The experiment will build on a set of Arctic sea ice cores recently collected during the largest Arctic expedition in history (MOSAiC,’Multidisciplinary driftingObservatory for the Study of Arctic Climate’, October 2019-September 2020, http://www.mosaicexpedition.org). These cores contain important and still unraveledinformation about the sea ice properties during the MOSAiC drift. We aim for phase-contrast tomography and hierarchical scanning of several full sea ice coresfrom MOSAiC, exploring the unique possibility at BM18 to scan large sample volumes with locally high resolution. The experiment will set a benchmark in 3Dimage quality of sea ice microstructure and produce an open source dataset be relevant for many polar research topics and groups within and outside theMOSAiC consortium.

Sea-ice thickness, salinity, temperature, density, and stable water isotope composition were measured during surveys at the Ridgey McRidgeFace (RMRF) coring site during the MOSAiC expedition (leg 3). RMRF was a first-year ice ridge formed in March 2020. The ice cores were extracted either with a 9-cm (Mark II) or 7.25-cm (Mark III) internal diameter ice corers (Kovacs Enterprise, US). This data set includes data from coring site visits performed on 22 April 2020 and 05 May 2020 at RMRF in the MOSAiC Central Observatory (Kanzow and Damm, 2023). During each coring event, ice temperature was measured in situ from a separate temperature core, using Testo 720 thermometers in drill holes with a length of half-core-diameter at 5-cm vertical resolution. Ice bulk practical salinity was measured from melted core sections at 5-cm resolution using a YSI 30 conductivity meter. Ice density was measured using the hydrostatic weighing method (Pustogvar and Kulyakhtin, 2016) from a density core in the freezer laboratory onboard Polarstern at the temperature of -(16–18)°C. Relative volumes of brine and gas were estimated from ice salinity, temperature, and density using Cox and Weeks (1983) for ice colder than -2°C and Leppäranta and Manninen (1988) for ice warmer than -2°C.The data contains the event label (1), time (2), and global coordinates (3,4) of each coring measurement, coring site (5), and core type (6). Each core has its manually measured ice thickness (7), ice core length (8), and mean snow height (9). Each core section has the total length of its middle (10), top (11), and bottom (12) measured in situ. Each core section has the value of its practical salinity (13), as well as sea ice temperature (14), laboratory temperature (15), and ice density at the laboratory (16) and in situ (17) temperatures, brine volume fraction estimates (18), and gas volume fraction estimates at the laboratory (19) and in situ (20) temperatures. Some core sections have stable water isotopic values (21, 22). The location of ice sections relative to the nearby ridge void is described in comment (23). The locations of the coring sites are shown on the digital elevation map (Hutter et al., 2023).The stable oxygen isotopic compositions of the melted snow samples (δ18O) were determined in the central laboratory of the Swiss Federal Institute for Forest, Snow and Landscape, Birmensdorf, Switzerland with an Isotopic Water Analyzer IWA-45-ER (ABB - Los Gatos Research Inc., US). Measurement uncertainty for δ18O is ±1‰, the precision ± 0.5‰. All samples were measured in duplicate and averaged. The quality control was conducted with three standards for δ18O at 0.00‰, -12.34‰ and -55.50‰ and are presented as per mil difference relative to VSMOW (‰, Vienna Standard Mean Ocean Water).

The WFJ_DailySMP dataset contains daily SnowMicroPen (SMP) measurements conducted throughout the winter seasons 2015-2025 (ongoing) at the Weissfluhjoch research site, Davos, Switzerland. The measurement program DailySMP started in winter 2015/2016 during which the measurement protocol was established as part of the RHOSSA campaign, described in the related publication.Since then, measurements have been repeated every winter as part of the standard snow monitoring program by the PhD students at SLF. All published SMP files have undergone manual quality control by an expert, including a supervised detection of the snow surface, which is included in the dataset. In addition to the raw data, we provide a rudimentary post-processing workflow in which the derived SSA and density profiles are adjusted by the total snow height obtained from the adjacent IMIS snow station (https://www.slf.ch/en/avalanche-bulletin-and-snow-situation/measured-values/description-of-automated-stations/) and their evolution is plotted over time. Please refer to the README for detailed information on how to get started.

Data of snow and sea ice in the McMurdo Sound, October-December 2022. The data was collected as part of the New Zealand Marsden Fund Research Grant 21-VUW-103 "Can Snow Change the Fate of Antarctic Sea Ice?" The dataset includes raw data of the manual snow and sea ice measurements from snow pits and ice cores (temperature, density, salinity, dO18), measurements of snow water equivalent (SWE), spatial information of snow height (MagnaProbe) and sea ice thickness (EM-31), AWS (air temperature, wind speed, wind direction, relative humidity, pressure), radiations stations (shortwave, longwave, thermal IR, spectral shortwave), differential GPS data (3 fixed stations on different sea ice thicknesses, + 1 rover station for georeferencing UAV measurements), SIMBA buoy temperature (+heated temperature) data (3 buoys during November, 1 buoy for 15 months), UAV data: RGB, thermal IR, broadband albedo, spectral albedo, Chlorophyll-a from ice cores (bottom 10 cm), NIR reflectivity data of snow at 850 nm, and 940 nm (snow surface, profile, ice surface), photographs (1. overview of field sites, 2. for Structure from Motion for surface roughness, 3. macrophotos of snow) surface impurity concentrations, microCT data of snow microstructure, Denoth probe (density) and InfraSnow (specific surface area - SSA). See the README file in each dataset for detailed information.

