Task 4e
Estimating cue rates when the detector/classifier influences the cue definition
The ACCURATE project will investigate which ‘cue’ is the most stable for deep diving animals using a combination of empirical and simulated data. The choice would be driven by which cue as defined from the detection system is more stable and easier to detect with a high efficiency and low false alarm rate with a minimum of operator input. A simulation tool will be used to examine the cue passive detection system and cue characteristics that might affect detectability, starting with a deep diver species. The detection system will model the cue detectors of the field to identify the underlying relation between cue production rate and cue detection rate, other than the relationship of decreasing detectability due to distance.
We have addressed these issues via two separate threads:
- Sensitivity of density estimates on off-axis beam pattern assumptions in odontocetes
- Estimating cue rates when the detector/classifier influences the cue definition
Sensitivity of density estimates on off-axis beam pattern assumptions in odontocetes
The ACCURATE project will investigate how assumptions on sound production model affect density estimation. Identifying possible areas of uncertainty on parameters used during density estimation using PAM is important to reduce biases on animal density estimation and can guide future research to better quantify these parameters. Here we look at biases introduced by uncertainties in the source level of off-axis echolocation clicks (i.e., the beam pattern) for two deep diving species, sperm whales (Physeter macrocephalus) and beaked whales (specifically Blainville’s beaked whales, Mesoplodon densirostris), when using a simple cue counting density estimation method.
Approach
To estimate the effect of beam pattern on density estimation, a simulated cue counting approach was used, with the use of real biologging data from tagged animals and a hypothetical array of acoustic receivers. The biologging data consisted of movement data (depth and orientation) and regular click production times are recorded on acoustic and movement tags DTAGs.
Figure 1 – Beam pattern assumptions for two deep diving species
Receivers at 20 m and at 800 m depth were looked at, and different detection thresholds assuming different environment noise conditions.
Figure 2 – Histogram of transmission angle of generated clicks for sperm whales and beaked whales for all receivers in the simulation and for receivers at close distances.
Progress
Simulations and data analysis have finished. While we are not presenting any new real measurements on beam pattern of sperm whales and beaked whales, we show case the fact that assumptions about the beam pattern can affect density estimates.
Estimating cue rates when the detector/classifier influences the cue definition
Task under progress
Team members
- Douglass Gillespie
- Kalliopi Gkikopoulou
Progress
Watch this space for task updates.
