Joint Hypocenter Relocation Studies with Sparse Data Sets--a Case History: Virginia Earthquakes by Marc Viret,

This study considers the use of the joint location techniques for sets of Virginia earthquakes recorded by different station distributions. These input data sets consist of microearthquakes (mb < 3) recorded by local networks (post-1978) and larger, felt events (m~ _-< 4.6) that occurred in the same areas prior to network installation (pre-1978). Very different epicentral geometries (both linear and diffuse) and host geologic terranes (one of folded and faulted sedimentary rocks and one of metamorphosed crystalline rocks) are involved. The main purpose of the investigation was to test if some characteristics of the spatial distribution of the original hypocenters might be due to bias resulting from the different networks used to locate the earthquakes and thereby not reflect the configurations of the source zones of the earthquakes. Specifically, we relocated some 50 earthquakes (1959 to 1981) using both joint hypocenter determination (JHD) and a two-stage location method (JED/SE77) based on joint epicenter determination (JHD with depth fixed in the computation). Seismic zones in southwestern (Giles County) and central Virginia were examined. We did not find evidence that the use of HYPOELLIPSE with nonuniform networks has either generated spurious "structure" in the pattern of hypocenters or hidden a real structure. The Giles County lineation and the scattered distribution of sources in central Virginia that appeared in the HYPOELLIPSE calculations also appear in the JHD and JED/SE77 calculations. We were also not successful in convincingly improving the relative locations of regionally recorded earthquakes with respect to locally recorded events, because too few of the local stations were installed when the regional events occurred. In the central Virginia zone, however, the joint location methods uncovered inconsistent arrival-time data that had not been identified in the original HYPOELLIPSE computations. On the whole, the principal benefit derived from the joint location calculations was corroboration of the characteristics of the hypocenter distributions that had been inferred from the original individual locations. INTRODUCTION Earthquake occurrence in Virginia is concentrated spatially in the southwestern (Giles County in the Valley and Ridge province) and central (Piedmont province west of Richmond) portions of the state (Bollinger, 1973; Figure 1). For this investigation, the earthquake data base for those areas is partitioned into three subsets according to the seismograph stations available at a given time. The first such data set is referred to herein as "Regionally Recorded Seismicity," and consists of nine earthquakes that occurred from 1959 until the Virginia Tech Seismic Network was installed (1977). Each event in this subset had previously been jointly relocated by Dewey and Gordon (1980, 1983) using a general regional velocity model, arrival-times from seismograph stations at regional or teleseismic distances and the JHD method (Dewey, 1971). The second subset, "North Anna Seismicity," is a group of 11 earthquakes recorded during a seismic monitoring program at the North Anna nuclear power plant in central Virginia (1974 to 1977; Dames and Moore Corporation, 1977). Arrival-time data for those shocks were for distances from 15 to 97 km to a 17-station semi-permanent seismic network sited within a 15 km 2297 2298 M. VIRET, G. A. BOLLINGER, J. A. SNOKE, AND J. W. DEWEY