ROBUST METHODOLOGY FOR DETECTION OF SPIKES IN MULTIBEAM ECHO SOUNDER DATAROBUST METHODOLOGY FOR DETECTION OF SPIKES IN MULTIBEAM ECHO SOUNDER DATA

Resumo

Currently, during the operation in shallow waters, scanning systems, such as multibeam systems, are capable of collecting thousands of points in a short time, promoting a greater coverage of the submerged bottom, with consequent increase in the detection capacity of objects. Although there has been an improvement in the accuracy of the depths collected, traditional processing, that is, manual, is still required. However, mainly due to the increased mass of data collected, manual processing has become extremely time-consuming and subjective, especially in the detection and elimination of spikes. Several algorithms are used to perform this task, but most of them are based on statistical assumptions hardly met and/or verified, such as spatial independence and normality. In this sense, the goal of this study is to present the SODA (Spatial Outlier Detection Algorithm) methodology, a new method for detection of spikes designed to treat bathymetric data collected through swath bathymetry systems. From computational simulation, promising results were obtained. SODA, in some cases, was capable to identify even 90% of spikes inserted on simulation, showing that the methodology is efficient and substantial to the bathymetric data treatment.