TermAbbreviationDescriptionFurther information
absorptionSound is absorbed during its passage through seawater at a rate that is higher for higher frequencies of sound. The characteristics of broadband clicks consequently changes during transmission with the higher frequencies being reduced more than the lower frequencies. This difference increases with the distance from the source.
Acoustic Doppler Current ProfilerADCPThese typically operate at1Mhz or higher but often produce lower frequencies that may affect the distribution of cetaceans and be detected by C-PODs.
amplitudeThe height of the pressure wave of a sound. Also expressed as SPL = sound pressure level. This is used here as a simple instantaneous pressure scale value, but in some systems, it refers to a value in decibels of the RMS value of the pressures of the whole wave.
amplitude profileThe sequence of heights of the waves in a click. All clicks have three amplitudes logged – the loudest and the one before and after. Clicks with full waveform data have more.
The amp profile of NBHF clicks and boat sonars is fairly flat. That of dolphin clicks and many non-cetacean sources is much further from flat.
AmpReversalCountThis is the number of times the cycle amplitude trend (whether a wave peak is higher or lower than the one before) within a click switches from rising to falling. It is assumed to be rising at the start.
It is lower in cetacean clicks than noise and lowest in NBHF clicks.
bandwidthA measurement of the frequency width of the FFT spectrum. The width at half the maximum is often used = the 3dB bandwidth. Concatenation of reflections of a click, or of refracted replicates will make the click longer and this reduces the bandwidth.
The F-POD exploits a different approach to bandwidth by logging a high precision measure of the variation in length of the waves around the loudest wave in the click. These values are least affected by ambient noise and are independent of the length of the click.
beam structureClicks are emitted from the cetaceans’ melon as a narrow beam of sound that is loudest near the center. In dolphins the mix of frequencies changes strongly across the beam, with low frequencies dominating the outer edges of the beam, and vice-versa. This can often be seen in POD data, with surface echoes revealing the frequency content of a different part of the beam from that forming the direct path to the POD. For NBHF clicks there is little change in frequency content across the beam.
Broad Band Transients BBTShorter clicks of greater bandwidth
burst pulseShort sequence of rapidly repeated clicks, generally more broad band than normal clicks
calibrationRefers to measuring sensitivity
clickClick is normally taken to imply a short episode of increased amplitude. Whistles, by contrast, are not short and are not heard as a click but as a tone that may sweep up and down in pitch.
click rateThe number of clicks per second. Also called PRF – pulse repetition frequency. It is the reciprocal of inter-click-interval.
click train
clippingThe pressure of a loud clicks may reach the maximum output level of an amplifier or ADC (analogue to digital converter) before it has reached the peak of that cycle. It is then ‘clipped’. The time taken between crossing zero and clipping can be used to extrapolate the peak value.
cropMake a new shorter file that is a copy of part of a file with time removed from one or both ends.
cycleOne wave of sound pressure going from zero up, then down below zero then back up to zero.
date formatMinuteN refers to the minute number in FPOD time – it is the number of minutes since the start of the year 1900. Divide this by 1440 to get the time used in Excel and Access, which is a floating point number representing the number of days and fraction of the last day since the same start date.
DecibelsdBDecibels are intended to represent intensity (= power) of a sound and is 10 times the log to base 10 of the ratio of rms values. It is a long out-dated unit that is rightly deprecated by the SI system of units. It is not a physically precise unit of comparison because that depends on the acoustic impedance of the media being identical, it is troublesome to learn and it contributes to errors!
The alternative is to report pressures directly as Pascals or rms pressures.
Detection Positive MinutesDPMWidely used as a measure of density of animals or habitat use
Detections and EnvironmentThis adds information to each logging period about the angle, deadband, time lost (to ‘maxing out’ = reaching the limit for the minute) and some measures of the soundscape.
Detection Positive Ten Minute periodsDP10MThese are 10minutes set by the clock, so their boundaries always correspond with the hour ends. This unit reduces the effect of variations between loggers but will saturate (reach the maximum possible) before DPM if the density of animals is high.
Detection Positive MinutesDPMThis is a good general purpose measure of the level of acoustic activity. It’s not the best for every purpose and more information is on the website.
durationdurDuration of clicks in FPOD data is often represented as the number of cycles in a click. The duration in 250 nanosecond unit is also stored and used to calculate the frequency of the click.
Effective Detection RadiusEDRThis is the radius of a circle that would contain as many animals as were detected in total i.e. the number missed inside this radius is the same as the number detected outside it. There should be a time period specified i.e. ‘the 1 minute EDR’ as it will vary with this time period.
