LARGE-ARRAY SIGNAL AND NOISE ANALYSIS. SUBARRAY PROCESSING
Abstract
A theoretical Wiener multichannel filter was designed and applied to all operating subarrays for 14 noise samples and three signals. The filter system chosen had a disk signal model (11 km/sec to infinite velocity) and a noise annulus (2 to 6 km/sec) and exhibited good wavenumber response to 0.2 cps. Preliminary analysis showed that it was necessary to equalize the noise at the low-frequency peak (0.2 to 0.3 cps) prior to processing to obtain consistent noise rejection at low frequency. Because of the peaked spectrum, equalization could be accomplished by adjusting the 25 channels in a subarray to have the same RMS noise level (i.e., 1-point equalization). Two measured-noise Wiener filters were designed using an infinite-velocity signal model (with 30-percent gain fluctuation added). Their noise rejection was about 2 db better than the theoretical system over most of the 0- to 5-cps band. The theoretical Wiener filter performed about as well as the maximum-likelihood filter that had been applied to a noise sample not used in its filter design. The measured-noise Wiener filter performed almost as well as the maximum-likelihood filter that had been applied to the noise sample from which it was designed. Within a subarray, the noise at LASA was more than 99-percent predictable at the microseismic peak (0.2 to 0.3 cps). At 1.0 cps, the noise was still 65-percent predictable; but above 2.0 cps, it was essentially unpredictable. These results are similar to those at TFO and WMO but significantly lower than CPO.
Document Details
- Document Type
- Technical Report
- Publication Date
- Oct 16, 1967
- Accession Number
- AD0822715
Entities
People
- Terence W. Harley
Organizations
- Texas Instruments