No abstract

More than 1000 seismic events in northern Europe at distances of up to 400 km from the detecting network are located using an optimization method in which the global minimum of the traveltime function residuals is searched for using an Interval Arithmetic (IA) method. Epicentres are determined using P waves detected by the Finnish national seismic network: up to 15 stations were used in the analysis. The IA results coincide with locations provided by the University of Helsinki bulletins with a median location bias of 7.6 km.  A second data set of 59 explosions in the Siilinjärvi mine in central Finland was examined in detail, because the locations of the explosions were known exactly. In this case, the median difference of IA locations was 3.8 km from the average location of mine explosions, while all 59 events were located within 9 km of the ‘true’ epicentres. The corresponding median error of the University of Helsinki locations was smaller (3.2 km), but some Helsinki locations were well over 10 km from the mine. The convergence towards the global optimum using interval arithmetic was fast when compared with the conventional least‐squares approaches for epicentre determinations.

No abstract

No abstract

The data-adaptive autoregressive (hereafter DA) method was used to detect local and regional seismic events using digital data from the Vaasa (VAF) station with co-ordinates (62.3°N, 22.2°E) in western Finland. The seismic signal and the noise were assumed to have been normally distributed stochastic processes with a zero mean. The parameters of these processes were then adapted on the change of the registered signal as a function of time within a predefined detection window. The accuracy of the method presented is compared with the STA/LTA and visual methods. When the same detection threshold was used with the DA detector and the STA/LTA detector, it was found that the DA detector was more precise in detecting the onsets of seismic events. Bandpass (1.5 to 20 Hz) filtering was used in all the events discussed. This was done to reject the long-period microseismic noise. In one case, the detector was used on nonfiltered as well as filtered data, in order to show coinciding results.