Passive acoustic recordings from sonobuoys deployed during the Antarctic Circumnavigation Expedition 2017

This dataset contains acoustic recordings from Directional Frequency Analysis and Recording (DIFAR) sonobuoys that were deployed from 22 January – 18 March 2017 during the Antarctic Circumnavigation Expedition. During the 52 days at sea 301 sonobuoys were deployed yielding 492 hours of acoustic recordings. Two models of sonobuoys were used during the voyage: 1 was a bespoke reusable DIFAR buoy based on a sensor and radio from an AN/SSQ-53F sonobuoy (Ultra Electronics: SonobuoyTechSystems, USA) and 300 were re-lifed AN/SSQ-955-HIDAR (deployed in DIFAR compatibility mode; Ultra Electronics Sonar Systems, UK). Two dedicated acousticians monitored round-the-clock for blue, fin, sperm, humpback, minke, killer, and sei whales, and crabeater, leopard, Ross, and weddell seals and in all weather conditions.

During ACE, we conducted a broad-scale, passing-mode passive acoustic survey for marine mammals in the Southern Ocean. Listening stations were conducted by deploying SSQ955 HIDAR sonobuoys in DIFAR (standard) mode to monitor for and measure bearings to vocalising whales while the ship was underway (Gedamke and Robinson 2010, Miller et al. 2015). During transit, listening stations were conducted every 30-60 nmi in water depths greater than 200 m. Sonobuoys were occasionally deployed with spacing less than 30 nmi in an attempt to more precisely determine spatial extent and vocal characteristics of calls that were believed to be coming from animals relatively close to the ship’s track. During terrestrial stopovers and marine science stations, sonobuoys were deployed approximately 2-4 nmi prior to stopping in order to attempt to monitor them for the full six-hour duration of their operational life. This distance ensured good radio signal while minimising acoustic interference from the vessel. The sampling regime was chosen to balance spatial resolution with the finite number of sonobuoys available for this study.

Instrumentation, software, and data collection

At each listening station, a HIDAR sonobuoy was deployed with the hydrophone set to a depth of 140 m. Sonobuoys transmitted underwater acoustic signals from the hydrophone and directional sensors back to the ship via a VHF radio transmitter. Radio signals from the sonobuoy were received using an omnidirectional VHF antenna (PCTel Inc. MFB1443; 3 dB gain tuned to 144 MHz centre frequency) and a Yagi antenna (Broadband Propagation Pty Ltd, Sydney Australia) mounted on the top of the helicopter control room at a height of 23.0 m. The antennas were each directly connected to a WiNRADiO G39WSBe sonobuoy receiver via low-loss LMR400 coaxial cable.
The radio reception range on the Yagi antenna was similar to previous Antarctic voyages, and was adequate for monitoring and localisation typically out to a range of 12-14 nmi, provided that the direction to the sonobuoy was close (i.e. within around 30o) to the main axis of the antenna. The radio reception on the omnidirectional antenna typically provided 5-8 nmi of omnidirectional reception from sonobuoys. At transit speed (14-15 knots), the Yagi antenna provided about 55 minutes of acoustic recording time per sonobuoy Using both antennas together were able obtain radio reception for up to six hours (i.e. the maximum life of a 955 sonobuoy) when sonobuoys were deployed within 5 nmi of a marine science station.
Received signals were digitised via the instrument inputs of a Fireface UFX sound board (RME Fireface; RME Inc.). Digitised signals were recorded on a personal computer as 48 kHz 24-bit WAV audio files using the software program PAMGuard (Gillespie et al. 2008). Data from both the Yagi and Omnidirectional antenna were recorded simultaneously as WAV audio channels 0 (left) and 1 (right). Each recorded WAV file therefore contains a substantial amount of duplication since both antennas and receivers were usually receiving the same signals from the same sonobuoy.

Directional calibration

The magnetic compass in each sonobuoy was calibrated/validated upon deployment as described by Miller et al. (2015, 2016). Calibration procedure involved measuring the mean bearing error and standard deviation of errors between the GPS-derived bearing from the sonobuoy to the ship and the magnetic bearing to the ship noise detected by the sonobuoy. 15-20 bearings were used for each calibration as the ship steamed directly away from the deployment location.

