Time evolution of long-period noise levels at IRIS GSN and additional stations
Many of the seismographic stations in the IRIS Global Seismographic Network (GSN) have now been operating for several years and some have generated data for a decade or more. Though background seismic noise spectra for these stations have been determined, frequently soon after installation, data on the temporal evolution of background noise levels are not directly available. The purpose of this page is to present the results of a systematic analysis of long-period background signals recorded by IRIS GSN stations and by similar stations operated by other organizations.
The approach taken is to analyze continuously recorded data from a given station and define a background noise level based on the rms signal level. The signal power in narrow frequency bands is calculated hourly from the rms amplitude and converted to a decibel deviation from Peterson's (1993) new low noise model. For each station, the hourly values for each month are ranked. The background noise level is estimated as the recorded signal level not exceeded a small fraction (usually 10%) of the time.
Background ground noise is generally quite stable over time. Seasonal components are common, but are usually only a few db (at long periods). Large variations in the background noise can be caused by several factors:
The table below provides links to figures that represent the background noise level at some period (usually 100 seconds) over several months. The graphs show individual components: Vertical, Horizontal 1 (usually N-S), or Horizontal 2 (usually E-W). Frequently data from both the long period (LH) and very long period (VH) channels are shown in the same figure. The noise levels derived from different streams should be very similar, though they need not be identical owing to data availability.