Ocean current profiles were collected during the RV Polarstern PS131 cruise, ATWAICE, from drifting ice floes using an RDI 300 kHz Workhorse acoustic Doppler current profiler (ADCP, serial number 15085). Processed and quality-controlled measurements are provided here as 1-minute time averages in the upper 80 meters and cover ten drift stations, each lasting 2 to 23 hours, from 13 July to 1 August 2022. The raw data in instrument native files are also available from PANGAEA, see below. The aim of this expedition was to investigate sea ice summer melt processes, focusing on the contribution of the Atlantic water inflow into the region. These ocean current measurements were recorded to complement simultaneous ocean microstructure profiles by a microstructure profiler (MSS). The ADCP was installed in a standard in-line frame attached to a pole suspended through a hole in the ice. The pole was secured at the ice surface using a set of buoyancy "boxes." In total, ten short deployments were made during the visits to the experiment floes (three visits each to floes North, Middle, and South, and a 24-hour process station). Sampling was started and stopped on board the ship pre-deployment and post-recovery. During the deployments, beam 3 of the instrument was marked and aligned with the ship's bow direction to allow cross-referencing with the ship's navigation data. Except for the second deployment (Floe South, Visit 1), the instrument recorded in beam coordinates with 1-second pings in broadband mode (high resolution) and recorded all profiles with 2-meter vertical bins. The vertical range with good data was typically 80 meters. In the Floe South, Visit 1 deployment, the measurements were processed on-board the instrument in Earth coordinates as 1-minute averages. For the other deployments, the ship's heading was used to calculate an offset correction for the ADCP's heading, which was used to convert measurements to Earth coordinates. The ship's heading was interpolated to ADCP time, and then an offset correction was obtained as the median value of the differences between the ship's heading and the ADCP heading. The offset, rounded to 1 degree, was added to the ADCP heading. In full resolution (1 Hz sampling), beam profiles are screened using a correlation threshold of 60 counts and an echo intensity difference threshold of 15 counts for fish detection. This low value was necessary to detect and exclude data points corrupted by the nearby microstructure profiling instrument. Beam-wise velocities are converted to Earth coordinates using the instrument transformation matrix and the corrected heading. Additionally, magnitudes of the horizontal velocity components exceeding 1 m/s, vertical velocities exceeding 0.3 m/s, and error velocities exceeding 0.1 m/s are flagged as bad. Spikes and outliers are further flagged using the 1-second time series of velocity components and three successive applications of a Hampel filter (180 seconds and 3 standard deviations; 60 seconds and 2.5 standard deviations; 30 seconds and 2 standard deviations). The 1-second profiles in Earth coordinates are then averaged to 1-minute intervals (60 pings). Flagged records are used to calculate the fraction of good data points within the 1-minute averaging window. If the fraction of good data points in the 1-minute windows is less than 50%, the data point is considered bad. Additionally, points where the error velocity exceeds 2.5 standard deviations above the smoothed error velocity (smoothing windows: 6 meters vertical and 5 minutes in time) are flagged as bad. Finally, gaps less than 10 minutes are linearly interpolated, and short segments (less than 2 bins in vertical and 10 minutes in time) are removed. As the instrument drifted with the ice, the observed ocean current is relative to the ice. The ice-relative currents are converted to absolute currents by adding the ice drift velocity interpolated to the ADCP times. Eastward and northward ice velocity components are obtained from GPS measurements deployed on the ice, except for Floe South, Visit 3, when on-ice GPS measurements were not available. The 1-second GPS position data are bin-averaged in 1-minute intervals and converted to eastward and northward distance in local projection. Ice velocity is then obtained by centered first differencing using averaged records (10-point moving average). For Floe South, Visit 3, we used the position from the ship's navigation system, obtained as 1-minute averages. The vertical coordinate is provided as depth from the sea surface, derived from the mid-distance of each measurement cell from transducer and the median value of the pressure record when the ADCP was in water. Note that the instrument was deployed under sea ice, so the distance to the ice-ocean boundary is less than the provided water depth. To account for this, we supply ice draft measurements from the hole from which the microstructure sonde was deployed, which is in the vicinity of the ADCP deployment hole. Due to expected spatial variability of ice thickness, we estimate an uncertainty of approximately 0.2 meters for the ice draft at the ADCP hole. The raw data are available from https://doi.org/10.1594/PANGAEA.978463 .

The data are organized in netCDF files, one for each deployment. These files contain additional metadata, and are named corresponding to the respective event (with underscores replacing dashes), for example RDI300_OnIce_PS131_49_1.nc. We added one text file (ncdump_93_1.txt), containing the metadata of the RDI300_OnIce_PS131_93_1.nc file, as example for the users.The authors are grateful to the captain, crew, and technical/scientific staff of the expedition PS131 onboard RV Polarstern.