Improved Quantification of the Rate of Ocean Warming

Abstract The increased concentrations of greenhouse gases in the atmosphere create an increase Earth’s thermal energy, which is mainly stored ocean. Quantification rate ocean heat content (OHC) vital for understanding current and future climate Earth. Linear trend lines have been frequently used to quantify long-term rates change, but are inappropriate because they cannot capture nonlinearity trends, large start- end-point sensitivity, assumption linearity nonphysical. Here observed model-based linear regressions with higher-order polynomial (quadratic), piecewise linear, locally weighted scatterplot smoothing (LOWESS) compared. Piecewise LOWESS perform best depicting multidecadal trends. It shown that valid up about 15-yr segments (i.e., it compute within a time window). Using recommended methods, warming upper 2000 m increases from 0 0.06 ± 0.08 W −2 1958–73 0.58 2003–18, indicating acceleration happens all four basins near sea surface m. There consistency between multimodel-mean historically forced models observations, implies contribution internal variability small global 0–2000 OHC. Notable OHC 0–300 m) after 1980 deeper (300–2000 late 1980s also evident. This study suggests alternative methods those currently estimate provide more accurate quantification energy changes. Significance Statement Quantifying change needed understand evolution assess changing budgets. sensitivity. Based on analysis statistical features series, this proposes two including fit LOWESS. Robust detected through observational records 1958 2020, indicate robust warming. Slow penetration into