Preliminary Relocation of the M_L 3.9 Seismic Event

The hypocenter reported on the University of Utah Seismograph Stations (UUSS) Recent Earthquakes Web pages for the August 6, 2007, M_L 3.9 seismic event was determined from P-wave arrival time data at stations out to 160 km distance using standard UUSS velocity models and location procedures. This hypocenter (Table 1; Figure 1, open star) is 1.5 km WSW of the estimated center of the Crandall Canyon Mine collapse. The 95% confidence limits for the epicenter, as calculated by the location program Hypoinverse (Klein, 1978), overlap the western edge of the Crandall Canyon Mine near the collapse area (Figure 1). These confidence limits represent the random error in the solution, but do not fully account for possible systematic errors caused by inadequacies of the one-dimensional velocity models used. Therefore, it is likely that the confidence limits shown underestimate the actual uncertainty in the location.

In order to improve the accuracy of this location, we carefully repicked the P-wave arrival times and relocated the event using stations within 100 km of the epicenter only, a velocity model specific to the mine region, and a preliminary set of empirical station corrections. The selection of the 100 km distance cutoff, with full weight to stations within 75 km and a cosine taper distance weighting between 75 and 100 km (see Klein, 1978), was based on the station distribution in the area. The velocity model was modified from one constructed for the Trail Mountain Mine area ~20 km to the south (Arabasz et al., 2005) by increasing the datum from 2600 m to 3100 m to account for the higher surface and stratigraphic elevations in the Crandall Canyon Mine area. To calibrate the station corrections, we used data from the largest event (M_L 1.7) to occur in the mine area since the portable seismic network became operational. First, we relocated the M_L 1.7 “master event” using the revised distance weighting, the modified Trail Mountain velocity model, and regional network data supplemented by data from the three telemetered portable stations (the data from the two non-telemetered stations were not yet available). Then, we set the station corrections equal to the travel-time residuals, which are the differences between the observed and predicted arrival times. The trial hypocenter for the relocation is that of the M_L 1.7 event. Of the modifications we made to the standard UUSS location procedures, the use of the station corrections had by far the largest effect. The station corrections reduced the weighted root-mean-square travel-time residual for the main shock by a factor of two, suggesting that these corrections serve their intended purpose of partially compensating for lateral velocity variations.

The relocated epicenter (Table 1; Figure 1, solid star) is 1 km south of the estimated collapse center. The 95% confidence limits for this epicenter lie entirely within the mine boundaries and come within 400 m of the minimum estimated collapse area shown by the box on Figure 1. Note that there is minimal overlap between the 95% confidence limits for the two epicenters shown on Figure 1. This observation provides evidence that the 95% confidence limits underestimate the true uncertainties in the locations. The focal depth for the relocated hypocenter is 0.5 km, which is very close to the mine depth of 0.6 km below the datum. However, fixed focal depths anywhere between 0 and 8 km provide essentially equivalent fits to the data.

Based on the preliminary results presented in Figure 1, we consider the epicentral location of the M_L 3.9 seismic event to be consistent with the hypothesis that the source of this event was the Crandall Canyon Mine collapse. Further examination of the location of the seismic event relative to the mine collapse awaits more definitive information on the extent and location of the collapse, refinement of the station corrections using more calibration events and stations, and application of other tools for improving earthquake locations such as the double difference method.