Seismic noise characterization for the Buddusò-Ala dei Sardi wind park (Sardinia, Italy) and its impact on the Einstein Telescope candidate site

Wind turbines generate considerable seismic noise and interfere with sensitive instruments, such as permanent and temporary seismic sensors installed nearby, hampering their detection capabilities. This study investigates the seismic noise emission from one of Italy's largest wind farms, consisting of 69 turbines (2 MW each), located in northeastern Sardinia. Characterizing the noise emission from this wind farm is of particular importance due to its proximity to the Italian candidate site for hosting the Einstein Telescope (ET), the third-generation observatory for gravitational waves. We run a passive seismic experiment, "Wind turbIne Noise assEsSment in the Italian site candidate for Einstein Telescope" (WINES), using a linear array of nine broadband stations, installed at increasing distances from the wind farm. Spectral analysis, based on the retrieval of spectrograms and power spectral densities at all stations, shows a significant increase in noise amplitude when the wind farm is in operation. The reconstruction of noise polarization points out that the noise wavefield originates from a direction consistent with the wind farm's location. We recognize four dominant fixed spectral peaks at 3.4, 5.0, 6.8, and 9.5 Hz, corresponding to the modes of vibration of the wind turbine towers. While decreasing in amplitude with distance, the 3.4 Hz peak remains detectable up to 13 km from the nearest turbine. Assuming an amplitude decay model of the form r-alpha, where r is the distance, we estimate a damping factor of alpha similar to 2, which remains rather constant for each of the four main peaks, an observation that we relate to the good geomechanical characteristics of the local terrain, consisting of granitoid rocks. To better evaluate the possible impact of the wind farm noise emission on the ET, we also analyze the seismic data from two permanent stations bordering the ET candidate site area, each equipped with both a surface sensor and a borehole sensor at approximately 250 m depth. Power spectral density analysis for the surface and borehole sensors exhibits similar results and very low noise levels. When the wind farm operates at full capacity, the borehole sensors show an effective noise suppression at depth in the frequency range of interest (1-10 Hz). However, small residual spectral peaks at 3.4 Hz and between 4-6 Hz remain detectable.