Seawater temperature profiles from Expendable Bathythermograph (XBT) probe deployments during the Antarctic Circumnavigation Expedition (ACE).

***** Dataset abstract *****

This data set contains vertical seawater temperature profiles measured by Expendable Bathythermograph (XBT) probes that were deployed in the Southern Ocean during the Antarctic Circumnavigation Expedition (ACE) on board the R/V Akademik Tryoshnikov. 40 XBT probes were deployed during legs 2 and 3 of the expedition in the period 25th January, 2017 to 17th March, 2017. The XBT probes are manufactured and distributed by T.S.K./Sippican Tsurumi-Seiki Co. Ltd., Yokohama, Japan (http://www.tsk-jp.com) and are of the type T-07, which is rated at a ship speed of up to 15 knots. These probes have a measuring time of 123 seconds and maximum measurement depth of about 789 m. Probes were launched from a handheld device from the stern of the ship either on the port or starboard side while the ship was moving. The deck unit recorded the temperature and the time since the probe was launched. This time was then converted to depth using the known fall rate of the probe in seawater and the coefficients provided by the manufacturer (WMO standards; Hanawa et al., 1995). The profiles were corrected for known surface biases (Kizu and Hanawa, 2002; Uehara et al., 2008). We provide the raw data, the data produced by using the coefficients provided by the manufacturer, and a corrected version in which we apply an empirical correction based on a comparison with CTD data (Henry et al., 2019), where XBT profiles were launched alongside the CTD deployment. The data has been quality controlled by comparing it to a number of CTD profiles. Data is provided at full vertical resolution and a 1-m averaged resolution. In addition, we provide derived variables such as surface mixed layer depth (temperature threshold) estimates. We are grateful to the crew of the R/V Akademik Tryoshnikov and AARI for donating these probes to our project. Their use-by date had expired, however this was not seen as an issue. This data set provides insights into the hydrography of the Southern Ocean during one austral summer season and complements the CTD temperature profiles measured during ACE by filling in the gaps between CTD stations.

***** Original data collection *****

All deployed XBT probes are manufactured and distributed by T.S.K./Sippican Tsurumi-Seiki Co. Ltd., Yokohama, Japan (http://www.tsk-jp.com) and are of the type T-07 (http://www.tsk-jp.com/index.php?page=/product/detail/2/2). The crew of Akademik Tryoshnikov (AARI) donated these XBT probes. We are truly grateful to them for providing these probes. The use-by date of the probes had expired, however this was not seen as an issue and our quality control (see below) did not reveal any issues. After deployment the probe measured for 123 seconds every 0.05 seconds, i.e. a total of 2461 recordings. Using the standard (WMO) fall rate equation, this resulted in a maximum measurement depth of about 789 m. The probes are rated for a ship speed of up to 15 knots. The probes are rated for a temperature range between -2 degC and about 35 degC and the temperature accuracy is +/-0.2 degC. The accuracy of the calculated depth is about +/-25 m at 750 m (Hanawa et al., 1995).

The hand-held T.S.K. XBT/XCTD launcher system, which was connected to the probe with a copper wire, was connected to a deck unit recorder TS-MK150N (http://www.tsk-jp.com/index.php?page=/product/detail/27/2). Data was archived and post-processed on a desktop computer (Windows XP) using the MK-150 software.

