Chelonia Limited

  Cetacean Monitoring Systems

 

Quality


Hardware

The key issues are accurate calibration and long term stability.

Calibration

C-PODs are rotated in a hypo-echoic temperature-controlled underground test tank to measure radial uniformity and to set each unit to the same standard sensitivity. For a full description see C-POD standardisation and calibration.

Stability

Re-testing of C-PODs is done on all units returned to Chelonia, and shows stable values over time. An early C-POD was returned after heavy impacts with the side of a ship and tested exactly as at calibration 3 years before. Fortunately the ship was also OK!

Software

The key issues are low false positive rates and the ability to view, analyse and export data rapidly. The software, CPOD.exe, is free for use with C-POD data and includes the KERNO classifier currently the most advanced train detection and classification algorithm and some location-specific secondary encounter classifiers. To download the software, see the C-POD software and manuals page.

In the screen below, from CPOD.exe, the lower panel shows the frequency distribution of the raw data over 2 days, while the upper panel shows porpoise detections in purple and dolphins in orange.

The screen below shows two years of classified detections at one site that has a strong seasonal pattern:

High resolution views allow rapid visual validation of detections. Here the sound pressure level of raw and processed clicks is shown, colour coded by their dominant frequency:

Behavioural information can be derived from the inter-click intervals within trains, shown here as the click rates from the same raw data as above:

The distribution of frequencies is also available at a single key press:

A wide range of data export options is available:

To develop and verify the performance of automated detection processes requires fast software that allows visual and numerical access to data across multiple loggers and a range of time scales from microseconds to months.

False positive rates:

Four powerful methods have been used to establish the validity of classifications:

  1. Visual monitoring during deployment. Several published papers report such studies.
  2. Visual validation of detections displayed in CPOD.exe.
  3. Assessment of the clustering of the different train quality classes allocated by the KERNO classifier.
  4. Assessment of the spatial and temporal clustering of detections within arrays of PODs in very low density areas.

In most projects the excess of true positives over false positives is so large that, following rapid visual checks on a sample of detected trains, there is no point in taking any action to remove false positives as the impact on operational statistics is very far below the level of statistical noise from other sources such as inter-annual variations, changing distributions, etc.

Extreme monitoring: False positives are very significant where very low densities of animals are being monitored. Here secondary encounter classifiers are possible. The 'Hel1' classifier was developed out of an international workshop at the Hel Marine Station in Poland in 2010, at which 60.8 years of POD data containing 5 billion clicks was evaluated. Hel1 was built to process this Baltic data to detect harbour porpoises and achieves false positive rates of far less than 1 minute per year of logging falsely identified as 'porpoise positive'.

Array assessment

CPOD.exe also provides graphical tools to examine the detections made on arrays of C-PODs and these can be run at user-controlled speeds or you can move rapidly to a specific time.