western gray whale advisory panel sstf-4 4-d seismic ... · pdf file4th report of the sstf 1...

4th report of the SSTF

1

WESTERN GRAY WHALE ADVISORY PANEL

4-D SEISMIC SURVEY TASK FORCE

SSTF-4

REPORT OF THE 4-D SEISMIC SURVEY TASK FORCE

AT ITS 4TH MEETING

6 DECEMBER – 8 DECEMBER 2009

GENEVA, SWITZERLAND

CONVENED BY INTERNATIONAL UNION FOR CONSERVATION OF NATURE


2

Workshop to resolve outstanding issues in preparation for Astokh 4-D seismic survey scheduled for June-July 2010

Contents 1. INTRODUCTORY ITEMS ..................................................................................................................................................................... 3

1.1. Introductions of participants ............................................................................................................................................................ 3 1.2. Summary of progress since WGWAP-6 (April 2009) ..................................................................................................................... 3 1.3. Identification of outstanding issues to be resolved .......................................................................................................................... 3 1.4. Aims of workshop and expected product ........................................................................................................................................ 3

2. DOCUMENTS AND PROGRESS REPORTS ....................................................................................................................................... 4 2.1. Review of available documents ....................................................................................................................................................... 4

2.1.1. Existing documents made available ......................................................................................................................................... 4 2.1.2. New documents prepared for the workshop ............................................................................................................................. 4

2.2. Results of tasks requested at WGWAP 6 ......................................................................................................................................... 4 2.3. Other new information ..................................................................................................................................................................... 4

3. OPERATIONAL PLANS AND EXPECTATIONS .............................................................................................................................. 4 3.1. Changes in Sakhalin Energy plans for 2010 .................................................................................................................................... 4 3.2. Information on other relevant industrial or research activities (e.g. other seismic surveys, satellite tagging effort) .................... 5

4. PERIMETER MONITORING LINE (PML) .......................................................................................................................................... 6 4.1. Summary of status from Small Group Meeting ............................................................................................................................... 6 4.2. Final definition of the PML .............................................................................................................................................................. 9 4.3. Incorporation into operational plans including consideration of truncated or ‘partial’ A lines (Sakhalin Energy) ....................... 9

5. SHORE-BASED DISTRIBUTION/DENSITY AND BEHAVIOUR MONITORING ...................................................................... 10 5.1. Station locations, survey schedules, protocols and logistics ......................................................................................................... 11 5.2. Data to be recorded and protocols for near real-time processing and analyses ............................................................................ 11 5.3. Shore-based platforms: construction and contingency plan .......................................................................................................... 11

6. ACOUSTIC MONITORING ................................................................................................................................................................ 12 6.1. Equipment issues outstanding ........................................................................................................................................................ 12 6.2. Spatial and temporal deployment of buoys (sampling locations and time periods) ..................................................................... 12 6.3. Collection, transmission, storage and processing of data .............................................................................................................. 13 6.4. Source verification survey .............................................................................................................................................................. 13 6.5. Propagation modelling to be conducted before and during the survey ......................................................................................... 13

7. AT-SEA OPERATIONS ....................................................................................................................................................................... 13 7.1. Vessels – descriptions and responsibilities .................................................................................................................................... 13 7.2. Procedures for seismic source verification/calibration (SSV) ....................................................................................................... 14 7.3. Personnel and equipment (including e.g. ‘big eye’ binoculars, range finders) ............................................................................. 15 7.4. Use of inflatable for focal follows ................................................................................................................................................. 15 7.5. Protocols for behavioural follows .................................................................................................................................................. 15 7.6. Formalized operational decision tree ............................................................................................................................................. 15 7.7. Servicing DT-AUARs .................................................................................................................................................................... 16 7.8. Definition of ‘single-point authority’ for shut-downs, including communication links between all observation platforms that can affect a shutdown ............................................................................................................................................................................ 16

8. IMPLICATIONS OF OTHER INDUSTRIAL ACTIVITIES OVERLAPPING IN TIME AND SPACE WITH ASTOKH SURVEY .................................................................................................................................................................................................... 18

8.1. Lebedenskoye – timing and considerations/implications for Sakhalin Energy survey including acoustic monitoring considerations ........................................................................................................................................................................................ 18 8.2. Analytical framework for assessing impacts of Sakhalin Energy vs. Lebedenskoye survey, i.e. can we construct a framework that will allow us to assess the effects of one or both? ......................................................................................................................... 18

9. NEXT STEPS ........................................................................................................................................................................................ 19 9.1. Specification of tasks to be completed after meeting, with timeline and responsibilities ............................................................ 19

9.1.1. Behaviour data – focal follow ‘at sea’ protocol ..................................................................................................................... 19 9.1.2. Cooke analysis on PML .......................................................................................................................................................... 19

9.2. Arrangements for monitoring progress with tasks......................................................................................................................... 19 9.3. Need for additional SSTF meeting in early 2010, scheduling etc. ................................................................................................ 19 9.4. Follow-up analyses (forum for follow-up, data sharing, data management plan) ........................................................................ 20

REFERENCES........................................................................................................................................................................................... 21


3

The 4th meeting of the Seismic Survey Task Force was held at the Hotel Royal and Hotel Epsom, Geneva, Switzerland from 6 to 8 December 2009 under the chairmanship of Reeves.

1. INTRODUCTORY ITEMS

Reeves welcomed participants (Annex A) and noted that he was standing in for Greg Donovan who had chaired most previous meetings of the Seismic Survey Task Force (SSTF) but was unable to be present in Geneva on this occasion. After some discussion and a few additions to the draft, a final agenda was adopted (Annex B).

1.1. Introductions of participants

Participants introduced themselves. Sarah Humphrey served as meeting rapporteur and Finn Larsen and Laura Riddering of IUCN provided various kinds of support to the Task Force. Reeves thanked them for their help.

1.2. Summary of progress since WGWAP-6 (April 2009)

With assistance from other participants, Reeves summarised progress on SSTF work since the last WGWAP meeting in April 2009. Most of that work was based on a work plan developed and circulated soon after WGWAP-6, with the following subjects highlighted: (a) vessel numbers, types and strategies (discussion of various ‘scenarios’), (b) shore-based observation platforms, (c) estimation of the area containing 95% of the whale density (perimeter monitoring line, PML) and (d) estimation of the 156 dBSEL (agreed proxy for 163RMS dB) contour.

With regard to (a), a series of teleconferences organised and led by Nowacek had taken place during October-November 2009. This subject is addressed under item 7.

With regard to (b), Broker had provided specifications and, after receiving comments from other SSTF members, proceeded with construction of the two required platform towers in August. These towers were inspected by Larsen and three Panel members (Tsidulko, VanBlaricom, Dicks) during their site visits to Sakhalin (for other purposes) in September 2009. This subject is discussed under item 5.5.

With regard to (c), considerable work had taken place, as summarised under item 4.1.

With regard to (d), again some progress had been made, as summarised by Racca under item 6.

1.3. Identification of outstanding issues to be resolved

The outstanding issues are reflected in the agenda for this meeting (Annex B).

The group was advised by Sakhalin Energy that contracting for the seismic survey is moving ahead and a decision on the PML needs to be made imminently. This in turn will have a bearing on the numbers of A and B lines anticipated. Therefore, reaching agreement on the PML at this meeting was considered of utmost importance.

Operational details for monitoring and mitigation efforts also need to be resolved and agreed as soon as possible. There is some discretion in implementation – for example according to weather conditions – but principles and key parameters need to be agreed. Sakhalin Energy noted that monitoring and operational aspects need to be integrated and that much work remains in terms of operations and mobilising teams between now and mid-June. Developing operational plans for the observation vessel was identified as another critical issue. The other vessels on the water during the survey will be largely under the control of the seismic contractor, but the observation vessel will be responsible for most of the monitoring and mitigation implementation when onshore fog or other conditions preclude monitoring from shore.

1.4. Aims of workshop and expected product

A number of discussions since WGWAP-6, including the small group meeting on the PML in Gland in October (see item 4.1) and the ‘scenarios’ teleconferences in October and November (see item 7), represent


4

background work for this 4th formal meeting of the SSTF. The reports of the three previous SSTF meetings are posted on the IUCN website (http://www.iucn.org/wgwap/wgwap/task_forces/4_d_seismic_task_force/).

Although a draft partial report of the Gland small group meeting (without conclusions) had been circulated to the TF by Donovan in November, it had not been possible at that meeting to resolve all outstanding issues. Therefore it was agreed that instead of completing a separate report of the Gland meeting, its main elements should be integrated or subsumed within this report. Among other critical issues needing resolution at this meeting were a) an approach to determine A and B lines and b) procedures for source verification (during which no A lines should be shot), including where and to whom the data should be transmitted for evaluation and sign-off.

2. DOCUMENTS AND PROGRESS REPORTS

2.1. Review of available documents

The documents available for the meeting are listed in Annex C.

2.1.1. Existing documents made available Vancouver SSTF report – circulated by email, available on the public website, and hard copies were

made available for consultation at the meeting.

Paxton report – version 5 had been circulated by e-mail but Muir noted that a version 6 incorporating analyses undertaken after the Gland meeting was available and had been circulated by Donovan to participants in that meeting but not to the entire Task Force. Arrangements were made for Larsen to circulate to the TF at the beginning of this meeting.

Muir and Joy’s analysis – recirculated at this meeting, having previously been circulated by Donovan. This report shows the 95% maximum density line. A second figure with a line showing 95% of the population was circulated by Broker.

WGWAG-7/19 on ice conditions was circulated and was also on the portal.

WGWAP-7/21 on the construction the shore towers was too large to circulate but copies were made available on paper or as e-files at this meeting (and on the portal).

All previous meeting documents were available in electronic version at this meeting.

2.1.2. New documents prepared for the workshop New documents prepared for the workshop or for the work leading up to it included the draft partial report of the Gland small group meeting referenced above, to be subsumed into this report. The ‘teleconference’ group organized by Nowacek produced several working documents (e.g. the ‘scenarios’ document), but none of these is explicitly included in this report. Their content is instead incorporated here under agenda items 6 and 7.

2.2. Results of tasks requested at WGWAP 6

Results of work completed since WGWAP 6 are summarized under Agenda item 1.2.

2.3. Other new information

The new information prepared for this workshop is described in the relevant sections below and includes material covered in several slide presentations (e.g. ice cover, PML derivations).

3. OPERATIONAL PLANS AND EXPECTATIONS

3.1. Changes in Sakhalin Energy plans for 2010

There are no major changes to the Sakhalin Energy plans for the 4D seismic survey in 2010. Two key elements that have been pursued by Sakhalin Energy since WGWAP 6 are


5

A seismic source array experiment for acoustic model calibration and optimisation was carried out, successfully measuring noise levels in the area of the Astokh 4D seismic survey. This is discussed further under agenda item 6.4.

Analyses were conducted of statistics and predictions of ice-free dates in the context of Sakhalin Energy’s plans to start acquisition as early as possible once the sea in the survey area can be declared ice-free. This is intended to minimize both downtime and potential impacts on whales in the feeding areas.

