Initial Planning

Before a ship even leaves a dock, before any items are shipped, and before any operator is hired , any organization hired to conduct PAM needs to determine whether PAM is even appropriate for the task at hand, and if so, what the fundamental performance requirements of the system must be.  Often, the biggest mistakes in using PAM start at the planning stage.  By explicitly defining procedures for initial planning, the standard recognizes that no two PAM surveys can or need be exactly alike, and thus avoids the single biggest concern about standardization: that it forces a “one size fits all” mentality.

The working group will define a checklist of minimum information that needs to be researched by any organization attempting a PAM operation.  Whenever feasible the standard would identify databases or other resources that will allow this information to be obtained.  Examples of checklist items that the working group may consider include the following:

(1) Determining what time period and what region needs to be covered by the towed PAM system.

(2) Identifying what marine mammal species would be reasonably be expected to be present during this time in this region.  Flagging databases and/or literature reviews that condense this information.

(3) Obtaining best-informed estimates of source levels and frequency bandwidth of expected call types from the species identified in (2).  For many years, the book “Marine Mammals and Noise,” by Richardson et al, provided a useful reference, but more recent sources may be available.

(4) Extracting information about the acoustic propagation environment in the expected operating area.  This information may include bottom bathymetry, bottom composition, and sound speed profiles.  The standard will list resources where this information can be obtained, or describe how to make best-practice estimates of parameters when no such information is available.

(5) Establishing estimates of the expected “self” noise spectra of the towing vessel and other expected contributing sources (e.g., other vessels or sources which may be acting in concert with the PAM vessel), which may vary with time or operational state of the tow or support vessels.

This step needs to recognize a variety of potential situations facing a  PAM organization.  For example, the organization may be required to deploy off a certain platform, but in other cases  the organization can select its own platform.  An unresolved question is what to do if the characteristics of the towing and/or other vessels are unknown during the planning stage.  Would it be sufficient to use reference measurements for certain size and classes of vessels, if no direct measurements of the specific vessel in question are available?  Can a PAM organization comply with an ANSI standard if the towing vessel remains unknown until shortly before the deployment?

(6)Defining and setting signal-to-noise ratios (SNR) or other detection threshold criteria for various categories of sounds. For example, one set of SNR values might be provided for situations where a human would be reviewing a scrolling spectrogram, while another SNR value might be provided for the case where an automated detector is monitoring the system. The definition of SNR for impulsive and repetitive transient sounds may also need clarification, perhaps by exploiting related ANSI standards on noise measurement.

(7) In addition to these six items, the standard will assume that a minimum required detection range for PAM has been independently specified.  There are three reasons why the current standard would not tackle the topic of how to determine a minimum detection range.:

  1. First, setting a minimum detection range would require knowledge about what sound levels at which marine mammals would be expected to be affected (either physiologically or behaviorally) by an activity, which in turn would require knowledge or estimates about the hearing sensitivities of marine mammals, a subject where a community census may not yet be possible.
  2. Second, the steps required in determining a minimum detection range (i.e. knowledge of physiological and behavioral thresholds, environmental propagation conditions, and anthropogenic noise sources) are so involved that they deserve an entire standard to themselves.
  3. Finally, for better or worse, federal guidelines already exist for minimum detection ranges for a variety of environments, and to encourage rapid adoption of the standard by industry and regulators, it is best to use those ranges.

 

Once this research phase has been completed, the standard will then describe procedures and models (e.g. the sonar equation, standard acoustic propagation models) for determining whether the required detection range can be reasonably obtained for the species required and under the conditions expected.  If a towed PAM system cannot obtain the required detection range, given the knowledge available, then the standard will recommend that PAM should not be used in that situation.  The working group might also include criteria for determining to what degree localization capability is required of the system.  It is anticipated that this subject may take a substantial amount of the working group’s time to achieve consensus, as there are several tradeoffs involved.  For example, the ability to localize sounds to some degree would provide independent confirmation that the system is attaining its design goals; however, any localization capability will undoubtedly increase the complexity and cost of the subsequent system.