This dataset provides access to pre-trained models that were used for SnowMicroPen profile classification and segmentation. The models were trained on a part of the MOSAiC SMP dataset, available on https://doi.pangaea.de/10.1594/PANGAEA.935554. The labeled training data consists mostly of profiles from leg three of the expedition (January - May 2020), some profiles from leg one and two, and no profiles from leg four. Please refer to the snowdragon GitHub repository (https://github.com/liellnima/snowdragon) to access the models' training code and be directed to current publications. The following trained models are available here (alphabetically ordered): Artificial neural networks Bi-directional long short-term memory (blstm.hdf5) Encoder-decoder (enc_dec.hdf5) Long short-term memory (lstm.hdf5) Baseline Majority vote classifier (baseline.model) Semi-supervised models Cluster-then-predict models: Bayesian Gaussian mixture model (gmm.model) Bayesian mixture model (bmm.model) K-means clustering (kmeans.model) Label propagation (label_spreading.model) Self-trained classifier (self_trainer.model) Supervised models Balanced random forest (rf_bal.model) Easy ensemble (easy_ensemble.model) K-nearest neighbors (knn.model) Random forest (rf.model) Support vector machines (svm.model)
Loading Instructions:
The models with the file-ending ".model" are pickeled Python objects and can be loaded with pickle.load(your_model.model). The random forest must be loaded with joblib.load(rf.model). All artificial neural networks are h5py.File objects (tf.keras models) and can be loaded with tf.keras.models.load_model(your_ann.model).

This dataset provides access to pre-trained models that were used for SnowMicroPen profile classification and segmentation. The models were trained on a part of the MOSAiC SMP dataset, available on https://doi.pangaea.de/10.1594/PANGAEA.935554. The labeled training data consists mostly of profiles from leg three of the expedition (January - May 2020), some profiles from leg one and two, and no profiles from leg four. Please refer to the snowdragon GitHub repository (https://github.com/liellnima/snowdragon) to access the models' training code and be directed to current publications. The following trained models are available here (alphabetically ordered): Artificial neural networks Bi-directional long short-term memory (blstm.hdf5) Encoder-decoder (enc_dec.hdf5) Long short-term memory (lstm.hdf5) Baseline Majority vote classifier (baseline.model) Semi-supervised models Cluster-then-predict models: Bayesian Gaussian mixture model (gmm.model) Bayesian mixture model (bmm.model) K-means clustering (kmeans.model) Label propagation (label_spreading.model) Self-trained classifier (self_trainer.model) Supervised models Balanced random forest (rf_bal.model) Easy ensemble (easy_ensemble.model) K-nearest neighbors (knn.model) Random forest (rf.model) Support vector machines (svm.model)
Loading Instructions:
The models with the file-ending ".model" are pickeled Python objects and can be loaded with pickle.load(your_model.model). The random forest must be loaded with joblib.load(rf.model). All artificial neural networks are h5py.File objects (tf.keras models) and can be loaded with tf.keras.models.load_model(your_ann.model).

Dielectric permittivity (ε′) and dielectric loss (ε″) measurements were made of discrete snow layers and at fixed vertical intervals. Measurements were made using the Stevens Water Monitoring Systems Hydra Probe (a.k.a. hydraprobe). The hydraprobe uses a tine assembly consisting of a central waveguide and three outer rods, each 4.5 cm in length and 3 mm wide, to measure the impedance of the sample at 50 MHz over a cylindrical area of 5.7 cm in length by 3 cm in diameter. The sensor was calibrated using isopropyl alcohol for ε' (±0.6 %) and a saline solution of known conductivity for ε'' (±0.7 %). Samples were obtained by horizontally inserting the probe tines into the snow, at a given layer/interval (every 3cm), to their maximum depth and collecting a response. 0 cm is the snow-ice interface. For related temperature and density measurements see https://doi.pangaea.de/10.1594/PANGAEA.940200 and https://doi.pangaea.de/10.1594/PANGAEA.940214 in this bundled dataset.Please direct inquiries to; Aikaterini Tavri (PS122/4), Ruzica Dadic (PS122/5).

Overview photos were taken when arriving at the snowpit site of interest and before the snowpit was excavated. A standard digital camera (Olympus tough TG-5) was used to document the surface conditions and environment for each snowpit event. These images should be used to understand surroundings of the snowpit and conditions on the day of measurements. During winter months the measurement area of a snowpit was indicated using two orange flags. However, a transect snowpit event and snowpits during the summer seasons will not have this marking.

The metadata txt files can be used to obtain a quick overview of one snowpit event. They give details on date of the event, location of the snowpit, start time, GPS file name taken at the snowpit location, snow depth, weather during the event, participants conducting the measurements, instruments included in the measurements and chemical samples collected. Please direct inquiries to; David Wagner (PS122/1), Martin Schneebeli (PS122/2), Amy Macfarlane (PS122/3 and PS122/4), Ruzica Dadic (PS122/5).

For the majority of snowpits the snow surface type was recorded by the observer. The documented surface types include; new snow, rime, surface hoar, glazed, drifted snow, wet, visible dust, frost flowers, surface crust, surface scattering layer, jewel snow, frozen. Please direct inquiries to; David Wagner (PS122/1), Martin Schneebeli (PS122/2), Amy Macfarlane (PS122/3 and PS122/4), Ruzica Dadic (PS122/5).