Electronic Supplement to
Relocation of the 5 December 2004 Waldkirch, Germany, Earthquake with Regional 1D and 3D Seismic Velocity Models in the Presence of Upper Mantle Anisotropy

by Thomas Willi Münch, Manfred Koch, and Jörg Schlittenhardt

Tables of phase data and of statisticals results of various variants of the Waldkirch event relocation

In Table S1 the original phase-data of the initial Waldkirch event location as performed by the BGR is listed, whereas Tables S2- S11 summarize statistical results obtained with the various relocation variants as discussed in the main text.

Figures showing the effects of phase corrections, the 3D tomographics models, and the results of two Monte Carlo relocations

Figure S1 shows the effects of the various phase-correction variants as discussed in the main text. Figure S2 illustrates 3D tomographic isotropic and anisotropic seismic models for the crust and upper mantle underneath Germany computed by means of 3D SSH inversion of regional travel times (Münch, 2009). Figures S3 and S4 show the results of the MC relocations with randomly shifted initial hypocenters and of the velocity-perturbed MC-relocations, respectively.


Tables

Table S1. Header and output phase-list of the initial Waldkirch event location performed by the BGR (Bundesanstalt für Geowissenschaften und Rohstoffe) and fixing the hypocenter depth at 10 km. In the second column the original phases, as provided by the different contributing institutions are listed, whereas in the third column those phases which were corrected are marked with their new label. Indicative of these initially wrongly labeled phases are the relatively high residuals (column Res). Further abbreviations are described below.

Table S2. Statistical results for the various stepwise corrected relocation variants, with P+S-phases, with (=ani) and without (=iso) anisotropic Pn-phase correction before and after the first iteration of the relocation. Parameter "corr" stands for corrected phases, I and K represent the nonlinear and linear inversion steps, respectively, TSS is the squared sum of residuals, RMS is the root mean square of the TSS, i.e. the usual measure of hypocentral RMS. The initial TSS of the original raw (HYPO-71/HYPOELLIPSE) BGR-location of the event, called TSSHypo, is 1018 s2.

Table S3. Waldkirch hypocentral solutions as obtained by different institutions. An "f" behind the depth indicates that the latter has been manually fixed, i.e. could not be computed by the standard location procedure. Abbreviations are: LGRB= Landesamt für Geologie und Rohstoffe; LED: Landes Erdbeben Dienst of Baden-Württemberg, BGR: Bundesanstalt für Geowissenschaften und Rohstoffe; SED: Schweizer Erdbeben Dienst; BNS: Erdbebenstation Bensberg; NEIC: National Earthquake Information Center.

Table S4. Statistical details of the nonlinear iteration process of the 1D-relocation of the Waldkirch event with (a) the original (uncorrected) and (b) phase-corrected dataset, each with and without anisotropic correction for Pn phases, after initial ray-tracing (Iter = 1) and the final nonlinear inversion step (Iter=4). Initial depth is at z0 = 10 km, starting TSS = TSShypo = 816 s2. Only P-Phases are used here. Values for Δ-Lat, Δ-Lon, Δ-z are with respect to the 2006-LED solution (Benn, 2006) (see Table S3).

Table S5. Effects of different average Moho depths on the relocation of the Waldkirch event using only P-phases. Starting TSS = TSShypo = 816 s2.

Table S6. Hypocenter results using five different combinations of VP/VS -ratios of the crust (denoted by VP/VS -M) and of the upper mantle (denoted by VP/VS -K). The optimal hypocentral solution for which the TSS and the RMS are minimal is written in bold. I and K denote the outer and inner iteration numbers of the LM-iteration process. K denotes the solutions for different increasing damping factors with the same equation system, whereas for the next I, the equation system is recalculated and solved again for the same values of K. Starting TSS = TSSHypo = 816 s2.

Table S7. Summary of results of the Waldkirch relocations using the isotropic (iso) and anisotropic (ani) 3D lateral heterogeneous seismic velocity models for the crust and upper mantle beneath Germany of Münch (2009). Relocations with only P-phases and both P+S-phases are listed where for the latter case different VP/VS-ratios are used. The ?-values of the hypocenters are relative to the original Waldkirch location by LED as listed in the first row. Other parameters are described in previous tables.

Table S8. Resolution and covariance matrices for the final Waldkirch event locations obtained with the optimal isotropic and anisotropic 3D seismic velocity model of Münch (2009).

Table S9. Lengths of the principal axes (x=lon; y=lat) of the 95 % confidence ellipsoids for the various variants of the Waldkirch event relocation as shown in Figures 4 and 6 of the text.

Table S10.Statistical results for the mean and standard deviations Σx, Σy and Σz of the three hypocentral coordinates obtained with 1000 Monte Carlo relocations (see Figure 7 of the text) with randomly perturbed arrival-times (with noise as specified by Σ). The coordinates of the initial hypocenter are lat= 48.11o, lon= 8.07o and z= 14.7 km.

Table S11. Statistical results for hypocentral coordinates with the standard deviations Σx, Σy = 4 km and Σz = 2 km obtained with 1000 Monte Carlo relocations (Figure S3) with randomly perturbed initial hypocenters.


Figures

Figure S1. Polar plot of the travel-time residuals (in sec) as a function of the azimuth after ray-tracing for the various phase-correction variants discussed in the main text. Residuals for uncorrected phases are shown in the upper left panel, for corrected phases in the upper right panel, where the lines indicate the shifts from the original, wrongly assigned, to the new, corrected phase. Residuals after the first inversion for the isotropic and anisotropic cases are shown in the lower left and lower right panel, respectively.

Figure S2. 3D tomographic isotropic and anisotropic seismic models for the crust and upper mantle underneath Germany computed by means of 3D SSH inversion of regional travel times (Münch, 2009). The models consist of four layers, each discretized into 15 x 15 blocs, with vertical extension of 10 km and lateral extension of 650 km. Left panel: Isotropic model (no anisotropic Pn -travel-time correction included); right panel: anisotropic (with upper mantle Pn -travel-time correction) model. See Münch (2009) for further details.

Figure S3. RMS-surfaces resulting from 1000 MC hypocentral relocations with randomly shifted initial hypocenters (see text). Left panel: x-y-plane; right panel: x-z-plane. Details of the statistical analysis are listed in Table S11.

Figure S4. TSS-surface obtained with the 1000 velocity-perturbed MC hypocentral relocations of model MKS-2005 as a function of the epicentral shifts Δ-x, Δ-y (left panel) and of Δ-x and depth Δ-z (right panel). All values are relative to the initial hypocenter set at latitude = 48.11o, longitude = 8.075o and z= 13.0 km. The arrow in each plot indicates the (Δ-x, Δ-y) - location of the minimal TSS.

References

Benn, N.E. (2006). Seismologische Untersuchung des Waldkirchbebens vom 5.12.2004, Diploma thesis, Geologisches Institut der Albert-Ludwigs-Universität, Freiburg, Germany, 118 pp. (in German).

Brüstle, W. (2004). Erdbeben bei Waldkirch, 5.12. 2004, Communiqué, Landesamt für Geologie, Rohstoffe und Bergbau, Baden-Württemberg, Freiburg, Germany, 1pp. (in German)

Münch, T.W. (2009). 3D Simultaneous inversion for seismic structure and local hypocenters in Germany under consideration of Pn-anisotropy in the upper mantle, PhD thesis, University of Kassel, Kassel, Germany, 280pp. (in German).

[ Back ]