The SSTF has repeatedly emphasized that the most important mitigation measure for the Astokh 4D seismic survey is to conduct it as early in the season as possible, when fewer gray whales are likely to present in the Piltun feeding area, particularly mother-calf pairs. The definition of “as early in the season as possible” is directly tied to the timing of ice-free conditions in the survey area.

Sakhalin Energy confirmed its intention is to use all available information to ensure that the seismic survey starts as early possible. This differs from the previous year when the mobilisation date was fixed as 15 June. Using the ice data, Sakhalin Energy will now be in a position to take advantage of the earliest feasible start window, which should minimize impacts on the whales as well as reduce overall survey duration.

WGWAP-7/19 provided an analysis of ice data designed to predict when ice-free conditions in the Astokh survey area can be expected to occur in 2010. The overarching objective is to provide the earliest possible date for mobilization of the seismic vessel to the survey area. To this end, satellite images of sea ice coverage and concentration in the northern Okhotsk Sea during 1982-1999 and 2004-2009 were used for the analysis. More specifically, satellite images collected between January and March of each year were used to measure maximum sea ice coverage. Based on these measurements, winters were categorized as Severe, Normal and Mild.

In general, the severity of a given winter, as determined by satellite images from the February to March period, were useful in determining the probability of the Astokh survey area being ice-free by a certain date. Mild winters result in the earliest ice-free dates while severe winters delay clearing of the ice. For example, mild winters in 2008 and 2009 resulted in the Astokh survey area being ice-free on 8 June and 9 June, respectively.

Sakhalin Energy will use the approach outlined to the SSTF to determine the mobilization date. In brief, this approach will entail: (1) review of satellite images from February and March 2010 to categorize the winter as Severe, Normal or Mild, (2) statistical analysis of historical data to estimate an ice-free date with an associated probabilistic uncertainty range, and (3) continued qualitative monitoring of satellite and temperature data from the northern Okhotsk Sea and off Sakhalin between April and mid-May to track the ice break-up and, to the extent possible, narrow the uncertainty range around the predicted ice-free date. No later than mid-May, Sakhalin Energy will notify the PGS seismic vessel of a definite day on which it is required to be on-site and ready for data acquisition (i.e. the start-up date), taking into account the 21-day mobilization time needed by the vessel to reach Sakhalin from Singapore so that the survey can begin as soon as the area is ice-free.

The Task Force welcomed the above approach and acknowledged that such use of ice data had potential value in helping to ensure that the 2010 Astokh 4D seismic survey be initiated at the earliest possible date.

3.2. Information on other relevant industrial or research activities (e.g. other seismic surveys, satellite tagging effort)

Other industrial or research activities in 2010: Satellite tagging of gray whales – timing: mid-August at the earliest

o This work will occur later in the year than the seismic survey, so should not impinge. o At present, this is an informational matter on the Panel’s agenda since the Panel recommended

that tagging take place and agreed to delegate detailed planning to an IWC Scientific Committee/WGWAP coordination group.

o Funding is in place and arrangements are ongoing, with Tsidulko playing a lead role. Lebendinskoye seismic survey (Rosneft) – May/June?


6

o This survey is planned for as early in the season as possible based on similar advice to that received by Sakhalin Energy – see agenda item 8.

Sakhalin Energy offshore pipeline inspection – end June to September o By survey vessel and ROV.

Sakhalin Energy notional additional rock dumping on pipelines – end June to September o Only if required for stabilization (due to winter scouring on sections of unburied pipeline found

further offshore). Rocks are from a land-based source and delivered by barge. o Ecological survey (benthic and water column) - end June to mid-July and from October to

November to be conducted by Pavel Gordienko.

Others items noted: Pile driving at Exxon site is finished (confirmed by Broker). Other information obtained by Tsidulko at ‘Sakhalin Oil and Gas Conference’

o Exxon Arkutun-Dagi seismic survey prior to production startup in 2014 – not confirmed and unlikely in 2010. The Exxon drilling/production from Arkutun-Dagi will involve a platform close to the offshore feeding area.

IUCN (Larsen) had received information from Knizhnikov (head of oil and gas programme at WWF-Russia) about other activities, some of which may be seismic surveys but no details were provided.

IUCN (Larsen) has subscribed to the IHS industry scouting data provider.

4. PERIMETER MONITORING LINE (PML)

4.1. Summary of status from Small Group Meeting

Background Introduction by Muir At the 3rd SSTF Workshop (Vancouver, January 2009) it was noted that the density analyses to date ran the risk of underestimating density offshore and outside the visual range of the shore stations and dedicated survey vessels. It was agreed that analyses should be performed to determine whether the opportunistic data gathered by MMOs could be used to supplement the shore station and dedicated survey vessel sightings data.

On the recommendation of the Vancouver workshop, Charles Paxton of St Andrews University was contracted by IUCN to estimate a density surface using opportunistic MMO data. The contract was issued in March, a preliminary report was submitted to IUCN in June and this was circulated to the Task Force in July 2009. Following circulation of several intermediate versions, a final version of the analysis was provided to the TF as document SSTF4/6.

Paxton’s analysis used line transect methodology to estimate a detection function which was combined with a GAM-based spatial modelling approach to estimate a relative density surface, using the likelihood-based ‘count’ method developed at St Andrews (Hedley and Buckland 2004).

In order to apply line transect methods, it was considered necessary to exclude time periods when either a) the vessel was moving at less than 5 km/h, b) it was before sunrise or after sunset or c) Beaufort Sea state was >4. Track length was used as the measure of effort, with the tracks divided into segments of maximally 1.2 hr or 9 km, with an average track length of approximately 3.2 km. The resultant total effort was approximately 16 000 km.

A total of 2760 sightings were initially available, but approximately half of the sightings were excluded after application of the restrictions on effort and the detection function right truncation distance. These data were further reduced to 453 sightings by eliminating all 2007 data and 2005-2006 shoreward sightings made from near-shore, north-south transects.

The density was estimated at the midpoint of each effort segment, with covariates also estimated at the segment midpoints. This data set was then used to create the relative abundance density surface. The candidate PML was calculated as the density contour that contained 95% of the estimated maximum density (an earlier version of the analysis calculated the line differently). Uncertainty was estimated by bootstrapping with day as the bootstrapping unit.


7

The work plan for the SSTF drawn up at WGWAP-6 provided for a workshop on the PML to be held in October 2009 at which a small group of specialists would review the Paxton analysis and make recommendations for the PML.

The PML workshop reviewed a version of the Paxton analysis that had been revised following extensive written comments from Muir and Joy. Apart from some issues of detail, which were largely addressed in the final version, the main problem with the analysis was that in using only opportunistic sightings, coverage was insufficient in the near-shore area where whale densities are known to be highest. As a result, the analysis did not estimate near-shore densities reliably and thus the estimated density surfaces failed to reflect the high near-shore densities adequately. The method tended to place the candidate PML too far offshore.

The PML workshop concluded that an analysis that combined both the opportunistic and systematic (vessel and shore station) data was required, in order to ensure sufficient coverage of both near-shore and offshore areas. The PML workshop received two proposals for such an analysis and recommended that both be pursued:

1. Hybrid approach (Joy and Muir)

Based on geostatistical methods and kriging, using average whale densities estimated from Sakhalin Energy/ENL joint research program systematic shore-based and vessel-based surveys, and the sightings and effort segments defined by Paxton (described in document SSTF-4/4).

2. Integrated likelihood approach (Cooke)

A predictive model that uses both the opportunistic and systematic data sets.

Hybrid Approach (Joy and Muir) The aim of this approach was to develop a simple method that uses the available opportunistic MMO data to ‘fill in’ the ‘zero effort’ gaps in the June-July 2005 to 2007 density maps obtained from systematic data (from shore-based distribution and behaviour teams, the systematic vessel survey data and (prior to 2005) the systematic aerial survey data). The ‘segmented’ effort data from the Paxton analysis was used to allocate ‘effort’ (expressed as times visited) to a particular 1 km × 1 km square, taking into account the approximate search width of 4.5 km. This was then related to the position of sightings by 1 km × 1 km square (i.e. those sightings that were made in conjunction with the segmented effort) to obtain an estimate of sightings per unit effort per 1 km × 1 km square, calibrated with the densities from systematic data using the area of overlap. There was a strong correlation between the two surfaces (systematic vs. opportunistic) in the area of overlap.

The combined density estimates by grid cell were extrapolated to cells without effort using simple kriging or its alternatives. The analysis and results using this approach were presented in document SSTF-4/4 and are included as Annex F to this report. Co-kriging with depth as a covariate was tried in addition to simple kriging but failed to provide much improvement, probably because of the non-linearity of the depth-density relationship (co-kriging assumes a linear relationship). Joy and Muir recommended the predicted relative density surface based on the simple kriging.

The PML had been based on densities in the Piltun feeding area in June-July period. The Gland meeting discussed whether to include the offshore feeding area. Both approaches were used with little difference in results: using the offshore feeding area brings the line out by some 275 m. Including the offshore feeding area data does not seem to best serve the whales because moving the PML eastward based on these data does not provide additional protection to whales in the near-shore feeding area, nor does the inclusion of these data provide further protection to animals in the offshore area.

The new Muir/Joy PML is located farther eastwards along most of its length when compared to earlier Q1 2009 evaluations. It thus appears that the recommendation to include opportunistic sightings data was a useful one.


8

Figure 1. The Perimeter Monitoring Line as calculated by the 'hybrid approach' and explained in Annex F to this report. This PML is the one the Task Force recommends to be used for the 2010 Astokh seismic survey.


9

Integrated Likelihood Approach (Cooke) The integrated approach used all three datasets: from shore-based distribution surveys, dedicated vessel surveys, and opportunistic vessel surveys. The motivation was to obtain greater coverage than any one of the datasets alone would provide. A density function and a detection function were estimated in a combined fitting approach to maximise the likelihood as specified in Annex E. The detection function was taken to be a function of radial distance, so that both moving and stationary platforms could be accommodated.

Cooke has filtered data based on low visibility and other factors. He has not filtered out data for speed below five knots so has a much larger dataset than that used by Paxton. He has accounted for different sighting rates according to vessel speed.

Results from this approach were not available in time for this meeting.

4.2. Final definition of the PML

In discussion of the hybrid method, the SSTF noted that kriging is a well-established method for application to data consisting of measurements taken at discrete points. However, the data points used in the present application derive from sightings per unit effort where the sightings are point counts and the effort is spread out across an area, but these data points are treated for the purpose of kriging as if they were measurements taken at discrete points. The statistical properties of the original data are not taken into account. In theory, the integrated likelihood approach would overcome this limitation as well as the need for linearity with covariates such as depth.

The TF would have preferred to have results from both approaches to compare, but in view of the time constraints, the TF agreed that the results of the hybrid approach should be used to designate the PML. The TF considered that the hybrid approach is superior to the approach used to determine the PML on previous occasions, thanks mainly to the improved coverage offshore (i.e. from the opportunistic MMO observations).

