Strain From Small-Scale GPS Networks: Differences in Apparent Noise Levels

Time series from permanent GPS stations are usually shown as variations in station positions. Such series usually show correlations between stations in a region; these correlations are often removed using regional filtering. Another mode of analysis is to solve for a time series of strain changes across a network (as well as a mean displacement) from the series of station displacements. This automatically removes common-mode signals, and produces an output (strain change) of immediate geophysical relevance. Applying this method to time series for small (10-20 km span) subnetworks of the SCIGN and BARGEN networks shows that the resulting strain series show interesting signals not obvious in the original time series (such as local annual signals). The level in strain (as judged by day-to-day scatter) ranges from 40 to 100 nanostrain for a network size of 10 km, with the higher values for subnetworks in more coastal areas, and lower ones in desert sites. A large variation comes from the mode of processing used to create the displacement series, with point-positioning giving much larger apparent strain noise than baseline processing. This suggests that the sensitivity of GPS data to many geophysical signals will depend on the way in which the original phase data are processed.