Deployment of New Strong Motion Seismographs of K-NET and KiK-net

Part of the Geotechnical, Geological, and Earthquake Engineering book series (GGEE, volume 14)


Following the occurrence of the 1995 Hyogoken-Nanbu (Kobe) earthquake, National Research Institute for Earth Science and Disaster Prevention (NIED) has constructed two strong motion seismograph networks, K-NET and KiK-net. These networks cover uniformly the country with an inter-station interval of about 20–25 km, and a total number of the stations is about 1,700. To grasp the hazard (ground motion) of urban or downtown areas, most stations of K-NET are located in public offices, schools and parks, and a three-component accelerometer is installed on the free-surface. On the other hand, KiK-net stations are located in quiet places to avoid the artificial noise. Each KiK-net station has a borehole of 100 m or more in depth and strong motion seismographs have been installed both on the ground surface (uphole) and at the bottom of the boreholes (downhole). Recently, based on the request for quicker hazard information, all instruments of the K-NET and KiK-net have been renewed by including the change of accelerometers at the surface, new recorders, and also a new data collection system between stations and the Data Management Center (DMC) in Tsukuba. The newly developed instruments, which are state of the art in strong motion instrumentation, have several advantages such as real-time capability, larger measurable range and lower noise. This paper explains the new generation system of K-NET and KiK-net.

12.1 Introduction

Strong motion observation in Japan dates back as far as 1950s. Since then, many organizations have put a lot of effort to construct and maintain strong motion networks. The Japan Meteorological Agency (JMA; has operated an earthquake observation network for the monitoring of earthquakes. The Building Research Institute (BRI; has installed strong motion instruments in major cities throughout Japan to enhance the seismic safety of buildings. Strong motion observations at port facilities have been performed since 1962 under the Ministry of Land, Infrastructure, Transport and Tourism (MLIT; Some universities and research institutes including NIED have had local area strong motion networks.

After the 1995 Hyogoken-Nanbu earthquake, the development of the Japanese strong motion observation was drastically stimulated. Prompted by the fact that an insufficient coverage of the strong motion observation networks in Japan did not facilitate to grasp the overall characteristics of earthquakes, the Japanese government decided to improve the quality and quantity of the strong motion observations in Japan. JMA have increased the number of strong motion stations and the central government subsidized the local governments to install several thousands of observatories for seismic intensity observation. Since 1996, NIED has been in charge of constructing of two strong motion networks, K-NET and KiK-net (Fig. 12.1). These dense seismograph networks have successfully recorded many near source ground motions, and nearly 290,000 digital records from more than seven thousands events are now available through a public access web-site (Table 12.1 and Appendixes 1 and 2). These data are significantly contributing to disaster mitigation programs, earthquake resistant designs and scientific research in Japan and abroad.
Fig. 12.1

Distribution of aK-NET and b KiK-net stations. Open and closed circles in (a) show stations where K-NET02 and K-NET02A are installed

Table 12.1

Number of events and records released in the K-NET and KiK-net web sites




K-NET and KiK-net

















































































































12.1.1 K-NET(Kyoshin NETwork)

Just after the 1995 Hyogoken-Nanbu earthquake, construction of K-NET began and was completed in just 1 year [6]. The term K-NET stands for “Kyoshin network” where “Kyoshin” means “strong motion” in Japanese. At its inception, K-NET consisted of 1,000 stations whose average station-to-station distance was about 20 km. All the stations were outfitted with the same type of strong motion accelerographs, K-NET95, which were installed on the free surface (Fig. 12.2). Afterwards, some stations were added to K-NET, such as the existing stations within the Kanto-Tokai area or the cable type ocean-bottom strong motion accelerometers at the Sagami Bay [3]. The number of stations currently is 1,032.
Fig. 12.2