The TF agreed that the integrated analysis should nevertheless be completed by as soon as possible. Once Cooke’s work is completed, Joy and Muir will review and quality-control the work before any firm conclusion is made. If the integrated analysis shows that the PML can be moved closer to shore, this would be beneficial in that it should enable the survey to be completed earlier in the season when fewer whales are present.

The TF recommends that the PML as computed by the method described in Annex F and shown in Figure 1 be used for the Sakhalin Energy seismic survey in 2010.

4.3. Incorporation into operational plans including consideration of truncated or ‘partial’ A lines (Sakhalin Energy)

Broker introduced a conceptual framework for refining the area defined to constitute the A-line zone and presented Figure 2 for consideration by the TF. As illustrated, survey activity along portions of some lines designated as “A” lines would, as the survey vessel moves to the southeast, no longer result in noise predicted to exceed the 156 dBSEL (~163 dBRMS) received level behavioural criterion inside the PML (shaded yellow on Figure 2). Broker proposed that segments of lines shot in these portions of the “A-line area” (shaded gray) actually be treated as “B” lines and thus be subject to the corresponding monitoring and mitigation requirements (i.e. not requiring shutdown for whales inside the PML).


10

Figure 2. Schematic showing the proposed 'truncation' of survey lines designated as 'A' lines. This scheme effectively

allows for reclassification of portions of 'A' and 'B' lines as one or the other depending on the modelled or, more importantly, measured received levels at the PML.

Overall, the TF acknowledged that this conceptual approach was likely to have the benefit of shortening the survey. There was fairly extensive discussion and examination of this proposal, and the underlying predictions and propagation modelling required to determine with some level of confidence the portion of the ‘A’ zone in which lines would be classified as B-lines. While the pre-survey modelling will make the initial determinations for the reclassification or ‘truncation’ of A-lines, the TF agreed that such reclassification or truncation cannot be developed purely on the basis of the model. It was agreed that a 2-pronged strategy would provide a precautionary way of implementing Broker’s proposed approach: i) establishment of safety buffers around the modelled transition points (i.e., the point along the survey line where it changes from ‘A’ to ‘B’ or vice-versa) and ii) experimental verification of the transition points and buffers during the survey. (NB: This verification should happen de facto as the PML will be monitored in real time.) The quantitative bounds of the buffers described in (i) were not agreed, but 3 dB was put forward as an initial figure. It was agreed that Sakhalin Energy would provide the TF with an operational plan by 28 February 2010.

The task force agreed with this more sophisticated definition of A- and B-lines, given the expected benefit of executing the survey more rapidly, but noted the importance of making these determinations with sufficiently conservative margins of error to preclude needing to make revisions in the field. As such, the Task Force recommends that Sakhalin Energy implement this revised definition of A and B lines with agreed precautions.

5. SHORE-BASED DISTRIBUTION/DENSITY AND BEHAVIOUR MONITORING

At the beginning of the discussion on this subject, information was presented on weather conditions in June and July, with a particular focus on the frequent occurrence of foggy days. Fog, and sometimes wind and rain, precluded observations from the southern shore stations 60-70% of the time during the 2007 and 2009 seasons (there was no shore-based effort for this time period in 2008). With the boat-based observational protocols in place for the Astokh survey, poor weather conditions near shore will not necessarily be as problematic as they would be if monitoring were entirely dependent on the shore stations. However, observers on the seismic survey vessel itself must be able to monitor the safety radius, as described and defined in reports from the second and third meetings of the SSTF.


11

5.1. Station locations, survey schedules, protocols and logistics

Station locations had already been determined and agreed at previous SSTF meetings and therefore they were not discussed further at this meeting.

The discussions of behaviour protocols had not been concluded (e.g. watch intervals, data to be collected, focal follows, possible use of 0-1 sampling). Issues that still needed to be addressed included what data would be collected according to which protocols, e.g. behavioural state/pattern (travelling, feeding) and respiration rate can be measured with either continuous or discrete methods. Also, there was still a need to consider the current protocols and determine what, if anything, should be changed. Some protocols (e.g. an ethogram) were submitted during the meeting and the TF agreed that refining these should be a fairly minor issue. Southall provided the protocols from the behavioural response studies (BRS), though he cautioned that these had been developed to measure the response of beaked whales to sonar-type signals (i.e. continuous sounds as opposed to the impulsive sounds in the seismic survey case). A small group consisting of Broker, Southall, Nowacek, Gailey and Weller met between sessions to discuss the shore-based behaviour monitoring protocols and concluded that the current protocols are well designed. The small group also noted the value of maintaining consistency in procedures to the greatest extent possible to ensure data comparability between years. The TF agreed that there should be no changes at this point to the shore-based behaviour monitoring protocols.

The TF also agreed that it will be important to collect data after the seismic survey ends. Bell indicated that there needs to be a break between any post-survey monitoring and the beginning of the ENL/Sakhalin Energy joint programme monitoring, likely on the first of August. Although it is expected that the seismic survey will be completed before mid-July, if there is any delay it could mean that there is little time for both post-survey monitoring and a break. Bell noted that the joint programme involves other actors so adjusting the start date would be complicated, though not impossible. The shore and boat-based elements could be separated. Broker suggested that various potential scenarios for different survey periods should be developed and considered in advance. It was agreed that the logistics of how to accomplish the post-survey monitoring would be left to Sakhalin Energy to work out, but the SSTF recommends that visual data continue to be collected from the behaviour and distribution stations during all workable weather conditions for a period of 14 calendar days following the completion of the seismic survey. The recommended 14-day duration is based on results from the 2001 ENL seismic survey, when shifts in whale distribution continued for up to ten days following completion (Gailey et al. 2007).

5.2. Data to be recorded and protocols for near real-time processing and analyses

There was a brief discussion on whether any modifications were needed to the existing protocols for shore-based distribution/density and behaviour monitoring. The TF noted the importance of consistency so that data can be compared between years, and agreed that the current data being obtained and the protocols currently being used are well designed and adequate and that no major alterations are needed (see Vladimirov et al. 2008; Gailey et al. 2007).

The task force recommends that a small group compile the agreed-upon and established shore-based distribution/density and behaviour monitoring protocols into a single, relatively simple, publicly available document explaining the field methods and the advantages and limitations of the resulting data.

5.3. Shore-based platforms: construction and contingency plan

Sakhalin Energy has constructed platforms from which some of the shore-based teams will conduct their observations during the seismic survey. The platforms are located close to the shoreline and therefore some damage or shifting may occur due to winter storms. Sakhalin Energy gave assurances that the integrity of the platforms would be checked and any necessary repairs made after the winter storm season. Concern was expressed regarding the stability of the platforms because theodolites require stable (i.e. low-vibration) platforms. Weller noted that even small amounts of vibration or movement may compromise theodolite operations. The SSTF stresses the importance of adequate visual observation platforms and recommends that the structural integrity of the platforms be tested under realistic field conditions and that the tests be conducted with sufficient lead time to allow for any necessary corrective measures to be implemented. The term ‘adequate’ is meant to include: 1) the ability of observers to see the required areas to be monitored, i.e.


12

out to the PML, and 2) confirmation that only infrequent rebalancing of theodolites (i.e. not more than once every hour) will be needed under normal operating conditions. If the testing reveals that the platforms are inadequate, it is left to Sakhalin Energy to assemble contingency plans, with the understanding that monitoring from these platforms is an integral part of the mitigation program (e.g. to watch for whales shoreward of the PML during A-line acquisition).

6. ACOUSTIC MONITORING

6.1. Equipment issues outstanding

Racca reported on his recent visit to POI in Vladivostok to review the quality status of the digital telemetric acoustic buoys (DT-AUARs) to be used in 2010 and to discuss with the POI team any issues of reliability that had arisen during the 2009 season. Upon ENL’s request and with Sakhalin Energy’s concurrence, POI had made available to ENL for the 2009 season two of the ten units developed for the upcoming Astokh seismic survey monitoring so they could be used to measure water-propagated signals from an industrial operation. This provided a valuable test of the autonomy and reliability of the units in a fully operational setting, albeit at shorter radio telemetry ranges (< 4 km) than would be encountered monitoring the PML from shore. To address the latter consideration, an additional DT-AUAR was deployed by POI at a radio range of 12 km exclusively as a test unit for Sakhalin Energy. The operation of the three units during the 2009 season was satisfactory, with no design or assembly problems arising from the trials other than two specific instances of problems associated with human error or oversight in assembly of the devices. These issues were discussed in detail during Racca’s visit, their causes were analysed, and steps to address them were formulated. Based on Racca’s evaluation and the evidence from the field trials, POI has a mature monitoring station technology in place from the standpoints of resilience, autonomy and range, and only potential human errors now need to be addressed.

The tested radio range of 12 km was adequate in terms of PML telemetry because the radio receiving and acoustic data processing site will be relocated from the Piltun lighthouse to a camp much further to the south, almost exactly halfway along the length of the monitoring line. The DT-AUARs are capable of telemetry ranges well beyond the test distance, but their transmission power output can be reduced to provide extended battery life for shorter radio links. The aim is to minimize the potential need for transmitter battery servicing and ideally to run the full survey over a single autonomy span of the stations. The proposed new location of the shore receiving station will be on the ocean shoreline near the southern tip of Piltun Lagoon (the exact position remains to be defined), a site that is conveniently served by an access road to the interior, adjacent to one of the new behaviour observation towers, and far from radio interference from any industrial operation. The shore station infrastructure will consist of an 18 m military-type telescoping mast supporting the directional dipole receiving antennas, a rack of radio receivers and decoders, a data acquisition subsystem and various processing computers connected by local network. It is envisaged that all components save for the antennas will be housed in a container-type prefabricated laboratory space. Neither the shore station nor the transmitter buoys should raise any objection from either government or military officials as all required permits are in place.

Racca indicated that the configuration of the telemetric monitoring system is fully defined and, aside from the revised location of the PML and the repositioning of the land station farther to the south, does not deviate from any plans previously presented. The system is completely self-sufficient on the basis of the POI stations alone and will provide all the data required to implement the agreed acoustic monitoring plan and start-of-survey sound source verification measurements. Any additional contribution of underwater acoustic recording and/or telemetric systems by parties such as IFAW (using Vedenev’s technology) would provide a valuable addition to the documentation effort but will not be required to implement the plan.

6.2. Spatial and temporal deployment of buoys (sampling locations and time periods)

Deployment of both transmitting and recording buoys has been discussed extensively in previous TF meetings and documented in those reports (specifically the 2nd meeting of the Task Force, Section 7.5.1 of that report). The basic strategy has not changed with the Sakhalin Energy T-AUARs to be deployed at 2500 m intervals along the PML and then ≥ 3 record-only stations along the 10 m isobath within the feeding area.


13

That all of the telemetric systems are operating correctly must be verified prior to the start of the seismic survey.

Discussions at previous Task Force meetings also included deployment of acoustic recording stations by independent observers, specifically IFAW-supported monitoring using systems from Vedenev’s institute. However, at this meeting Vedenev indicated that funding from IFAW was doubtful. Practically, the monitoring and mitigation plan for the Astokh survey has always been premised to rely only on Sakhalin Energy assets, so the non-availability of equipment from IFAW/Vedenev would not affect the operational plans for the Astokh survey.

