Rapid moment tensor, shaking intensity and tsunami simulation: the M8.1 Mexico earthquake on 08 September 2017

Sept. 8, 2017

Categories:
tsunami
gempa Modules
Research

We have simulated the tsunami wave heights and observed an unusual source mechanism for the M8.1 earthquake in the Pacific off-coast Mexico on 08 September, 2017. The full suite of SeisComP3 and gempa modules including TOAST and AUTOMT/MTV were applied to analyze this exceptional event. Modelling PGA and MMI for the region by our upcoming module SIGMA for Seismic Intensity and Ground Motion Analysis we demonstrate the extreme ground motions that have effected large areas of Mexico and Guatemala.

The Pacific coast of Central America has been hosting a number of large subduction zone earthquakes that indicate compression. The compression results from the convergence of tectonic plates. Off-coast Mexico the Cocos plate subducts underneath the North American plate and the Caribean plate. Where the three plates meet they form a triple junction. Close to this triple junction the giant M8.1 earthquake has occurred.

We have automatically located the event, determined the magnitude and inverted for the source mechanisms using SeisComP3 and AUTOMT . TOAST was used to simulate the tsunami waves resulting from vertical movement of the ocean bottom. Unlike many other quakes in the region such as the disastrous Nicaragua earthquake in 1972, the event on 08 September seem to show normal faulting. The mechanism and the very large magnitude make the event unusual for this kind of plate boundary. The tsunami simulation predicts wave heights of more than 1 meter along the coast of Mexico threatening the population of the region.

Our AUTOMT module automatically determines the earthquake moment tensor at the centroid position of the event based on pre-defined inversion profiles. The profiles depend on magnitude and allow to choose the desired wave type and the control parameters that guide the inversion. The first solution was available and published in our catalog only 17 minutes after this earthquake happened. The interactive MTV allows the operator to adjust any of the pre-configured parameters and to chose any wave type. In this case we have chosen to use W-phases as well as body and surface waves. Both inversions result in very similar and stable solutions consistent with GEOFON and USGS . Our solutions have a very good fit and a high double couple. Both are characteristic for tectonic earthquakes and stable solutions.

We have used our latest version of TOAST to simulate the tsunami due to the earthquake. TOAST allows to calculate the tsunami propagation instantaneously and efficiently on the graphics card (GPU) and to select pre-computed simulations. Different earthquake scenarios are considered to trigger or select the simulation. These earthquake scenarios comprise different source mechanisms, assumptions on known faults and fault types as well as on hypocenter location and rupture geometry. Our simulation predicts tsunami wave heights of more than 1 meter along the coast of Mexico threatening the population of the region. We expect that smaller tsunami waves will be observed across large parts of the Pacific.

Our upcoming new SIGMA module for Seismic Intensity and Ground Motion Analysis is based on the SeisComP3 framework extending it in the field of seismic hazard assessment and engineering seismology. SIGMA may work with or independently of SeisComP3 by supporting FDSN Web services for importing earthquake or station information and waveforms. It provides a user-friendly and modern graphical interface for semi-automatic and interactive strong motion data processing. SIGMA provides intensity and (P)SA maps based on GMPE's or recorded data. It calculates the most common strong motion parameters, e.g.

  • PGA/PGV/PGD,
  • Arias intensity and duration,
  • Tp, Tm, CAV, SED,
  • Fourier-, power- and response spectra.

GMPE's are configurable. Supporting C++ and Python plug-ins, standard and customized GMPE's including the OpenQuake Hazard Library can be easily integrated and compared. 

The aftershock heat map shows the PDF of about 40 aftershocks with magnitude 4.5 and higher, indicating the rupture area of the main shock. A first comparison with the available GMPEs implemented in OpenQuake shows the best match for Abrahamson and Silva, 2008. Although underestimating the near field PGAs, it shows a good match for stations with distances larger 300km. A PGA and a MMI map for the Mexico 8.1 are shown below which match well with the results of USGS. OpenQuake considers site effects in terms of  Vs30 information at the target sites. For this shaking intensity simulation we used the Vs30 grid generated and provided by USGS analyzing the world wide elevation model based on SRTM data. In case real Vs30 measurements at the sites are available, these can also be configured and used. The GMPE PGA plot impressively shows simulated and observed values considering Vs30.