The China Digital Seismograph Network (CDSN) is being designed and installed to provide the People's Republic of China with the facilities needed to create a national digital database for earthquake research. The CDSN, which is being developed jointly by the PRC State Seismological Bureau and the U.S. Geological Survey, will consist initially of nine digitally-recording seismograph stations, a data management system to be used for compiling network-day tapes, and a depot maintenance center. Data produced by the network will be shared with research scientists throughout the world.

A national seismograph network must be designed to support a variety of research objectives. From this standpoint, the choices and tradeoffs involved in specifying signal bandwidth, resolution, and dynamic range are the most important decisions in system design. As in the case of the CDSN, these decisions are made during the selection and design of the seismic sensor system and encoder components. The purpose of this report is to describe the CDSN sensor systems, their important signal characteristics, and the results of preliminary tests that have been performed on the instruments.

Four overlapping data bands will be recorded at each station: short period (SP), broadband (BB), long period (LP), and very long period (VLP). Amplitude response curves are illustrated in Figure I. Vertical and horizontal components will be recorded for each data band. The SP and LP channels will be recorded with sufficient sensitivities to resolve earth background noise at seismically quiet sites. The BB channels will have a lower sensitivity and are intended for broadband recording of moderate-to-large body-wave signals and for increasing the effective amplitude range in the short- and long-period bands. The VLP channel does not provide additional spectral coverage at long periods; its purpose is to make use of on-site filtration and decimation to reduce post processing requirements for VLP studies. Early plans also included a triaxial set of low-sensitivity accelerometers for recording strong signals from large local and regional earthquakes. The accelerometers are not being installed; however, they may be added in the future.

The short-period signals will be derived from a three-component set of PRC-supplied Model DJ-I SP seismometers and US-supplied SP amplifiers. The seismometers will be installed in surface or shallow subsurface vaults, except at two of the stations where they will be installed in boreholes. The BB, LP, and VLP signals will be derived from Streckeisen STS-1 broadband sensor systems installed in vaults, except at one site where the LP signals only will-be derived from a KS-36000 borehole seismometer installed at a depth of 100 meters.

Analog signals will be sampled and quantized by an analog-to-digital converter (ADC) that is part of the recording system. Sampling rates chosen for the CDSN are as follows:

* SP 40 samples/second

* BB 20 samples/second

* LP 1 sample/second

* VLP 6 samples/minute

The ADC 16-bit data word format makes use of 14 bits to quantize the signal and 2 bits to specify an automatically ranged gain of 1, 8, 32, or 128. This will provide 84 dB of resolution and up to 42 dB of gain ranging for a total opera- ting range of 126 dB peak to peak.

Magnetic tape cartridges, each having a capacity of 67 megabytes, will be used for recording the digital data. LP and VLP data will be recorded continu- ously. SP and BB data will be processed through an automatic signal detector of the type described by Murdock and Hutt (1983), and only detected events will be stored on tape. Detection parameters, such as turn-on sensitivity and mini- mum recording duration for the SP and BB channels, will be fully programmable and easily changed. One or more of the data channels may also be recorded on analog recorders.

A CDSN recording system was not available at the time that the preliminary tests were performed on the CDSN sensor systems.  This did not interfere with the principal goals of the testing which were to determine the best sensor installation techniques and to demonstrate the operational performance of the sensor systems, especially with regard to instrument noise and detection capability.  When the CDSN recording system is available, additional tests will be performed to measure distortion levels in the full system and components.  Distortion tests have been performed on the Streckeisen STS-1 broadband seismometers, the most critical component from the standpoint of linearity, and the results of the tests are reported by Wielandt and Streckeisen (1982).  Calibration accuracy and stability are also important sensor system characteristics that cannot be defined without additional testing.  These tests will be performed as the instruments are installed at the stations and periodically during operation.