Improved visualization of large temporal series for the evaluation of good environmental status

Abstract

[EN] Passive acoustic monitoring of underwater sounds is an emerging discipline that can be used to guarantee that anthropogenic noise meets acceptable limits, to detect the presence of cetacean species, and to ensure sustainable exploitation of our oceans and seas. In this scenario, graphical representation techniques play a key role in helping to reveal seasonal structures of human made noises. Nevertheless, for very long temporal series, it might be challenging to find a graphic visualization technique that allows representing a time range that is long enough to capture these seasonal events, while at the same time preserving short isolated events. We propose a framework for the creation of such visualization techniques and analyze the different stages involved: data reduction, color encoding, and signal processing on graphs. All of this is applied to data from deployments in two marine protected areas in order to provide an acoustic panorama and identify seasonal events.