DMCA
Automatic Detection of Outliers in Multibeam Echo Sounding Data
| Citations: | 1 - 0 self |
BibTeX
@MISC{Hou_automaticdetection,
author = {Tianhang Hou and Lloyd C Huff and Larry Mayer},
title = {Automatic Detection of Outliers in Multibeam Echo Sounding Data},
year = {}
}
Share
 |
 |
 |
 |
||
OpenURL
Abstract
The data volumes produced by new generation multibeam systems are very large, especially for shallow water systems. Results from recent multibeam surveys indicate that the ratio of the field survey time, to the time used in interactive editing through graphical editing tools, is about 1:1. An important reason for the large amount of processing time is that users subjectively decide which soundings are outliers. There is an apparent need for an automated approach for detecting outliers that would reduce the extensive labor and obtain consistent results from the multibeam data cleaning process, independent of the individual that has processed the data. The proposed automated algorithm for cleaning multibeam soundings was tested using the SAX-99 (Destin FL) multibeam survey data [2]. Eight days of survey data (6.9 Gigabyte) were cleaned in 2.5 hours on an SGI platform. A comparison of the automatically cleaned data with the subjective, interactively cleaned data indicates that the proposed method is, if not better, at least equivalent to interactive editing as used on the SAX-99 multibeam data. Furthermore, the ratio of acquisition to processing time is considerably improved since the time required for cleaning the data was decreased from 192 hours to 2.5 hours (an improvement by a factor of 77).
Keyphrases
automatic detection multibeam echo sounding data processing time graphical editing tool shallow water system data volume new generation multibeam system sgi platform eight day destin fl consistent result multibeam sounding field survey time extensive labor recent multibeam survey multibeam data cleaning process automated approach sax-99 multibeam data important reason apparent need survey data multibeam survey data large amount interactive editing