Intensity calibration

Obtaining calibrated intensity measurements from sonobuoys not only requires knowledge of the sensitivity of the hydrophone, but also the calibration parameters of the radio transmitter and radio receiver. Throughout the voyage, a hydrophone sensitivity of -122 dB re 1 V/micro Pa was applied to recordings via the Hydrophone Array Manager in PAMGuard. This value is defined in the DIFAR specification as the reference intensity at 100 Hz that will generate a frequency deviation of 25 kHz (Maranda 2001), thus the specification combines the hydrophone sensitivity and transmitter calibration. In line with manufacturers specifications, the WiNRADiO G39 WSB had a measured voltage response of 1 V-peak–peak (approximately -3 dB) at 25 kHz frequency deviation (Miller et al. 2014), and this was subtracted from the hydrophone sensitivity to yield an total combined factor of 125 dB re 1 V/µPa. The gain of the instrument input on the Fireface UFX was set to 20 dB, yielding a maximum voltage input voltage range of 8.36 V peak–peak. These calibration settings, along with the shaped filter response provided by Greene et al. (2004) make it possible to obtain calibrated pressure amplitude from the recorded WAV audio files.

Sonobuoy deployment metadata

The PAMGuard DIFAR Module (Miller et al. 2016) was used to record the sonobuoy deployment metadata such as location, sonobuoy deployment number, and audio channel in the HydrophoneStreamers table of the PAMGuard database (PamguardBlueWhale-2015-02-03.mdb). A written sonobuoy deployment log (Sonobuoy deployment logbook - 2015 Tangaroa.pdf) was also kept during the voyage, and this includes additional notes and additional information not included in the PAMGuard Database such as sonobuoy type, and sonobuoy end-time.

Real-time monitoring and analysis (Acoustic event log)

Aural and visual monitoring of audio and spectrograms from each sonobuoy was conducted using PAMGuard for at least an hour at each listening station. Two different spectrograms were typically viewed, one for low-frequency sounds with the following parameters: 250 Hz sample rate; 256 sample FFT; 32 sample advance between time slices. The other spectrogram was used to view mid-frequency sounds with the following parameters: 8000 Hz sample rate; 1024 sample FFT; 128 sample advance between time slices. Monitoring was conducted in real-time as data were being acquired, and the intensity scale of the spectrogram was adjusted by the operator to suit the ambient noise conditions.
When detections from marine mammals, ice, and other sources were detected, they were classified manually, and their time and frequency bounds marked on the spectrogram. The PAMGuard DIFAR module (Miller, Calderan, et al. 2016) was then used to measure the direction of arrival and intensity of suitable calls such as tonal, frequency-modulated, and pulsed calls of baleen whales, whistles and trills from pinniped, and some whistles from toothed whales. Echolocation clicks from sperm whales (Physeter macrocephalus) were noted in the PAMGuard UserInput (free form notes stored in the PAMGuard Sqlite database), but could not be localised with the DIFAR module due to limitations inherent in directional sensors in the sonobuoy. Detection, bearing, and intensity measurements were saved both within a PAMGuard binary file and within the DIFAR_Localisation table of the PAMGuard database. In addition to PAMGuard binary files and audio files, the PAMGuard settings and metadata were saved to the PAMGuard Sqlite database.

During Leg 3, some experimental trials were conducted with sonobuoys deployed in pairs with one hydrophone set to a depth of 140m and the other set to either 300m or 30m (the other two depth options available in the sonobuoy settings). The aim of these experiments was to investigate any differences with received level and the depth of the receiver. Recordings collected over a range of received levels as the vessel headed away from vocalising whales can also allow estimates of bearing accuracy for weak calls (by comparing bearings to the same call from different buoys) and the relative detection probability for calls under different noise conditions (by using the signals from each buoy in a similar way to independent observer experiments).

References

Greene, C.R.J. et al., 2004. Directional frequency and recording ( DIFAR ) sensors in seafloor recorders to locate calling bowhead whales during their fall migration. Journal of the Acoustical Society of America, 116(2), pp.799–813.

Maranda, B.H., 2001. Calibration Factors for DIFAR Processing,

Miller, B.S. et al., 2014. Accuracy and precision of DIFAR localisation systems: Calibrations and comparative measurements from three SORP voyages. Submitted to the Scientific Committee 65b of the International Whaling Commission, Bled, Slovenia. SC/65b/SH08, p.14.

Miller, B.S. et al., 2016. Software for real-time localization of baleen whale calls using directional sonobuoys: A case study on Antarctic blue whales. The Journal of the Acoustical Society of America, 139(3), p.EL83-EL89. Available at: http://scitation.aip.org/content/asa/journal/jasa/139/3/10.1121/1.4943627.

Miller, B.S. et al., 2015. Validating the reliability of passive acoustic localisation: a novel method for encountering rare and remote Antarctic blue whales. Endangered Species Research, 26(3), pp.257–269. Available at: http://www.int-res.com/abstracts/esr/v26/n3/p257-269/.

