Many years of experience in operating Teledyne Geotech KS borehole seismometers has established that the horizontal components are usually noisier than the vertical component at periods larger than about 30 seconds despite the fact that the instruments are installed quite deep beneath the surface. This phenomenon is quite frustrating because the purpose of operating the instruments below the surface is to reduce the horizontal noise to lower levels than would be measured on the surface. Several procedures have been attempted to control air motion within the borehole, with varying degrees of success. These include wrapping the seismometer with a layer of foam insulation, wrapping the centralizer assembly with foam insulation, and installing foam plugs just above the sensor and near the top of the borehole. Throughout the years, it has been suggested that the remnant horizontal noise in borehole installations is tilt noise generated by air motion in the vicinity of the instrument package itself.
Experiments aimed at controlling air motion near the sensor by filling the empty space with sand have been underway at ASL for about a year and a half. Initial investigations were conducted in a shallow borehole (about 25 feet deep) and were aimed at confirming that the seismometer could be removed from the hole if it was installed in sand; removing the sensor from dry sand proved to be quite easy. During calm atmospheric conditions, the horizontal noise levels observed in a shallow borehole with sand installation indicated significant improvement over those normally observed in a conventional installation. A deeper borehole experiment was deemed necessary.
Therefore, between days 81 and 87 of 1995, ANMO was converted from a traditional KS borehole installation to a sand installation as follows. The holelock was removed from the borehole and discarded; the pilot on the bottom of the sensor package was replaced by a new assembly designed to increase the area of the load bearing surface at the bottom of the seismometer. Ordinary playground sand was poured into the borehole to a depth of approximately two feet above the bottom of the hole. The seismometer was lowered into the hole until it rested on the sand, at a depth of 498 feet below the surface. Sand was then poured into the hole, using a volume calculated to fill up the space between the sensor and the walls of the borehole to the top of the seismometer.
Figure 1. One hour of particle motion before and after sand installation in the 20 to 600 second pass band.
Despite the fact that the original horizontal noise levels were not abnormally high, ANMO long period noise levels using the sand installation are significantly lower than previously, as is illustrated by the particle motion plots in Figure 1. This figure illustrates the relative levels of particle motion in the 20 to 600 second band during an event-free hour. It is evident in this figure that the sand installation produces considerably less noise than the conventional installation. The long-term relative noise levels are illustrated in the time domain in Figure 2 over approximately seven days for both installation methods. Figure 3 demonstrates the improved noise levels in the frequency domain.
Figure 2. Seven day time domain records in the 20 to 600 second pass band of relative noise levels in sand and no-sand installations.
In the future, ASL plans to convert several of the noisier borehole sites in the IRIS/USGS GSN to sand installations. If successful, the sand should reduce the noise level on the horizontal sensors in noisy boreholes.
Users of GSN data should be warned that horizontal borehole sensors installed in sand will probably not be oriented north-south and east-west because a holelock orienting system is not being used. However, the sensor orientation will be known and will be so noted in the appropriate SEED header blockett. As a precaution against inadvertent confusion by data users, the seed names of the horizontal components of non north-south and east-west oriented KS sensors will no longer end in N and E respectively; instead, they will end in 1 and 2 to warn the user to refer to the header for the true sensor orientation azimuth.
Figure 3. Power spectral denisties of SAND and NO SAND installations. LNM is the USGS Low Noise Model.
Go to next Article
Return to: Newsletter Title Page
Return to: Information on IRIS