6.3. Collection, transmission, storage and processing of data

Sakhalin Energy indicated that its acoustic data collection, archiving and processing plan had been fully defined and standardized. The TF recognized the requirement that raw acoustic data remain within Russia, but noted that it will require, through agreed procedures and methodology, access to sufficient information (including acoustic, distribution, and behaviour data) to conduct a thorough post-survey analysis and thus assess any effects on whales.

6.4. Source verification survey

Consideration of this item was combined with and thus subsumed under agenda item 7.2.

6.5. Propagation modelling to be conducted before and during the survey

As discussed during previous TF meetings, JASCO will prepare a library of propagation model results from which the one that most closely matches the data measured during the source verification will be chosen for the assignment of ‘A’ and ‘B’ survey lines. If any of the measured data fall ≥ 3 dB outside the range (|1 SD|) that any of the predetermined model scenarios predict, then the propagation models will be recalculated using the empirical data.

Racca indicated that the plan is to provide charts to observers with information on the area(s) they need to consider when observing for whales that may or may not be in an ‘A’ zone. These charts will be generated after the source verification and model tuning/selection occurs (see Section 7.2).

7. AT-SEA OPERATIONS

7.1. Vessels – descriptions and responsibilities

The following definitions and general responsibilities apply to vessels involved in the seismic survey and in the associated monitoring and mitigation efforts:

Observation vessel (OV): vessel responsible for behavioural follows and servicing of acoustic monitoring buoys. The current plan is for this vessel to be the M/V Pavel Gordienko, which can serve both to conduct observations and to service buoys. Onboard the OV will be a rigid-hulled inflatable boat (RHIB), equipped with one or more 4-stroke outboard engines, that can be deployed from the OV to conduct focal follows and also service the PML monitoring buoys as needed. Characteristics and responsibilities of this platform include:

o ‘Clears’ (i.e. verifies the absence of whales) and then patrols the 163 dBRMS ‘overlap’ area during A-line surveys.

o Crew consists of MMOs and individuals trained to conduct behavioural follows plus any members of the POI acoustics group required to perform buoy maintenance and to monitor and process the radio-transmittted acoustic data in case of disastrous failure of the shore-based data receiving station.

o Conduct behavioural follows from this vessel as well the RHIB, launched from the OV when weather permits; extend focal follows on whales displaying aberrant movements and/or behaviour and monitor potential long-term responses.


14

o Make CTD measurements that are to be used on a daily basis to inform the selection of appropriate modelling scenarios.

o Conduct acoustic recording from the RHIB, i.e. obtain recordings of seismic shots at the locations of whales.

o Equipment to be included:

‘Big eye’ binoculars

Hand-held reticle binoculars for estimating range and position of whales

o Maintain distance of ≥ 1 km from focal whales to minimize disturbance.

Seismic survey vessel (SV): industry seismic vessel; crew includes MMOs

Scout vessel (SCV): vessel responsible for ‘clearing’ areas of whales in advance of the SV as it acquires survey lines. The SCV is under the control of the seismic survey company and is not available to be re-purposed to follow whales. This vessel will be in contact with the OV and SV so that whales sighted can be tracked for mitigation purposes.

7.2. Procedures for seismic source verification/calibration (SSV)

A targeted acquisition of seismic data was performed in the Astokh area in October 2009, with three acoustic buoys (mini AUARs) deployed at 1, 2.5 and 3.5 km distance from the SV sail line, to allow validation of the acoustic noise modelling results against a good-quality set of measurements.

The data from the three mini-AUARs were downloaded by POI scientists upon return to Vladivostok and found to be of excellent quality. A blind cross-comparison of results from analysis of the acoustic data using POI’s and JASCO’s custom software was performed, with excellent results. The two processing codes were found to agree on the values of various per-pulse sound level metrics, with differences of < 0.1 dB.

The agreement between model and measurement was found to be reasonable considering that no special effort had been made to tune any of the parameters of the acoustic environment to improve the quality of the fit. The model tends to overestimate the received levels in the broadside lobe direction, in particular at larger distance, providing a precautionary result.

Details of this experiment and validation results can be found in Annex G.

With respect to the 2010 SSV and how the Panel reviews the data from the verification, Racca suggested that this would be similar to any SSV reporting to an environmental agency:

Commit to results within 48 hours of instruments being taken from the water (previously 72 hours had been agreed, the standard is up to 90 hours)

Report as plots of amplitude for RMS and SEL vs. range for the following frequency ranges: 10-100 Hz, 100-1000 Hz, 1000-2000 Hz, and 2000-5000 Hz.

Send to agreed reviewers

Write up results – to be considered as final document.

Racca reported there are no requirements to report to any agency so this is an internal WGWAP process. It was agreed that Vedenev, Nowacek and Southall should be provided with the data. Normally, according to Racca, data checking and report back is a straightforward process. The reviewing team can simply report back with a ‘yes’ or ‘no’ response. Only if the data were unexpected and unusual would there be a need for substantial discussion.

There was considerable discussion regarding the text of the Vancouver SSTF report (SSTF-3) that refers to the issue of retuning the model and what that would mean for the exclusion zone and the assignment of A- and B-lines. The TF agreed that nothing had changed in this regard since the Vancouver meeting and that JASCO, using the model parameters most closely matching the SSV experiment described earlier in this section, will make the initial assignments of A- and B-lines. If the results of the 2010 SSV and subsequent model selection (see Section 6.5) and consultation indicate that some lines need to be reclassified, this will


15

be done immediately before any putative A-lines are acquired. Furthermore, there are measures in place to reclassify lines if necessary (either ‘A’ to ‘B’ or vice-versa) based on real-time measurements during the survey. Finally, the safety exclusion zone will initially be set with a 2 km radius around the SV, as agreed at previous TF meetings, and will be redefined after the SSV, if appropriate.

7.3. Personnel and equipment (including e.g. ‘big eye’ binoculars, range finders)

Addressed under agenda item 7.1.

7.4. Use of inflatable for focal follows

Addressed under agenda item 7.1.

7.5. Protocols for behavioural follows

The TF agreed that the specific behavioural data collection protocols employed by the both the OV and RHIB are to be developed by Gailey, with assistance from Nowacek, Southall and Weller, and ready for review (via e-mail) by the TF no later than 28 February 2010.

7.6. Formalized operational decision tree

The OV is the only vessel that will operate at the discretion of Sakhalin Energy, and its functions will vary depending on the weather conditions, the area being surveyed by the SSV and the number of whales detected in the area. The TF recommends that the OV follow the operational decision trees shown in Figures 3 and 4. When an A-line is to be acquired by the SSV, the OV should follow the operational scheme shown in Figure 3.

Figure 3. Operational decision tree for OV when an A-line is to be acquired by the SV.


16

During acquisition of B-lines, the OV should follow the operational plan set out in Figure 4. If the A-line is one that is subject to partial truncation (see Section 4.3), then the OV should switch modes during the line, being informed by the lead MMO on the SV and the shore station acoustics team as to when the levels have dropped to where they are acceptable.

Figure 4. Operational instructions for the OV during the acquisition of a B-line.

7.7. Servicing DT-AUARs

The TF agreed that to minimize the number of vessels operating in or near the feeding area and thus the potential impacts on whales, the OV would also be responsible for the servicing the DT-AUARs. The PML real-time buoys are to be spaced at intervals of ≤ 2500 m so that if one becomes inactive, the survey can continue, i.e. a gap of ≤ 5000 m is considered acceptable. If two adjacent buoys fail, however, the survey must be halted until at least one of them has been repaired. In an attempt to prevent this latter scenario, the OV will, according the following operational plan, service DT-AUARs that fail:

Option #1, the preferred option, is to conduct repairs during a ‘down’ day

Option #2, if the repairs must be conducted during an ‘active’ day, they should be done

o preferably while the SV is turning from an A-line to a B-line

o otherwise, while the easternmost B-line of the day is being acquired.

7.8. Definition of ‘single-point authority’ for shut-downs, including communication links between all observation platforms that can affect a shutdown

The TF identified the need for a communications plan that will allow for all necessary communications between the teams/vessels in the field: i) acoustics team; ii) distribution teams; iii) behaviour teams; iv) OV; v) SCV; vi) SV. The most important connection is for the lead individual in any given team to have unencumbered contact with the lead MMO on the SV who can immediately effect a shutdown if called for under the monitoring/mitigation plans. The behaviour teams (shore- and vessel-based) also need to be in contact and coordinate their activities in order to determine which team is following which whale. The


17

acoustics team also needs to be able to communicate with the distribution and behaviour teams so that it can inform them of changes in zones to be monitored. Figure 5 lays out a scheme for the communications plan.

Inter-distribution team

Inter-behaviour teams (shore and offshore)

Inter-vessels

Between shore teams and seismic vessel

Between shore teams and coordinator

Between vessels and coordinator

Figure 5. The scheme shows the communications links necessary during the survey. The different coloured arrows indicate dedicated lines or radio channels. The SSTF recommends that these channels be separate as having all communications on a single channel would be untenable and likely result in confusion and/or inability to maintain appropriate contact.

Lead MMO

Seismic vessel

MMO

Observation vessel

Acoustic team

Distribution teams

Behavioral teams

MMO

Scout vessel

Zodiac of observation

vessel

Survey monitoring coordinator


18

8. IMPLICATIONS OF OTHER INDUSTRIAL ACTIVITIES OVERLAPPING IN TIME AND SPACE WITH ASTOKH SURVEY

8.1. Lebedenskoye – timing and considerations/implications for Sakhalin Energy survey including acoustic monitoring considerations

The Lebedenskoye block is operated by Rosneft Far-east Shelf. The planned seismic survey is to be onshore and in the offshore transition zone, in the general vicinity of Whale Observation Station 3 (Figure 6). Due to the very shallow water depth, conventional seismic vessels cannot be used. It is not known which contractor will carry out the seismic survey work.

The survey has an onshore and an offshore component, and the timing is likely ‘early’ May (onshore) + June (marine). The onshore part of survey is purportedly to be conducted using a buried sound source with offshore ocean bottom cable to record the acoustic returns. The offshore part of survey will reportedly be done using small boats with seismic guns array utilizing ‘APG Bolt Technology’ with recording of the pulse returns presumably done with the same ocean bottom cable.

Rosneft has received VNIRO advice to carry out the survey as early in the season as possible to reduce the impacts on gray whales. Vedenev stated that water depth conditions are similar to those in the area of the seismic survey at the Veny field (near Lunskoye) in 2009. The Veny operation took several months and was not completed.

The following concerns were raised:

Concurrent seismic acquisition at Lebedenskoye and Astokh could increase the impacts on feeding whales although it was unclear which would be worse for the whales – concurrent or consecutive surveys.

It was unclear whether Rosneft intends to take the same precautions and implement the same monitoring and mitigation measures as Sakhalin Energy has developed with the SSTF.

Concurrent acquisition may confound any effort to identify and characterize the responses of whales to the Sakhalin Energy survey.