本数据集包含2017年1月22日至3月18日南极环极航行探险（Antarctic Circumnavigation Expedition, ACE）期间部署的定向频率分析与记录（Directional Frequency Analysis and Recording, DIFAR）型声呐浮标的声学录音数据。
本次海上航行共计52天，共部署301台声呐浮标，累计获得492小时声学录音。本次航次使用两款声呐浮标：1台为定制化可重复使用DIFAR浮标，其传感器与无线电模块源自AN/SSQ-53F型声呐浮标（美国Ultra Electronics: SonobuoyTechSystems公司）；剩余300台为翻新AN/SSQ-955-HIDAR型浮标（以DIFAR兼容模式部署；英国Ultra Electronics Sonar Systems公司）。两名专职声学研究员在全天气况下开展24小时值守监测，目标物种包括蓝鲸、长须鲸、抹香鲸、座头鲸、小须鲸、虎鲸、塞鲸，以及食蟹海豹、豹海豹、罗斯海豹和威德尔海豹。

在南极环极航行探险期间，团队于南大洋开展了大规模过境式海洋哺乳动物被动声学调查。航行过程中，通过以DIFAR（标准）模式部署SSQ955 HIDAR型声呐浮标，实现对发声鲸类的监测与方位测量（Gedamke与Robinson 2010，Miller等人2015）。航行阶段，每间隔30~60海里且水深超过200米的区域即布设监听站位。为更精准确定疑似靠近航线的动物发声信号的空间范围与声学特征，团队偶尔会将浮标布设间距缩短至30海里以内。在陆地停靠与海洋科学站位作业阶段，团队会在停靠前约2~4海里处部署声呐浮标，以确保可在浮标完整6小时作业周期内完成监测。该距离既能保证良好的无线电信号传输，又可最大限度降低船舶自身带来的声学干扰。本次采样方案旨在平衡空间分辨率与本次研究可使用的声呐浮标数量之间的关系。

## 仪器设备、软件与数据采集
每个监听站位均部署1台HIDAR型声呐浮标，其水听器布设深度为140米。声呐浮标通过甚高频（Very High Frequency, VHF）无线电发射机，将水听器与定向传感器采集的水下声学信号回传至母船。浮标发射的无线电信号由两套天线接收：一套是全向VHF天线（PCTel公司 MFB1443；增益3 dB，中心频率调谐至144 MHz），另一套是八木天线（澳大利亚悉尼Broadband Propagation Pty Ltd公司），安装于直升机控制室顶部，高度23.0米。两套天线均通过低损耗LMR400同轴电缆直接连接至WiNRADiO G39WSBe型声呐浮标接收机。

八木天线的无线电接收范围与此前南极航次的结果相近，当浮标方位接近天线主轴线（即偏差约30°以内）时，可实现12~14海里范围内的监测与定位，满足常规需求。全向天线则可提供5~8海里的全向浮标信号接收范围。在航行速度为14~15节的情况下，单台八木天线可实现每台浮标约55分钟的声学录音时长。当声呐浮标部署于海洋科学站位5海里范围内时，同时使用两套天线可实现最长6小时的无线电信号接收（即955型浮标的最大作业寿命）。

接收信号通过Fireface UFX声卡（RME Fireface；RME公司）的仪器输入接口完成数字化。数字化后的信号通过PAMGuard软件（Gillespie等人2008）存储至个人计算机，格式为48 kHz采样率、24位精度的WAV音频文件。八木天线与全向天线采集的数据分别以WAV音频通道0（左声道）和1（右声道）的形式同步录制。由于两套天线与接收机通常会接收到来自同一浮标的相同信号，因此每个录制的WAV文件均包含大量重复数据。

## 定向校准
每台声呐浮标的磁罗盘均在部署时完成校准/验证，校准方法参照Miller等人（2015、2016）的描述。校准流程包括测量浮标基于GPS获取的至母船的方位，与浮标探测到的母船噪声的磁方位之间的平均方位误差及误差标准差。当母船直接驶离部署点位时，每次校准需采集15~20组方位数据。

