Reference Level Stability of the Canadian Superconducting Gravimeter Installation

The Canadian Superconducting Gravimeter Installation (CSGI) operates a GWR TT70 superconducting gravimeter at a site near Cantley, Quebec. An adjacent pier is the reference point for the Canadian Absolute Gravity Station (CAGS) JILA-2 absolute gravimeter. The co-location of the two instruments provides an opportunity to examine the drift of the superconducting gravimeter and to search for spurious signals in either instrument. It is known that the superconducting gravimeter suffers from occasional tares, but the extent to which there are spurious signals on any time-scale is unknown. Nine separate experiments have been conducted since February 1998, in which the absolute gravimeter was dedicated to rapid sampling of gravity for about a week. From these we have calibrated the superconducting gravimeter, established the drift, and assessed the level of spurious gravity signals in both instruments. The calibration factor recovered from these experiments is -78.3 +/0.1 microgal/V. During the period from early 1998 to late 1999, the superconducting gravimeter maintained a constant reference level to better than three microgals. However, in 1 early 2000, the level of the absolute gravimeter rapidly diverged from that of the superconducting gravimeter achieving a maximum recorded difference of fourteen microgal during April 12-13 2000. As of June 2000 the absolute gravimeter had recovered and the offset between the two is again only two microgal. There is occasionally a rough correlation between the (tide and pressure corrected) residuals of the two gravimeters at the microgal or better level on time-scales down to about a day. However, in several experiments no correlation was apparent and excursions of the absolute gravimeter of several microgal persisting for several hours were noted. Most of the variance of the absolute gravimeter residuals seems to be at periods of hours to a day, and there are occasionally episodes of several hours duration when the instrument reads low by several microgal. Introduction The co-location of an absolute gravimeter (AG) and a superconducting gravimeter (SG) is advantageous to both, because the SG offers rapid, precise samples of relative gravity while the AG delivers absolute gravity measurements although of less precision. Indeed, absolute gravimeters are often used to calibrate superconducting gravimeters (Hinderer et al, 1991), and superconducting gravimeters have been used to assess the precision of absolute gravimeters (Okubo et al, 1997). SG’s are subject to tares, or sudden offsets in reference level. These are routinely detected and repaired, but the only way of assessing the reliability of this procedure is by reference to an absolute gravimeter. Accumulated tares of about 14 μgal have been removed from the SG data used here. The small residual offset between AG and SG of ±3μgal suggests that the detection and correction procedure is working well. In this work, we examine the residuals of both AG and SG to try to establish the level of spurious, or instrumental, signals in both. The co-location of an absolute gravimeter (AG) and a superconducting gravimeter (SG) is advantageous to both, because the SG offers rapid, precise samples of relative gravity while the AG delivers absolute gravity measurements although of less precision. Indeed, absolute gravimeters are often used to calibrate superconducting gravimeters (Hinderer et al, 1991), and superconducting gravimeters have been used to assess the precision of absolute gravimeters (Okubo et al, 1997). In this work we examine the residuals of both to try to establish the level of spurious, or instrumental, signals in both. The SG can be calibrated against the AG, because the earth tide provides a large signal which both instruments easily detect. The SG samples are at one second intervals and these are filtered and re-sampled to the times of the AG samples (nominally at one minute intervals). A least squares solution for a calibration 2 factor, an offset, and a drift is then performed. Since the Earth tide is about 300μgal, and the one minute samples of AG have a standard deviation of about 3μgal, this procedure provides a calibration factor good to about 0.1 percent, with a few days observations. The calibration experiments are usually scheduled when the Earth tide is near a maximum to take full advantage of the range. Figure 1 shows a typical calibration, performed in Dec 1999. The dots are the AG observations and the calibrated SG observations are shown as a solid line. After correcting for the earth tide and atmospheric pressure, the residuals (Figure 2) show that the noise in AG is considerably larger than in SG σAG ≈ 3μgal σSG ≈ 0.2μgal. The one minute AG samples are the average of ?? individual drops, so only the smoothing supplied by averaging has been applied. The SG data are one second samples filtered and decimated to one minute samples. Thus, the SG data has benefited from a more aggressive filtering than the AG data and these numbers somewhat overstate the case for lower noise levels in the SG. 1086 1087 1088 1089 109