The iceberg is now being asked to file its own census
the iceberg is now being asked to file its own census: a resunet with multi temporal checks counts 39,619 grounded icebergs around antarctica with an f1 above 0.91 by enforcing bathymetry and sea ice constraints, a dual branch unet separates ice, open water and bergs at 0.964 mean intersection over union, and yolo spots open ocean bergs in sentinel wave mode.
the frozen island now files its own registry.
Context
The census is Grounded icebergs around Antarctica: a high-resolution dataset derived from deep learning (Earth System Science Data, 2026). The abstract says the method uses bathymetry and sea ice concentration to mitigate environmental false positives, reaches a detection F1 score above 0.91 and lowers the minimum detectable iceberg size to 0.016 km2. A dataset record describes 39,619 stationary iceberg targets under conditions of the annual landfast sea ice minimum, late February to early April 2025, from SAR imagery.
The 0.964 figure is from a Prydz Bay study using a DenseNet-enhanced Dual-Branch U-Net (DBU-Net) to separate sea ice, open water and icebergs, with mean IoU of 0.964 and overall accuracy of 0.9810, over 2016 to 2024. The open ocean detector is IBYOLO, adapted from YOLOv11, for kilometre-scale icebergs in Sentinel-1 Wave Mode SAR data.
The 39,619 targets are described as stationary icebergs in the dataset record, and the paper separates truly grounded icebergs from those held motionless by fast ice, so the count and the word grounded need the paper's caveat. The 0.964 is a mean IoU for a three-class segmentation in one region, Prydz Bay, and it is not comparable with the F1 of 0.91, which is a detection score over the continent. The excerpt read does not give IBYOLO accuracy, so none is claimed.
Watch next
- How the paper splits grounded from fast-ice-held icebergs in the 39,619 count. IBYOLO detection scores.
Sources
- Grounded icebergs around Antarctica: a high-resolution dataset derived from deep learning (ESSD)essd.copernicus.org
- A Deep Learning-Based Framework Reveals Sea-Ice And Iceberg Dynamics in Prydz Bayexa.ai
- Iceberg detection from global open ocean Sentinel-1 wave mode SAR data with YOLO deep learning (Ifremer Archimer)archimer.ifremer.fr
Provenance
The note above is reproduced unedited from the original post, first published on Threads on 4 October 2026 at 09:59 IST. Sources are the papers and datasets the note draws on.
View the original post ↗Embed this note
More notes
The air is now being asked to keep its own ledger
the air is now being asked to keep its own ledger: ecmwf’s aifs compo becomes the first ai model to forecast atmospheric composition globally every three hours, cleanair simulates 365 days of pm2.5 over china in ten seconds, and a unified framework maps six pollutants at one kilometer across the whole country. the air now files its own composition report.
read the note →The current is now being asked to draw its own map
the current is now being asked to draw its own map: china’s langya 2.0 predicts six ocean phenomena including internal waves and mesoscale eddies, a deep net called wenhai resolves eddies globally with air sea flux formulas built in, and scripps infers surface currents from the way temperature patterns deform in satellite images. the ocean now files its own circulation report.
read the note →The soil is now being asked to report its own carbon
the soil is now being asked to report its own carbon: a nix color sensor paired with generative data augmentation predicts soil organic carbon without a lab, random forest drives 74 percent of soil health mapping studies, and sentinel 2 tracks five year carbon change across france and italy from 922 samples. the dirt now files its own carbon account.
read the note →