## 强度校准
从声呐浮标获取校准后的强度测量值，不仅需要知晓水听器的灵敏度，还需掌握无线电发射机与接收机的校准参数。本次航次全程通过PAMGuard软件中的水听器阵列管理器（Hydrophone Array Manager），将水听器灵敏度设置为-122 dB re 1 V/μPa。根据DIFAR技术规范，该值对应100 Hz频率下可产生25 kHz频偏的参考声强（Maranda 2001），因此该参数同时整合了水听器灵敏度与发射机校准结果。根据厂商技术规范，WiNRADiO G39 WSB接收机在25 kHz频偏下的实测电压响应为1 V峰峰值（约-3 dB）（Miller等人2014），将其从水听器灵敏度中扣除后，得到总组合因子为125 dB re 1 V/μPa。Fireface UFX声卡的仪器输入增益设置为20 dB，对应的最大输入电压范围为8.36 V峰峰值。结合上述校准设置与Greene等人（2004）提出的整形滤波器响应，即可从录制的WAV音频文件中获取校准后的声压幅值。

## 声呐浮标部署元数据
使用PAMGuard DIFAR模块（Miller等人2016）记录声呐浮标部署元数据，包括部署位置、浮标编号与音频通道等信息，存储于PAMGuard数据库的HydrophoneStreamers表中（数据库文件：PamguardBlueWhale-2015-02-03.mdb）。航次期间同时保留纸质声呐浮标部署日志（文件：Sonobuoy deployment logbook - 2015 Tangaroa.pdf），其中包含PAMGuard数据库未收录的额外备注与信息，例如浮标型号与浮标终止工作时间。

## 实时监测与分析（声学事件日志）
每个监听站位均使用PAMGuard软件对每台浮标的音频与语谱图进行至少1小时的听觉与视觉监测。通常同时查看两种语谱图：一种用于分析低频声音，参数设置为：采样率250 Hz；快速傅里叶变换（Fast Fourier Transform, FFT）点数256；时间切片步长32个采样点。另一种用于分析中频声音，参数设置为：采样率8000 Hz；FFT点数1024；时间切片步长128个采样点。监测在数据采集过程中实时开展，操作人员会根据环境噪声条件调整语谱图的强度标尺。

当监测到海洋哺乳动物、海冰或其他声源的信号时，操作人员会手动进行分类，并在语谱图上标记信号的时间与频率范围。随后使用PAMGuard DIFAR模块（Miller、Calderan等人2016）测量符合要求的发声信号的到达方向与强度，这些信号包括须鲸的单音、调频脉冲声，鳍足类动物的哨声与颤音，以及部分齿鲸的哨声。抹香鲸（*Physeter macrocephalus*）的回声定位咔哒声被记录于PAMGuard用户输入模块（存储于PAMGuard Sqlite数据库的自由格式备注），但由于声呐浮标定向传感器的固有局限，无法通过DIFAR模块完成定位。检测结果、方位与强度测量值同时存储于PAMGuard二进制文件与PAMGuard数据库的DIFAR_Localisation表中。除PAMGuard二进制文件与音频文件外，PAMGuard的软件设置与元数据均存储于PAMGuard Sqlite数据库。

在第3航段，团队开展了成对部署声呐浮标的试验：其中一台浮标的水听器布设深度为140米，另一台则设置为300米或30米（声呐浮标设置中提供的另外两种深度选项）。本次试验旨在探究接收声级与接收机布设深度之间的关联。当母船驶离发声鲸类时，采集不同接收声级的录音数据，还可实现以下分析：通过对比不同浮标对同一信号的方位测量结果，估算弱信号的方位精度；通过以类似独立观察者实验的方式利用各浮标的信号，评估不同噪声条件下信号的相对检测概率。

## 参考文献
Greene, C.R.J. et al., 2004. Directional frequency and recording ( DIFAR ) sensors in seafloor recorders to locate calling bowhead whales during their fall migration. Journal of the Acoustical Society of America, 116(2), pp.799–813.
Maranda, B.H., 2001. Calibration Factors for DIFAR Processing,
Miller, B.S. et al., 2014. Accuracy and precision of DIFAR localisation systems: Calibrations and comparative measurements from three SORP voyages. Submitted to the Scientific Committee 65b of the International Whaling Commission, Bled, Slovenia. SC/65b/SH08, p.14.
Miller, B.S. et al., 2016. Software for real-time localization of baleen whale calls using directional sonobuoys: A case study on Antarctic blue whales. The Journal of the Acoustical Society of America, 139(3), p.EL83-EL89. Available at: http://scitation.aip.org/content/asa/journal/jasa/139/3/10.1121/1.4943627.
Miller, B.S. et al., 2015. Validating the reliability of passive acoustic localisation: a novel method for encountering rare and remote Antarctic blue whales. Endangered Species Research, 26(3), pp.257–269. Available at: http://www.int-res.com/abstracts/esr/v26/n3/p257-269/.