Comparing Approaches to Network Design and Drift Correction for Relative-Gravity Surveys

Relative gravimeters, despite their limitations, remain the primary instrument for mapping the spatial changes in Earth's gravity field that are important in hydrology, volcanology, and other fields. High-precision surveys require special field procedures to account for gravimeter drift, namely, by making repeated measurements. Different survey (or network) designs are possible, as are different methods for drift correction. The efficiency of three network designs—double-looping, dense-network, and ladder surveys—and three approaches to drift correction—as a term in the network adjustment, the Roman method, and the continuous method—are compared. A range of gravimeter behavior, including high and low drift rates, linear and non-linear drift, and stochastic tares, was simulated under each network design, and evaluated as to how well the true values, and their uncertainty, could be recovered from least-squares network adjustment. The optimal network design depends on gravimeter behavior, travel time between stations, and the presence and number of absolute-gravity stations. For small networks with short travel times between stations, the dense-network design and continuous-method drift correction worked best in most cases. For larger networks with travel times of 30 minutes or more between stations, a double-loop approach is more suitable. Most suitable for nonlinear or inconsistent gravimeter drift is the Roman method, which interpolates drift between a repeated station and any intervening station. The different network designs and drift correction approaches are demonstrated using the freely available software GSadjust.