On the seafloor horizontal displacement from cGPS and compass data in the Campi Flegrei caldera

Seafloor deformationmonitoring is nowroutinely performed in themarine sector of the Campi Flegrei volcanic area (Southern Italy). The MEDUSA infrastructure is formed by four buoys deployed at a water depth ranging from 40 to 96 m, and equipped with cGPS receivers, accelerometers and magnetic compasses to monitor the buoy status and a seafloor module with a bottom pressure recorder and other onboard instruments. The analysis of the time series data acquired by the MEDUSA monitoring infrastructure system allows to study the seafloor deformation in the Campi Flegrei caldera with geodetic accuracy. In a previous work, we show that the time series acquired by the Campi Flegrei cGPS onland network and MEDUSA over the period 2017-2020 are in good agreement with the ground deformation field predicted by a Mogi model which is widely used to describe the observed deformation of an active volcano in terms of magma intrusion. Only for one of the buoys, CFBA (A), the data differ significantly from the model prediction, at a level of similar or equal to 6.9 sigma and of similar or equal to 23.7 sigma for the seafloor horizontal speed and direction, respectively. For this reason, we devised a new method to reconstruct the horizontal sea bottom displacement considering in the analysis both cGPS and compass data. The method, applied to the CFBA buoy measurements and validated also on the CFBC (C) buoy, uses compass data to correct cGPS positions accounting for the pole inclination. Including also systematic errors, the internal consistency, always within similar to 3 sigma for the speed and similar to 2 sigma for the angle, between the results derived for different maximum inclinations of the buoy pole (up to 3.5 degrees) indicates that the method allows to significantly reduce the impact of the pole inclination which, if not properly taken into account, can alter the estimation of the horizontal seafloor deformation. In particular, we find a good convergence of the retrieved velocity and deformation angle as we include in the analysis data from increasing values of the buoy pole inclination. Taking the result derived assuming the maximum allowed cutoff and accounting for statistical and systematic errors, we found a speed v = (3.521 +/- 0.039 (stat) +/- 0.352 (syst)) cm/yr and a deformation direction angle alpha = (-115.159 +/- 0.670 (stat) +/- 7.630 (syst))degrees (statistical errors at 1 sigma quoted from the rms of their values, main systematic errors added linearly). The relative impact of themain potential systematic (statistical) effects increases (decreases) with the cutoff. Our analysis provides a horizontal speed consistent with the model at a level of similar or equal to 5.2 sigma (stat only) or of similar or equal to 0.5 sigma (stat and syst added linearly), and a deformation angle consistent with the model at similar or equal to 4.3 sigma level (stat only) or at similar or equal to 0.3s level (stat and syst added linearly). Correspondingly, the module of the vectorial difference between the velocity retrieved from the data and the velocity of the adopted Mogi model diminishes by a factor of similar or equal to 7.65 +/- 1.23 (stat) or +/- 5.78 (stat + syst) with respect to the previous work. A list of potential improvements to be implemented in the system and instruments is also discussed.