Predict the mean noise level you would measure at detections

Given a candidate true noise distribution, predict the mean of the noise levels that would be measured at detected calls. Those measurements are biased low, because a quiet period gives a high SNR and a high SNR gives a detection, so detections over-represent quiet periods.

Usage

predictSampledNL(
  mu,
  sigma,
  detFun,
  SL,
  TL,
  truncationDistance = max(TL[[1]]),
  nNodes = 121,
  ...
)

Arguments

mu

Candidate true mean noise level in dB.

sigma

Standard deviation of the true noise distribution in dB.

detFun

Detection function. See pDetGivenNL.

SL

Source level distribution, with elements named mean and sd.

TL

Transmission loss table. See pDetGivenNL.

truncationDistance

Scalar or one value per transect, in metres.

nNodes

Number of nodes used to integrate over the noise distribution.

...

Passed to pDetGivenNL, e.g. binWidth.

Value

Scalar. Always less than or equal to mu.

Details

The weighting is pDetGivenNL: if the chance of detection at 80 dB is 2.5 times the chance at 84 dB, then 80 dB periods appear among the detections 2.5 times more often than they occur.