Oral Presentation Society for Freshwater Science 2026 Annual Meeting

Rapid GPU-Accelerated Fish Track Detection from Split-Beam Hydroacoustics for Riverine Lake Sturgeon Monitoring (135312)

Eamonn Powers 1 , Eric Snyder 1 , Stephanie Ogren 2 , Marty Holtgren 3 , Bill Flanagan 4
  1. GVSU, Grand Rapids, MICHIGAN, United States
  2. Grand Rapids Public Museum , Grand Rapids, Michigan, US
  3. Encompass Socio-ecological Consulting, Manistee , Michigan , United States
  4. John Ball Zoo , Grand Rapids, Michigan, US

The purpose of this study is to generate a population estimate for Lake Sturgeon in the Grand River, Michigan, a threatened population, especially given the presence of a dam in downtown Grand Rapids. Hydroacoustic surveys monitoring adult sturgeon were coupled with juvenile pit tagging/genetic surveys. Both surveys allow for a greater understanding of the threatened Lake Sturgeon population in the Grand River. Genetic parentage analysis indicates a small population of spawning adults, a conclusion in the process of being evaluated by fixed hydroacoustic surveys. 

Hydroacoustic surveys are a valuable, non-invasive method of monitoring fish in larger rivers, yet split-beam datasets can be incredibly slow to turn into usable information when processing and interpretation requires a substantial amount of manual effort. Our Lake Sturgeon research team presents a significantly simpler workflow that speeds up the path between raw split-beam recordings, to summarize fish tracks that are ready for review, export, and further analysis.  

We used an HTI, Hydroacoustic Technology Inc. 241 split beam echosounder with a 2.5×10° elliptical beam transducer. The workflow we have developed is designed to support routine use during the field season by producing track candidates quickly, as well as organizing them in a fashion that makes it easy to confirm positive observations. 

The outputs we generate are formatted for relatively common monitoring questions tied to Lake Sturgeon monitoring, such as where and when fish are encountered and how movement patterns vary by both season and environmental conditions. Overall, this approach demonstrates an advanced path towards making non-invasive split-beam hydroacoustics scalable for riverine monitoring of fishes.