STRATASkip to the research
Metal Orca · Immersive, original XLVII plate

Metal Orca · Immersive

An acoustic pulse can reach a receiver by several paths, producing delayed arrivals from one source. Event counts depend on whether the recording and analysis can distinguish those arrivals from independently emitted signals.

For a rectangular observation, a 0.25-second record has a Fourier-bin spacing of 4 Hz; a one-second record has a spacing of 1 Hz. These values specify the frequency grid, while separation of nearby tones also depends on the window, sampling, amplitudes, phase, and noise.

Historical record

  1. Bandwidth becomes a condition of distinguishable transmission.

    Nyquist connects a communication system’s frequency response to whether signals can be recovered without ambiguity. The same mathematical concern appears when an acoustic recording must distinguish nearby tones rather than merely register that sound is present.

  2. Noise enters a general model of communication.

    Shannon’s theory provides a language for asking what information a noisy channel can carry. Applying it to an acoustic setting requires an actual model of the source, the water and the receiver; the general theory does not supply those measurements.

  3. An acoustic habitat becomes a subject of coordinated observation.

    NOAA releases its Ocean Noise Strategy Roadmap on 23 September. The document treats sound through its sources, propagation and consequences for marine life, broadening the question from detecting a signal to understanding the environment in which it is heard.

A longer recording can separate tones that a short window blurs.

Construct two steady tones at 100 Hz and 102 Hz. A 0.25-second rectangular observation has a Fourier-bin spacing of 4 Hz; a one-second observation has a spacing of 1 Hz. The longer record can show their separation more clearly under suitable sampling and noise conditions. Bin spacing is not a universal promise of resolution: window shape, amplitude, phase and background noise still matter. Adding a delayed copy creates a second arrival from one source, so an additional peak in time need not mean an additional animal or event.

  • Keep the sample rate and source amplitudes fixed while changing the recording duration. Show both the waveform and the spectrum.
  • Add a synthetic echo at a known delay, then compare the result with two independently timed sources. Label the construction so it is not mistaken for field evidence.
  • Review an event count when the detector cannot separate an echo from an independent call. Preserve the original recording and the detector settings for that review.

Recordings and archives

Research sources

  1. Certain Topics in Telegraph Transmission Theory

    Harry Nyquist / IEEE reprint hosted by Washington University · Original treatment of transmission bandwidth and unambiguous signal interpretation.

  2. A Mathematical Theory of Communication

    Claude E. Shannon / Bell Labs · 1948 communication model, entropy and coding framework.

  3. New Guidelines Show NOAA’s Commitment to Address Effects of Ocean Noise on Marine Mammals

    NOAA Fisheries · 23 September 2016 release of the Ocean Noise Strategy Roadmap.

  4. Multimedia

    NOAA Pacific Marine Environmental Laboratory · Official underwater sound, video and animation archive; recordings retain their own context.