Waves in-ice observations made during the SIPEX II voyage of the Aurora Australis, 2012

Antarctic sea-ice is highly influenced by the dynamic nature of the Southern Ocean. Ocean waves can propagate from tens to hundreds of kilometres into sea-ice, leaving behind a wake of broken ice sheets. As global climate change intensifies, storm intensity will increase in the Southern Ocean. Increased storm intensity will bring stronger winds and bigger waves, which has the potential for waves to travel deeper into the ice pack and increase the likelihood that ice floes break apart. To enhance our understanding of this system, our aim during SIPEXII was to improve on the scarce Antarctic waves-in-ice dataset by collecting a set of wave observations in the MIZ.

In order to achieve this, we designed and produced eight custom made wave sensors. The sensors were deployed in the Antarctic marginal ice zone along a transect line perpendicular to the ice edge and spread over approximately 200 km. Every three hours, the sensors simultaneously woke and recorded their location and a burst of wave acceleration data. Each sensor performed on-board data quality control and spectral analysis before returning the wave spectrum via satellite. The sensors were powered via lithium batteries and had enough battery power to last a minimum of 6 weeks.

This project involved collaboration between the Australian Antarctic Division (AAD) and the NZ's National Institute for Water and Atmospheric Research (NIWA). The work was funded by a New Zealand Foundation of Research Science and Technology Postdoctoral award to A.L.K.; the Marsden Fund Council, administered by the Royal Society of New Zealand; NIWA, through core funding under the National Climate Centre Climate Systems programme; the Antarctic Climate and Ecosystems Cooperative Research Centre; and Australian Antarctic Science project 4073.

Instruments were designed and built by Inprod PTY LTD. Below is a summary of the design and hardware:
Accelerometer: Kistler ServoK-Beam accelerometer. Model 8330B3.
IMU: Razor IMU (3 axis acceleration, 3 axis magnetometer and 3 axis gyro)
ADC: TI ADS1247 Analog-to- Digital converter
CPU (main): Arduino Mega 3.3V
CPU (maths): BeagleBone from BeagleBoard.org who use Texas Instruments (TI) ARM processors
GPS: Skytraq Venus634FLP
Temperature readings: SHT15 from SparkFun
Transmission: Iridium 9602
Battery: Lithium batteries (enough to survive a minimum of 6 weeks)
Inner housing: Explorer 1908OE
Outer housing: The case is fitted in a fork lift tyre ( .53 m diameter and .165 m height) with an inner tube to enable floating.
Aerial housing: The aerial is housed in a plastic spherical container on top of a .5 m tube attached to the tyre.
Feet: 3 screws stick out of the bottom to create friction with the ice.

Onboard processing:
Every 3 hours, the instruments wake and record wave accelerations for 35 mins.
An initial low pass analogue filter is used.
We over sample at 64 Hz and decimate down to 2 Hz. Downsampling from 64 Hz to 2 Hz is achieved through a multistage decimation of 8 followed by 4, to achieve a total decimation of 32. Prior to each downsampling stage, a second order lowpass Butterworth filter is applied to remove all components above the nyquist frequency. We first apply the Butterworth filter with a cut off of 1 Hz and sample at 8 Hz and secondly with a cut off of 0.5 Hz and sample at 2 Hz. A high-pass filter was then applied and the acceleration double-integrated to provide displacement. Welch's method, using a 10% cosine window and de-trending on four segments with 50% overlap, was applied to estimate the power spectral density.

Sample frequency: 2 Hz
Sample duration - raw: 2048 sec
Sample duration - fft: 1280 sec
No. of discrete bins of fft: 512
No. of segments: 4

Below is a detailed description of each line of the raw output.

