The best robust estimation method to determine local surface

Abstract

Geoid/quasi-geoid modelling in a local (small) area is often made with polynomials. The optimal version of a polynomial is found by testing its successive versions using different statistical parameters, i.e. with different degree and number of coefficients. In this publication, the authors presented 3 approaches to search for an optimal version of the polynomial. Two of them were based on one of the robust estimation methods, i.e.The Danish method. Therefore, the question arises whether this method is suitable for this type of work. The publication analyses 6 different methods of robust estimation and the method of least squares. The control parameters of the methods were selected in such a way as to obtain a comparable accuracy of fitting the polynomial into the points each time. Then, the actual accuracy obtained for each of them was compared. The result of the study is a ranking of methods in terms of the most accurate fitting results.