Throughout legs 2 and 3, we complemented the CTD stations (Henry et al., 2019) with 40 Expendable Bathythermograph (XBT) probes of which 26 were launched on leg 2 and 14 on leg 3 (see ace_xbt_deployment_summary.csv). Each deployment was assigned an event number as part of ACE and these numbers fit in with the entire expedition. XBT probes were strategically launched when crossing fronts, CTD stations were cancelled, or when spacing between stations was large. The recorded temperature profile can be used to determine the depth of the mixed layer, and the layer thickness of water masses if they are identifiable from the temperature profile, e.g. the extent of the winter water layer. During leg 3, seven of the 14 profiles were recorded in polar waters; one profile was recorded when entering the polar waters in the west at the Polar Front (event 2280) and one when leaving the polar waters in the east at the Polar Front (event 3013). This change in water masses can be depicted from the vertical temperature profile. During leg 2, two probes (event 1110 and 1111) were launched at CTD station 36 together with the CTD cast number 13 (event 1096). One probe (1110) was launched with the downcast and one (1111) with the upcast. During leg 3, one probe was launched at CTD station 92 (cast 006, event 2652) right after the upcast for temperature sensor intercomparison. All XBTs (except for three) are associated with a surface sample of salinity (Haumann et al., 2020) and oxygen isotope (Haumann et al., 2019) from the underway line. Additional information on each XBT deployment can be found in the file “ace_xbt_deployment_summary.csv”.

XBT probes were left outside on the main deck at the stern for at least one day before launch to reduce an existing surface temperature bias. Nevertheless, some surface biases remained and were accounted for by post-processing the data (see below). This surface bias is a well-known feature of these probes and is most likely associated with the XBT probe adjusting to the temperature of the surrounding seawater as it enters the sea (referred to as "start-up transient"; Kizu and Hanawa, 2002). However, interference from the ship, waves and bubbles, or the unknown age of the probe could lead to potential issues near the surface. Therefore, near-surface data need to be treated with caution.

Prior to each launch the USB connection between the computer (Windows XP) in the CTD laboratory and the TS-MK 150N unit was checked. The MK-150 software was executed and checked for functioning, correctness of the method (T-7), and a new measurement was started. Probes were launched from the stern of the ship either on the port or starboard side depending on wave and wind direction. They were launched using the T.S.K. hand-held XBT/XCTD launcher system. The probes were loaded to the launcher and dropped at an arm-length distance from the railing. During the unspooling of the copper wire, which transmits the data from the probe to the launcher and then to the recording unit, the launcher was held at an approximate 45 degree angle with respect to the horizontal, facing to the water surface. After the wire had fully unspooled the wire was ripped apart. Subsequently, the raw data file (as .RAW files) stored by the MK-150 software was post-processed to an .XBT-file and to a 1-m binned file, and saved. Raw data files can be found in the folder ace_xbt_raw and are named in the format ace_xbt_YYYYMMDD_xxxx_uuuuuuuuuuuu.RAW, where YYYYMMDD is the date of deployment (UTC), xxxx is the ACE event number and uuuuuuuuuuuu is the XBT number assigned by the software.

Latitude and longitude were entered manually to the system and are the locations at which the deployment started. They are obtained from the operator recording the time of deployment (in UTC) and then getting the corresponding location data from the ship’s GPS. However, these locations were corrected with the post-cruise corrected GPS track (Thomas and Pina Estany, 2019) during the post-processing (see below). The folder into which the data were saved during the expedition was copied to the expedition data storage on an hourly basis.

***** Data processing *****

Since XBTs do not measure depth directly, the depth is inferred using the instrument dependent fall-rate equation (Hanawa et al., 1995). This conversion is performed by the software (MK-150) during the conversion of the '.RAW' files to the '.XBT' files. The software uses the World Meteorological Organization (WMO) standard coefficients for the TSK T-7 conversion from the WMO Code Table 1770 code number 222 (https://www.wmo.int/pages/prog/www/WMOCodes/WMO306_vI2/LatestVERSION/WMO306_vI2_CommonTable_en.pdf). These coefficients are a=6.691 (m s^-1) and b=-0.00225 (m s^-2). The corresponding fall-rate equation reads:

d(t) = a*t + b*t^2 = 6.691*t - 0.00225*t^2,

where d is the depth in meters as a function of time t in seconds. The constants a and b are derived from theoretical and empirical methods and are based on Hanawa et al. (1995). They are stored in the header of the files (see below).

