A paper in JGR: Oceans compares how the 2014-2016 and 2023-2024 El Niños moved global mean sea level. In both, the dominant term was not warming seawater but water moving off the land. The totals, the durations and the regional fingerprints were not the same, and the distinction matters more than the headline number.
The paper is Jin, Zhong, Yang and Feng (2026), published online in Journal of Geophysical Research: Oceans on 31 May 2026 (volume 131, article e2025JC023533). It is not research published this week. What is new is the coverage: a run of write-ups appeared in mid-August, roughly eleven weeks after publication. That gap is worth knowing when a study is presented to you as breaking.
The authors combine satellite gravimetry with Argo float products to split the sea-level change into two parts. Barystatic change is mass: water arriving in the ocean from land, including terrestrial water storage and ice. Steric change is volume without mass: seawater expanding as it warms, or contracting as it cools. The finding is that the mass term dominated both events.
All figures in this table are the paper’s, for the development phases it defines.
| 2014-2016 event | 2023-2024 event | |
|---|---|---|
| Development phase | October 2014 to December 2015 | May 2023 to December 2023 |
| Interannual GMSL rise | 9.16 mm | 7.70 mm |
| Barystatic (mass from land) | 6.20 mm | 6.26 mm |
| Steric (thermal and haline) | 3.11 mm | 1.60 mm |
| Barystatic share (IAMElNino calculation) | ~68% | ~81% |
The 68% and 81% figures are not stated in the abstract. They are 6.20 ÷ 9.16 and 6.26 ÷ 7.70, computed here from the paper’s published values.
Here is the comparison that makes the paper interesting. The ocean gained nearly the same barystatic mass-equivalent amount during both development phases: 6.20 mm and 6.26 mm, which the authors call comparable. Barystatic change covers mass exchange between ocean and land in general, including land water and land ice; the paper attributes the accelerated 2023 rise specifically to terrestrial water storage depletion.
But the second event delivered it over a much shorter window. The paper defines the 2014-2016 development phase as October 2014 to December 2015, and the 2023-2024 phase as May 2023 to December 2023, and describes the later surge as “significantly more rapid, occurring over a much shorter duration.” It attributes the acceleration to terrestrial water storage depletion in the Americas and Africa, and notes that South America in particular lost storage at a significantly faster rate than during the earlier event.
The authors attribute that faster 2023 rise primarily to terrestrial water storage depletion in the Americas and Africa, with particularly rapid losses in South America. That is the paper’s own explanation, and it is a stronger statement than any ratio computed from the outside.
A note on why there is no rate table here. It is tempting to divide each total by the number of months and print a millimetres-per-month figure, and I drafted exactly that before discarding it. The abstract gives the phase endpoints but not the interval convention, and fifteen inclusive calendar months are not the same as fifteen elapsed monthly intervals. Choosing one convention over the other changes the barystatic ratio by roughly seven percent, which is enough to matter in a table that looks authoritative. Without the methods section, the honest move is to report what the authors report and leave the arithmetic alone.
The steric side is where the two events diverge most clearly. The 2014-2016 phase produced 3.11 mm of steric rise; the 2023-2024 phase produced 1.60 mm, roughly half.
The geography differed too. The paper finds the 2023-2024 steric rise was predominantly thermosteric, meaning driven by warming rather than salinity, with a substantial contribution originating from the Indian Ocean. The authors state that this regional anomaly was likely associated with the simultaneous occurrence of a peak positive Indian Ocean Dipole and El Niño.
That hedge is the authors’ own and it should survive translation. Some coverage of this paper has firmed it into a definite cause. “Likely associated with” is a different claim from “driven by,” and the difference is exactly the sort of thing that gets lost between a journal and a headline. We wrote about what a positive IOD does and does not imply in The Other Ocean Is Watching.
This section is context, not a finding of the paper, and the two should not be blended.
NOAA CPC’s August 2026 strength table puts the probability of a very strong El Niño during October to December 2026 at 95%, where very strong means a three-month RONI value at or above +2.0°C. Its 13 August Diagnostic Discussion separately gives a 69% probability of what it calls a historic event, defined as RONI at or above +2.5°C. Those are two different questions and we set out the difference here.
One correction worth making, since it appears in coverage of this very paper: at least one write-up closes by saying NOAA gives an 81% chance of a strong El Niño for October to December. That figure is CPC’s July issuance, for a very strong event, and the August table revised it to 95%. Quoting a superseded probability alongside a research finding gives the whole package a false currency.
If the 2026 event does reach the magnitude CPC is forecasting, it becomes a third case for exactly the method this paper applies. Whether its sea-level fingerprint looks like 2014-2016, like 2023-2024, or like neither is an open question, and the same types of satellite-gravimetry and Argo observations could support a comparable analysis after the event.
Note what would not settle anything. A single further event outside the roughly 68 to 81 percent barystatic range would demonstrate event-to-event variability, not overturn the finding that mass dominated these two. And an Indian Ocean anomaly arriving without a positive IOD would not weaken the paper’s explanation for 2023, since the authors never claim a positive IOD is necessary for one. CPC’s next Diagnostic Discussion on 10 September 2026 does not test any of the sea-level findings either; it only updates whether the event is developing strongly enough to become a useful comparison.
During two strong El Niño development phases the ocean gained a similar barystatic mass-equivalent amount, 6.20 and 6.26 mm, and in both the mass term outweighed thermal expansion. The second phase was much shorter, and the authors attribute its faster rise primarily to terrestrial water storage depletion in the Americas and Africa, with particularly rapid losses in South America. Its smaller thermal signal was concentrated in the Indian Ocean, likely associated with a coincident positive Indian Ocean Dipole. None of this is the long-term sea-level-rise rate, none of it is a general law about El Niño, and none of it is about 2026. What it gives is a method and two documented cases, against which a third could be compared once the current event has run.