An alternative method to evaluate earthquake detection from synthetic Wood–Anderson seismograms: an application in Italy

Abstract

Summary Defining the regional variability of minimum magnitude for earthquake detection is crucial for planning seismic networks. Knowing the earthquake detection magnitude values is fundamental for the optimal location of new stations and to select the priority for reactivating the stations of a seismic network in case of a breakdown. In general, the assessment of earthquake detection is performed by analysing seismic noise with spectral or more sophisticated methods. Further, to simulate amplitude values at the recording sites, spectral methods require knowledge of several geophysical parameters including rock density, S wave velocity, corner frequency, quality factor, site specific decay parameter and so on, as well as a velocity model for the Earth's interior. The simulation results are generally expressed in terms of Mw and therefore a further conversion must be done to obtain the values of local magnitude (ML), which is the parameter commonly used for moderate and small earthquakes in seismic catalogues. Here, the relationship utilised by a seismic network to determine ML is directly applied to obtain the expected amplitude (in mm, as if it were recorded by a Wood-Anderson seismometer) at the recording site, without any additional assumptions. The station detection estimates are obtained by simply considering the ratio of the expected amplitude with respect to the background noise, also measured in mm. The seismic noise level for the station is estimated starting from 4 waveforms (each signal lasting one minute) sampled at various times of the day for a period of one week. The proposed method is tested on Italian seismic events occurring in 2019 by using the locations of 16.879 earthquakes recorded by 374 stations. The first results indicate that by evaluating the station noise level with 5-second windows, a representative sample of the variability in expected noise level is generated for every station, even if only 4 minutes of signal per day over a week of recordings is used. The method was applied to define the detection level of the Italian National Seismic Network (RSN). The RSN detection level represents a reference for the definition and application of guidelines in the field of monitoring of subsurface industrial activities in Italy. The proposed approach can be successfully applied to define the current performance of a local seismic network (managed by private companies) and to estimate the expected further improvements, requested to fulfil the guidelines with the installation of new seismic stations. This method has been tested in Italy and can be reproduced wherever the local magnitude ML, based on synthetic Wood-Anderson (WA) records, is used.