Geophysical parameter estimation - Application to denoising of seismological data

Abstract

Geophysical data modelling involves parameter estimation of the modelled system using mathematical relationship describing the physical process. Most of these relationships are inherently non- linear and requires solving them through a process of linearization or using any of the nonlinear search algorithm. Estimating model parameter from the geophysical data is not only unique, but also dependent on the initial model. Apart from these, the data error adds to the parameter estimation complexity. Some of these issues have been addresses through robust statistics incorporating apriori information related to data and model covariance through their probability density function and search through global optimisation. In this thesis, we look at the various denoising algorithm and implement the best approach to model gravitational field data from India in terms of Earth parameters. Using this approach, we can retrieve the original signal from a noised signal upto a great extent with a decent signal to noise ratio.