Signal-to-noise Ratio (SNR)

The echo level (the signal) must be higher than the noise level to be detected and to obtain interpretable data.  This is referred to as the signal-to-noise-ratio (SNR).  Although noise levels do not change with time (≡ depth in standard applications), the signal decreases with depth due to spreading and absorption.  Therefore, the signal to noise ratio decreases with depth, and at some depth, the signal will be too weak to be reliably separated from noise.  This depth defines a detection range for the target size in question.  How small the SNR can be depends on the consistency of the noise.  As a rule of thumb, a SNR of 10 dB, (a factor 10), is acceptable.  At lower SNR, the chances increase that spurious noise spikes are included in the Sv estimates (Simmonds and MacLennan 2005).  Low SNR also causes problems for in situ TS measurements.  Noise often change over time (≡ from ping to ping) and there are methods being developed to dynamically adjust noise levels when analyzing data (DeRobertis and Higginbottom 2007).  Note that if noise levels vary during the survey, so do the detection limits.

In practice, the noise levels relative to the signal can be measured by observing the increase in noise levels with depth as it is amplified through the time-varied gain.  We recommend collecting passive data during the conditions of the survey to measure noise level. It is important to know the depth limitations for detecting targets of interest to avoid biased results for fish abundance and distributions.