Bias Problems in Existing Teleseismic Travel Time Databases: Ignore or Repair?

Abstract

An increasing number of earth scientists use teleseismic data for investigations of local, regional, and global geoscientific problems. The open access to global data makes this possible also for scientists outside the formal and informal communities of seismologists who produce and quality‐control these global data sets. In order to investigate the crust and upper mantle structure in the region of southern Scandinavia, Europe, we acquired waveform data from temporary stations contributed by several European instrument pools. The mobile station arrays were deployed in different time intervals (Fig. 1). One of the investigation methods was absolute travel time tomography. Here, we discuss some specific problems and their remediation in connection with tomographic inversion of teleseismic travel times when data from both permanent and moving temporary arrays are combined. Figure 1. Study area of southern Scandinavia with permanent and temporary seismological stations analyzed for tomographic velocity modeling (Medhus et al. , 2012). The non‐permanent arrays, marked with closed‐symbol markers, were deployed in different projects for 1–2 years covering about 10 years all together. The bold dashed outline shows the first case area of data for Figures 2 and 4. Stations studied in the second case are indicated by name boxes. Teleseismic travel times, defined as the difference between phase arrival time at the seismometer and the hypocenter time, are prone to errors of several kinds. These errors contaminate velocity modelling, such as tomographic inversion. Uncorrelated errors tend to average out, and therefore it has been very popular to take in massive data sets like the International Seismological Centre (ISC) (2001), Bulletin data (e.g., Widiyantoro et al. , 1997, Bijward et al. , 1998, Weidle et al. , 2005). However, bias errors do not average out. We discuss two sources of bias error in teleseismic …