The main indicator of the quality of a seismic station is the level of the noise recorded in the absence of seismic signals. Noise sources include anthropogenic activity, but also ocean microseisms, temperature and pressure fluctuations, as well as noise from the sensors and dataloggers themselves. The level of the noise recorded on a seismometer can be reduced by locating the station away from human activity, selecting good geology and ensuring good vault design, including thermal and pressure insulation and excellent ground coupling. Modern scientists are now not only interested in recording earthquake signals at the best quality, but also in recording ‘high quality noise’! Since modern seismic stations have very broadband capabilities, and network analysts and scientists are interested in scientific analysis of signals spanning an increasingly wide frequency range, its critical to document the actual performance of a seismic station across the frequency range. The background noise at a site can also vary over time, as human activity in the vicinity can change, and tracking the noise allows network managers to understand the performance of station through time.
The standard tool to analyse the noise performance is the Power Spectral Density (PSD), where the frequency content of the waveform from a period of time is evaluated. PSDs can be calculated in time steps to cover the entire time a station records data. The individual PSDs can be combined to produce a Probability Density Function (PDF) for the PSDs. These PDFs are a very powerful tool that can be used to rapidly evaluate the quality of each seismic station. They are also crucial to verify basic data quality issues, such as faulty hardware, and incorrect station metadata (the PSDs are calculated on acceleration time series, so the instrument response must be known and is removed during processing). At the SED, we use a commercial software, SQLX, to compute a database of PSDs for every channel recorded for every station in our archive. All new data is added on a daily basis. An interactive GUI is available to allow the analyst to track the PSD PDFs over time. SQLX follows the methodology of McNamara and Buland (2004).
On these pages, for each channel, we show noise PDF snapshots over different periods of time - the entire period for which the station is operational, the last day, and the last week. Changes on these timescales can be used to identify recent trends in station performance. Different pages are available for networks (CH - the permanent Swiss network, as well as temporal networks such as the AlpArray Seismic Network Z3 and the targeted EASI XT and CASE 8X Networks), sensor types (broadband HH, strong motion, HG and short period EH) and sampling rates (HH for >100sps, LH for 1sps).