After the cruise, the files were separated into three folders ('ace_xbt_raw', 'ace_xbt_wmo_hanawa95', 'ace_xbt_wmo_hanawa95_1m') and renamed to reflect their recording date (UTC; YYYYMMDD), event number (xxxx), and XBT number assigned by the software (‘uuuuuuuuuuuu’) for convenience and handling. The corresponding file names are ‘ace_xbt_YYYYMMDD_xxxx_uuuuuuuuuuuu’.

Since the longitude and latitude were entered manually by the operator during the expedition, they were not always fully accurate. Therefore, the latitude and longitude information contained in the files was replaced after the cruise by using the GPS record at one-second resolution (Thomas and Pina Estany, 2019) for the deployment time recorded by the deck unit.

A first visual inspection of the data confirmed the surface temperature bias associated with the start-up transient (see above; Kizu and Hanawa, 2002) in the approximately upper 4 m of the profiles, where the recorded temperature was adjusting to the surrounding seawater temperature. In order to address this issue, the .XBT-files and _1m.XBT-files were reprocessed using the recommendation by the Japan-Hawaii Monitoring Program (JAHMP; http://www.pol.gp.tohoku.ac.jp/~kizu/jahmp/notes-e.htm) who use a similar system (TSK T-7) to ours. Therefore, we first replace all temperature data from the surface to 19th temperature data entry of the .XBT-file with the 20th temperature data entry. Using the original fall-rate equation, the 20th temperature entry occurs at 6.02 m. Then, we recompute the _1m.XBT-file temperature using a linear interpolation (shown to reproduce the software derived estimates within the given precision). This procedure eliminated the effect of transient behaviour of the temperature sensor upon entry to the water and automatically reduced the effects associated with a possible interference from the ship, waves and bubbles.

One profile (event 1111) suffered from false temperature values at depth. We removed this obviously bad data by applying a filter to all profiles (.XBT-files only, not the .RAW files) that removed data rows with temperatures smaller than -3 degC from the file. We then recalculated the 1-m files (_1m.XBT). Both the full resolution and 1-m reprocessed files after these post-processing steps using the original fall rate equation are available in the data set in the folders ace_xbt_wmo_hanawa95_fullres and ace_xbt_wmo_hanawa95_1m, respectively.

Comparing these data to the quality-controlled CTD temperature profiles at the three locations (Henry et al., 2019) where they were launched right before or after the CTD cast, we found a mean bias of -0.167 degC (-0.133 degC, -0.143 degC, and -0.225 degC for XBT events 1110, 1111, and 2684, respectively) for the XBT temperature compared to the CTD temperature (XBT minus CTD data). This bias indicates that further corrections might be needed. A number of issues with XBT profiles have been identified previously and if not corrected, they affect estimates of the global ocean heat content changes (e.g. Kizu et al., 2011; Hamon et al., 2012; Cheng et al., 2014). However, as opposed to most global estimates, our XBT profiles exhibit a cold bias. Since biases could result from both a temperature or depth error, we explore both possibilities. We first test the corrections suggested by Kizu et al. (2011) and Cheng et al. (2014). None of these corrections reduced the biases. We then fit a linear regression through the XBT and CTD data (Figure 1), obtaining the following empirical correction equation: T_corr=T*1.028+0.197. After applying this correction, the mean bias is smaller than 0.001 degC (0.013 degC, 0.003 degC, -0.013 degC, for each of the above CTD stations, respectively). Following Hamon et al. (2012), we then compute the median depth bias. However, no significant depth bias is found when comparing the corrected temperature profiles to the CTD profiles. Therefore, we do not apply any further corrections. The empirically corrected data is available in the folders ace_xbt_corrected_fullres and ace_xbt_corrected_1m. It is left to the user whether to use the official fall rate equation data (“wmo_hanawa95”) or the empirically corrected data (“corrected”).