Observation facility for K-NET. A basement for cold areas is shown at the inset

12.1.2 KiK-net(Kiban Kyoshin network)

The Japanese government established “the Headquarters for Earthquake Research Promotion” on July 18, 1995, and the “Fundamental Survey and Observation for Earthquake Research” plan was developed under their direction. In short, this plan is called KIBAN, which is a Japanese word meaning fundamental or infrastructure. The main part of the “Kiban” project encompasses many types of observation networks [12], such as the high sensitivity seismic network (Hi-net [9]), the strong motion observation network (KiK-net; Kiban Kyoshin network [1]), the broadband seismic observation network (F-net [4]), and the continuous GPS observation network (GEONET [5]). Each KiK-net station (Figs. 12.3 and 12.4) has an observation borehole of more than 100 m deep (Table 12.2), and a pair of tri-axial accelerometers (V404 manufactured by Akashi Corporation) is installed on the ground surface and also at the bottom of the observation boreholes together with high sensitivity velocity seismometers of Hi-net. The acquisition system, SMAC-MDK, was also manufactured by Akashi Corporation.
Fig. 12.4

a Observatory building of KiK-net. b Acquisition system of KiK-net and Hi-net installed in a rack c Accelerograph of KiK-net installed in the pit

Fig. 12.3

a Observation facility and b a downhole instrument for KiK-net

Table 12.2

Distribution of the KiK-net borehole depths

Depth (m)

Number of stations















K-NET and KiK-net are pioneer networks in Japan to freely releasing all digital data through the Internet immediately after an earthquake, and nowadays this open-data policy is becoming a common practice. Although our initial policy was to release all data within 1 week after the occurrence of an earthquake, the request for a quicker hazard information from many local governments who are responsible for the initial response after an earthquake, prompted us to update our network. Based on these requests, we conducted the replacement of all the instruments at about 1,700 K-NET and KiK-net stations (excluding the downhole sensors) between 2003 and 2008.

12.2 New Instruments ofK-NETand KiK-net

New instruments were developed by introducing new technologies including information technologies and high-performance sensors. The networks are equipped with three types of instruments (K-NET02, K-NET02A and KiK-net06) which have similar function and performance. K-NET02 and K-NET02A systems are installed in the stations indicated by open and closed circles in Fig. 12.1a and KiK-net06 in the all KiK-net stations (Fig. 12.1b). Summary of the new K-NET and KiK-net systems are shown in Tables 12.3 and 12.4, respectively, by comparing them with the old systems, K-NET95 and SMAC-MDK.
Table 12.3

Comparison of old and new instruments of K-NET





Maximum measurable acceleration

2,000 gal

4,000 gal

4,000 gal

Dynamic range (RMS noise/full scale)

114 dB (19 bit)

132 dB (22 bit)

132 dB (22 bit)





Calculation of JMA seismic intensity


Calculation of response spectrum


Recording capacity

8 MB

512 MB

768 MB

User’s programmability


∘(Linux OS)

∘(Linux OS)

Continuous data recording


Data communication




aAkashi Corporation.

bKinemetrics Inc.

cJapan Aviation Electronics Industry, Ltd.

Table 12.4

Comparison of old and new instruments of KiK-net




Maximum measurable acceleration surface/borehole

2,000/2,000 gal

4,000/2,000 gal

Dynamic range (RMS noise/full scale)

114 dB (19 bit)

132 dB (22 bit)

Accelerometer surface






Calculation of JMA seismic intensity


Processing for EEW (B-delta method)


Calculation of strong motion indexesc


Recording capacity

80 MB

768 MB

User’s programmability


∘(Linux OS)

Continuous data recording


Data communication



aAkashi Corporation.

bJapan Aviation Electronics Industry, Ltd.

cContinuous calculation of PGA, PGV, PGD, real-time intensity and response spectrum.

12.2.1 Sensor Module

At each station, a tri-axial accelerometer whose maximum measureable range is ±4,000 gal, is installed on the free surface. K-NET02 employs Episensor FBA-ES-DECK (Kinemetrics Inc.), and K-NET02A and KiK-net06 use the JA-40GA sensor [16] manufactured by Japan Aviation Electronics Industry Ltd. Though the Episensor is a high performance accelerometer for precise strong motion observation, sometimes step-wise noise, which is a rather common phenomenon, appears. This kind of noise is brought by the release of the stress accumulated in the spring (flexure). Though the amplitude of the stepwise noise is very small (much less than 1 gal in most cases), they have harmful influence when one integrates the accelerogram into velocity or displacement and it is very difficult to remove as it overlaps the data (e.g. [2]). To avoid this type of noise, the spring of the new sensor, JA-40GA, is made of quartz and therefore the occurrence of stepwise noise in the data is significantly reduced as compared to the old sensor.

