GSN Quality Issues: Motivation for 2010 Assessment

Although installed as a state-of-the-art seismic network, the GSN is constantly aging.  A major rejuvenation of the acquisition systems and ancillary sensors and systems began in 2010.  The enhanced capabilities of the upgraded field systems significantly improved the ability to track data quality, through remote calibrations and inter-sensor comparisons. However, the GSN Operators faced a serious situation with the workhorse primary sensors in the network, and in particular, the Streckeisen STS-1. There is, quite literally, no equal to this sensor, and as such, no replacement (the STS-1 is no longer produced).

From 2005-2010, the STS-1 sensors at some stations started to show degradation in the stability of their long period response. As early as 2007, the Lamont Waveform Quality Center (WQC) noted issues with several STS-1 sites, but the source of the sensitivity anomalies was unclear.WQC intersensor comparison for KIP  Work conducted from 2007-2010 at the USGS Albuquerque Seismological Laboratory (publication in the works) shows that aging of the seals on the STS-1 feedback electronics (FBEs) can introduce a non-linearity in the instrument response (related to humidity in the electronics) near the long-period corner of the instrument (around 360 sec), that appeared to be the primary source of the problem.  This did not apparently affect either the short period (microseism or above) or the ultra-long period (tidal) ends of the spectrum, and therefore, had gone undetected by past standard QC techniques of the network operators.  Further studies by the WQC and the network operators suggest that ~15-20% of the STS-1's deployed may have been affected. The extensive analyses carried out to identify this problem also revealed other (possibly unrelated) issues with the absolute gains at a few other STS-1 installations. In the STS-1’s that were affected, the sensitivity anomaly appeared to increase with age of the sensor, and stations in high humidity environments were the most susceptible.  The period band near the STS-1 corner frequency (360 s) was most strongly affected, The degradation in response is time-dependent and non-linear, so that it was difficult to make adjustments to the metadata and transfer functions to describe the actual system response, yet the IDA group had some success with this approach at a few sites. 

With humidity identified as the apparent source of FBE degradation, a re-design and retro-fit of the FBE package was a viable solution.  Through a successful development effort funded by IRIS/NSF, Metrozet collaborated with UC Berkeley in the creation of a replacement set of FBEs for these sensors. With augmentation funding (as a result of the ARRA), GSN Operators acquired replacement FBEs for the entire network.

The design for GSN stations calls for redundant VBB sensors and these are currently installed at many locations (mostly STS-2's, Guralp CMG-3T's and Trillium T-240's). With NSF augmentation funds, we intend to install redundant secondary sensors at all GSN stations and as of 2010, this work is underway.  This provides not only a reliable backup to the aging primary sensors, but also allows for inter-sensor comparison to help resolve any transducer issues that may arise.

During 2010, the GSN operators collaborated with the Lamont WQC to produce an overview of known sensor response problems, past and present, with GSN sensors and began the task of publishing revised metadata, where possible. To help users avoid sensors with known problems and to help users identify problems they may observe, a brief snapshot of the current knowledge of problems was developed based on a thorough analysis of the primary sensor data returned from the M8.8 Chilean Earthquake from 2010.

Both GSN Network Operations centers also publish a list of current station problems on their local Data Collection Center webpages (USGS/ASL and IRIS/IDA). This does not include historic problems, but may include problems that have been recently updated during the major field effort currently underway. It is, however, a snapshot of the current issues. Note that the "damped response" issues with many STS-1's relates to the FBE response problems noted above.

In the past, we have encouraged contributing data centers and the WQC to provide us with information on any problems they encounter with GSN data quality using the Data Problem Report.  We are considering options to allow other data users to have input on identification of station problems.

The STS-1 FBE problem represents only one example of several aging issues.  The FBE replacement does not address issues with the mechanical assembly for the STS-1.  In addition, we reviewed reliability issues with our primary borehole sensor (KS-54000).  Several problems pointed to the need to find the next generation primary sensors, and IRIS initiated work with potential developers and vendors to provide long-term solutions.  In addition, IRIS also needed to ensure that the metadata is adequately represented for the aging instruments to the best of our ability, and inform the data user community of those instances when the parametric data descriptions are inadequate in describing the true response.

In addition to sensors, the infrastructure at the GSN stations is aging right along with the electronics.  Many of the sites are decades old, and there are a number of instances where buildings are crumbling and leaking, vaults and piers are eroding, boreholes leak, etc.  These can all effect data quality and are very difficult to capture in metadata as they result in a slow and gradual deterioration of signal quality.  As renewal began with the electronic base of the GSN, IRIS also planned to rebuild the infrastructure and worked to understand any loss in quality that infrastructure aging problems may have produced.