High-Performance Compression of Multibeam Echosounders Water Column Data

Abstract

Over the last few decades, multibeam echosounders (MBES) have become the dominant technique to efficiently and accurately map the seafloor. They now allow to collect water column acoustic images along with the bathymetry, which is providing a wealth of new possibilities in oceans exploration. However, water column imagery generates vast amounts of data that poses obvious logistic, economic, and technical challenges. Surprisingly, very few studies have addressed this problem by providing efficient lossless or lossy data compression solutions. Currently, the available options are only lossless, providing low compression ratios at low speeds. In this paper, we adapt a data compression algorithm, the Fully Adaptive Prediction Error Coder (FAPEC), which was created to offer outstanding performance under the strong requirements of space data transmission. We have added to this entropy coder a specific pre-processing stage tailored to the Kongsberg Maritime water column file formats. Here, we test it on data acquired with Kongsberg MBES models EM302, EM710, and EM2040. With this bespoke pre-processing, FAPEC provides good lossless compression ratios at high speeds, whereas lossy ratios reach water column file sizes even smaller than bathymetry raw files still with good image quality. We show the advantages over other lossless compression solutions, both in terms of compression ratios and speed. We illustrate the quality of water column images after lossy FAPEC compression, as well as its resilience to datagram errors and its potential for automatic detection of water column targets. We also show the successful integration in ARM microprocessors (like those used by smartphones and also by autonomous underwater vehicles), which provides a real-time solution for MBES water column data compression.