Regarding KiK-net stations, in addition to installation of new free surface sensors, a tri-axial accelerometer (V404) whose maximum measureable range is ±2,000 gal, was also installed at the bottom of the borehole, stored in a pressure-resistant tube made of stainless steel together with the Hi-net high-sensitivity seismometers. The downhole sensors which were originally installed at the time of construction of the stations have not been replaced at this time. The orientation of horizontal components of downhole seismometers used in the KiK-net has some uncertainty due to difficulties during installation. Orientations are therefore estimated by evaluating correlation of teleseismic waveform data [15]. Sensor orientations may change every time the sensor is re-installed for maintenance or some other reason. The updated orientations are also available through the Internet.

12.2.2 Acquisition System

The acquisition system consists of a measurement module and a communication module (Fig. 12.5), and each of them is controlled by individual built-in Linux boards which allow an easy update of the system program through the network. The measurement module functions as a conventional strong motion seismograph with high precision of observation. The communication module processes advanced tasks, such as the calculation of the JMA seismic intensity, the recording of continuous data and a real-time data transmission. When a power shortage continues more than an hour, the measurement module stops power supply from the backup battery to the router and the communication module to save battery life for recording data. Internal clock is synchronized with GPS (Global Positioning System) and the accuracy of absolute timing is within 0.1 ms. as long as the GPS signal is received. To protect the acquisition system from the damages owning to lightning, arresters are installed between the system and outside modules such as power supply, telephone line and GPS antenna. Arresters are also installed inside the sensor module.
Fig. 12.5

Block diagram of K-NET02 which consists of a sensor module and acquisition system (measurement module and communication module)

Time history of the ground motion is recoded by an event triggering system. When the ground motion exceeds the threshold, the system begins to record ground acceleration with 100 Hz sampling including a 15 s pre-trigger data. To avoid aliasing, a high-cut filter is applied to the data before recording. Thus the total response characteristic is almost flat from 0 (direct current component) to 30 Hz. At the same time of triggering, the communication module begins to make connection to the DMC through a digital telephone line (ISDN; Integrated Services Digital Network) using a dial-up router. It takes typically 5–7 s to establish the connection and a few seconds to send the pre-trigger data. Typically after 7–10 s the system is triggered, real-time data can be obtained. This data includes a short 1.5 s delay, which corresponded to the time spent for making 1 s packet data plus 0.5 s system delay.

The capability of the new system of making the connection from the station to the DMC has two advantages compared to the old DMC-to-station dialing-up system. First, the new system substantially reduces the time to obtain the data because it eliminates the need to wait for information of the earthquake parameters (location and magnitude), which was a requirement of the old system for deciding a station priority dial-up list. Second, it helps avoiding any overcrowding of telephone lines because the connection can be established before an eventual congestion, which may extend for several hours or days following an earthquake.

12.2.3 Performances of New Instruments

To examine the performance of old and new instruments, we installed K-NET95 and K-NET02 at Tsukubane strong motion observatory where noise level is much lower than the instrumental noise and obtained the record of instrumental noise by observing horizontal ground motion. Figure 12.6 shows the noise records of the whole system including sensor and acquisition systems in time domain and spectral domain, respectively. Spectra of noise were obtained by the standard method of evaluating the noise level of broadband stations [13]. For comparison we show the noise level of NHNM and NLNM, the noisiest and quietest broadband stations of IRIS all over the world. Comparison of the noise in time and spectral domains shows that the level of noise of the new instruments is roughly ten times lower than old one.
Fig. 12.6

a Records of instrumental noise in time domain obtained by K-NET95 and K-NET02 by observing vertical ground motion at Tsukubane strong motion observatory where noise level is much lower than the instrumental noise. b Records of instrumental noise in spectral domain obtained by K-NET95 (upper), and K-NET02 (lower). Broken lines show the quietest (NLNM) and noisiest (NHNM) of IRIS broadband stations

