We studied the ambient noise recorded at Irpinia Seismic Network (ISNet), a seismic network installed along the Campania–Lucania Apenninic chain (southern Italy), with the aim of characterizing the noise spectrum for each station as a function of time and the detection threshold of the network. For the latter purpose, we proposed a mixed indirect approach based on the signal‐to‐noise ratio (SNR) in the time domain, with parameterization in the frequency domain. The source signature is represented by the convolution of the Brune source time function with the Azimi attenuation curve. We found that 1.3 is the minimum magnitude an event should have to be detected at least at five stations with an SNR larger than five, wherever it occurs. We observed a space variability of the detection threshold as large as 0.3 units, ascribed to both the geometrical configuration of the network and the differences in the noise levels at the different stations. A sensitivity study indicates that the estimation of the detection threshold is robust for changes in the focal depths and stress drop, while it is strongly affected by the anelastic attenuation. In our case, changes of the reduced time t* in the range 0.015–0.035 s generate changes in the completeness threshold of 0.5 units.
Noise levels were obtained by a statistical analysis on the power spectral density curves along almost three years of continuous data from 22 stations. We found that, at short periods, major time variations are generated by diurnal changes in the wind intensity and other meteorological factors. At longer periods, we retrieved the microseismic peak, resulting from the constructive interference of oceanic waves. We also found an additional peak between 2 and 4 s, correlated with the sea wave height along the Tyrrhenian coast.