A Genetic Algorithm to Produce Optimal Three Dimensional Position Estimates of an Acoustic Tag Location Using All Available Hydrophones

The U.S. Geological Survey California Water Science Center Hydrodynamics project team (U.S.G.S.C.W.S.C. Hydrodynamics) uses multi-dimensional tracks of acoustically tagged juvenile Chinook salmon to observe outmigration behavior of fish emigrating from the Sacramento River through the Sacramento – San Joaquin River Delta.  Our team has found that the current available technology for computing the location of an acoustic pulse that is observed on multiple hydrophones lacks sufficient accuracy to resolve key outmigration behavior parameters such as net rheotaxis orientation and swim speed.  A primary failure of existing algorithms is their inability to incorporate information from more than 4 hydrophone observations concurrently; this is a significant limitation given that many U.S.G.S.C.W.S.C. studies observe fish with up to 40 hydrophones.  In order to gain the required positioning resolution, a genetic algorithm to directly estimate the location of an acoustic tag pulse using information from all available hydrophones was developed.  The algorithm can be configured to use information from any number of hydrophones, and produces a position estimate and an estimate of positioning accuracy for each acoustic pulse observed by the hydrophone array.  Tracks generated for acoustically tagged fish using the algorithm are compared to tracks generated using traditional techniques, and outmigration behavior parameters are calculated for each set of tracks and compared.