Search Results

You are looking at 1 - 2 of 2 items for

  • Author or Editor: Brian Polagye x
  • All content x
Clear All Modify Search
Emma Cotter and Brian Polagye

Abstract

Multibeam sonars are widely used for environmental monitoring of fauna at marine renewable energy sites. However, they can rapidly accrue vast volumes of data, which poses a challenge for data processing. Here, using data from a deployment in a tidal channel with peak currents of 1–2 m s−1, we demonstrate the data-reduction benefits of real-time automatic classification of targets detected and tracked in multibeam sonar data. First, we evaluate classification capabilities for three machine learning algorithms: random forests, support vector machines, and k-nearest neighbors. For each algorithm, a hill-climbing search optimizes a set of hand-engineered attributes that describe tracked targets. The random forest algorithm is found to be most effective—in postprocessing, discriminating between biological and nonbiological targets with a recall rate of 0.97 and a precision of 0.60. In addition, 89% of biological targets are correctly classified as either seals, diving birds, fish schools, or small targets. Model dependence on the volume of training data is evaluated. Second, a real-time implementation of the model is shown to distinguish between biological targets and nonbiological targets with nearly the same performance as in postprocessing. From this, we make general recommendations for implementing real-time classification of biological targets in multibeam sonar data and the transferability of trained models.

Restricted access
Trevor Harrison, Kristen M. Thyng, and Brian Polagye

Abstract

High-resolution, four-dimensional mapping of currents in tidally dominated coastal settings can be conducted with a range of instrumentation. Here, we assess four approaches to data collection: an X-band radar, a stationary (bottom mounted) acoustic Doppler current profiler (ADCP), a mobile (vessel based) ADCP, and a swarm of Lagrangian floats. Using the output from a hydrodynamic simulation, a virtual field campaign was performed at 24 locations in Admiralty Inlet, Puget Sound, Washington, during spring and neap tidal exchanges. A reconstruction of the volumetric currents was generated for each platform every 15 min and evaluated against the true currents to assess accuracy over a horizontal extent of 400 m × 500 m at 5 m resolution and vertically through the entire water column (20–80 m) at 2 m resolution. Results demonstrate that, for this survey extent and resolution, a vessel-based ADCP survey is most accurate, followed closely by the float swarm. The overall performance hierarchy persists over most locations and times. Thus, if mapping currents at high resolution (<10 m) and short time scales (<1 day) is the primary scientific objective, vessel-based ADCP surveys are likely the best option. For longer-duration surveys, a combined deployment with a stationary ADCP and X-band radar system is the best choice. Last, if in situ measurements of scalar properties (e.g., salinity, temperature, dissolved oxygen) are also desired, float swarms can simultaneously sample these while surveying currents with accuracy comparable to mobile ADCPs.

Free access