The noise level of the new instruments is mainly due not to the sensor but to the acquisition system because an effective dynamic range of the acquisition system is 132 dB and the dynamic range of the sensor is larger than this value. The practical advantage of the new sensors, especially the JA-40GA, is a larger maximum measureable-range and the occurrence of few stepwise noise. Maximum measurable range of the accelerometer for the surface was increased from 2,000 to 4,000 gal because recent dense strong motion seismograph network revealed that extremely large shaking exceeding gravity is not very rare. Figure 12.7a shows ground acceleration recorded just above the reverse fault during the 2008 Iwate-Miyagi Nairiku earthquake. The maximum acceleration of UD component on the surface was 3,866 gal which would not have been possible to record without saturation by the old sensor. Permanent displacements as well as time history of displacements (Fig. 12.7b) are easily obtained by double integration of these accelerograms because they are practically stepwise noise-free.
Fig. 12.7

Time history of a acceleration and b displacement recorded by the surface and borehole sensors during the 2008 Iwate-Miyagi Nairiku earthquake at KiK-net IWTH25 station (West Ichinoseki). Displacement waveforms are derived by the double integration of the original acceleration records with baseline correction. Rightmost arrows indicate the static displacement observed by the GPS station (ICNS) of Tohoku University, which is located a few hundred meters from IWTH25  station

12.3 Continuous Observations

The acquisition system of the new instruments also has a continuous observation capability. Continuous data are stored in the pre-reserved area of a compact flash memory on the communication modules. The recording capacity of continuous data is of several days, and when the pre-reserved area reaches to full the oldest data are automatically over-written by new data. Several kinds of strong motion indexes such as real-time intensity [8], peak values of acceleration, velocity and displacement, and response spectra are continuously calculated every 1 s and stored in the memory. Those data and indexes are easily downloaded from DMC through the network of the data collection system by specifying the required time window by the user.

If a continuous communication line is available, continuous telemetric observation is also possible using the new instrument. As a feasibility study for continuous strong motion observation, and taking advantage of the fact that each KiK-net station shares the same seismic observatory with Hi-net, we set up the continuous telemetric observation of the downhole UD component of KiK-net through a shared use of the EarthLAN. The EarthLAN, by which Hi-net continuous data is transmitted, is a continuous telephone line service provided by a communication company [10]. The reason for limiting the continuous transmission of KiK-net to only one component is because most of the bandwidth of the EarthLAN is occupied by Hi-net data, and the extra capacity is not enough for all six components, three components both for surface and downhole accelerometers. The continuously calculated strong motion indexes above explained, as well as the UD component of the downhole data, are sent every 1s using the surplus bandwidth of the EarthLAN. And also, the epicentral distance and magnitude estimated by the B-Delta method [11] and P-wave arrival time are promptly estimated at each station and sent to the DMC at optional timing through the EarthLAN. These indexes have the potential to be implemented in earthquake early warning systems.

From the viewpoint of disaster mitigation, it will become necessary in the near future to provide the data that is required to estimate the spatial distribution of not only the hazard (ground motion) but also the risk (damage) of the earthquake in real-time. To meet this need, more rapid data transmission may be required. Because of the infrequent occurrence of large earthquakes, strong motion has been commonly observed by event triggering system which requires connection of the telephone-line only during the data collection. To achieve more rapidity, continuous observation is one of the most likely options. Data recorded by an event triggering system provides important information of past earthquakes and helps to estimate the hazard and risk of a future earthquake. With a continuous observation system, owning to the rapid progress of information technologies, soon we would be able to fully monitor in real-time and thus directly contribute to the mitigation of ongoing seismic disasters. Our strong motion network capability currently lies somewhere in between these two observation systems.

