Oxygen loss at 200-300 m off southern California did not continue in 2007-2020 and levels stayed low; change on density surfaces, not heave, gave 93% of the 1984-2006 loss (CalCOFI)

We tested if the 1984-2006 dissolved oxygen (O2) decrease off southern California continued after 2006. We used CalCOFI bottle data from 56 stations on six core lines (1949-2020) and a station-by-quarter fixed-effects annual index. We used 1984-2006 for training and 2007-2020 as an out-of-sample test. At 300 m, O2 decreased by 0.80 umol/kg per year in 1984-2006 (95% interval -1.28 to -0.33). In 2007-2020, the slope was +0.24 (-0.34 to 0.85), and the slope difference excludes zero. O2 continued to decrease to a minimum in 2012 and then stopped decreasing. It did not return to the earlier level: the 2007-2020 mean was 5 umol/kg below the 2002-2006 mean and 10 umol/kg below the 1950-1983 mean. The extrapolated 1984-2006 trend did not clearly forecast 2007-2020 better than a flat 2002-2006 mean (MSE ratio 0.86, 0.69 to 1.40). On sigma-theta 26.6, the slope changed from -0.90 to +0.01 per year. The hypoxic boundary (60 umol/kg) moved up by 1.82 m per year in 1984-2006, reached 67 m above its 1984-2006 mean in 2012, and then moved partly down. On sigma-theta 26.8, a slow decrease continued. A split of each cast into heave and along-isopycnal parts showed that O2 change on density surfaces gave 93% of the 1984-2006 decrease at 300 m and 87% of the annual variance. After 2012, downward motion of density surfaces gave most of the partial recovery at fixed depths. A titration-only sensitivity analysis gave the same conclusions. Limitations: one region, 14 test years, approximate block-bootstrap intervals, a break year chosen from the data, a change to CTD-sensor O2 values in 2018, and no direct test of the cause. The bundle contains code, per-cast derived data, all statistics, provenance, and figures.

dissolved oxygen · hypoxia · CalCOFI · California Current · Southern California Bight · isopycnal heave · out-of-sample test · deoxygenation · trend change · oxygen minimum zone