Unpredictable critters

Unpredictable Critters

Last time I wrote a post it was on the introduction of moving receivers in dBSea (link to post).
in this post I want to visit the topic of deciding what paths the moving receivers/animals should take in our scenario.

I start with a few assumptions:
  1. Animals only swim away if the noise is above a certain threshold.
  2. Animals swim directly away from the source at all times.
  3. Animals have some error in their heading (± 10 degrees), meaning that "2." above will have some random error to it.
  4. Animals have a max evasion swmming speed (you're free to complicate things by introducing a dose-response swimming speed!).
  5. Noisy vessels move in straight lines.
Constants for this post:
  1. The vessel moves at 2.5 m/s (think seismic survey vessel).
  2. Vessel source level is 220 dB RMS
  3. Tranmission loss is given by: TL = 15 x log(range).
For impact modelling in these types of surveys it's ofte assumed that the animals will swim perpendicular to the survey line from the moment they start moving.  For this example though I will assume that the animal is swimming away from the noise source (the jury is out on what's more realistic).

It turns out that the swimming speed of the animal is a very important predictor for the noise level experienced by that animal, while the range from the vessel at which the animal starts to swim away is less crucial. This is because a slow animal moving directly away from an approching vessel will get caught eventually, no matter how far away from the vessel they started swimming.

Below I'll present a few cases for different starting ranges and animal swimming speeds.

First a single case to introduce the plot type:
Figure 1. Main plot: Five animal tracks over 30 minutes (coloured lines), with the vessel going straight up (thick black line). Animals move at 1.5 m/s and start moving when vessel is 500 m away (~180 dB rms). Note that after the vessel catches up to the animals they will turn and swim in the opposite direction to maximise their distance to the vessel.
Upper left plot: RMS and SEL levels over time for the animals.
Lower left plot: The range to the vessel for the animals over time.

Even though the starting parameters of the animals are identical, I've decided that they have som error in their heading so that for each "timestep" they vary their heading randomly ± 10 degrees from the heading directly away from the vessel. With no "random" behaviour they would just keep swimming away from the vessel in a straight line until the vessel would catch up.

Below follows a small summary of 6 different starting conditions showing the variation in expected behaviour for the starting parameters (zoom or "rightclick-view image" to see full size):
Figure 2. Same as Figure 1, but with starting conditions varying for animal speed (1.5 & 2 m/s) and starting range (100, 500 & 1000 m). ("Right-click/view image" to see full size))

Another way to summarise this is to show 5 different swimming speeds in one plot:
Figure 3. All animals start to swim at range 500 m from the vessel (~180 dB), but have different swimming speeds (0.5, 1.0, 1.5, 2.0 & 2.4 m/s).
Note in upper left plot how important swimming speed is for received level.

I have added a "text-book" animal ("A1" red) in the following figure that simply swims perpendicular to the vessel's course at 1.5 m/s to make the comparison easier:
Figure 4. Same parameters as Figure 3 above, but with "A1" (red line) now swimming perpendicular to the vessel's course. This escape strategy is overall better (it has lower SEL and maxRMS than even the faster animals), but is initially worse (RMS and SEL levels increase the fastest).

This is by no means a great representation of what an animal would do in the wild, we'd need to include (at least):
  • Animal speed dependent on received level
  • Animals' intended heading (migration, foraging)
  • Flock cohesion (many of the animals are social)
But I think we should at least take steps in the right direction.
 
I made this in a spreadsheet for a small local job, and it surprised me how much worse a simple fleeing response (heading straight away from the noise) is when compared to an escape route that travels perpendicular to the route of the noise source.

(We've a few tools in the pipeline for dBSea to do this for you)

Thanks you for reading, feel free to comment below.

Comments