Implementation of a generalized method for the analysis of deformation surveys
Over the past three years, a generalized method for analyzing deformation surveys has been developed at the University of New Brunswick. A number of papers and a Ph. D. thesis have presented a rigorous and unified method which has successfully achieved the ability to apply the same computational procedure to any type of deformation; to be applied to one, two, or three dimensional data with spatial and temporal parameters; to utilize any type of data simultaneously; to be applied to any configuration, even when incomplete or defective; and to allow the use of any type of minimal constraints. The method of analysis follows several steps, from assessing the quality of the observations, through the pursuit of trend and the devising of possible models, to the estimation of model parameters with assessment of the model, its parameters and the choice of the “best” model and its depiction. Dealing with the implementation of the generalized method, this thesis provides detailed discussion on the inclusion of a variety of non-geodetic observables, on trend analysis, on model selection, and on model parameter estimation using a number of campaigns simultaneously. In doing so, two main examples are given. One is a reference triangulation network which is isolated with several of the reference stations being unstable. The other is a relative trilateration network for monitoring tectonic movement in California, with 12 years of at least annual campaigns and varying configuration. This method has been found to be flexible and adaptable to an assortment of applications and truly generalized.