NBHF

Methods Summary

  • Pamguard v.2.02.09 (https://www.pamguard.org/) is used to run click detector on full bandwidth data (384 kHz sample rate)

  • Data is high-pass filtered at 100 kHz with a click detector (12 dB threshold) 

  • Pamguard v2.02.09f (public release expected to be available in early 2024) is used to reclassify clicks

    • Click detections are classified within Pamguard based on peak frequency

    • The Matched Template Classifier module evaluates the similarity of each click detection to known templates

  • Potential acoustic events are automatically defined in R

  • PAMpal is used to calculate features of click detections in each event

Step-by-Step Instructions

Run echolocation click detector

  1. Define location to save database and binary files

  2. Define array configuration (hydrophone sensitivity and separation)

  3. Save configuration file with a descriptive filename

  4. Press Run

Reclassify clicks with click classifiers and matched template classifier

  1. Import Matched Template Classification module and Click Detector from NBHF_ClickClassifiers_and_MTC.psfx found in content/ToothedWhales/Pamguard settings directory
    • Click classifiers are saved within the Click Detector module in the default NBHF_ClickClassifiers_and_MTC configuration file. They are also saved as individual .pgccs files within content/ToothedWhales/Click Classifiers/NBHF directory.

    • In the Pamguard File Menu, select “Import PAMGuard Modules”

    • For Click Detector, select “Replace existing Click Detector” and for Match Template Classifier, select “Import as new module”

  2. Reclassify clicks with NBHF click classifiers
    • In the Pamguard Click Classifier menu, select “Reanalyse Clicks”

    • Select “All Data”, tick the “Delete old database entries” box, and tick the “Reclassify Clicks” box. Add a note “Reclassify based on NBHF peak frequencies”. Then click “Start”

  3. Reclassify clicks for Matched Template Classifier
    • In the Pamguard Settings menu, select “Matched Template Classifier” and “Reclassify clicks”

    • Select “All Data”, tick the “Delete old database entries” box, and tick the “Matched Template Classifier” box. Add a note “Reanalyze based on NBHF MTC”. Then click “Start”

  4. When finished, save data and close Pamguard

Automatically define potential acoustic events in R

Four thresholds are used to identify potential NBHF events, including “Pd_1” and “Pd_2” for Dall’s porpoise, “Pp” for Harbor porpoise, and “Ksp” for Kogia. Template thresholds: Pd_1=0.45, Pd_2=0.45, Pp=0.6, Ksp=0.45. These click templates were generated from acoustic encounters with known species. Thresholds were established in an iterative process to ensure no high quality acoustic events were missed.

The matchTemplateScript_NBHF.R script is used to identify potential NBHF events, add them to a database, and then generate an acoustic study using the PAMpal R package.

Events are automatically defined when the following conditions are met

  • 3 or more clicks exceed the MTC thresholds within 120 seconds
  • Events have a maximum duration of 120 seconds

Predict species classification using BANTER model

Review labels of low-scoring events