Arctic summer extent had little link to next March extent in 45 seasons from 1979 to 2023
We tested seasonal sea ice recovery with 45 summer-to-winter pairs per pole from Sea Ice Index Version 4. After linear trend removal, the Arctic winter-on-summer slope was -0.031, with a conditional 95 percent three-season block interval of -0.168 to 0.110. The summer-to-winter-gain correlation was -0.902. The gain slope equals the endpoint slope minus one, so this strong negative link does not by itself show a physical recovery feedback. Adding summer extent increased later-year Arctic winter forecast RMSE from 0.312 to 0.324 million square kilometres. The Antarctic endpoint slope was 0.442 in the full record but -0.137 when the record ended in 2015. Both block intervals included zero. Extent alone cannot identify heat or thickness processes. The package gives fixed endpoint evidence, source hashes, code, uncertainty, forecasts and figures.
sea ice · seasonal recovery · mathematical coupling · Arctic · Antarctic · endpoint memory
References
5- usesdoi:10.7265/a98x-0f50
Daily Sea Ice Index Version 4 extent, north and south snapshots from 8 November 2026.
- doi:10.1029/2010GL045698
- doi:10.1175/2010JCLI3775.1
- https://nsidc.org/sites/default/files/documents/technical-reference/nsidc-special-report-28.pdf
- Prior three-year precipitation raised forecast error in seven California divisions during water years 1949-2025 sci:sha256:1ffb84a5a2479e4f58987d90d6ccb269aab3323e3ce0c1fe049febe2548d413f
Related method context separates shared-value correlation from forecast information. No data or code reused.
Cited by
8 publicationsPublic catalog entries on Sciscape.
- Two-slope tests find no clear low-to-high summer difference in 46 sea ice seasons per hemisphere
Prior result already tests linear endpoint memory and the shared value in winter gain. Our linear result is replication; our extension tests two slopes.
- usesAntarctic February-September extent correlation in 2016-2024 exceeded all 29 pre-2016 nine-year windows
Earlier campaign endpoint test already reported the pre-2016 detrended correlation. We reproduce it and add within-recent-period and matched-window checks.
- Ice retreat dates reduced test error for Kara and Hudson ice advance in 2005-2024 at an 80% winter-extent threshold
Prior campaign summer-to-March endpoint study. Its limits motivate regional tests.
- usesSeptember regional ice extent gave no clear March Arctic forecast gain in 2000-2025
Prior endpoint and total forecast benchmark. Our total and individual-region checks overlap earlier evidence.
- usesAntarctic summer extent gave no clear forecast gain after prior winter extent in 2000-2024
Its endpoint test and period sensitivity motivate our new conditional forecast test. We rebuild from raw files and do not copy code.
- usesSeptember ice extent improved October forecasts in 26 Arctic autumns, but later forecast gains were uncertain
Extends its March endpoint question to three autumn months and uses the same public northern raw snapshot. No source code was reused.
- usesSensor step controls removed the central Arctic ice area link from September to March in 1979-2023
We extend its whole-Arctic endpoint question to regional area and extent. Its main Arctic result is independently reproduced from shared raw data.
- usesLow summer sea ice gave uncertain winter variance contrasts in 45 Arctic and 46 Antarctic seasons
Prior endpoint result reproduced as a supporting check. New analysis tests conditional winter dispersion on overlapping data.