Geophysical models constitute an important component of calibration for nuclear explosion monitoring. We will focus on four major topics: (1) a priori geophysical models, (2) surface wave models, (3) receiver function derived profiles, and (4) stochastic geophysical models. The first, a priori models, can be used to predict a host of geophysical measurements, such as body wave travel times, and can be derived from direct regional studies or even by geophysical analogy. Use of these models is particularly important in aseismic regions or regions without seismic stations, where data of direct measurements might not exist. Lawrence Livermore National Laboratory (LLNL) has developed the Western Eurasia and North Africa (WENA) model which has been evaluated using a number of data sets, including travel times, surface waves, receiver functions, and waveform analysis (Pasyanos et al., 2004). We have joined this model with our Yellow Sea - Korean Peninsula (YSKP) model and the Los Alamos National Laboratory (LANL) East Asia model to construct a model for all of Eurasia and North Africa. Secondly, we continue to improve upon our surface wave model by adding more paths. This has allowed us to expand the region to all of Eurasia and into Africa, increase the resolution of our model, and extend results to even shorter periods (7 sec). High-resolution models exist for the Middle East and the YSKP region. The surface wave results can be inverted either alone, or in conjunction with other data, to derive models of the crust and upper mantle structure. We are also using receiver functions, in joint inversions with the surface waves, to produce profiles directly under seismic stations throughout the region. In a collaborative project with Ammon, et al., they have been focusing on stations throughout western Eurasia and North Africa, while we have been focusing on LLNL deployments in the Middle East, including Kuwait, Jordan, and the United Arab Emirates. Finally, we have been exploring methodologies such as Markov Chain Monte Carlo (MCMC) to generate data-driven stochastic models. We have applied this technique to the YSKP region using surface wave dispersion data, body wave travel time data, and receiver functions.
