Improving the statistical correlations between low seismic events and CO2 variations subtracting the rain contribution

A correlation between low seismic activity and CO2 measurements variations was observed at the Gallicano thermomineral spring, Tuscany, Italy, where an automatic monitoring multiparametric geochemical station is operative since 2003 (Pierotti et al., 2015). The above-mentioned correlation reported a time delay of about 2 days of small earthquakes with respect to CO2 anomalies. Starting from this correlation a conditional probability of earthquake occurrence given the CO2 anomaly detection was calculated, with a probability gain near 4 (Pierotti et al., 2022). A statistical correlation was also calculated between rain events and CO2 anomalies which was observed for rain vents ahead CO2 anomalies of one days. This permitted to distinguish CO2 anomalies due to meteorological versus tectonic activities. Following this distinction, and subtracting the rain contribution to the CO2 variations, a new correlation was observed between small earthquakes and CO2 anomalies which confirmed the past results whit a better performance. The new correlation peak is better defined and concentrated in the time lag of 2 days. The p-values of both earthquake and rain to CO2 correlations were calculated. The correspondent probability gain in an earthquake forecasting experiment, taking into account the rain events, increased from less than 4 to 4.5.