Highly Efficient Bayesian Tomography for Multi-Phase 2-D Coda Calibration: Application to North Korea

Abstract

To accomplish a full 2-D calibration capability in the coda methodology, we developed a new Bayesian calibration technique for the 2-D envelope shape correction. First, we developed a hybrid envelope template modeling technique by integrating the hybrid scattering formula, forward- and wide- angle scattering coefficients derived on 3D von Karman media, and the Markov Chain Monte Carlo sampling method. The Bayesian interface of the template modeling technique allows us to sample the posterior distribution of the envelope shape parameters. Then, we developed a new calibration technique by adopting a trans-dimensional Bayesian tomography algorithm to map the sampled posterior of the shape parameters. In our new calibration approach, all modeling parameters for the hybrid template are physics-based and calibrated for laterally varying scattering/absorbing structures. The new hybrid envelope modeling technique includes both forward- and wide-angle- scattering effects in full-waveform envelope templates that fit the direct and scattered phases in local- and regional-distance observations of low-yield explosions. The newly developed tomographic calibration technique allows us to estimate optimal envelope shape parameters for the hybrid template for given source-receiver geometry and subsequent accurate yield estimates from small local explosions. Our combined R and D development efforts are producing significantly improved fits to shorter-length signals from smaller events, as well as more accurate yield estimates. The new processing technique is also shifting conventional coda-based analysis from a strictly empirical basis to a physical one. These improvements have been long-term goals of significant interest to U.S. monitoring agencies.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Apr 11, 2018
Accession Number
AD1123120

Entities

People

  • Seung-Hoon Yoo

Tags

Communities of Interest

  • Energy and Power Technologies

DTIC Thesaurus Topics

  • Air Force Research Laboratories
  • Algorithms
  • Arms Control
  • Bayesian Networks
  • California
  • Computational Science
  • Convolution Integrals
  • Department Of State
  • Forward Scattering
  • Monte Carlo Method
  • North America
  • North Korea
  • Scattering
  • Three Dimensional
  • Two Dimensional
  • United States
  • Wave Propagation

Readers

  • Acoustical Oceanography.
  • Adaptive Control and Estimation with Uncertainty in Dynamic Systems.
  • Seismology

Technology Areas

  • AI & ML
  • AI & ML - Bayesian Inference