Robust and intelligent bearing estimation
As the monitoring thresholds of global and regional networks are lowered, bearing estimates become more important to the processes which associate (sparse) detections and which locate events. Current methods of estimating bearings from observations by 3-component stations and arrays lack both accuracy and precision. Methods are required which will develop all the precision inherently available in the arrival, determine the measurability of the arrival, provide better estimates of the bias induced by the medium, permit estimates at lower SNRs, and provide physical insight into the effects of the medium on the estimates. Initial efforts have focused on 3-component stations since the precision is poorest there. An intelligent estimation process for 3-component stations has been developed and explored. The method, called SEE for Search, Estimate, and Evaluation, adaptively exploits all the inherent information in the arrival at every step of the process to achieve optimal results. In particular, the approach uses a consistent and robust mathematical framework to define the optimal time-frequency windows on which to make estimates, to make the bearing estimates themselves, and to withdraw metrics helpful in choosing the best estimate(s) or admitting that the bearing is immeasurable. The approach is conceptually superior to current methods, particular those which rely on real values signals. The method has been evaluated to a considerable extent in a seismically active region and has demonstrated remarkable utility by providing not only the best estimates possible but also insight into the physical processes affecting the estimates. It has been shown, for example, that the best frequency at which to make an estimate seldom corresponds to the frequency having the best detection SNR and sometimes the best time interval is not at the onset of the signal. The method is capable of measuring bearing dispersion, thereby withdrawing the bearing bias as a function of frequency. The lowest measurable frequency in the dispersion pattern is often a near error free bearing. These latter features should be helpful in calibrating the stations for frequency dependent biases induced by the earth. Future efforts will enhance the SEE algorithm and will also evaluate it using larger station data sets.
- Research Organization:
- Sandia National Labs., Monitoring Technologies Dept., Albuquerque, NM (United States)
- Sponsoring Organization:
- USDOE, Washington, DC (United States)
- DOE Contract Number:
- AC04-94AL85000
- OSTI ID:
- 656698
- Report Number(s):
- SAND-98-1095C; CONF-980920-; ON: DE98002997; BR: GC0402000; TRN: AHC29817%%255
- Resource Relation:
- Conference: 20. annual seismic research symposium on monitoring a comprehensive test ban treaty, Santa Fe, NM (United States), 21-23 Sep 1998; Other Information: PBD: [1998]
- Country of Publication:
- United States
- Language:
- English
Similar Records
Robust Bearing Estimation for 3-Component Stations
MatSeis and the GNEM R&E regional seismic anaylsis tools.