Skip to content


Open Access

Soft computing methods for geoidal height transformation

Earth, Planets and Space200961:BF03353193

Received: 18 June 2008

Accepted: 6 January 2009

Published: 31 August 2009


Soft computing techniques, such as fuzzy logic and artificial neural network (ANN) approaches, have enabled researchers to create precise models for use in many scientific and engineering applications. Applications that can be employed in geodetic studies include the estimation of earth rotation parameters and the determination of mean sea level changes. Another important field of geodesy in which these computing techniques can be applied is geoidal height transformation. We report here our use of a conventional polynomial model, the Adaptive Network-based Fuzzy (or in some publications, Adaptive Neuro-Fuzzy) Inference System (ANFIS), an ANN and a modified ANN approach to approximate geoid heights. These approximation models have been tested on a number of test points. The results obtained through the transformation processes from ellipsoidal heights into local levelling heights have also been compared.

Key words

Fuzzy inference systemsneural networkgeoid undulation