Figure 2 from Ordoñez, Alba, Ingrid Utseth, Olav Brautaset, Rolf Korneliussen, and Nils Olav Handegard. “Evaluation of Echosounder Data Preparation Strategies for Modern Machine Learning Models.” Fisheries Research 254 (October 1, 2022): 106411. https://doi.org/10.1016/j.fishres.2022.106411.
Echosounder data (echogram) from the Lofoten Vesterålen Ocean Observatory illustrating the improved resolution with broadband signals (FM) over conventional narrowband signals (CW). The same aggregation of marine organisms was observed first with CW before switching to FM.
Broadband echogram showing backscattering by a small natural seep of methane (red box) near Node 7 of the Lofoten Vesterålen Ocean Observatory. The broadband frequency response of the seep is also shown (lower right inset). The seep consists of a few small single bubbles rising from the seafloor at a depth of 220 meters, clearly observed in the broadband data.
Echogram of station 364 (38 kHz) with depth profile and predictions of blue whiting (blue), herring (red), mackerel (green), and mesopelagic fishes (orange). Images at top are from the positions indicated along the trawl’s path and show the bounding boxes calculated by the object detection model. The size of the bubbles in the centre panel are proportional to the number of fish predicted per minute. The stacked histogram at bottom similarly shows the number of each fish species per minute but avoids the problem of the blue whiting symbol obscuring mesopelagic fishes.
DOI 10.1093/icesjms/fsab227