https://doi.pangaea.de/10.1594/PANGAEA.996908Ringel, Maximilian HaraldMaximilian HaraldRingel0000-0002-0600-135XInstitute of Environmental Physics, University of BremenSpreen, GunnarGunnarSpreen0000-0003-0165-8448Institute of Environmental Physics, University of BremenKrumpen, ThomasThomasKrumpen0000-0001-6234-8756Alfred Wegener Institute, Helmholtz Centre for Polar and Marine Research, BremerhavenNeudert, MaraMaraNeudertAlfred Wegener Institute, Helmholtz Centre for Polar and Marine Research, BremerhavenPANGAEA2026Arcticco-locatedCONTRASTS expeditionGEM-2magnaprobemelt pond depthoceanPS149RV PolarsternSea iceSea ice thicknesssnow depthArctic Amplification (AC3)2025-07-10T13:51:22/2025-08-27T16:42:06Bundled Publication of Datasets10.57738/BZPM_0808_2026https://doi.pangaea.de/10.1594/PANGAEA.996869https://doi.pangaea.de/10.1594/PANGAEA.99690710 datasetsapplication/zipCreative Commons Attribution 4.0 InternationalThis dataset contains measurements of snow/SSL (surface scattering layer) depth or melt pond depth co-located with total ice thickness estimates from repeated walks across sea ice floes (called long transects) conducted during the PS149 CONTRASTS expedition. During each ice station (i.e., floe visit), a long transect was traversed across the ice floe to obtain statistically representative distributions of snow/SSL and melt pond depth and ice thickness. Each transect was marked with flags at its corner points to ensure that a similar track was measured during repeated visits to each ice floe. A GPS-equipped MagnaProbe was used to measure snow/SSL and melt pond depth, while a Geophex Ltd. GEM-2-730 instrument was used to measure total sea ice thickness. Total sea ice thickness refers to the distance between the instrument and the ice–ocean interface. Accordingly, if present, snow/SSL or melt pond depth must be subtracted from this total thickness to obtain the actual sea ice thickness estimate.GEM instrument & data processing:
The same frequency set as in the MOSAiC campaign (doi:10.1594/PANGAEA.943666) was used for the GEM measurements: 1.525 kHz, 5.325 kHz, 18.325 kHz, 63.025 kHz, and 93.075 kHz. The ice thickness estimates are derived for all these frequencies from the raw GEM measurements (doi:10.1594/PANGAEA.996869) using an empirical calibration procedure. For this procedure, the GEM was positioned at known heights using the steps of a wooden ladder standing on a flat area of the sea ice floe, where sea ice thickness was measured at five locations below the instrument. Based on these calibration measurements, an exponential function is derived and applied to the transect data to retrieve the total ice thickness. For all calibrations, the conductivity of the sea ice plus snow layer was assumed to be 0.05 S m⁻¹, and the conductivity of seawater was set to 2.55 S m⁻¹. Since all calibrations were performed on unponded sea ice, the conductivity of melt pond water was not directly accounted for. If, for some reason, no successful calibration was done during the ice station the temporally closest calibration from another ice station is used for processing. For the transects the GEM instrument was mounted within a Pulka sledge leading to a height offset of about 0.11m relative to the ground. The processing software is based on the gem2-seaice-toolbox (https://gitlab.awi.de/sitem/gem2-seaice-toolbox), and its application is detailed in the following tutorial: https://gitlab.awi.de/sitem/gem2-processing-tutorial. The calibration measurements, along with all YAML files used to derive the total ice thickness, are provided with the raw data (doi:10.1594/PANGAEA.996869).

Important notes:
• For ice station PS149_47-1 (floe visit 3d) the Geophex Ltd. GEM-2-556 instrument was used due to malfunctioning of the GEM-2-730. However, the GEM-2-556 data has a structural different output than the data of the GEM-2-730. Therefore, the processing method developed by Neudert et al. (2024) was used instead of the method by the gem2-seaice-toolbox. This method retrieves the ice thickness from all GEM frequencies combined instead of separately for each frequency.
• For the GEM-2-730 data we recommend using the ice thickness derived from the 18.325 kHz in-phase channel for further analysis, similar to the GEM processing performed for the MOSAiC campaign (see link above).
• If there is no derived sea ice thickness available for a certain channel (e.g. 93.075 kHz quadrature) this means that, for some reason, the ice thickness could not be retrieved for this channel.
