Page 34 - Summer2022

FEATURED ARTICLE Weird Data: The Element of Surprise in Underwater Acoustic Sensing Erin M. Fischell The Challenge of Underwater Acoustics Dive underwater and the world changes. The color you see shifts toward blue green, your ears pop and fill with water, and suddenly things sound different (e.g., Casper et al., 2022). The splashes and bird calls above the waves disap- pear and are replaced by groans, clicks, and pops (Dahl et al., 2007). What are these sounds? Where do they come from? Or maybe you are on a boat over a trench thousands of meters deep. Suddenly, the depth sounder seems to think that the water is only 500 meters deep, but the chart says it should be 5,000 meters. The instrument jumps from 500 to 5,000 meters and back for a few minutes and then settles. What happened? The commonality between a boat’s depth sounder and those mysterious sounds you hear when you dive beneath the waves is underwater acoustics, a field in which even the most experienced practitioners struggle with under- standing all of the many sources of interference, noise, and changing physics needed for data interpretation. Users of ocean acoustic instruments don’t control whale calls, shipping, snapping shrimp distribution, fish finders on other vessels, nesting creatures, or pile driving and cannot predict ahead of time all of the possible sources of interference in acoustic data. The complexity of under- water acoustic systems provides further challenges; is that unexpected signal a new source in the environment, a potential signal of interest, or system noise? This complex interaction of environment, uncontrolled and uncorrelated noise sources, internal noise sources, and unexpected reflections leads to a lot of “weird data” in underwater acoustics. These weird data are the seg- ments in any underwater acoustic time series that don’t answer the fundamental questions at the core of the experiment or are mysterious in origin. Humans are driven to find patterns in the chaos and to try to understand the whys and wherefores of our world. Everyone experiences trying to understand weird sound data in daily life. For example, you might hear an unex- pected squeak or beep at home and spend a few minutes walking around the house, turning your ears in different directions, sticking your head out windows, and pausing to listen, all to try to find the noise source. Or perhaps you apply sound pattern recognition while trying to diagnose a suspicious rattle in a car engine, pressing the gas pedal and then the brake, querying your partner to determine if the sound got quieter or louder with the change of variables. The quality of underwater acoustic measurements con- stantly changes based on uncontrolled, capricious factors. Therefore, interpretation of underwater acoustic data is like hunting that unexpected squeak, groan, rattle, or flutter, generally without the benefit of being able to stick your head out the window or adjust self-noise to see if the sound is still there. This sometimes frustrating pro- cess is the quintessential center of science: taking in the unexpected and using that new information to question the foundations of knowledge. There is a lot to be learned by our weird data, and in surprises that provide insight into systems, biology, and oceanographic processes in the ocean. Categories of Interference Surprises in underwater acoustic signals occur when our a priori understanding of the environment, ambient-noise sources, and the paths of transmission of acoustic energy are incorrect or incomplete. The ocean environment is stochastic in nature (Colosi, 2016), with properties of both signal and noise varying constantly in ways that are difficult to predict (Miksis-Olds et al., 2018). Understand- ing underwater acoustic data becomes difficult when the ©2022 Acoustical Society of America. All rights reserved. 34 Acoustics Today • Summer 2022 | Volume 18, issue 2 https://doi.org/10.1121/AT.2022.18.2.34