According to Vedenev, Rosneft will probably develop mitigation measures of some kind, with at least MMOs onboard the seismic skiffs. Acoustic monitoring apparently be limited, but the Lebedenskoye field is inside the feeding area so whales there at the time will certainly be exposed to considerable sound energy.

It was agreed that the SSTF was not in a position to recommend either simultaneous or consecutive surveys because too little was known at the time of the meeting to support a position on this matter.

8.2. Analytical framework for assessing impacts of Sakhalin Energy vs. Lebedenskoye survey, i.e. can we construct a framework that will allow us to assess the effects of one or both?

The distance between the Lebedenskoye survey area and the Astokh survey area is at least 40 km. Signals received at offshore monitoring buoys from both surveys can likely be distinguished due to distinct differences in signal.

Monitoring options discussed by the SSTF included:

Lebedenskoye

Astokh

Exact survey outline unknown

Figure 6. Lebedenskoye (green) and Astokh (yellow) survey areas. The exact plan for the Lebedenskoye survey is unknown, so this figure is approximate.


19

During the Astokh 4D seismic acquisition, three distribution teams will be covering the South and North monitoring stations, so if the surveys are concurrent those teams will be collecting data throughout both surveys; however it will be difficult to tease apart the effects of one survey from the effects of the other.

Vedenev noted that any scientific programme of Rosneft Far-east would involve the Shirshov Institute of Oceanology (his home institute), but no decision on such involvement had been taken at the time of this meeting. If such a collaboration were to emerge, there may be opportunities to collect data that the Panel could examine.

Joint programme acoustic recorders could provide some data about the Lebedenskoye survey, though those recorders do not go into the water until late July or early August and even then, the release of any data would be subject to ENL approval.

Weller reminded the group that data on behaviour that has been and continues to be collected by the joint programme cannot be considered as a true ‘baseline’. Seismic surveys and other sources of noise operate without consistent monitoring and often without reliable information on their locations, on/off timing etc. The situation that could emerge in 2010 – a possible concurrent seismic survey by another operator, with no coordination or sharing of information between the companies – is simply one more example of a chronic or recurrent problem faced by those trying to use joint programme data to understand effects of industrial noise on western gray whales.

9. NEXT STEPS

9.1. Specification of tasks to be completed after meeting, with timeline and responsibilities

9.1.1. Behaviour data – focal follow ‘at sea’ protocol Gailey, with input from Weller, Nowacek and Southall, will develop this protocol, by e-mail exchange (and teleconference if needed). This will consider both the on-shore data collection procedures already well established by the joint programme and the BRS protocols provided by Southall.

Timing: Deadline 28 February 2010. Larsen will send a reminder by 15 February.

9.1.2. Cooke analysis on PML Cooke will complete his modelling analysis on distribution data relative to the perimeter monitoring line. The implications for the results of these analyses are discussed under agenda item 4.2 (above).

Timing: Originally agreed as 4 January 2010; Cooke agreed to notify of any delays. Distribution will be via Larsen to the full SSTF.

Related actions and follow-up: Muir requested to see Cooke’s prediction surface (coordinates and values), for visualization in GIS relative to the Muir/Joy analysis.

9.2. Arrangements for monitoring progress with tasks

Discussed above.

9.3. Need for additional SSTF meeting in early 2010, scheduling etc.

No particular need for a full SSTF meeting before the seismic survey in 2010 was foreseen. However, it was generally agreed that a SSTF meeting prior to WGWAP-8 (April 2010) would be useful. Among the topics likely to need further attention are: i) the unknowns about the Lebedenskoye survey and any changes in planning it may require on the part of Sakhalin Energy; ii) the details of the A-line truncation proposal; and iii) a review of the ice cover data. Given this, it was agreed that the SSTF would meet for ~1 day before WGWAP-8.


20

9.4. Follow-up analyses (forum for follow-up, data sharing, data management plan)

The TF discussed the need for follow-up assessment and analyses. Among the things that would be valuable are: 1) a post-mortem on general processes (e.g. lessons learned from the planning and execution of monitoring and mitigation) and 2) an analysis of the data collected in conjunction with the seismic survey. At a minimum, the acoustic data should be ready for WGWAP-9 in December 2010; hopefully other analyses will be at least partially completed in time for that WGWAP meeting.

Given that Sakhalin Energy will have sole control of the data collected during the seismic survey (and therefore such data will not be part of the ENL/Sakhalin Energy joint programme), it should be easier than usual to complete post-survey analyses. A caveat is that raw acoustic data will still have to be processed in Russia. The TF recommends the formation of a working group to ensure that retrospective analyses are carried out on, among other things, the effectiveness of the mitigation measures and the effects on gray whales.

There was a general discussion on the need for a carefully designed and implemented data management plan. The SSTF acknowledged that all of the data would be under Sakhalin Energy’s control, and the intent was not for the TF or the Panel to prescribe the details of such a plan. Rather, the group wished to emphasize that the plan should be sufficiently detailed and coordinated so that all biologically relevant data are made available to the SSTF as soon as possible after the survey, particularly since this will be the first acoustic-behaviour-distribution dataset free of the constraints and limitations attached to joint programme data.


21

REFERENCES Gailey, G., Würsig, B. and McDonald, T.L. 2007. Abundance, behavior, and movement patterns of western gray whales

in relation to a 3-D seismic survey, Northeast Sakhalin Island, Russia. Environ. Monit. Assess. 134(1-3):75-92.

Hedley, S.L. and Buckland, S.T. 2004. Spatial models for line transect sampling. J. Agric. Biol. & Environ. Stat 9(2): 181-199.

Vladimirov, V.A., Starodymov, S.P., Afanasyev-Grigoryev, A.G., Muir, J.E., Tyurneva, O.Y., Yakovlev, Y.M., Fadeev, V.I. and Vertyankin, V.V. 2008. Distribution and abundance of western gray whales off the northeast coast of Sakhalin Island (Russia), 2007. Document SC/60/BRG9 presented to the IWC Scientific Committee, June 2008, Santiago, Chile. 9pp.


22

Annex A

List of participants

Doug Bell (Sakhalin Energy)

Koen Broker (Sakhalin Energy)

Justin Cooke (Panel)

Vladimir Efremov (Sakhalin Energy)

Glenn Gailey (Sakhalin Energy)

Judy Muir (Sakhalin Energy)

Doug Nowacek (Panel)

Roberto Racca (Sakhalin Energy)

Randall Reeves (Panel)

Brandon Southall (Associate Scientist)

Dorine Terwogt-de Jonge (Sakhalin Energy)

Grigory Tsidulko (Panel)

Dave Weller (Panel)

Alexander Vedenev (Panel)

IUCN

Sarah Humphrey (IUCN) Finn Larsen (IUCN)

Laura Riddering (IUCN)

Interpreters

Alex Danilov Grigory Shkalikov


23

Annex B

Agreed Agenda

1. Introductory items

1.1. Introductions of participants

1.2. Summary of progress since WGWAP-6

1.3. Identification of outstanding issues to be resolved

1.4. Aims of workshop and expected product

2. Documents and progress reports

2.1. Review of available documents

2.1.1. Existing documents made available (e.g. Vancouver SSTF workshop report)

2.1.2. New documents prepared for the workshop (e.g. report of Gland PML small group meeting in October 2009)

2.2. Results of tasks requested at WGWAP-6

2.3. Other new information

3. Operational plans and expectations

3.1. Any changes in Sakhalin Energy plans for 2010

3.2. Information on other relevant industrial or research activities (e.g. other seismic surveys, satellite tagging effort)

4. Perimeter Monitoring Line (PML)

4.1. Summary of status from Small Group Meeting (Cooke, Muir)

4.2. Final definition (or, at worst, determination of route for getting there)

4.3. Incorporation into operational plans including consideration of truncated or ‘partial’ A lines (Sakhalin Energy)

5. Shore-based distribution/density and behaviour monitoring

5.1. Station locations, survey schedules and protocols, logistics

5.2. Data to be recorded, protocols for near real-time processing and analyses

5.3. Communications and coordination among shore observers

5.4. Communications and coordination with vessels

5.5. Shore based planning contingency plan

6. Acoustic monitoring

6.1. Equipment issues outstanding


24

6.2. Spatial and temporal deployment of buoys (sampling locations and time periods) (Sakhalin Energy, IFAW)

6.3. Collection, transmission, storage and processing of data

6.4. Source verification survey

6.4.1. Methods for execution beyond SOP conducted by JASCO, e.g., additional acoustic buoys near SSV

6.4.2. Communication of results to select panel members and/or agreed upon experts

6.4.3. Determination of the RMS-SEL offset during SSV

6.5. Propagation modelling to be conducted before and during the survey

7. At-sea operations

7.1. Vessels – number and responsibilities of each

7.2. Procedures for source verification/calibration (seismic vessel), including transmission of data to experts and incorporating their feedback, final designation of A and B lines, etc.

7.3. Personnel and equipment (including e.g. Big Eyes, range finders)

7.4. Use of inflatable for focal follows (deployed from observation vessel?)

7.5. Protocols for behaviour follows

7.6. Formalized operational decision tree (‘what if scenarios’)

7.7. Servicing TAUARs

7.8. Definition of ‘single-point authority’ for shut-downs, including communication links between all observation platforms that can affect a shutdown

8. Implications of other industrial activities overlapping in time and space with Astokh survey

8.1 Lebedenskoie – timing and considerations/implications for Sakhalin Energy survey including acoustic monitoring considerations

8.2. Analytical framework for assessing impacts of Sakhalin Energy vs. Lebedenskoie survey, i.e., can we construct a framework that will allow us to assess the effects of one or both?

9. Next steps

9.1. Specification of tasks to be completed after meeting, with timeline and responsibilities

9.2. Arrangements for monitoring progress with tasks

9.3. Need for additional SSTF meeting in early 2010, scheduling etc.


25

Annex C

List of documents

DOCUMENT NUMBER

SUBMITTED BY

TITLE STATUS

PRIMARY DOCUMENTS

SSTF-4/1 IUCN Provisional agenda Circulated

3 DEC

SSTF-4/2 IUCN List of documents Circulated

6 DEC

SSTF-4/3 No document

SSTF-4/4 Muir/Joy Hybrid approach for the estimation of a perimeter monitoring line

Circulated 19 NOV

SSTF-4/5 Cooke PML analysis Not yet

available

SSTF-4/6 Paxton A Relative Density Surface with 95% Population Mass Contours for Sakhalin Grey Whales (ver. 6)

Circulated 19 NOV

OTHER DOCUMENTS

SSTF-SG Rep. Donovan Draft report of the small specialist workshop to determine the perimeter monitoring line

Circulated 27 OCT

WGWAP-7/19 Sakhalin Energy

Ice Data Relevant for the Astokh 4D Seismic survey in 2010

Circulated 6 DEC**

WGWAP-7/21 Sakhalin Energy

Report on Installation of Shore-based Observation Platforms IUCN Portal*

SSTF-3 Rep. IUCN 3rd report of the Seismic Survey Task Force (Vancouver)

Circulated 3 DEC*

SSTF-2 Rep. IUCN 2nd report of the Seismic Survey Task Force (Lausanne)

IUCN webpage

SSTF-1 Rep. IUCN 1st report of the Seismic Survey Task Force (Den Haag)

IUCN webpage

WGWAP-6 Rep. IUCN 6th report of the WGWAP IUCN webpage



WGWAP-3 Rep. IUCN 3rd report of the WGWAP IUCN webpage

WGWAP-2 Rep. IUCN 2nd report of the WGWAP IUCN webpage

WGWAP-1 Rep. IUCN 1st report of the WGWAP IUCN webpage


26

Annex D

(from Vancouver SSTF Meeting, January 2009)

Monitoring and mitigation measures for the 2009 seismic survey

MONITORING (NUMBER, DISTRIBUTION, BEHAVIOUR)

The monitoring measures proposed here are integrally related to the mitigation measures proposed or likely to be proposed for future surveys. Indeed, most of the monitoring measures provided below are essential for implementation of the mitigation measures proposed for the 2009 seismic survey.

