Automated offset detection approaches: case study in IGS Repro2 and 3

Traditional automated offset detections on global navigation satellite system (GNSS) station coordinate time series still cannot fully replace manual detections in practical applications due to their high false positive detection rates. We developed preliminary and enhanced offset detection approaches and tested them against the solutions from the International GNSS service 2nd and 3rd data reprocessing campaigns (Repro2 and Repro3). Their manually detected offset recordings in International Terrestrial Reference Frame (ITRF) 2014 and ITRF2020 are used as evaluation criteria. In the preliminary approaches, stochastic models based on covariance matrix, white noise model, and white noise plus flicker noise model of both univariate and multivariate are studied. Although we achieved true positive, false positive, and false negative (TP, FP, FN) rates of (0.44, 0.40, 0.16) for Repro2 and (0.42, 0.44, 0.13) for Repro3, the preliminary automated detections still lead to many false positive detections. Thus, based on the preliminary approaches, and ancillary data, an enhanced detection approach is proposed. Enhanced detections significantly reduce 56% ~ 80% false positive detections compared to preliminary approaches. As a result, for Repro3, the optimal overall performance is attained with (TP, FP, FN) rates of (0.57, 0.25, 0.18), along with a detection rate of 75%; for Repro2, the rates are (0.58, 0.20, 0.22), accompanied by a 73% detection rate. The current enhanced approach may serve as a supplementary or reference to manual detection, although still not being perfect. Furthermore, 20 manually detected unknown offsets in ITRF2020 are found to correspond to some known events (13 earthquakes and 7 equipment changes); 34 automated detections that correspond to known events but are not collected in ITRF2020 are manually checked as offsets (14 earthquakes and 20 equipment changes).