12.4 Data Flow and Data Release

Real-time ground motion maps (acceleration, real-time intensity and response spectrum) of Japan are always generated and uploaded to the web site ( every 5 s, by using the strong motion indexes which are continuously calculated and sent from every KiK-net stations through the EarthLAN (Fig. 12.8).
Fig. 12.8

A real-time ground motion map, which is generated from real-time strong motion indexes (PGA, real-time intensity, response spectrum) continuously calculated and telemated from every KiK-net stations every 1s. The map is routinely shown on web site and automatically re-loaded every 5 s

When an earthquake occurs and the data receiving system of DMC in Tsukuba receives an event data from the earliest station, an event number is assigned for making a group of data which is expected to belong to same earthquake. Event data receiving within 600 s are considered to be the same group of data. Each data are sent to the web site one by one. This procedure does not require to wait until the transmission of data from all the station have finalized, and make it possible to release the data from the near source of an earthquake in a few minutes for most cases. Since this prompt release of data do not go through any manual check, some inappropriate records such as noise data (ground notion of artificial noise or data due to observational trouble), or the data from a different earthquake occurring at almost the same time. Figures such as waveforms and response spectrum at each station, and maps of peak ground motion (PGA, PGV, response spectrum and seismic intensity) distributions are uploaded at almost the same time. These maps are created using interpolations by an optimal Delaunay triangulation [17] from all data observed by K-NET and KiK-net.

Seismic intensities (e.g. [7, 14]) of an event observed by K-NET are promptly transmitted to DMC (within 2 min), as soon as their values are calculated by using 60 s of ground motion data counted from arrival time by the communication module of the acquisition system at each station. The seismic intensities are sent to the local governments and the mass media through JMA and this information is immediately run as a telop on television. The seismic intensities are also directly sent to the local governments by request.

All data are manually checked in office hours. The operators take away noise data, and records are related to source information (occurrence time, location and magnitude) and formally upload the archives, digital data and a set of related figures, to the web site.

12.5 Summary

All the instruments (excluding downhole sensors) of the Japanese nationwide strong motion networks, K-NET and KiK-net, have been replaced recently, to answer to the request for a quicker hazard information immediately after an earthquake. The instruments have been newly developed by introducing new technologies including information technologies and high-performance sensors, which make them state of the art in strong motion instrumentation.

The new instruments have the capability of automatically calling the DMC at NIED in Tsukuba, several seconds after being triggered and are able to transmit waveform data even while recording. This function not only significantly reduces the time for data collection but also helps avoiding any overcrowding of telephone lines. K-NET02 and K-NET02A have been officially approved as a seismic intensity meter by the JMA. Measured seismic intensity, which is the most popular and important index for national and local government during an earthquake, is automatically transmitted to them through JMA within 2 min after being triggered. This information is greatly contributing to the decisions and actions by the administration during an earthquake, and is widely broadcasted through television and radio.

The basic performance of the strong motion instruments has been greatly improved. Measureable range of accelerometers on the ground surface has been extended from ±2,000 to ±4,000 gal and the dynamic range has been improved by about a factor of 10 as compared to the old system. The newly introduced sensor, the JA-40GA, which has a quartz spring (flexure) significantly reduces the occurrence of stepwise noise which very frequently appears in data observed by a sensor with a metal spring. This will improve the accuracy of data analysis especially in the long-period range such as waveform inversion analysis of source processes and real-time seismology.

New instruments also have a continuous observation capability. Continuous data are stored in the memory. Several kinds of strong motion indexes such as real-time intensity, peak values of acceleration, velocity and displacement, and response spectra are continuously calculated by the communication modules. Those data and indexes are easily downloaded from DMC through the network by specifying the required time window by the user. Continuous telemetric observation is also possible using the new instrument if a continuous communication line is available. The communication module sends the data by packets of 1-s. duration. Indeed, UD components of the downhole sensor of KiK-net06 are continuously sent to DMC by using leeway bandwidth of continuous line prepared for Hi-net, as a feasibility study for continuous observation of strong motion. From the viewpoint of disaster mitigation, data from continuous strong motion observation will become important both for the earthquake early warning, and for grasping the spatial distribution of the hazard and the risk (damage) immediately after an earthquake.

Digital data and related figures and movies are available soon after the earthquake (typically within 15 min) without any manual check. This data is checked by an operator in the next business hour and the new archive is re-uploaded as the formal data.

