-
The global upper 2000 m OHC changes since 1958 (Fig. 2) show that, regardless of the processing techniques, there has been an unequivocal ocean warming trend in recent decades. The upper 2000 m of the world’s ocean has warmed on average by 6.6 ± 0.3 ZJ yr−1 during 1958–2023 (IAP/CAS) and by 5.4 ± 0.4 ZJ yr−1 during 1958–2020 (NCEI/NOAA pentadal estimate). The 95% confidence levels are calculated using the approach of Cheng et al. (2022b). However, these trends do not match within the error bars, probably because of (1) conservative assumptions by NOAA when there are no data (relax to climatology in data gaps), especially in the presence of global warming trends; and (2) the difference in XBT/MBT bias correction and the new inclusion of the bottle data bias correction (Gouretski and Cheng, 2020; Gouretski et al., 2022).
Figure 2. Global upper 2000 m OHC from 1958 through 2023 according to (a) IAP/CAS and (b) NCEI/NOAA (1 ZJ = 1021 J). The line shows (a) monthly and (b) seasonal values, and the histogram presents (a) annual and (b) pentad anomalies relative to a 1981–2010 baseline.
Regardless of which estimate is used, there has been a two- to three- fold increase in the rate of increase in OHC since the late 1980s. For example, according to the IAP analysis, the OHC trend for 1958–1985 is 3.1 ± 0.5 ZJ yr−1, and since 1986, the OHC trend is 9.2 ± 0.5 ZJ yr−1 (Fig. 2). The IAP trend within 1958–1985 of 3.1 ± 0.5 ZJ yr−1 is higher than the previous release in Cheng et al. (2023) (2.3 ± 0.5 ZJ yr−1), mainly because the new inclusion of the bottle data bias correction.
After 2007, with better global coverage of ocean subsurface data, OHC uncertainty is reduced. There is a significant warming trend of 10.8 ± 1.2 ZJ yr−1 and 10.3 ± 0.8 ZJ yr−1 from 2007–2023 for IAP/CAS and NCEI/NOAA (seasonal time series), respectively (Fig. 2). The NCEI three-month OHC estimate has a slightly stronger trend than the pentadal time series from 2005 to 2020, indicating the impact of sampling changes associated with the mapping approach.
OHC tends to peak shortly before and then decline during and after an El Niño event, associated with ocean heat release into the atmosphere, mainly through increased evaporation (Cheng et al., 2019). In 2023, OHC was at the highest level ever recorded in the world’s ocean, and the El Niño effects may not yet be fully evident. The 2023 upper 2000 m OHC exceeds that of 2022 by 15 ± 10 ZJ according to IAP/CAS data, and by 9 ± 5 ZJ according to NCEI/NOAA data (for the 0–2000 m layer; 95% confidence interval is presented; Table 1). A ranked ordering of the hottest five years for global OHC is provided in Table 1. The annual OHC values from 2019 to 2022 updated in this paper (Table 1) are collectively higher (~20 ZJ) than the numbers in the previous release (Cheng et al., 2023), because of the update of the IAP/CAS dataset that led to higher OHC anomalies after 2019 relative to the 1981–2010 baseline. Preliminary analyses suggest the difference is likely attributed to the replacement of the WOD-QC system (used in previous IAP analyses) by the new CODC-QC systems. Tan et al. (2023) indicated that the WOD-QC system has removed more positive anomalies than CODC-QC. The bias corrections to data collected by marine animals play a secondary role. The difference in 2023 OHC values between the two groups is also primarily attributed to the data QC, which relates to how the outliers are defined and flagged with a secondary contribution from the mapping approach. A careful investigation is warranted to reconcile the two groups’ estimates.
During an El Niño event, there is a heat redistribution from the 100–500 m layer into the upper ~100 m layer, yielding higher SST than normal (Cheng et al., 2019). The anomalously high SST leads to a higher global mean surface temperature (GMST) (Trenberth et al. 2002; Li et al., 2024). In 2023, the SST became the highest on record after April, and the annual mean was 0.23°C higher than in 2022 and an astounding 0.54°C higher than the 1981–2020 average (Fig. 3). By June 2023, global monthly SSTs were already ~0.2°C above those of any prior year, an exceedingly large value (Fig. 3) that also meant GMSTs were the highest on record. The monthly SST anomaly in 2023 relative to 1981–2010 grew from 0.35°C in January to 0.67°C in September, making September 2023 the hottest month on record for global SSTs. Normally, the hottest month for SST in a particular year occurs in March, at the end of the southern summer, because there is a large ocean area in the Southern Hemisphere (Fig. 3 inner box plot). Although SST has increased dramatically in 2023, the OHC increase has been steady over time (Fig. 2). Therefore, it is the relatively small year-to-year natural variability in OHC relative to the warming trend that makes OHC such a good indicator of climate change.
