NEES: The KiK-net database processed using an automated ground motion processing protocol

By Haitham M Dawood, Adrian Rodriguez-Marek, Jeff Bayless, Christine Goulet and Eric Thompson



Published on



The authors of this database are Haitham M Dawood (Virginia Tech), Adrian Rodriguez-Marek (Virginia Tech), Jeff Bayless (URS Corporation), Christine Goulet (Pacific Earthquake Engineering Research Center) and Eric Thompson (San Diego State University). The database was published at NEEShub on July 08 2014 and has been reproduced here to preserve the data and ensure continued access.


doi: 10.17603/DS2K081

The Kiban-Kyoshin network (KiK-net) is one of two strong motion seismograph networks constructed in Japan following the 1995 Kobe earthquake. The KiK-net network has more than 650 stations that cover most the Japanese territory. Each station has two collocated instruments at the ground surface and at a borehole at depths that are typically 100 m or 200 m. This allows for an empirical evaluation of the site response at each station. The ground motions recorded by the KiK-net stations are available to researchers all over the world through the following website ( The availability of such database is extremely important for research related to earthquake engineering and engineering seismology because of: 1) the large number of records in the database; 2) the uniform distribution of the recording stations over a large area; and 3) the availability of measurements for the shear wave velocity profiles at most stations. Nevertheless, using the data is not an easy task because the motions are uploaded to the website in a raw format and many of the important metadata associated with the ground motions are not provided (e.g., moment magnitude). Hence, processing these records and collecting its metadata are necessary steps to enable their use in engineering applications.

The database presented here consists of about 157,000 KiK-net strong ground motion records that were automatically processed using a protocol described in detail by Dawood et al. (2014). The automated protocol includes the selection of corner frequencies for band-pass filtering. In addition, a comprehensive set of metadata was compiled for each record. As a part of the metadata collection, two algorithms were used to identify dependent and independent earthquakes. Earthquakes are also classified into active crustal or subduction type events; most of the GM records correspond to subduction type earthquakes. The compilation of the metadata is described in detail in Dawood et al. (2014) and Dawood (2014). Some of the most relevant metadata includes:

  • Moment magnitude (Mw), which was obtained from the F-net seismic catalog ( after conducting a thorough matching process between the KiK-net records and the earthquakes listed in the catalog.
  • Earthquake type (e.g., active crustal). This was estimated using two different algorithms developed to find the earthquake type given its different characteristics (e.g., source location).
  • Different source to site distance measurements (e.g., closest distance to fault rupture, Rrup). This was calculated using published finite fault models for the large earthquakes and simulating fault rupture using the hypocenter location, Mw, and focal mechanism for small magnitude earthquakes.
  • Different shear wave velocity parameters (e.g., Vs30). These were calculated using the shear wave velocity profiles at the different stations.

Cite this work

Researchers should cite this work as follows:

  • Haitham M Dawood, Adrian Rodriguez-Marek, Jeff Bayless, Christine Goulet and Eric Thompson (2015), "NEES: The KiK-net database processed using an automated ground motion processing protocol,"

    BibTex | EndNote