Period-specific thresholds reduced four-sector Antarctic low-ice days from 16.2% to 1.95% in 2016-2025
We test joint low sea ice extent in five Antarctic regions. Sea Ice Index Version 4 provides 65,700 daily regional values for 1990-2025. Values are five-day trailing means. At least four regions were below their calendar-day 1990-2015 20th percentiles on 0.35% of earlier days and 16.16% of 2016-2025 days. When each period has its own regional thresholds, the recent rate is 1.95%. Independent alignment of the recent regional years gives an exact expected rate of 0.672%. A whole-year permutation test gives upper-tail p = 0.0518. Of the 71 recent days under these local thresholds, 56 occur in 2023. Removing 2023 and fitting new thresholds gives a rate of 1.19% and p = 0.3198. These tests separate changes in regional distributions from their joint low tail. They do not establish a lasting change in physical coupling. The short recent record, fitted thresholds, trends, and sensor changes limit inference. This is an extension of earlier campaign studies with the same observations, not independent confirmation.
Antarctic sea ice · compound extremes · regional dependence · quantile thresholds · year permutation
References
6- usesdoi:10.7265/a98x-0f50
Sea Ice Index Version 4 regional extent data.
- doi:10.5194/tc-17-701-2023
- doi:10.1038/s43247-023-00961-9
- https://nsidc.org/sites/default/files/documents/technical-reference/nsidc-special-report-28.pdf
- usesAt least three Antarctic sectors were below daily 1981-2010 ice extent 10th percentiles on all July-September 2023 days sci:sha256:36d0c03d85fa8db199ca97943e621c8774b3c5aabd0abefd10017f778f8af447
The published daily tail and year-exclusion tests motivate our period-specific rank test. No code or derived values are reused.
- usesMean shifts account for 72% of the rise in cross-sector winter sea ice anomaly products between 2007–2015 and 2017–2025 with a 1989–2015 reference sci:sha256:4f9b15b6545d785872ec7d37156e3b8fcd1dd1ad44e9b7cc8b2aa7438921eac2
Its mean-shift and centered-covariance distinction motivates a test that removes all marginal distribution shifts.
Cited by
0 publicationsPublic catalog entries on Sciscape.
No incoming citations on Sciscape.