One of the new capabilities of the NIED strong motion web site are the maps of real-time ground motion all over Japan. One can always see the intensity, PGA and response spectrum which are continuously sent from all KiK-net station every 5 s and observe the wave propagation in real-time.


  1. 1.
    Aoi S, Obara K, Hori S, Kasahara K, Okada Y (2000) New strong-motion observation network: KiK-net. EOS trans. Am Geophys Union 81:F863Google Scholar
  2. 2.
    Boore DM (2001) Effect of baseline corrections on displacements and response spectra for several recordings of the 1999 Chi-Chi, Taiwan, earthquake. Bull Seismol Soc Am 91:1199–11211CrossRefGoogle Scholar
  3. 3.
    Eguchi T, Fujinawa Y, Fujita E, Iwasaki S, Watabe I, Fujiwara H (1998) A real-time observation network of ocean-bottom-seismometers deployed at the Sagami trough subduction zone, central Japan. Mar Geophys Res 20:73–94CrossRefGoogle Scholar
  4. 4.
    Fukuyama E, Ishida M, Hori S, Sekiguchi S, Watada S (1996) Broadband seismic observation conducted under the FREESIA Project. Rep Nat’l Res Inst Earth Sci Dsas Prev 57:23–31Google Scholar
  5. 5.
    Geographical Survey Institute (1998) Crustal deformation of Japan detected GEONET. Accessed 29 Jan 2010
  6. 6.
    Kinoshita S (1998) Kyoshin net (K-NET). Seismol Res Lett 69:309–332CrossRefGoogle Scholar
  7. 7.
    Kunugi T (2000) Relationship between Japan meteorological agency instrumental intensity and instrumental modified mercalli intensity obtained from K-NET strong-motion data. Zisin2 53:89–93Google Scholar
  8. 8.
    Kunugi T, Aoi S, Nakamura H, Fujiwara H, Morikawa N (2008) A real-time processing of seismic intensity. Zisin2 60:243–252Google Scholar
  9. 9.
    Obara K (2002) Hi-net: high sensitivity seismograph network, Japan. Lect Notes Earth Sci 98:79–87CrossRefGoogle Scholar
  10. 10.
    Obara K, Shiomi Y, Haryu Y, Matsumura M, Shimanuki T (2008) Development of network platform for data transmission and practical use for NIED Hi-net system. Japan geoscience union meeting 2008:S144–P012Google Scholar
  11. 11.
    Odaka T, Ashiya S, Tsukada S, Sato S, Ohtake K, Nozaka D (2003) A new method of quickly estimating epicentral distance and magnitude from a single seismic record. Bull Seismol Soc Am 93:526–532CrossRefGoogle Scholar
  12. 12.
    Okada Y, Kasahara K, Hori S, Obara K, Sekiguchi S, Fujiwara H, Yamamoto A (2004) Recent progress of seismic observation network in Japan – Hi-net, F-net, K-NET and KiK-net – . Earth Planets Space 56:15–28Google Scholar
  13. 13.
    Peterson J (1993) Observations and modeling of seismic background noise. USGS Open File Rep. 93-322Google Scholar
  14. 14.
    Shabestari KT, Yamazaki F (2001) A proposal of instrumental seismic intensity scale compatible with MMI evaluated from three-component acceleration records. Earthq Spectra 17:711–723CrossRefGoogle Scholar
  15. 15.
    Shiomi K, Obara K, Aoi S, Kasahara K (2003) Estimation on the azimuth of the Hi-net and KiK-net borehole seismometers. Zisin2 56:99–110Google Scholar
  16. 16.
    Tomioka T, Yamamoto S (2006) Development of Low Noise Accelerometer (JA-40GA), Japan Aviation Electronics technical report 29:14. Accessed 29 Jan 2010
  17. 17.
    Wessel P, Smith W (1995) New version of the generic mapping tools released. EOS trans. Am Geophys Union 76:329CrossRefGoogle Scholar

Copyright information

© Springer Science+Business Media B.V. 2011

Authors and Affiliations

  1. 1.National Research Institute for Earth Science and Disaster PreventionIbarakiJapan

Personalised recommendations