ESA CCI v3 and buoys show no 1982-2023 change in the slow decay of daily SST anomalies; DOISST v2.1 variance steps cause half its memory trend, its slow-decay rise and a false variance fall

Question: Two studies found that the reported rise of daily SST-anomaly memory in NOAA DOISST v2.1 is mostly a loss of fast variance at satellite input changes. Over the full ocean, which statistics does this artifact reach, and what real change remains? Approach: We compared DOISST v2.1 with the satellite-only ESA SST CCI Analysis v3.0.1 on 28,458 ice-free 1-degree cells (60S-60N), on the same points and days (1982-01-01 to 2024-06-22). We then compared both products with in situ SST at 169 moorings (121 GTMBA, 48 NDBC; 2,766 buoy-years) on matched days. Results: Relative to CCI, DOISST day-to-day variance fell in steps on 1984-10-09 (x0.68), 2007-01-02 (x0.67), 2017-03-04 (x0.69), 2020-06-28 (x0.67) and 2021-11-01 (x0.59), in 99.9% of the ocean area. In CCI, the slow decay time (tau_s, lags 5-20 days) had no trend (-0.01 days per decade, 95% interval -0.66 to +0.63). DOISST gave +0.90 (+0.35 to +1.44). The tau10 trend was +2.23 days per decade in DOISST and +1.08 in CCI, and all of the CCI rise came from the fast-variance term. At 123 long-record buoys, the buoys' own day-to-day variance had no trend (-0.009 ln units per decade). Relative to the buoys, it fell by -0.32 per decade in DOISST and by -0.135 in CCI, with a CCI step near 2002. The buoy tau10 trend was +0.50 days per decade (per-buoy median; +0.12 to +1.06) and was not significant in a sample-robust series. The DOISST fall of total anomaly variance (-0.098 ln units per decade) is absent in CCI (-0.018) and at the buoys (+0.032). Detrended marine heatwave counts fell twice as fast in DOISST as in CCI. Before 2021, detrended DOISST event durations were 3.4-3.7 days shorter than in CCI, and this offset closed after 2021. Conclusion: The slow decay of SST anomalies did not change detectably. Most of the DOISST memory rise, its slow-decay rise and its variance fall are product artifacts. A real tau10 rise is at most about 0.5-1 day per decade. Limitations: CCI has its own fast-variance drift of up to 0.3 ln units; buoys are DOISST inputs and sample only the tropics and US waters; CCI data end in mid-2024; the products disagree on the mean warming trend.