Determination of Crust and Upper Mantle Structure Beneath Africa Using a Global Optimization Based Waveform Modeling Technique
Abstract
During the initial years of the project, a waveform modeling program based on the reflectivity method and incorporating a global Optimization algorithm, was developed. During the inversion process, the method implements a global Optimization algorithm using very fast simulated annealing (VFSA). Global optimization allows for searching within a broad spectrum of models and eliminates dependency on the starting model. VFSA is advantageous because it allows for larger sampling of the model space during early stages, and much narrower sampling in the model space as the inversion converges and the temperature decreases. Ability of VFSA to perform different perturbations for different model parameters allows for individual control of each parameter and the incorporation of priori information. The method also uses statistical tools such as computation of the posterior probability density (PPD) function and the parameter correlation matrix, so as to evaluate the uniqueness of a particular model and the trade-offs between individual model parameters therein. The waveform modeling approach has now been applied to ground truth data recorded spanning the African continent. The waveforms modeled included S, Sp, SsPmP, and shear-coupled PL(SPL) phases from deep (200-800 km) earthquakes of magnitude 5.5 to 7.0 located at distances of 30 to 80 degrees. Overall, the method is successful in determining the crust and upper mantle structure of the African continent. Notable advantages include ability to simultaneously model all the waveforms, assess uncertainties in model parameters, finding a range of acceptable models that explain the data, and avoiding a dependence of the final model on the starting model. Furthermore, it differs from other commonly used methods in the fact that it provides a direct measurement of P and S wave velocities siniultaneously. The use of the SPL phase wherever available improves constraints on the models of the lower crust and upper mantle.
Document Details
- Document Type
- Technical Report
- Publication Date
- Sep 29, 2006
- Accession Number
- ADA468916
Entities
People
- Abhijit Oangopadhyay
- Jay Plilliam
- Mrinal K. Sen
Organizations
- University of Texas at Austin