We provide additional derived data such as the surface temperature, upper 15 m averaged temperature, vertical mean, maximum, and minimum temperatures, etc. We also computed the mixed-layer depth (MLD) as the depth at which the temperature is 0.2 degC below the reference value closest to 10 m (de Boyer Montégut, 2004). These values are also listed in the appended profile figures (“ace_xbt_figures/ace_xbt_figures/ace_xbt_YYYYMMDD_xxxx_1m.pdf”) and provided in the file “ace_xbt_mld_tavg.csv”. We additionally, provide a merged product of these vertical statistics from both XBT and CTD data (Henry et al., 2019) in the file “ace_merged_ctd_xbt_mld_tsavg.csv”.

***** Quality checking *****

Using the standard error of the residuals from the regression line (see above), we estimate the uncertainty in the corrected XBT temperature data to be +/-0.075 degC. The file “ace_xbt_deployment_summary.csv” lists possible issues with the data in the “time_location_flag”. A flag of “0” indicates no known issues. A flag of “1” indicates potential issues in deployment or data that showed no clear effect on the results. A flag of “3” indicates possible issues with the time and location recorded.

The reported temperature accuracy of the sensor is +/-0.2 degC. The accuracy of the calculated depth is about +/-25 m at 750 m (Hanawa et al., 1995).

While a correction to available CTD data has been performed, the true quality of the resulting XBT data remains unknown. For a more insightful quality assessment, we would need more coinciding CTD profiles. It is left to the user, whether the empirically corrected or the post-processed WMO standard is being used. Both temperature data are plotted in the figures provided for each profile.

***** Standards *****

This data set uses WMO Code Table 1770 code number 222.

***** Dataset contents *****

Data:
- ace_xbt_raw/ace_xbt_YYYYMMDD_xxxx_uuuuuuuuuuuu.RAW, data file, comma-separated values
- ace_xbt_wmo_hanawa95_fullres/ace_xbt_YYYYMMDD_xxxx_uuuuuuuuuuuu.XBT, data file, comma-separated values
- ace_xbt_wmo_hanawa95_1m/ace_xbt_YYYYMMDD_xxxx_uuuuuuuuuuuu_1m.XBT, data file, comma-separated values
- ace_xbt_corrected_fullres/ace_xbt_YYYYMMDD_xxxx_uuuuuuuuuuuu.XBT, data file, comma-separated values
- ace_xbt_corrected_1m/ace_xbt_YYYYMMDD_xxxx_uuuuuuuuuuuu_1m.XBT, data file, comma-separated values

Auxiliary data:
- ace_xbt_mld_tavg.csv, data file, comma-separated values
- ace_merged_ctd_xbt_mld_tsavg.csv, data file, comma-separated values

Figures:
- figure1.pdf, metadata, portable document format
- ace_xbt_figures/ace_xbt_YYYYMMDD_xxxx_1m.pdf, metadata, portable document format

Metadata:
- ace_xbt_deployment_summary.csv, metadata, comma-separated values
- data_file_header.txt, metadata, text
- README.txt, metadata, text

***** Dataset contact *****

F. Alexander Haumann, Princeton University, USA; British Antarctic Survey, UK; ETH Zurich, Switzerland. ORCID: 0000-0002-8218-977X. Email: alexander.haumann@gmail.com
Jenny Thomas, Swiss Polar Institute. ORCID: 0000-0002-5986-7026. Email: jenny.thomas@epfl.ch, jen@falciot.net

***** Dataset license*****

This seawater temperature profile dataset from ACE is made available under the Creative Commons Attribution 4.0 International License (CC BY 4.0) whose full text can be found at https://creativecommons.org/licenses/by/4.0/

***** Dataset citation *****

Haumann, F. A., Thomas, J., Tsukernik, M. and Leonard K. (2020). Seawater temperature profiles from Expendable Bathythermograph (XBT) probe deployments during the Antarctic Circumnavigation Expedition (ACE). (Version 1.0) [Data set]. Zenodo. doi:10.5281/zenodo.3836648.