GPS Landslide Monitoring: Single Base vs. Network Solutions — A case study based on the Puerto Rico and Virgin Islands Permanent GPS Network

摘要

This study demonstrated an approach to using permanent GPS stations from a local continuous GPS network as no-cost references in conducting long-term millimeter-level landslide monitoring. Accuracy and outliers from a series of single-base and network GPS measurements of a creeping landslide were studied. The criterion for accuracy was the weighted root-mean-square (RMS) of residuals of GPS measurements with respect to true landslide displacements over a period of 14 months. This investigation indicated that the current Puerto Rico and Virgin Islands GPS network, as a reference frame, can provide accuracy of 1 to 2 mm horizontally and 6 mm vertically for local 24-hour continuous landslide monitoring with few outliers (1%). The accuracy degraded by a factor of two for 6-hour sessions, and more for shorter sessions. This study indicated that adding a few reference stations to GPS data processing can reduce the number of outliers and increase the accuracy and robustness of landslide surveying, even if these references are far from the study site. This improvement was particularly significant for short sessions and vertical components. The accuracy of network solutions depended slightly on the number of reference stations, but the dependence on the distance and geometric distribution of the references was weak. For long-term landslide monitoring, accuracy under 5 mm horizontally and 15 mm vertically are often expected. Accuracy at this level can be stably achieved in the Puerto Rico and Virgin Islands region by performing field observations for 4 hours or longer, and applying 3 or more reference stations for solving a network solution. This study also indicated that rainfall events can play a crucial rule in high-precision GPS measurements. GPS data collected during heavy rainfall events should be cautiously analyzed in landslide studies.