Boreal CO2 seasonal amplitude records (NOAA, 1974–2024) cannot detect a 42–74% faster increase, and the t-test that reported it gives p < 0.05 for 70–87% of series with no acceleration

Guo et al. (2026, Earth's Future) report that the increase of the northern high-latitude CO2 seasonal cycle amplitude (SCA) accelerated by 74.24% (42.1% for station data, 51.64% at Barrow). They subtract early moving-window trends from recent ones and apply a one-sample t-test to the differences. We did a test of this claim with NOAA GML flask data from 79 sites (version 2026-07-17) and in-situ data from Barrow and Mauna Loa, through 2025. We calculated the annual SCA with three methods, which include the NOAA CCGCRV filter that Guo et al. used. In series with a constant linear trend and no acceleration, the Guo-style t-test gave p < 0.05 in 74–87% of AR(1) series and 70–84% of white-noise series, for all tested settings. The reason is that the windows overlap, so the differences are not independent. The Guo-style percentage depends on the start year: at Barrow in situ, it goes from 49% (start 1974) to 126% (start 1978) and then to 0% (start 1992). The 95% range with no acceleration is −52% to +96%, and our 1974–2021 value is 49.3%. With calibrated AR(1) null tests, no boreal site had a significant slope change with CCGCRV (p > 0.15). At six long boreal sites (≥ 50°N; noise model with the correlation between sites), the SCA slope after 2005 was +7% of the earlier slope (95% interval −77% to +373%, p = 0.92). Two other SCA methods gave +16% and +44% (p > 0.6). A scan of break years 1995–2012 found no break. The power to detect a true 74% acceleration was only 0.22, and 80% power needs an increase of about 180%. Thus these station records neither confirm nor exclude an acceleration of 42–74%, and the reported significance is an artefact of the test. As a secondary result, South Pole and Palmer Station had positive slope changes (p = 0.004 and 0.029). This agrees with the Southern Ocean SCA increase reported before, but 2 of 35 tests is near the number that chance gives. Limits: we built the statistic from the text of Guo et al. because their supplement was not available to us, and we did not test their XCO2, MBL or model results.

atmospheric CO2 · seasonal cycle amplitude · boreal ecosystems · carbon cycle · NOAA GML · CCGCRV · trend acceleration · moving-window trends · false positive rate · statistical power · replication · Barrow · Alert