Arctic Open-Water Window

Not for navigation. Use at your own risk. This is a historical climate indicator. It does not show current ice conditions in a channel and must not be used for route planning or navigation safety. Use national ice service charts for operations. Contains modified Copernicus Sentinel data 2015-2026. Open-water days each season at fixed gates on the Bering Strait, Northwest Passage and Northern Sea Route, from the NOAA/NSIDC passive-microwave sea-ice concentration record since 1978. A historical indicator, not a navigation product. Each gate is a fixed polygon across a passage. For each day, sea-ice concentration is averaged over the grid cells inside the gate, excluding cells contaminated by land. A day is open water when that average is below the threshold; three thresholds (15, 30 and 50 percent) are published so users can see how sensitive the answer is. Seasonal counts and first and last open dates follow from the daily series. Near-real-time values are marked provisional and replaced when the final record is released. Imagery: Weekly Sentinel-1 gate imagery for 3 gates, 2015 to 2026: 1,663 of 1,842 gate-weeks have an image (128 no scene, 51 low coverage). Resolution 80 m, EPSG:3995. Cloud-Optimized GeoTIFF and WebP in 36 gate-year ZIP bundles under `imagery/bundles/` on the Zenodo record (https://doi.org/10.5281/zenodo.23021978); `imagery/index.csv` and `imagery/IMAGERY.md` are unzipped in this package. Intended use: Long-term change in how many days Arctic passages are open, season timing, and background for shipping, emergency response and security research. Not intended for: Route planning or navigation safety, ice thickness or type, or conditions in channels narrower than the 25 km grid. Use national ice service charts for operations. Limits: Narrow straits span only one to three 25 km cells, so gate values describe the surrounding sea, not the channel itself. The two concentration algorithms in the record disagree at low concentrations; algorithm choice shifts seasonal counts by several days on average. Users should treat count differences of less than 5 days as within-algorithm uncertainty. Years 1978 and 2026 are partial seasons (partial=true); exclude them for trend and median analysis. Documentation, API and MCP access: https://heimdallresearch.com/data/arctic-window/. Published by Heimdall Research.