It is not the same as the mean distance at detection, which is typically much larger.
envelopeAn imaginary line stretched over the amplitude peaks of the waves in a click.
event‘Clicks’ are detected by PODs as short periods in which successive wavelengths are similar. These are tonal ‘events’, and there is no requirement that they must be louder than the sound stream before or after, although they usually are.
Export SPL as PaTicking this box will show the amplitude of each click in Pascals by correcting for the frequency-dependent sensitively of the system and the non-linearity of the amplitude scale. However, it makes it less clear when a frequency has reached and may have exceeded the maximum response for that frequency.
F-PODChelonia suggest using the following description of an FPOD for a paper or report:
False negativeClicks made by cetaceans that were missed by the detector. Because it includes clicks from cetaceans at all distances from the detector the idea is better represented by the ‘detection function’ that describes the proportion of false negatives in relation to the distance from the detector.
False positiveClicks identified as a coming from a cetacean that did not. In the case of trains the concept is a bit different as the inclusion of a single click that did come from a cetacean means that it correctly represent the presence of a cetacean.
Fast Fourier TransformFFTThis takes in a set of amplitudes, measured at uniform intervals, and generates half as many complex numbers. One of the pair of values in each complex number represents the intensity of one frequency within the set of amplitudes. The set typically has 256, or 512, or 1024 (or other power of 2) values.
Field Programmable Gate ArrayFPGA
frequency domainThe information within the frequency spectrum. This includes no information on the order in which frequencies appear in a sound.
frequency profileThe F-POD stores the wavelengths of the successive waves in a click. These give information about the frequencies in the click and their time sequence, and can be graphically displayed as the frequency that would have that wavelength in a pure sine wave. Frequency profiles cohere in trains.
As with the amplitude profile it is flatter in clicks from NBHF sources and boat sonars than cetaceans and noise sources.
generalization performanceThe success of a classifier when applied to data that was not part of its training set.
Some papers have based advice on better POD settings on limited studies providing a training set only. This is risky, at best, predictably bad at worst.
guard bandA simple technique of detecting a narrow band signal is to compare the frequency of interest with a ‘guard band’ of higher or lower frequencies. The very first POD used 3 guard bands with used defined threshold ratios for each.
Inter-click-intervalICIThe time between the starts of successive clicks in a train. For cetacean click trains this ranges between 1 or a few seconds and less than 1ms
Inter-peak-intervalIPIThe time interval between the peaks of the waves in a click. In the F-POD this is measured in 250ns units.
KERNOChelonia’s automated classifier for data from C-PODs.
KERNO-FChelonia’s automated classifier for data from the F-POD.
specificityhow well a detector picks the real target and rejects false positives
Narrow Band High Frequency clicksNBHFLong, tonal clicks, around 125 kHz Characteristic of all porpoises and a few dolphin species and Kogia sp.
Passive acoustic monitoringPAMThe detection of cetaceans by listening for the noises they make. Active acoustic monitoring means sending out very loud sounds and listening for the weak echoes from animals, and is used in fish finders and depth sounders. PODs, however, are PAMs and are silent when logging.
Programable Interface ControllerPICa type of micro-controller
sensitivityhigh sensitivity = able to detect very weak sounds. Sensitivity can often be increased at the cost of reduced specificity
standardisationsetting the sensitivity to a common standard
Static Acoustic MonitoringSAMpassive acoustic monitoirng by a static logger of any type
toothed whalesOdontocetes - the toothed whales - all the cetaceans except the Baleen Whales that do not have teeth but filter small prey
train detectionToothed cetaceans make long series of similar clicks - called trains - made with similar time gaps between successive clicks. The gaps usually show continuous variation - getting shorter or longer, and dolphins also vary the sound characteristic of the clicks. Trains are also produced by boat sonars and can arise by chance from random sources of clicks. The software filters that extract cetacean trains are essential to the POD's ability to accurately identify the presence of the animals and work by identifying coherent trains - those in which the variation in timing and character between successive clicks is lower than might occurs by chance among the many background noise clicks that are logged.
uniform thresholdall types of logger tend to be more senstive in quiet conditions, and this can bias detection rates. Fixing a f=relatively high detection threshold is a way of avoiding this variability.
Weak Unknown Train SourcesWUTSSound interference. Weak unknown train sources exist in some water bodies. Their identity is not known but small crustaceans that may colonise the transducer housing surface are on the list of suspects. IN some places large numbers occur and dominate teh acoustic record but their identity is not yet known.