Figure 3. Global SST changes from 1955 through 2023 according to first level (1 m) data in the IAP/CAS temperature gridded analysis (°C). The black line is the annual value, and the red is the monthly value. The anomalies are relative to a 1981–2010 baseline. The within-year variation of SST is shown in the inner box, with 2023 values shown in black.
Rank Year OHC (IAP/CAS) (units: ZJ) OHC (NCEI/NOAA) (units: ZJ) SST anomaly (IAP/CAS) (units: °C) 1 2023 286 247 0.54 2 2022 271 238 0.31 3 2021 254 229 0.28 4 2020 237 211 0.38 5 2019 228 210 0.40 Table 1. Ranked order of the five hottest years of the world’s ocean since 1955. The OHC values are for the upper 2000 m in units of ZJ. The SST values are in °C. Both OHC and SST anomalies are relative to the 1981–2010 average. Note the IAP/CAS values are collectively higher (~20 ZJ) than the previous release (Cheng et al., 2023) because of the update of the IAP/CAS dataset that led to higher OHC anomalies relative to the 1981–2010 baseline.
-
Substantial changes are also seen in other oceanic metrics. The upper 2000 m SC index time series since 1958 (Fig. 4) reveal a robust increase in the SC index in the past half-century, indicating an amplification of the 0–2000 m salinity pattern (Cheng et al., 2020). The SC index reached 7.2 mg kg−1 in 2023, the fourth-highest value since 1958. However, the difference between the top 5 years 2017, 2022, 2021, 2023, and 2019) is not statistically significant because of the large inter-annual variability and data uncertainty; for instance, there are more real-time Argo salinity data recently that have not undergone careful quality-control and bias adjustment. This ocean-based metric is generally consistent with many atmosphere-based estimates and strengthens the evidence that the global water cycle has been amplified with global warming (Cheng et al., 2020). On land, the amplified water cycle means stronger and longer dry spells and more heavy rainfall events with the potential for flooding, as observed (Fischer et al., 2021).
Figure 4. Monthly (black line) and annual (color bars) changes in (a) SC index, (b) stratification, and (c) spatial inhomogeneity index of temperature in the upper 2000 m of the global ocean from 1958 to 2023 [data updated from Cheng et al. (2017)]. The units for salinity are g kg−1 (using absolute salinity).
Ocean density stratification has also increased since the late 1950s (Fig. 4b) because of the change in vertical temperature and salinity structure (Li et al., 2020a). The stratification index shows stronger interannual to decadal variability than the OHC and SC-index because it reveals more upper-ocean changes, which shows stronger anomalies than the deeper ocean. In 2023, the upper 2000 m stratification increased to (6.93 ± 0.39) × 10−7 s−2, reaching record high values in 2023 mainly because of the development of the strong El Niño.
The spatial inhomogeneity index of ocean temperature has also increased since the 1950s (Fig. 4c), with a trend of 0.020 ± 0.003°C (10 yr)−1. This index reached a record high of 0.093°C in 2023 relative to a 1981–2010 baseline, indicating a substantial increase in ocean temperature spatial variance. The non-uniform upper-ocean warming, which was more rapid at mid-to-low latitudes, was mainly responsible for this index increase in 2023 (Ren et al., 2022).
Rank | Year | OHC (IAP/CAS) (units: ZJ) | OHC (NCEI/NOAA) (units: ZJ) | SST anomaly (IAP/CAS) (units: °C) |
1 | 2023 | 286 | 247 | 0.54 |
2 | 2022 | 271 | 238 | 0.31 |
3 | 2021 | 254 | 229 | 0.28 |
4 | 2020 | 237 | 211 | 0.38 |
5 | 2019 | 228 | 210 | 0.40 |