@ARTICLE{echofilter,
    AUTHOR={Lowe, Scott C. and McGarry, Louise P. and Douglas, Jessica and Newport, Jason and Oore, Sageev and Whidden, Christopher and Hasselman, Daniel J.},
    TITLE={{E}chofilter: {A} Deep Learning Segmention Model Improves the Automation, Standardization, and Timeliness for Post-Processing Echosounder Data in Tidal Energy Streams},
    JOURNAL={Frontiers in Marine Science},
    VOLUME={9},
    YEAR={2022},
    URL={https://www.frontiersin.org/articles/10.3389/fmars.2022.867857},
    DOI={10.3389/fmars.2022.867857},
    ISSN={2296-7745},
    ABSTRACT={Understanding the abundance and distribution of fish in tidal energy streams is important for assessing the risks presented by the introduction of tidal energy devices into the habitat. However, tidal current flows suitable for tidal energy development are often highly turbulent and entrain air into the water, complicating the interpretation of echosounder data. The portion of the water column contaminated by returns from entrained air must be excluded from data used for biological analyses. Application of a single algorithm to identify the depth-of-penetration of entrained air is insufficient for a boundary that is discontinuous, depth-dynamic, porous, and varies with tidal flow speed.Using a case study at a tidal energy demonstration site in the Bay of Fundy, we describe the development and application of deep machine learning models with a U-Net based architecture that produce a pronounced and substantial improvement in the automated detection of the extent to which entrained air has penetrated the water column.Our model, Echofilter, was found to be highly responsive to the dynamic range of turbulence conditions and sensitive to the fine-scale nuances in the boundary position, producing an entrained-air boundary line with an average error of 0.33 m on mobile downfacing and 0.5–1.0 m on stationary upfacing data, less than half that of existing algorithmic solutions. The model’s overall annotations had a high level of agreement with the human segmentation, with an intersection-over-union score of 99% for mobile downfacing recordings and 92–95% for stationary upfacing recordings. This resulted in a 50% reduction in the time required for manual edits when compared to the time required to manually edit the line placement produced by the currently available algorithms. Because of the improved initial automated placement, the implementation of the models permits an increase in the standardization and repeatability of line placement.}
}