The monitoring measures fall into two categories:

(1) real-time (or near real-time) monitoring required to trigger appropriate action where sound levels approach or exceed defined thresholds (i.e. essential for mitigation);

(2) additional monitoring (involving the collection of some data that do not need to be analysed in real time) to obtain data on the effects of the seismic survey on whales, especially western gray whales, to add to the existing knowledge base, and to contribute to the design of mitigation strategies for future seismic surveys.

Acoustic monitoring (perimeter and within area)

Along the perimeter of the feeding area (the perimeter monitoring line) (1) Real-time monitoring of acoustic levels using sea-bottom receivers will be undertaken during all

periods of seismic source activity.

(2) A total of at least nine receivers will be positioned at 2500m intervals along the edge of the feeding area to ensure adequate redundancy. There will thus never be more than 5000m between active buoys (considered the range of reliable model-based interpolation of recorded sound levels).

(3) Receivers will be in place and verified to be functioning properly before activity starts and for the duration of the survey.

(4) There will be a direct radio link between the real-time monitoring acoustician and the Senior MMO on the active seismic vessel.

On the coastal side of the perimeter monitoring line (1) All necessary efforts will be made to obtain archival acoustic data within the feeding area using

bottom-mounted receivers.

(2) During the seismic survey, ≥ 3 acoustic monitoring buoys will be deployed in the feeding area on or near the 10m isobaths and near the centre of the field of view of the shore stations. Verification that these buoys are operational during the survey should be undertaken, at least at the start of the survey.

General visual monitoring (shore-based and vessel-based)

Note that whilst the monitoring below focuses on the area within the feeding area (i.e. on the coastal side of the perimeter monitoring line) near to the seismic operations, it is also important to maintain the observation effort throughout the rest of the area as in previous years. This is important for analysing and interpreting the data with respect to actual or potential effects of seismic surveys on the whales, and for maintaining the longer-term monitoring data series that will be a valuable resource when future seismic operations occur.


27

On the coastal side of the perimeter monitoring line (shore-based) (1) Shore-based scan surveys will be undertaken by two teams at the five pre-existing vehicle scan

observation points south of the mouth of the Piltun lagoon (i.e., vehicle scan survey observation stations 9 to 13) to enhance the resolution of potential changes in whale density and distribution. The timing of scans will be scheduled to ensure the monitoring of whales pre-, during and post seismic line acquisition. Pre- and post surveys will be conducted 1 hour prior/post of the acquisition whilst ‘during’ scans will be conducted 1 hour from the onset of line acquisition. A third team will conduct daily surveys, weather permitting, at the eight pre-existing vehicle scan observation stations (numbers 1 to 8) using the same survey protocols as in previous years.

(2) Behaviour will be monitored by two behavioural monitoring teams. The location of behavioural platforms will be directly inshore of the seismic activity in areas that are expected to have the highest exposure levels. Scans conducted by behavioural monitoring teams will augment information on whale numbers collected by the shore-based distribution surveys.

On the coastal side of the perimeter monitoring line (vessel-based) (1) Behaviour will also be monitored from a vessel platform. This will cover whales that may occur near

or slightly outside the defined feeding habitat that are not being monitored effectively by shore-based teams (e.g. due to low station heights, onshore fog etc.). The monitoring location will be within regions of maximum predicted ensonification. The vessel type should be selected based on requirements of minimal sound output with an effective observation height (5-10m) to increase the range of whale observations. Focal follow observations will be conducted from this platform to monitor respiration patterns and the general movement of the whales in the specified region. Extended focal follows should be conducted on whales displaying aberrant movements and/or behaviour to monitor potential long-term responses. The vessel will maintain a distance of at least 1 km from the whale being observed.

(2) Gray whale distribution will also be monitored from a vessel platform in the event of inclement weather (e.g. onshore fog) that prevents monitoring.

(3) The observation vessel will have a direct radio link to the Senior MMO on the active seismic vessel.

Within the proximity of the seismic related vessel(s) (1) Experienced MMOs will be stationed on all vessels (i.e. seismic, scout and supply vessels) for the

duration of the survey.

(2) MMOs will be limited to a maximum 2-hour continuous shift with a minimum of 1 hour between shifts.

(3) Single-point authority for operational shutdown will lie with the on-shift Senior MMO on the seismic vessel.

(4) All vessels and real-time acousticians will have direct radio access to the on-shift Senior MMO.

(5) MMO observation platforms should be located at the highest elevation available on each vessel with the maximum viewable range from the bow to 90˚ port/starboard of the vessel. Optimal locations might be on the ‘flying bridge’. Use of the bridge should be avoided due to obscured views and potential distractions.

(6) An extended visual search (20 minutes) will be conducted prior to start-up of the seismic source.

(7) There will be a minimum of two MMOs on watch on the seismic vessel at any given time during ramp-up, shooting and for the 20 minutes before start of ramp-up.

(8) Occurrence and behaviour of whales will be documented in accordance with existing MMPP (Marine Mammal Protection Plan) and MMO procedures.


28

MITIGATION MEASURES

Timing of surveys

(1) The seismic survey will commence and be completed as early in the season as logistically possible. Logistics include ensuring that all mitigation and monitoring procedures are in place.

(2) The duration of the seismic survey will be as short as technically and logistically feasible. Logistics includes ensuring that all mitigation and monitoring procedures are implemented fully.

(3) Lines in Zone A (see definition below) should be acquired at the earliest possible opportunity given visibility, mitigation and monitoring requirements.

General design and conduct of surveys

The most stringent mitigation measures should be applied in Zone A as defined below. The monitoring measures defined above must be in place and operational for the acquisition of lines.

Definition and updating of A and B zones (1) Initially, the survey area for which the additional mitigation measures are in effect (A zones) will be

defined by the overlap of the ‘feeding area’ and the maximum shoreward extent of the 163 dBRMS isopleths for that particular shot line.

(2) Before any lines are shot within the range currently predicted to exceed 156 dBSEL at the perimeter monitoring line, received sound levels at the line will be compared with model predictions. If received sound levels exceed model predictions, then the model shall be retuned to match the observed levels. Based on the updated model predictions, shot lines for which an overlap is predicted between the 163 dBRMS contour and the monitoring line will be reclassified as A lines, for which the additional mitigation measures specified below apply.

(3) The comparison between observed and expected sound levels at the perimeter monitoring line, and, where indicated, retuning of the acoustic model, shall be repeated at regular intervals during the survey.

(4) In the event that the 163 dBRMS threshold is exceeded at any receiver on the edge of the feeding ground while shooting a B line, operations shall be suspended immediately or shifted away from the feeding ground until a recalibration exercise has been conducted as described above, and the lines have been reclassified accordingly.

Measures within the proximity of the seismic vessel – entire survey (1) After more than 20 minutes of inactive source, ramp-up procedures will be adopted such that the

individual air guns will be activated in a progressively larger combination over a period of several minutes (6 dB increments every 5 minutes over 20 minutes).

(2) The Senior MMO will initiate source shutdown if a gray whale is observed within defined exclusion radius of the source array.

(3) The Senior MMO will initiate a precautionary shutdown if a gray whale is observed to be on a course that will result in its entering the shut down zone.

(4) Low level single (smallest) gun operations will be conducted during line changes. Ramp-up procedures will furthermore be implemented 20 minutes prior to the sequential line acquisition. As long as the single gun operation is uninterrupted during the line change, this period will not be interpreted as ‘source inactivity’ for the purposes of clause (5) below.


29

(5) For operations in conditions that preclude effective visual monitoring of the defined exclusion radius of the source array (e.g. night, fog, poor visibility1).

(a) Prior to a seismic acquisition, the line of interest will have been surveyed (if necessary using a second vessel) at least 6 hours preceding the start time of acquisition of the line to ensure that no gray whales have been sighted in the vicinity of the line. If poor visibility hampers survey of the entire line, then the line will not be acquired.

(b) Operations will shut down for the night period if whales are sighted in the pre-dusk scan.

(c) After more than 20 minutes of source inactivity, operations will not be re-commenced, due to the inability to conduct an adequate visual scan.

Additional restrictions for Zone A (1) No acquisition during periods of poor visibility3 or at night.

(2) No acquisition unless the feeding area perimeter line is within the effective sighting distance of a shore station or an additional vessel.

(3) No acquisition if any gray whales have been observed in Zone A over the preceding 6 hours.

(4) No acquisition if mother-calf pairs have been observed in Zone A in the preceding 12 hours.

1 “Poor visibility” means any conditions under which the estimated distance at which a gray whale can be reliably sighted is less than the defined exclusion radius. A prohibition on night shooting could increase the survey duration by up to 50%, which would be undesirable, both in terms of increased costs and in terms of extending the duration of the seismic survey into the period of peak whale abundance. The Workshop therefore agreed that night shooting or shooting during fog or poor light should only be conducted when the line has been surveyed (either by a separate “scout” vessel or while shooting an adjacent line) in good conditions during the preceding six hours and no gray whales have been sighted within this period. This would be the first time such a measure has been tried: the period of 6 hours, while considered practical to implement, remains, in the absence an analysis of the effectiveness of the measure for different choices of period, arbitrary.


30

Annex E

Proposed analysis for the estimation of a perimeter monitoring line using systematic and opportunistic data

Justin Cooke

Data sets

Sightings data are available from the following platform types:

(1) shore-based;

(2) research vessels on dedicated surveys;

(3) research vessels, opportunistic sightings.

For all three platform types, it was agreed to use data from June through August collected during 2005-2009 for the purpose of fitting the model, but that predictions would be made for mid-July (see below). Data shall be truncated to the following box (with respect to the estimated position of sightings, not the platform position).

51.75 – 53.25 Lat

143.2 – 144.1 Long

Data collected during Beaufort > 3 or visibility < 2.5 km shall be excluded for all platforms.

For the shore-based platforms, only data collected for distribution (not behaviour) shall be used. A platform- and time-specific truncation distance based on the penultimate reticle and taking account of eye height and tide level was calculated by Muir and Bychkov and incorporated into the data file.

