The assessment of the optimal time window for prediction of seismic hazard for longwall coal mining: the case study

Abstract

The dynamic nature of rock mass damage during mining activity generates seismic events. This article shows, how the time window for the database influences on the actual status of seismic hazard for the longwall mining area in one of Polish coal mines using Gutenberg–Richter law. A time window of 10–90 days was assumed with similar or shorter prediction times forecast on its basis. Additionally, for each seismic database the hazard prediction accuracy was determined. The analysis shows that the 10- and 20-day base periods are too short for prediction purposes. The higher-energy seismic events sometimes do not occur within such a short period of time, preventing regression analysis and parameter b determination. The best time window for the seismic hazard prognosis in given geological and mining conditions seems to be 30–50 days. The shorter periods cause the underestimation of the seismic hazard prognosis. Low range of tremor energies and the relatively low number of seismic events with high energy cause the low probability of prediction of the seismic mining events (10–40%) of the energy of min. 106 J, even for longer day base periods. The accuracy of hazard prediction, obtained from each seismic database period, was determined, using the developed coefficient of hazard autoregression CN. The analysis of the Gutenberg–Richter distribution should serve as complementary tool of seismic hazard prediction only.