The first result of complying with the Initial Planning section of the standard would be a go/no-go decision on using PAM.  If a “go” decision is obtained, then the reward resulting from all the time spent by the standard on the planning phase will be greater flexibility and simplicity in defining requirements for hardware, software, and operations.

 

13 thoughts on “Initial Planning”

  1. (from Stanley Labak, BOEM):
    “Before a ship even leaves a dock, before any items are shipped, and before anyone is hired, a fundamental decision needs to be made whether PAM is even appropriate for the task at hand, and if so what the fundamental performance requirements of the system must be. ”

    I understand the tone that you are trying to get across here, but the fundamental decision may (and probably does) need some data provided by someone hired to plan the activity. You may want to define “hired” for what? And, realistically, those hired may do both the initial planning and overall execution so the wording is tricky.

  2. (from Stanley Labak, BOEM)
    “Establishing estimates of the expected “self” noise spectra of the towing vessel and other expected contributing sources (e.g., other vessels or sources which may be acting in concert with the PAM vessel), which may vary with time or operational state of the tow or support vessels”…
    I
    believe you are talking about the masking of the animal’s signal as received by the towed array here, but others may confuse it with masking at the animal.

  3. (from Stanley Labak, BOEM)

    “This step may need to be divided into situations where the PAM organization is required to use a certain platform, and where the organization can select its own platform. In some cases, a sound profile of the deployment platform may be required. In other cases, reference resources for certain size and classes of vessels should be used. ”

    I am a little confused on what you mean here. What I think you are saying is that there may be multiple ways, including different vessels or arrays to use for PAM and that each configuration should/needs to be checked to see what provides the best PAM coverage.

  4. (from Stanley Labak, BOEM)

    “Setting signal-to-noise ratios (SNR) [SJL9] or other detection threshold criteria for various categories of sounds. For example, one set of SNR values might be provided situations where a human would be reviewing a scrolling spectrogram, while another SNR value might be provided for the case where an automated detector is monitoring the system.”

    1. It may be a relatively simple matter to establish minimum SNRs to ensure certain and clean PAM detections, but this raises the issues of probability of detect, false alarm rates and essentially ROC curves for these “systems”. These are much more complex and difficult subjects, especially for PAM systems which in my mind are just “riding” on other systems which may (and probably do) have entirely different objectives. For example, a streamer array may be positioned vertically in the water column to optimize seismic reception, but that is probably not the best location for shallow dolphins, deep sperm whales, and animals that are in between these two extremes. This issue also extends to into many other practical considerations like array hydrophone spacing, which if it is optimized for receipt of say 200 Hz seismic signals, certainly is not optimized for a 20 kHz dolphin whistle. So what is the SNR for PAM in this case? Does the operator really have a good handle on his beam patterns, beam noise levels, etc. Can he even solve the right/left bearing ambiguity issue if he is only using a single streamer of a multiple streamer tow? If a PAM system is subsampling a larger array, how and who decided which hydrophones to use and what is the PAM “array’s” gain?

  5. (from Stanley Labak, BOEM)

    “In addition to these six items, the standard will assume that a minimum detection range has been independently specified. There are three reasons why the current standard would not tackle the topic of how to determine a minimum detection range…”

    I am not sure that this discussion is correct, or I am really missing the point. To me, the minimum detection range of a PAM system is not driven by an animal’s sensitivity, but by the source level of its transmitted signal (along with TL, the ambient noise, and the SNR needed to detect signal against background). I don’t see how a minimum detection range could be independently specified, except in the crudest/roughest way, without specific knowledge of the PAM system, the whale’s signal, and the ambient environment they are operating in. I can understand that Regulators or industry may have specified minimum ranges that represent when an airgun’s transmissions must be shut down (if an animal enters within that range), but that is not a PAM range. They just how PAM will cover out to that and more.