Header info
L1: Longitude (decimal degrees)
L2: File name of attachment emailed via Iridium
L3: Temperature inside the box (degrees Celsius)
L4: Sensor identification number
L5: Time wave record starts (24 hr format HHMMSS)
L6: Date of wave record (yyyy-mm-dd)
L7: Current voltage
L8: Elevation (cm)
L9: Latitude (decimal degrees)

Wave spectrum
L10-L64: The power spectral density for wave period bins (secs) centred on
[24.38 19.69 18.96 18.28 17.65 17.06 16.51 16.00 15.51 15.05 14.62 14.22 13.83 13.47 3.12 12.80 12.48 12.19 11.90 11.63 11.37 11.13 10.89 10.66 10.44 10.24 10.03 9.84 9.66 9.48 9.30 9.14 8.98 8.82 8.67 8.53 8.39 8.25 8.12 8.00 7.64 7.31 7.01 6.73 6.48 6.24 6.02 5.81 5.50 5.22 4.97 4.74 4.53 4.33 4.16]

Spectral moments
L65-L70: m-2 - m4

Quality control
L71: mean roll (degrees)
L72: mean pitch (degrees)
L73: mean yaw (degrees)
L74: wave direction (degrees)
L75: directional spread (degrees)
L76: ratio term to evaluate quality of wave direction approximation (should be close to 1)
L77: standard deviation of acceleration (m/s2)
L78: standard deviation of gyro x axis (radians/s)
L79: standard deviation of gyro y axis (radians/s)
L80: standard deviation of gyro z axis (radians/s)
L81: standard deviation of yaw (radians)
L82: Accelerometer quality flag. 0 = good, 1 = accelerometer bad, 2 = accelerometer and imu bad
L83: IMU quality flag. 0 = good, 1 = pitch/roll bad, 2 = yaw bad, 3 = both bad
L84: mean acceleration removed (m/s2)
L85: no. of flat spots in raw acceleration data
L86: the maximum number of consecutive flat spots
L87: no. of spikes (data point greater than 6 standard deviations of data set)
L88: the maximum number of consecutive spikes
L89: Quality flag indicating whether the total power in the time domain and frequency domain are equal. 0 = difference less than 0.01, 1 = difference greater than 0.01.

Deployment method:
The Helicopter Resources team, lead by Leigh Hornsby, and the Aurora crew, lead by Murray Doyle, were a crucial component to the success of the deployment. The first three sensors were deployed via helicopter. The sensor was lowered via a rope onto floe whilst the helicopter hovering about 2 m above floe. Due to weather constraints, the remaining five were deployed via crane. The ship pulled up beside a chosen floe and the sensors were lowered onto it via crane. Once deployed, the ship slowly moved forward until the floe was clear of the turbulence generated by the ship. Both the helicopter and crane deployment methods proved to be successful. See /Waves/Wave Observations/wiios_deployment.pdf for more details on the deployment procedure.

Approximate floe dimensions in metres based on the images in
/Waves/Ice Observations/Ice_floe/Sensor ID):

Sensor ID,Freeboard,Width,Length
1,0.15,28,28
2,0.33,10,12.5
3,0.15,10,15
4,.1,12,12
5,0.15,10,16.5
6,1,10,16.5
7,0.5,11.5,24
8,1,28.5,9

Ice observations:
A collection of images and movies of the ice conditions are provided in Waves/Ice Observations. The folders include:
Aerial: This folder contains aerial images taken with a gopro hero 2 fixed to the underside of the helicopter. Note that the date stamp on the GoPro is incorrect. Use the following for calibration:
20121022 13:52:00 - GPS - Australian eastern standard (no daylight savings)
20110707 14:00:07 - GoPRO
Ice floe: Images of floes the sensors were deployed on.
Ship: Images of the ice conditions taken from the ship.

/Waves/Wave Observations/raw/sensorID_yyyy-mm-dd_hhmmss.raw
Maps and shapefiles.zip - contains an ArcGIS map and shapefiles containing track data.
KML.zip - contains KML files (point data) showing point-in-time snapshots of the buoy positions.
Raw_NIWA_data.zip - contains the raw data files.

