Empirical measures of accuracy using small sensor arrays to localize animal vocalizations

Field studies using 3 and 4 sensor arrays were conducted to determine the accuracy and range of sound localizations. Under-ice and in-air trials examined the range/precision problems associated with deploying small arrays (77 m underwater, 14 m in air). The errors of the localized sound sources were determined within and outside of different array configurations in reverberant conditions. Using four receivers resulted in greater sound source localization accuracy than three. Variability of the sound source location errors within and outside of the arrays was high. The location accuracy and precision decreased when the sound source was located outside of the array. The measurement accuracy of the receiver positions was less important than the ability to determine very accurate time of arrival differences (TOAD) for each of the receiver pairs. The ability to cross correlate signals to determine TOADs appears to be the major difficulty. Also, there is a trade-off between the size of an array and the dynamic range of the call source levels that can be located. Only high amplitude calls can be localized using large (km wide) arrays. When using passive acoustic monitoring to locate fish and mammals, small hydrophone arrays will only provide accurate information when the sound source is within, or very close to the array.