This dataset encompasses data produced in the study ''Seasonal Arctic sea ice forecasting with probabilistic deep learning'', published in Nature Communications. The study introduces a new Arctic sea ice forecasting AI system, IceNet, which predicts monthly-averaged sea ice probability (SIP; probability of sea ice concentration > 15%) up to 6 months ahead at 25 km resolution. The study demonstrated IceNet''s superior seasonal forecasting skill over a state-of-the-art physics-based sea ice forecasting system, ECMWF SEAS5, and a statistical benchmark. This dataset includes three types of data from the study. Firstly, IceNet''s SIP forecasts from 2012/1 - 2020/9. Secondly, the 25 neural network files underlying the IceNet model. Thirdly, CSV files of results from the study. The codebase associated with this work includes a script to download this dataset and reproduce all the paper''s figures. This dataset is supported by Wave 1 of The UKRI Strategic Priorities Fund under the EPSRC Grant EP/T001569/1, particularly the "AI for Science" theme within that grant and The Alan Turing Institute. The dataset is also supported by the NERC ACSIS project (grant NE/N018028/1).

We present the Weddell Sea Benthic Dataset (WSBD), a computer vision-ready collection of high-resolution seafloor imagery and corresponding annotations designed to support automated analysis of Antarctic benthic communities. The dataset comprises 100 top-down images captured during RV Polarstern Expedition PS118 (cruises 69-1 and 69-6) in 2019, using the Ocean Floor Observation and Bathymetry System (OFOBS) in the Weddell Sea, Antarctica. A subset of this imagery was manually annotated by ecologists at the British Antarctic Survey (BAS) to support ecological analyses, including benthic community composition and species interaction studies. These annotations were subsequently standardised into 25 morphotypes to serve as class labels for object detection tasks. Bounding box annotations are provided in COCO format, alongside the training, validation, and test splits used during model development at BAS. This dataset provides a benchmark for developing and evaluating machine learning models aimed at enhancing biodiversity monitoring in Antarctic benthic environments. This work was funded by the UKRI Future Leaders Fellowship MR/W01002X/1 ''The past, present and future of unique cold-water benthic (sea floor) ecosystems in the Southern Ocean'' awarded to Rowan Whittle.

Model weights for the optimal object detection model, trained on the Weddell Sea Benthic Dataset. Trained 2025-05. Weights should be used with a Deformable-DETR architecture. This work was funded by the UKRI Future Leaders Fellowship MR/W01002X/1 ''The past, present and future of unique cold-water benthic (sea floor) ecosystems in the Southern Ocean'' awarded to Rowan Whittle.