SOURCES OF ERROR IN A SEISMIC DIGITAL RECORDING SYSTEM*)

The use of digital recorders and computers in seismic exploration promises major enhancement of the quality of final documents available to interpreters. The ultimate objectives of recording and processing remain what they always have been:

1 Record the reflection wavelet as a function of time; this requirement has been met with satisfactory accuracy for a number of years.

2. Record the reflection wavelets with sufficient fidelity to permit the interpreter to recognize them.

Various factors affect our ability to achieve this second objective. Certain recording errors are associated with digital recording systems. However, an understanding of the sources of error will enable the operator to use his system properly and to estimate the noise level or inaccuracy of field recordings. Field operations do not require rigorous error analysis; in most cases a satisfactory approximation can be obtained from simple calculations.

Three types of “noise”–seismic, instrument and power line–introduce errors. Factors which contribute to over‐al recording system error include specifically input noise, power supply ripple, crosstalk, A‐D conversion error, quantizing noise, aliasing, distortion. Examination of each component of a recording system, permits the determination of its ultimate effect on the over‐all noise level–or error level–of the entire system.

Many of the error sources produce statistically independent noise which is not correlative. Where this is true, error voltages from various sources may be combined by taking the square root of the sum of the mean square noise voltages, giving a result slightly greater than the largest single voltage if one source is much greater than any other source. This simplification can be used to estimate over‐all system noise levels. Distortion and crosstalk depend on signal amplitude and should be added algebraically in each category. Each final sum should be used as a statistically independent noise source with respect to other system noise sources.

Using the foregoing examples and simplified system for estimating over‐all system noise, and assuming that much of the distortion (which limits signal/instrument noise ratio to 54 db) can be removed by filtering, we determine that the combined effect of all sources of error is to reduce the system S/N ratio to approximately 74 db.

With proper care digital field recording systems can produce very good field records, and exotic computer processes can enhance signal and reduce various forms of noise. However, one always must recall that the level of confidence which one can place in an interpretation of seismic data must be dependent on a knowledge of the accuracy of the basic data.