南极海冰受南大洋动力特性的强烈影响。海浪可在海冰中传播数十至数百公里，留下破碎冰盖的尾迹。随着全球气候变化加剧，南大洋的风暴强度将有所提升。风暴强度增强会带来更强的风力与更大的海浪，使得海浪能够更深入海冰区，并提升浮冰碎裂的概率。为加深对这一系统的认知，我们在SIPEXII科考期间的目标是，通过在海冰边缘区（Marginal Ice Zone, MIZ）采集海浪观测数据，完善当前稀缺的南极海冰内海浪数据集。

为达成这一目标，我们设计并制造了8台定制化海浪传感器。传感器被部署在南极海冰边缘区，沿垂直于冰缘的断面布设，覆盖范围约200公里。每3小时，传感器会同步唤醒，记录自身位置与一段海浪加速度数据。每台传感器均会先完成板载数据质量控制与频谱分析，随后通过卫星传回海浪功率谱。传感器由锂电池供电，续航时长至少可达6周。

本项目由澳大利亚南极局（Australian Antarctic Division, AAD）与新西兰国家水与大气研究所（National Institute of Water and Atmospheric Research, NIWA）合作开展。本项目的资金支持包括：授予A.L.K.的新西兰研究科学与技术基金会博士后奖学金；由新西兰皇家学会管理的马斯登基金委员会；新西兰国家水与大气研究所通过国家气候中心气候系统项目提供的核心经费；南极气候与生态系统合作研究中心；以及澳大利亚南极科学项目4073。

仪器由Inprod PTY LTD设计并制造，以下为该仪器的设计与硬件概要：
加速度计：Kistler ServoK-Beam加速度计，型号8330B3。
惯性测量单元（Inertial Measurement Unit, IMU）：Razor IMU，包含3轴加速度计、3轴磁强计与3轴陀螺仪。
模数转换器（Analog-to-Digital Converter, ADC）：德州仪器（Texas Instruments, TI）ADS1247模数转换器。
主中央处理器（Central Processing Unit, CPU）：Arduino Mega 3.3V。
数学运算CPU：BeagleBoard.org出品的BeagleBone，搭载德州仪器（Texas Instruments, TI）ARM处理器。
全球定位系统（Global Positioning System, GPS）：Skytraq Venus634FLP。
温度采集模块：SparkFun出品的SHT15。
通信模块：铱星9602（Iridium 9602）。
供电电池：锂电池，续航时长至少可达6周。
内部壳体：Explorer 1908OE。
外部壳体：安装于直径0.53米、高度0.165米的叉车轮胎内，搭配内胎以实现漂浮功能。
天线壳体：天线收纳于附着在轮胎上的0.5米长管材顶端的塑料球形容器内。
支撑脚：底部伸出3颗螺钉，用于与冰面形成摩擦力。

板载处理流程：
每3小时，仪器唤醒并记录35分钟的海浪加速度数据。首先使用初始低通模拟滤波器进行预处理。我们以64Hz的采样率进行过采样，随后将采样率降为2Hz。从64Hz到2Hz的降采样通过两级抽取实现：先抽取8倍，再抽取4倍，总抽取倍数为32。在每一级降采样前，均会应用二阶巴特沃斯低通滤波器，以滤除高于奈奎斯特频率的信号分量。首先将滤波器截止频率设为1Hz，采样率降至8Hz；随后将截止频率设为0.5Hz，采样率降至2Hz。随后应用高通滤波器，并对加速度进行两次积分以得到位移量。采用韦尔奇（Welch）法估算功率谱密度：使用10%余弦窗，将数据分为4段且段间重叠率为50%并进行去趋势处理。

采样频率：2Hz。
原始数据采样时长：2048秒。
FFT处理采样时长：1280秒。
FFT离散频段数：512。
分段数：4。

以下为原始输出每行数据的详细说明。

头部信息：
L1：经度（十进制度数）
L2：通过铱星邮件发送的附件文件名
L3：箱内温度（摄氏度）
L4：传感器标识号
L5：海浪记录开始时间（24小时制，格式为HHMMSS）
L6：海浪记录日期（格式为yyyy-mm-dd）
L7：当前电压
L8：海拔高度（厘米）
L9：纬度（十进制度数）

