Coseismic offset estimation using time series decomposition based on static and kinematic solutions from high-rate GNSS observations

Abstract Coseismic offsets can be estimated using GNSS data through differential methods or time-series decomposition. Displacement estimation using the differential method (average coordinate differencing) is frequently influenced by noise and long-term crustal movement trends. This research estimates pure coseismic offset values by isolating the effects of interseismic and early postseismic phases using GNSS time-series decomposition. The research focuses on two seismic events with distinct tectonic characteristics: the 2022 Cianjur Earthquake (Mw 5.6) in West Java and the 2021 Mamuju-Majene Earthquake (Mw 6.2) in West Sulawesi. Observation data were obtained from the Indonesian Continuously Operating Reference Stations (Ina-CORS) and processed using the Precise Point Positioning – Ambiguity Resolution (PPP-AR) method in both daily static and 1 Hz high-rate kinematic solution schemes. Displacement estimation was evaluated by comparing two primary approaches: the differential method (average coordinate difference) and the time-series decomposition method, which simultaneously models interseismic trends, coseismic jumps, and postseismic decay using a combination of linear, step, and logarithmic functions. The results indicate that in static solutions, the two methods show no statistically significant differences. However, in kinematic solutions, the selection of the observation window is crucial; the differential method yields optimal results within a short window (5 min), whereas the decomposition method demonstrates greater stability over wider data ranges (23 h) with smaller uncertainty values (standard deviation). Spatial analysis identified the largest coseismic offset at the CJUR station for the Cianjur Earthquake and the CMJU station for the Mamuju Earthquake. This study concludes that time-series decomposition is more representative for obtaining coseismic offset values that are free from the biases of other deformation phases.