LaCoste & Romberg gravity meter: System analysis and instrumental errors

Although the LaCoste & Romberg (L&R) gravity sensor can be used with great care to collect 1-mu Gal data, the sensor has a thermodynamic noise limit of 0.012 mu Gal-about two orders of magnitude below the present measurement precision. Hence, thermodynamic noise is not the limiting factor, and there is significant room for improvement by bringing the other perturbing influences under adequate control. We do not see any fundamental instrumental Limitations to substantially improving measurement accuracy at least down to 0.1 mu Gal. We have improved the noise level, stability, and reliability of the L&R borehole gravity meter sensor, with the goal of reducing the total system noise to below 1 mu Gal. In the process of these improvements, we made several fundamental observations about noise sources within the L&R sensor, particularly related to thermal noise, electronic noise, sources of mechanical and electrical rectification errors, temporal mass variations, and instrument tares and drift. It has always been assumed that the effective proof mass of the L&R sensor did not change over the time scale of a gravity survey. From our experiments we conclude that this assumption is incorrect below about 10 mu Gal. Large instrumental errors are produced by temporal proof-mass variations over time scales of a survey, resulting from contamination and migration of dust, oil, and other volatiles inside the sensor. The results are drift and tares, for it takes only 0.01 mu g of dust or oil transferred to the proof mass to offset the gravity reading by 1 mu Gal.