A new algorithm for automatic processing of bathymetric data

A sea-floor map can be obtained from sonar data using the supplied sonar system software or one of several public domain or proprietary systems. With any of these solutions, the process of production of a sea-floor map is unnecessarily time consuming in the absence of a systematic and efficient solution to the problem of identification of erroneous data (outliers). We present an algorithm to fit bathymetric data and to automatically deal with outliers. The most important characteristics of the algorithm are: production of a triangulated map in which the local spatial resolution and the local known accuracy are dependent both on the sea-floor topography and the data noise amplitude, automatic identification of outliers, intrinsic modularity of the program suitable for a multiprocessor environment, low computing cost, and minimal operator intervention. The algorithm is described and its parameters are defined on the basis of their effect on the results. The results of the application of the algorithm to synthetic data with and without outliers demonstrate its behavior on a known "sea floor". The results of the algorithm on real data and its performances (in term of speed and fitting error) are also tested.