Prediction and Elimination of Pressure Generated Ground Noise on Long Period Seismograms Using Optimum Filters.
Abstract
Seismic noise generated by surface winds severely limits the usefulness of long-period seismograms. A linear dependence is observed between the recordings of a long-period seismograph and a co-located microbarograph in the period range 20 to 100 seconds. It is possible to design a filter, based on the least-mean-square method, such that the effects of the wind generated noise on the seismograph may be predicted from the atmospheric pressure changes recorded by microbarograph located at the same site. The filter has been used to reduce the noise level for seismic data and hence increase the S/N ratio. However, only limited success has been achieved and the efficiency of the prediction tends to depend on the wind direction, wind speed and the variance of the wind speed. The turbulence cells associated with atmospheric pressure changes are believed to be broken down into smaller cells as the speed and variability of the wind increase. The increase of high frequency energy contributed by smaller cell disturbances also increases the nonlinear part of the transfer function and hence breaks down the prediction ability of the filter. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- Jan 01, 1977
- Accession Number
- ADA120247
Entities
People
- Eugene T. Herrin
- Tom Goforth
Organizations
- Southern Methodist University