Integrating diverse calibration products to improve seismic location
The monitoring of nuclear explosions on a global basis requires accurate event locations. As an example, under the Comprehensive Test Ban Treaty, the size of an on-site inspection search area is 1,000 square kilometers or approximately 17 km accuracy assuming a circular area. This level of accuracy is a significant challenge for small events that are recorded using a sparse regional network. In such cases, the travel-time of seismic energy is strongly affected by crustal and upper mantle heterogeneity and large biases can result. This can lead to large systematic errors in location and, more importantly, to invalid error bounds associated with location estimates. Corrections can be developed and integrated to correct for these biases. These path corrections take the form of both three-dimensional model corrections along with three-dimensional empirically based travel time corrections. LLNL is currently working to integrate a diverse set of three-dimensional velocity model and empirical based travel-time products into one consistent and validated calibration set. To perform this task, we have developed a hybrid approach that uses three-dimensional model corrections for a region and then uses reference events when available to improve the path correction. This Bayesian kriging approach uses the best apriori three-dimensional velocity model that is produced for a local region and uses this as a baseline correction. When multiple models are produced for a local region, uncertainties in the models are compared against each other using ground truth data and an optimal model is chosen. We .are in the process of combining three-dimensional models on a region-by-region basis and integrating the uncertainties to form a global correction set. The Bayesian kriging prediction combines this a priori model and its statistics with the empirical calibrations to give an optimal aposteriori calibration estimate. In regions where there is limited or no coverage by reference events the corrections will be based primarily on the model. The integrated a priori model is particularly important in these areas. In regions with adequate calibration events, we are demonstrating improvement in event location through the reduction of regional bias. In regions with sparse or no ground truth, the a priori model will need to be spot-validated with the use of dedicated calibration experiments or through the use of mining explosions, where available.
- Research Organization:
- Lawrence Livermore National Lab. (LLNL), Livermore, CA (United States)
- Sponsoring Organization:
- US Department of Energy (US)
- DOE Contract Number:
- W-7405-ENG-48
- OSTI ID:
- 15005957
- Report Number(s):
- UCRL-JC-138989; TRN: US200402%%84
- Resource Relation:
- Conference: 22nd Annual Department of Defense/Department of Energy Seismic Research Symposium, New Orleans, LA (US), 09/12/2000--09/15/2000; Other Information: PBD: 17 Jul 2000
- Country of Publication:
- United States
- Language:
- English
Similar Records
LLNL Seismic Locations: Validating Improvement Through Integration of Regionalized Models and Empirical Corrections
Grid-Search Location Methods for Ground-Truth Collection From Local and Regional Seismic Networks