  6. (from Stanley Labak, BOEM)

    “Once this research phase has been completed, the standard will then describe procedures and models (e.g. the sonar equation, standard acoustic propagation models) for determining whether the required detection range can be reasonably obtained for the species required and under the conditions expected”

    Again, I don’t believe that there is a “required detection range.” And I can’t believe that anyone went to the large amount of work to specify it for multiple different species, with different transmitted calls, with different signal characteristics, in even a single estimated ambient noise environment. And they certainly didn’t attempt to enter the discussion about the regularity of animals transmissions. A minimum detection range is useless if the animal is silent. So, some other metric will need to be determined to identify and quantify the efficacy of PAM in a given situation. Embedded in this discussion is a requirement to address efficacy of only some species being detected by PAM. How are ESA and non-ESA species viewed? What about sensitive species? Etc.. You are correct that the localization issue also a huge area to investigate, implement and write about. To my knowledge (which is slight as far as PAM implementation is concerned), there are probably many techniques that are available that might be very useful to PAM operations. One that I mentioned at the workshop is the so-called (by the USN) “ping steal” technique, where if a bottom-bounce path is present with enough SNR in the BB and direct signals, then the time difference may be used to determine range. I have seen some hyperbolic fixing in simple PAM situations, and there may be many more approaches. Finally, somewhere in this standard there will need to be a discussion of tracking animals as not only an aid/enhancement to detection, but also as a cue for potential shutdown initiation. Tracking has a whole new set of issues, techniques and standards.

  7. (from Marie Roch, SDSU):
    “In some cases, a sound profile of the deployment platform may be required.”

    It seems to me that this will be highly dependent upon tow speed, frequency bands of interest, etc.

    I think that there needs to be something about the sound profile of the environment in which the measurements are taken and the recognition and that can undergo radical changes based on wind speed, tidal flow, season, etc.

    Bioacoustic masking noises such as snapping shrimp for odontocetes etc should also be considered.

  8. (from Marie Roch, SDSU):

    “Setting signal-to-noise ratios (SNR) or other detection threshold criteria for various categories of sounds”

    The measurement of SNR should be standardized, or at least guidelines should be given.

  9. (from Marie Roch, SDSU):

    “There are three reasons why the current standard would not tackle the topic of how to determine a minimum detection range. First, it requires knowledge or estimates about the hearing sensitivities of marine mammals, a subject where a community census may not yet be possible. ”

    I don’t see how this is relevant. Could you have meant for measuring detection range of the PAM platform by cetacea? I know that you mention this later on.

    1. I should rephrase as “hearing and behavioral sensitivities,” in that the required detection range of an application will be derived from exposure criteria, which in turn are being derived from data collected from hearing studies, both behavioral and evoked potential.

  10. (from Marie Roch, SDSU):

    “If a towed PAM system cannot obtain the required detection range, given the knowledge available, then the standard will recommend that PAM should not be used in that situation”

    Unless localization is part of the PAM system, many operators might want to think about maximum detection range. For example, if you pick up a baleen whale call in a surface duct that’s quite far away, mitigation steps may not be appropriate. Without some type of localization or propagation model, I don’t think that there’s much that can be done about this but it’s probably something that folks should think about.

  11. “(3) Obtaining best-informed estimates of source levels and frequency bandwidth of expected call types from the species identified in (2).”

    I feel this is going to be difficult to standardize because new studies are frequently published with new information. In addition, in my opinion, specific-specific call information is still an area in progress. Except for a few species with typical stereotype calls (e.g., right whale upcalls, “gun-shots”; humpback whale songs; minke boing, etc.), many delphinids whistles and clicks are difficult to ID to species. I guess that (2) along would provide sufficient species information for making the call in initial planning.

  12. “(6) Setting signal-to-noise ratio (SNR)”

    I feel this might be difficult to implement during the initial planning because SNR varies frequency over time in different weather and sea states, etc.

Leave a Reply

Your email address will not be published. Required fields are marked *

We listen.

scripps oceanography uc san diego