海浪频谱：
L10至L64：对应以如下波周期区间（单位：秒）为中心的功率谱密度值：[24.38, 19.69, 18.96, 18.28, 17.65, 17.06, 16.51, 16.00, 15.51, 15.05, 14.62, 14.22, 13.83, 13.47, 3.12, 12.80, 12.48, 12.19, 11.90, 11.63, 11.37, 11.13, 10.89, 10.66, 10.44, 10.24, 10.03, 9.84, 9.66, 9.48, 9.30, 9.14, 8.98, 8.82, 8.67, 8.53, 8.39, 8.25, 8.12, 8.00, 7.64, 7.31, 7.01, 6.73, 6.48, 6.24, 6.02, 5.81, 5.50, 5.22, 4.97, 4.74, 4.53, 4.33, 4.16]

谱矩：
L65至L70：对应谱矩m₋₂至m₄。

质量控制：
L71：平均横滚角（单位：度）
L72：平均俯仰角（单位：度）
L73：平均偏航角（单位：度）
L74：海浪传播方向（单位：度）
L75：方向弥散度（单位：度）
L76：用于评估海浪方向近似质量的比值项（理论值应接近1）
L77：加速度标准差（单位：m/s²）
L78：陀螺仪X轴标准差（单位：rad/s）
L79：陀螺仪Y轴标准差（单位：rad/s）
L80：陀螺仪Z轴标准差（单位：rad/s）
L81：偏航角标准差（单位：rad）
L82：加速度计质量标记：0=正常，1=加速度计异常，2=加速度计与惯性测量单元均异常
L83：惯性测量单元质量标记：0=正常，1=俯仰/横滚异常，2=偏航异常，3=所有参数异常
L84：已去除的平均加速度（单位：m/s²）
L85：原始加速度数据中的平值点数量
L86：连续平值点的最大数量
L87：异常尖峰数量（数据点超出数据集6倍标准差范围）
L88：连续异常尖峰的最大数量
L89：质量标记，用于判断时域与频域总功率是否一致：0=差值小于0.01，1=差值大于0.01。

部署方式：
部署工作由Leigh Hornsby带领的直升机资源团队与Murray Doyle带领的“极光号”（Aurora）船员共同完成，二者为本项目部署工作成功的核心保障。前3台传感器通过直升机部署：直升机在浮冰上方约2米处悬停，通过绳索将传感器下放至浮冰表面。受天气条件限制，剩余5台传感器通过起重机部署：科考船停靠至选定浮冰旁，通过起重机将传感器下放至浮冰表面。部署完成后，科考船缓慢前行，直至浮冰脱离船体产生的尾流湍流。直升机与起重机两种部署方式均取得成功。部署流程的详细信息可参考/Waves/Wave Observations/wiios_deployment.pdf文件。

基于/Waves/Ice Observations/Ice_floe/文件夹内的影像，各浮冰的近似尺寸（单位：米，格式为Sensor ID）：
传感器ID,干舷高度,宽度,长度
1,0.15,28,28
2,0.33,10,12.5
3,0.15,10,15
4,0.1,12,12
5,0.15,10,16.5
6,1,10,16.5
7,0.5,11.5,24
8,1,28.5,9

冰况观测：
冰况的影像与视频合集存储于Waves/Ice Observations文件夹中，该文件夹包含以下子文件夹：
航拍子文件夹：该文件夹包含安装于直升机底部的GoPro Hero 2拍摄的航拍影像。请注意GoPro的时间戳有误，校准参照如下：
2012年10月22日13:52:00，GPS时间为澳大利亚东部标准时间（无夏令时）
2011年7月7日14:00:07，GoPro时间
浮冰子文件夹：包含部署传感器的浮冰影像。
科考船子文件夹：包含从科考船拍摄的冰况影像。

原始数据文件路径：/Waves/Wave Observations/raw/sensorID_yyyy-mm-dd_hhmmss.raw
Maps and shapefiles.zip：包含ArcGIS地图与航迹数据的矢量文件（shapefile）。
KML.zip：包含KML文件，用于展示浮标位置的瞬时快照点数据。
Raw_NIWA_data.zip：包含原始数据文件。