• For ice station PS149_25-1 (floe 2b) no co-location was done as the GEM and MagnaProbe long transect measurements have been conducted on two consecutive days causing large uncertainties in the spatio-temporal co-location. Therefore, the GEM and MagnaProbe data of this station are provided without co-location at doi:10.1594/PANGAEA.996907.
--
MagnaProbe instrument & processing:
The MagnaProbe (produced by Snow-Hydro LLC) instrument measures the depth of the snow and/or surface scattering layer (SSL) on top of the sea ice as well as the depth of melt ponds. The probe consists of a 1.5 m long measuring rod with a flat basket at the lower end. During the measurement, the rod is inserted into the snow, with the basket remaining on the snow surface. The basket was adapted with a flotation made of extruded polystyrene foam. This adaptation allows to measure the depth of melt ponds as the basket floats and stays on top of the water surface. The difference between the sensor tip (ice surface) and the basket (snow/water surface) gives the snow/SSL or melt pond depth. In addition, the GPS position is recorded for each measurement. All measured data is stored in a data logger in the backpack carried by the operator and can be read out later on a PC. MagnaProbe profiles along transects were always measured in conjunction with ice thickness measurements from the GEM (see above). The difference between the GEM measurements and the snow depth measured by the MagnaProbe then gives the actual sea ice thickness.

During transect measurements melt ponds were marked in the dataset by a double-measurement, i.e., pushing the button of the MagnaProbe two times at the same location. The MagnaProbe is calibrated for every transect by doing three measurements when the basket is at lowest and highest position, respectively. This calibration is done at the beginning and end of each transect.
The transects of the ice stations PS149_13-1, PS149_16-1, PS149_18-1, PS149_21-1, PS149_25-1 were performed with magnaprobe-marcel, the transects of the ice stations PS149_30-1, PS149_32-1, PS149_36-1, PS149_41-1, PS149_46-1, PS149_47-1 with magnaprobe-anja.
In a first step, melt pond measurements have to be identified in the dataset. Melt ponds were marked by a double-measurement. Thus, two criteria were used to identify melt ponds:
1) The time difference Δt between two measurements is smaller than a specified threshold value called MPDT and
2) The depth difference Δd between the two measurements is smaller than a specified threshold value called MPDD.
MPDT and MPDD were identified for every transect from histogram plots and vary for MPDT between 1.4 s and 3 s and for MPDD between 1 cm and 2.6 cm.
The six calibration measurements for the basket in lowest position (three at the beginning, and three at the end of the transect) were averaged and used as reference. The offset of the MagnaProbe is then determined by subtracting the reference value from the length of the metal tip of the MagnaProbe value. The offset is added to all MagnaProbe measurements.
Finally. the snow/SSL depth and melt pond depth measurement values were saved as CSV files together with the GPS position and time of each measurement.
--
Spatio-temporal co-location of both datasets:
The co-location of the MagnaProbe snow/melt pond depths and the GEM total ice thickness estimates is based on a two-step spatio-temporal nearest-neighbour approach. For each MagnaProbe measurement, all GEM measurements within a 20 m spatial radius and 3-minute time window were identified in the according GEM dataset and averaged. This procedure yields one matched pair per MagnaProbe point. A linear drift correction was applied to both datasets using the displacement vector between the first and last GEM GPS position. During the long transects both GEM and MagnaProbe always returned back to the same starting point, i.e., assuming a closed survey loop on a drifting ice floe. The rotational component of the drift vector is not accounted for as there is not sufficient reference data of GNSS stations on the respective ice floes available. It is also assumed that the drift speed is constant during the walk of the long transect. This is not necessarily the case and could be improved (e.g. using the RV Polarstern continuous GNSS measurements). However, as the same linear drift correction procedure is applied to both the GEM and MagnaProbe datasets, the impact on the co-location quality is small. The linearly drift-corrected matched pairs were then binned onto a regular 1 m × 1 m spatial grid, with all variables aggregated by arithmetic mean within each bin. Coordinates reported in the dataset represent the geographic centre of each 1 m bin in WGS84 (EPSG:4326), derived from the Arctic Polar Stereographic projection (EPSG:3413) used internally for all distance calculations.-17.9292570020049133.98359994447588482.3834215080998184.98347177Alfred Wegener Institute, Helmholtz Centre for Polar and Marine Research, Bremerhavenhttps://doi.org/10.13039/501100003207AWI_PS149_01Sea Ice Physics (CONTRASTS)German Research Foundationhttps://doi.org/10.13039/501100001659268020496TRR 172: ArctiC Amplification: Climate Relevant Atmospheric and SurfaCe Processes, and Feedback Mechanisms