For the vessel-based data, a radial truncation distance of 4.5 km shall be applied throughout, unless the data indicate that a smaller value is more appropriate. The time on effort shall be based on the MMO data files, but excluding times before sunrise or after sunset. Paxton supplied a formula for determining the times of sunrise and sunset.

The opportunistic effort from research vessels shall be divided into three speed categories:

(a) < 1 km/h (stationary); (b) 1-5 km/h (slow); (c) > 5 km/h (cruising). Speed is not taken into account further.

Density of whales

Modelled as a point process with log-intensity function (x, c) of location x and other covariates c. The covariates are:

Year (discrete, 2005-2009, as random effect) Date in year (1-D exponential-smoothed or Gaussian-smoothed) Distance offshore or depth (linear) (choice according to best fit) Location (2-D exponential-smoothed or Gaussian-smoothed)

Interactions between year and other factors will also be included if indicated by the model selection criterion. The model selection criterion is the mixed-model version of the AIC.


31

Detection function

Log-hazard rate (r, v) of radial distance I and other covariates v. The detection function can be modelled as flat (up to a platform-specific truncation distance), exponential, or Gaussian, i.e.

( , ) ( )r v v (flat)

( , ) ( ) ( ).r r v v v (exponential)

( , ) ( ) ( ). ²r r v v v (Gaussian)

where the parameters and are functions of the covariates. To obtain uniqueness, is fixed at zero for shore-based platforms. The choice of model will be based on the model selection criterion (MSC).

The covariates are:

Platform type (categorical: shore-based; dedicated vessel; and opportunistic vessel periods further subdivided into stationary, slow and cruising based on speed as described above)

Visibility, Beaufort (include if indicated by MSC) Interactions between visibility, Beaufort and Platform type (include if indicated by MSC).

Fitting the data

The data consist of times and positions (both absolute and relative to the observer) of WGW sightings. The sighting process is treated as an over-dispersed Poisson process. The log-rate factor for the sighting process from platform I at time t for potential whale position x is given by:

( ; ) ( ) , ( ) , ( , )i i it t t t x z x v x c x

where zi(t) is the position of platform I at time t, and vi(t) is the vector of other covariates for platform I at time t.

Ignoring constant terms, the likelihood of the data is proportional to:

( )1 1

, exp , ( )n m

j i j j j i i iAj i

L s t t I t R d dt

x x x z x

where sj is the group size of the jth WGW sighting, n is the number of WGW sightings, m is the number of platforms, i(j) is the platform index for the jth WGW sighting, tj is the time of the jth sighting, A is the area over which the density function is to be estimated, and Ri is the truncation distance for platform i.

Post-fitting calculations

After estimating the parameters by maximum likelihood, the predicted density surface for mid-July in a random year is computed over the box, and the density contour is found which contains 95% of the integrated density.


32

Annex F

Hybrid approach for the estimation of a perimeter monitoring line Ruth Joy and Judy Muir

Introduction

The hybrid approach for the estimation of the Perimeter Monitoring Line (PML) used geostatistical methods on data from both the SEIC/ENL WGW Research Program 2005-2007 Opportunistic vessel surveys and the Systematic surveys (vessel, and shore-based distribution and behaviour). The main objective of this approach was to apply a simple method that utilized all data that have been collected in the study area by different platforms. This approach allowed western gray whale (WGW) sightings and effort from the Opportunistic vessel data set used in the Paxton analysis (Paxton 2009) to supplement the systematic survey data. In particular, survey effort from the Opportunistic vessel surveys was used to try to ‘fill in’ the ‘zero effort’ gaps in the LGL density maps that were obtained from systematic quality data.

The hybrid approach first created a relative (WGW) average density surface for 1 km by 1 km grid cells for each data set. The two density surfaces were calibrated and combined. Geostatistical methods were then used to interpolate and extrapolate whale density for those grid cells with no effort and to provide a smoothed predicted relative density surface for the entire area of interest. Finally, the 95% contour of the final relative density surface was calculated and used as an estimate of the PML. These methods were applied to:

i) June-July 2005-2007 data

ii) June-August 2005-2007 data

iii) Each data set using a spatial extent that covered both the Piltun and Offshore feeding areas.

iv) Each data set using an irregular spatial extent that excluded the Offshore feeding area.

Methods

Estimating sightings per unit effort for the Paxton Opportunistic vessel survey data The ‘segmented’ effort data from the Paxton analysis was used to allocate ‘effort’ (expressed as times visited) to a particular 1 km × 1 km grid cell, taking into account the approximate search width of 4.2 km (see Paxton 2009 for details as to how the segments were created and search width was derived). Hawth’s tool “Add XY Line Data” (Beyer 2004) was used to create segments in ArcGIS v9.2 from the Paxton segment endpoints. The segments were then buffered by 4.2 km to represent search effort from each segment. Hawth’s tool “Enumerate Intersecting Feature” (Beyer 2004) was then used to count the number of times each segment buffer covered the centroid of each grid cell as an estimate of effort. This tool output a count of the number of intersecting segment buffers, and a list of the segment IDs that intersected each grid cell centroid. Because segment buffers were rounded at the ends of each segment, this potentially created areas with overlapping segment effort, and consequently, potential duplication of effort by adjacent segments if the overlapping buffers overlaid a grid cell centroid. These overlapping buffers were identified and removed with the reduced buffer count in each grid cell then used as an index of effort.

The WGW sightings associated with the Paxton segments were assigned to the grid cells in which the sightings were located using Hawth’s “Point Intersect” tool, and the number of gray whales in the sightings were summed for each grid cell. An estimate of relative density was then calculated for each sampled grid cell by dividing the total number of gray whales in each grid cell by that cell’s effort, i.e. the number of segment buffers (with duplicated effort removed).


33

LGL average WGW densities estimated from June-July 2005-2007 systematic survey data. Two WGW density surfaces were estimated based on average 2005-2007 June-July and average 2005-2007 June-August available systematic survey data. Methods to estimate these density surfaces are described in detail in Vladimirov et al. 2009.

Merging the Opportunistic and Systematic Datasets The estimates of relative densities per 1 km × 1 km square in the two datasets were assessed for compatibility by plotting opportunistic-related whale density against systematic-related whale density in log-scale (not shown), and by evaluating the associated Pearsons correlation coefficient for all those grid cells with at least 2 units of effort associated with them (97% when zero densities were included; 82% when zeros were excluded). The merge was deemed appropriate as the correlation coefficient was significantly positive (p<0.0001) on the logged effort-corrected whale densities in the two datasets. The recalibrating of the Opportunistic data was done by multiplying the logged data by the linear regression coefficient between the logged data in each dataset (β=0.974 June-August). A constant of 1 was added before the log was applied and this constant was selected as the log(0 + 1) is 0, which ensured that a zero in the uncalibrated space, was a zero in the calibrated space. Where there was overlap of effort in both data sets, a mean whale abundance was calculated such that the contribution to the grid cell mean was weighted proportionally to the associated effort for each data set. Figure 1 gives a visual representation of the Opportunistic and Systematic effort coverage for the study area, and the extents of coverage in the merged dataset. There were 7370 unique grid cells of effort in the Opportunistic, 4693 unique grid cells of effort in the Systematic datasets and when merged there were 7824 grid cells of effort for the June-August dataset.


34

Figure 1. Effort corrected effort in June-August 2005-2007. There were 7370 unique grid cells of effort in the Opportunistic dataset (left panel), and 4693 unique grid cells of effort in the Systematic datasets (centre panel). When merged there was effort in 7824 grid cells (right panel) for the June-August dataset.


35

These methods were repeated for calibrating between the opportunistic and systematic survey subset of June-July data (β=0.972 June-July). For these data, there were 6902 unique grid cells of effort in the Opportunistic, 4067 unique grid cells of effort in the Systematic datasets and when merged there were 7276 grid cells of effort (Figure 2). This allowed an assessment of the effect of exclusion of the August data on the location of the perimeter monitoring line.

Figure 2. Effort corrected effort in June-July 2005-2007. There were 6902 unique grid cells of effort in the Opportunistic, 4067 unique grid cells of effort in the Systematic datasets and when merged there were 7276 grid cells of effort.


36

Spatial Mapping of Abundance using Kriging The original univariate kriging method used to fit to these data was Ordinary Kriging. This method was deemed a priori to be the most likely candidate model for these data as there is no assumption for an underlying spatial mean using this method (Cressie 1993). However, this method did not perform as well as expected, due to large areas of data void coverage particularly in the upper east corner – with impossible extrapolations and errors in this region. Instead, Simple Kriging was used that allowed the assumption of a known mean of zero whales per unit effort for those areas where there was no effort (Figure 3, left panel), which is consistent with present knowledge based on systematic and opportunistic surveys of the offshore deep water regions of the study area. This assumption of a zero mean created a stable extrapolation outside the range of the data and produced abundance estimates that conformed well to the raw data (Figure of Ordinary Kriging surface not shown). In fitting the variogram to these data, an exponential covariance structure was used (other covariance models such as spherical and Gaussian were also tested but gave undesirable results in regions of abrupt changes in whale abundance).

A different kriging approach is contrasted in the right panel of Figure 3. Co-kriging was also considered for estimation as it allowed for the inclusion of a covariate (depth) with known values in areas where there was no whale survey effort. The idea behind co-kriging is that if the target and co-variable are correlated, and the co-variable is available over a broader spatial region, then estimation in those areas that are data void for sightings effort would be improved. The biological link between grey whale abundance and shallower waters provided the rationale for the use of these variables. However, the co-kriging results were disappointing (right panel of Figure 3). The cross-variogram indicated a weak negative correlation between abundance (the target) and depth (the co-variable). The subsequent kriging surface derived using this cross-variogram yielded a very smooth field for abundance that is inconsistent with the whale observations, and largely dependent on depth. A key assumption in co-kriging model is that the relation between whale abundance and depth is linear. In reality, whales have a preferred depth range where they are found, and their abundance cannot be modelled by the simplification of a linear drop off in abundance with increasing depth. Therefore, co-kriging was not used further to predict whale abundance with the merged Opportunistic and Systematic sightings data.

Figure 4 shows the two different spatial domains that were considered; one a rectangular polygon that included the offshore feeding area and the other an irregular shaped polygon that excludes this area. Figure 4 demonstrates that the positive whale abundances in the SE corner of the region, i.e. the Offshore feeding area, did not affect kriging estimates in the broader domain, and the contour lines that were constructed using the irregular or rectangular grid are interchangeable.


37

Figure 3. Maps of whale abundance using different kriging methods for the “irregular” domain for June-August data. The left panel shows results from Simple Kriging. The right panel shows results from Co-Kriging using whale abundance as the target variable, and depth as the co-variable. The 95% contour of whale abundance derived from each of these mapping results is plotted as a dashed line.


38

Figure 4. Kriging estimates for whale abundance in the two spatial domains for the June-August data. Simple kriging with an exponential covariance function was used.


39

Figures 5, 6 and 7 are various visualisations of the Simple Kriging surface for the June-August, and a subset of this dataset – the June-July data. Table 1 is a summary of the effort and whale sightings in each of these datasets. These are included to highlight the similarities and differences between the datasets.

