An overview on wavelet multi-resolution decomposition compared with traditional frequency domain filtering for continuous gravity data denoising

Abstract

Continuous gravity recordings in volcanic area could play a fundamental role in the monitoring of active volcanoes and in the prediction of eruptive events too. This geophysical methodology is utilized, on active volcanoes, in order to detect mass changes linked to magma transfer processes and, thus, to recognize forerunners to paroxysmal volcanic events. Spring gravimeters are still the most utilized instruments for microgravity studies because of their relatively low cost and small size, which make them easy to transport and install. Continuous gravity measurements are now increasingly performed at sites very close to active craters, where there is the greatest opportunity to detect significant gravity changes due to a volcanic activity. Unfortunately, spring gravity meters show a strong influence of meteorological parameters (i.e. pressure, temperature and humidity), especially in the adverse environmental conditions usually encountered at such places. As the gravity changes due to the volcanic activity are very small compared to other geophysical or instrumental effects we need a new mathematical tool to get reliable gravity residuals susceptible to reflect the volcanic effect. In the following we present and discuss a preliminary work about the confrontation between the traditional filtering methodology and the Wavelet transform. The overall results show that the performance of the wavelet-based filter seems better than the Fourier one. Moreover, the possibility of getting a multi-resolution analysis and study local features of the signal in the time domain makes the proposed methodology a valuable tool for gravity data processing.