smoothed Minima), Wind, and the ONRPI (Ozone to Nitrogen Oxides Ratio Proximity Indicator). While the Wind method is based on an algorithm specifically designed to consider LMT’s characteristics as a central Mediterranean site, BaDS and SM were already present in literature and their implementation at LMT involved minor changes. These methods are statistical in nature, while the ONRPI is based on atmospheric chemistry, i.e. the O₃/NO_x ratio. The ONRPI classifies as BKG (Background) data with a O₃/NO_x ratio higher than 100, attributed to very aged air masses. Multiple studies on LMT’s record of greenhouse and reactive gases, as well as aerosols, have expanded the ONRPI and turned LMT into a key hotspot for the implementation of this methodology and its correction factors.

The performance of BaDS, SM, and Wind on the LMT record of CO, CO₂, and CH₄ has been assessed using nearly a decade of continuous measurements, however no attempts have been made to compare these methods with the ONRPI. A direct comparison between the ONRPI and other methodologies has highlighted the presence of data in the LMT record attributed by BaDS, SM, and Wind to the AB of CO, CO₂, and CH₄, which are however characterized by very low O₃/NO_x ratios and therefore affected by local sources of emissions. For example, up to 49.79% of the data classified as URB (Urban, with O₃/NO_x ratio lower than 0.1) by the ONRPI, are flagged by BaDS as representative of CO’s AB. Consequently, the AB itself tends to be overestimated.

A multi-year analysis applied to nearly one decade of continuous measurements at LMT, integrated by statistical evaluations, has shown substantial differences between the seasonal oscillations and annual growth rates of CO₂ and CH₄ as computed by the ONRPI, and those resulting from the other methods. During the boreal summer, the AB of CO filtered by ONRPI is nearly 20% lower than that of the other methods, possibly due to contributions such as Mediterranean open fires which are not filtered out by BaDS, SM, and Wind. While these comparisons need to be further expanded, the findings underline the importance of integrating multiple methodologies for AB detection.