A range-dependent echo-association algorithm and its application in split-beam sonar tracking of migratory salmon in the Fraser River watershed

Fish sonars have long been used to survey and monitor migratory salmon in rivers and at sea. However, research has been lacking in the development of algorithms to extract fish tracks from data collected in a riverine or oceanic environment. Current fish trackers, based on a pulse-to-pulse tracking method, only work well under ideal conditions when targets are well separated and the signal-to-noise ratio is high. Fisheries biologists often have to identify fish traces visually from raw echograms, This approach is both labor-intensive and time-consuming, limiting the usefulness of hydroacoustic techniques for fisheries management. This paper presents a fish-tracking algorithm which sorts randomly distributed echoes into coherent fish traces. Fish counts obtained with the algorithm compare well with visual counts at two quite different sites in the Fraser River watershed. The key features of the algorithm are: 1) the linking mechanisms among sequential fish echoes are range-dependent; 2) the growth echo for a developing track depends not only on its space-time relation with the previous track echo (the pulse-to-pulse statistics) but also on its relation to the entire track being constructed; 3) there are a total of only five echo-association criteria in the algorithm; 4) the simplicity of the algorithm structure provides a convenient platform for implementing specific and sophisticated tracking criteria to meet specific needs; and 5) the user can fully control the performance of the algorithm by choosing values for the 11 well-defined tracking parameters.