Automatic offset detection using HECTOR
July 6, 2017 • 128 Views
Workshop 2017 - Rui Fernandes - We present here the offset detection scheme implemented at HECTOR and that uses the Bayesian Information Criterion (BIC) to determine the most likely time of an offset. The novelty of this scheme is that it takes the temporal correlation of the noise into account. Such approach significantly increases the computation time to ensure that the estimated BIC value is correct. However, we were able to create a fast algorithm by adopting the methodology implemented in HECTOR (Bos et al., 2013). We demonstrate the feasibility of the approach using the IGS network, where most of the offsets have been accurately determined. This permits to have an external evaluation of this new outlier detection approach that has been included in HECTOR.