Table 1. Effort and whales associated with the two datasets.

Time Period Number of Grid Cells with Effort

Number of Grid Cells without Whales Observed

Total Number of Effort Units*

June-July, 2005-2007 7276 6764 (93%) 135,611

June-August, 2005-2007 7824 7122 (91%) 276,195

* Total Number of Effort Units was calculated as the sum of all units of derived effort for all grid cells in each data set.

Table 1 highlights that there were a total of 7824 grid cells with effort for the time periods June to August 2005-2007, and 7276 grid cells when restricted to June, July data only, a difference of 548 grid cells. Many of the grid cells in both datasets were visited on more than one occasion, and in one grid cell in the June-August dataset there was effort associated 424 times over the period. By including the August effort in the kriging analysis, the sample size is doubled with a total of 276,195 effort units, or in other words, the restricted June-July dataset has half the amount of effort associated with it, with a total of 135,611 units of effort.

The Simple Kriging surfaces in the left and right panels of Figure 5 are intuitive representations of their respective data, and this is seen in Figure 6 where the data is over-plotted on the surface predictions. The main difference between the predicted values of the surfaces is that a higher maximum prediction was observed in June-July (0.24 whales per unit effort) than in June-August (0.19 whales per unit effort). This may suggest a difference in distribution of whales at different times of the season.

Figure 7 shows that the contour produced from using the subset of June-July data lies inside the contour produced using the full June-August dataset. The location of the June-July contour compared to June-August suggests that, although overall numbers are lower in this period, the distribution of whales may be more concentrated earlier in the season when the seismic surveys will be conducted.

Statistical Assessment of Kriged Surface and its Variance. The kriging standard deviation was computed as part of this analysis. The spatial distribution of the error was plotted and visually assessed (not shown). The kriged whale density predictions ranged from 0 to 0.215 whales per kilometres squared. Standard deviation was relatively constant at a value of 0.03 in areas near grids with effort. The de-correlation distance was optimized to be at a 12 km range in the fitted variogram, therefore the kriging error increased at distances greater than 12 km from observations. In areas that were data void, the standard deviation increased to a maximum of 0.04.


40

Figure 5. Prediction surfaces from simple Kriging using data sets from June-August (left panel) and from June-July (right panel). The associated 95% contour lines are plotted as dashed lines.


41

Figure 6. Same as Figure 5, but with non-zero abundance data included (over-plotted). Observed abundances over 0.1 are indicated as dark blue dots, while abundances between 0 and 0.1 are medium blue. 95% contours are indicated in each panel.


42

Figure 7. Overlaying the 95% contour lines from the 2 datasets for comparison.


43

Summary

The hybrid analysis estimated a relative average western gray whale density surface from opportunistic and systematic survey data. Kriging was used to interpolate and extrapolate densities in areas with no effort, and to produce a smoothed surface over the spatial extent of the study area. Three different kriging methods (simple, ordinary, and co- kriging) were applied over 2 different spatial prediction grids, using the June-August, and the subset of June-July data (hence 12 different cases were investigated, plus a number of different covariance models for each case).

It was found that simple kriging with an assumed zero mean was most suitable for spatial abundance estimation, due to its ability to transition over the sharp gradients in abundance, and extrapolate into data-void areas. Extending the analysis domain to include the areas of relatively high whale abundance in the southeast corner of the area of interest did not affect the kriging results (either the abundance maps or contour line) for the coastal areas with high whale abundance. This is due to the localized nature of kriging estimation. Comparison of the analysis surface that was based on June-August data with the surface from the subsetted June-July data showed that whale abundances where more concentrated near shore in June-July, and that the 95% contour was, for the most part, closer to shore.

Literature Cited

Beyer, H. L. 2004. Hawth's Analysis Tools for ArcGIS. Available at http://www.spatialecology.com/htools.

Cressie, N. 1993. Statistics for spatial data, revised edition. Wiley, NY. 900 pp.

Paxton, C. G. M. 2009. A relative density surface with 95% population mass contours for Sakhalin Gray Whales. Report to IUCN, October 2009.

Vladimirov, V.A., Starodymov, S.P., Afanasyev-Grigoriyv, A.G. and Muir, J.E. 2009. Distribution and abundance of Korean stock gray whales in the waters of northeastern Sakhalin Island during July – October 2008 (based on data from onshore and vessel based surveys). Final report to Exxon Neftegas and Sakhalin Energy Investment Company. Yuzhno-Sakhalinsk, Russia.


44

Annex G

Validation of seismic model against data from trial at PA-A in October 2009 Prepared by Roberto Racca, JASCO Applied Sciences

In preparation for the significant use of the airgun array sound footprint modelling that will be made in support of the monitoring and mitigation measures for the Astokh 4D survey, a targeted acquisition of seismic data was performed to allow validation of the model results against a good quality set of measurements.

Acoustic recording of the seismic source signal under controlled conditions was done as the seismic vessel transited through the Astokh area toward its next assignment. It was agreed that the seismic survey team would deploy both the array and the full streamers (thus allowing PGS to perform simultaneous assessment of the geophysical signal) for a short run on a direct north-south course just inshore of the PA-A platform starting about 5km north of the platform and ending about 2 km south of it.

A special purpose three-person team from POI was mobilized to Sakhalin Island from the Vladivostok based institute. Working from the Sakhalin Energy platform stand-by vessel Smit Sakhalin, this team deployed three mini-AUAR stations at nominal ranges of 2.5, 3.5 and 5 km from PA-A along a westward radial as shown in Figure 1. The maximum range was dictated by the safety and operational rules of the stand-by vessel that restrict it to sail no more than 5 km away from the platform.

The seismic vessel transited at about 1.5 km from the platform. This provided measurements of the propagated levels from the seismic array at ranges of about 1 km, 2 km and 3.5 km from the closest point of approach (CPA) as also shown in Figure 1. All distances were known precisely from GPS based documentation of the mini-AUAR drop points and the GPS logs of the seismic vessel which provide actual shot point coordinates and times. The sail-past with operating seismic source took place on 30 October 2009 between the hours of 10:00 and 10:40.

Figure 1. Layout of sail line and receivers


45

The data from the three mini-AUARs were downloaded by POI scientists upon return to Vladivostok and found to be of excellent quality. Taking advantage of a visit by Racca to the POI laboratories at the beginning of November, a blind cross-comparison of results from analysis of the acoustic data using POI’s and JASCO’s custom software was performed. Agreement of the metrics calculations between these unrelated processing codes is key to consistent data processing in the course of the Astokh 4D survey, when the POI software will likely be used in the analysis of the sound source verification (SSV) measurements from multiple receivers at fairly close range from the source whilst the JASCO software is slated to perform the real-time processing of the telemetred data from the Perimeter Monitoring Line instruments at the acoustics land camp on Sakhalin. In a comprehensive cross-check of the full data sets from all three stations the two processing codes were found to agree on the values of sound level metrics that included the per-pulse SEL and the 90% energy RMS to a very good precision, typically within 0.01 dB and never larger than 0.1 dB.

JASCO performed modelling of the point-source equivalent far field directional source levels from the PGS array and their propagation through the acoustic environment using the company’s proprietary AASM source model and MONM propagation model. The JASCO modelling approach has been discussed and characterized extensively in previous meetings of the Seismic Survey Task Force and the corresponding reports. For this validation study the same acoustic environment parameters were used as in all modelling work performed in the past to derive estimated propagation fronts from the array to be used in the Astokh 4D survey. It must be stressed, however, that the modelling or measurement absolute numbers presented here cannot in any way be compared to those for the Astokh survey or taken to be indicative of expected levels in that operation because of the different source array involved in this trial versus the source array that will be used during the Astokh 4D seismic survey. . The geo-acoustic parameters used in the modelling are listed for reference in Table 1 and the water sound velocity profile, obtained from typical CTD casts for the early part of the summer season, in Table 2.

Table 1 – Geo-acoustic parameters used in modelling Depth

(mbsf)

Density

(kg/m3)

P-wave

speed

(m/s)

P-wave

attenuation

(dB/λ)

S-wave

speed

(m/s)

S-wave

attenuation

(dB/λ)

0 1772 1652 0.14 150 13.6

500 1772 2152 0.14 150 13.6

>500 1772 2152 0.14 150 13.6

Table 2 - Water sound velocity profile used in modelling

Depth (m)

Sound Speed

in water

(m/s)

0.9 1469

2.5 1467

3.1 1466

5.1 1461

6.8 1456

8.0 1452

9.0 1448


46

10.2 1446

11.5 1444

32.0+ 1444

The model results were computed for every seismic shot position at each of the three receiver locations, at a depth just off the sea floor. The values of both the measured per-pulse SEL metric and its modelled estimate are plotted against the along-line offset from CPA in the figures that follow. Figure 2 shows the results for the receiver at 1 km off the sail line, Figure 3 for the receiver at 2 km, and Figure 4 for the receiver at 3.5 km.

Figure 2. Measured (individual dots) and modelled (continuous line) per-shot SEL metric in dB re µPa2 for a receiver 1 km off the sail line, plotted against along-line offset in m from CPA.


47

Figure 3. Measured (individual dots) and modelled (continuous line) per-shot SEL metric in dB re µPa2 for a receiver 2 km off the sail line, plotted against along-line offset in m from CPA.

Figure 4. Measured (individual dots) and modelled (continuous line) per-shot SEL metric in dB re µPa2 for a receiver 3.5 km off the sail line, plotted against along-line offset in m from CPA.

The agreement between model and measurement is reasonable considering that no special effort was made to tune the water SVP (or possibly any of the geo-acoustic parameters although this adjustment would be less justifiable in terms of consistency with past modelling and validations) to improve the quality of the fit. The model tends to overestimate the received levels in the broadside lobe direction, providing a precautionary result. Along the approach run (left of the CPA in the plots) the model matches well the trend of the


48

measured data at 1 km and 2 km from the sail line even though it fails to fully account for smaller scale oscillations that may depend on fine conformation of the seafloor topography not resolved in the modelling bathymetry database. The sudden drop in measured levels along the departing run is not fully recognized by the model, resulting in overestimation; again a local smaller-scale conformation of the sea floor (which causes an asymmetry in what should be an essentially mirror image pattern) may be the cause of the discrepancy. The only deviations that may raise concern in terms of precaution are found along the approach run for the 2 km and 3.5 km receivers. Here the received levels rise steeply above the model predictions starting at an along-line offset approximately equal to the receiver distance from the line and remain higher up to about 1.5 km before CPA, exceeding by as much as +3 dB the model estimates which follow a gradual, more conventional rising trend. No explanation for this phenomenon has been advanced at this time. In terms of the estimation of sound levels fronts, the observed broadening of the directional pattern should still not result in propagation farther inshore than predicted based on the broadside estimate. Further attention will nonetheless be dedicated to the potential repercussion of such discrepancies when carrying out the modelling of various propagation scenarios that is to be done in the lead-in to the Astokh survey to yield operational ensonification charts.