📄 earth_science

Detecting regional deep-ocean steric changes remains challenging with sea level budget analyses

This study evaluates the use of satellite altimetry, gravimetry, and Argo data to infer deep-ocean steric changes as a residual in sea level budgets, finding that robust regional detection remains challenging due to the method's lack of validation against long-term observations and the dominance of non-oceanographic gravity signals in certain regions like the Northeast Atlantic.

Original authors: Yang Zhang, Xinfeng Liang, Don Chambers, Igor Yashayaev

Published 2026-07-06✓ Author reviewed ⓘ

📖 4 min read☕ Coffee break read

CC BY 4.0

Original authors: Yang Zhang, Xinfeng Liang, Don Chambers, Igor Yashayaev

Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine the Earth's ocean as a giant, three-layered cake. The top layer is the "upper ocean" (the first 2,000 meters), which we can measure quite well using floating robots called Argo floats. The bottom layer is the "deep ocean," which is vast, dark, and incredibly hard to reach.

Scientists want to know if this deep bottom layer is getting warmer and expanding (which would raise sea levels). To figure this out without diving to the bottom, they use a clever accounting trick called the Residual Method.

The "Leftover" Trick

Think of the total sea level rise as a pie. Scientists have three ways to measure the ingredients of that pie:

Total Sea Level: Measured by satellites looking down from space (like a ruler measuring the whole cake).
Ocean Mass: Measured by satellites that weigh the water (like a scale measuring how much water is added).
Upper Ocean Expansion: Measured by the floating robots (like measuring how much the top layer of the cake swells due to heat).

The logic is simple: If you take the Total Pie and subtract the Weight and the Top Layer, whatever is left over must be the Deep Layer.

The Problem: The "Leftover" is Noisy

The authors of this paper tried to use this "leftover" trick to see what was happening in the deep ocean across different regions. They compared their "leftover" math against actual measurements taken by deep-sea moorings (like underwater weather stations) at seven different spots around the world.

The Result? The math didn't match the reality.

The Reality: The deep ocean was changing, but only a tiny bit (like a quiet whisper).
The Math: The "leftover" calculation showed huge, wild swings (like a screaming siren).

It turned out that the "leftover" wasn't actually the deep ocean. It was mostly noise—errors and static from the measuring tools themselves. Just like trying to hear a whisper in a room full of construction noise, the signal of the deep ocean was drowned out by the errors in the satellite data.

The Mystery of the Northeast Atlantic

The biggest confusion happened in the Northeast Atlantic. Here, the "leftover" math suggested the deep ocean was expanding rapidly, raising sea levels by nearly 4 millimeters a year. But the actual underwater sensors said it was only rising by about 0.4 millimeters. That's a huge difference!

The authors investigated why the math was so wrong. They looked at the "weight" data (satellite gravity) and found a strange, fake signal.

The Analogy:
Imagine you are trying to weigh a fish in a bowl. You put the bowl on a scale, but the scale is sitting on a shaky floor that is slowly sinking. The scale thinks the fish is getting lighter because the floor is dropping, not because the fish is losing weight.

In this case, the "shaky floor" was a non-oceanographic gravity signal. The satellites (GRACE/FO) were detecting a change in gravity caused by the solid Earth beneath the ocean (perhaps the ground shifting or a geological process), not by water moving. The scientists had mistaken a "ground signal" for a "water signal."

The Bottom Line

The Trick Failed: Using the "leftover" method to measure regional deep-ocean changes is currently too unreliable. The errors in the tools are too big compared to the tiny changes happening deep down.
The False Alarm: In the Northeast Atlantic, a major discrepancy wasn't caused by the ocean warming up; it was caused by the satellites picking up a signal from the solid Earth that looked like water loss but wasn't.

In short: We are currently trying to hear a pin drop in a hurricane. Until we can fix the "noise" in our satellite tools and understand the "shaky floor" beneath them, we cannot accurately use sea-level math to detect regional deep-ocean changes.

Technical Summary: Detecting Regional Deep-Ocean Steric Changes Remains Challenging with Sea Level Budget Analyses

Problem Statement
Detecting and quantifying deep-ocean changes below 2000 meters remains a critical challenge for the oceanographic and climate communities due to the sparse spatial and temporal coverage of in situ observations. While deep-ocean heat uptake contributes to global mean sea-level rise (approximately 0.11 ± 0.10 mm/year), existing estimates suffer from large uncertainties. A promising approach to infer deep-ocean steric (density) change is the "residual method," which calculates the difference between total sea level (from satellite altimetry), ocean mass (from satellite gravimetry/GRACE-FO), and upper-ocean steric change (from Argo floats). Despite increasing usage, this method has not been rigorously validated against long-term, direct deep-ocean observations at regional scales. Previous studies have yielded mixed results, with some suggesting the method cannot detect deep signals due to noise, while others claim agreement with hydrographic trends in specific regions.

Methodology
The authors evaluated the residual method's ability to detect regional deep-ocean steric variability by comparing residual estimates against directly observed Deep Steric Sea Level anomalies (DSSLa) at seven long-term hydrographic time-series sites spanning 2005–2022. These sites include open-ocean stations (BATS, HOT), western boundary current regions (RAPIDwest, MOVE3), and eastern boundary regions (RAPIDeast, RAPIDmarwest, MOVE1).

The sea-level budget framework decomposes total sea level ( $\eta_{total}$ ) into ocean mass ( $\eta_{mass}$ ) and steric components ( $\eta_{steric}$ ). The deep-ocean steric anomaly ( $\eta_{deep}$ ) is derived as the residual:
$\eta_{deep} = \eta_{total} - \eta_{mass} - \eta_{upper\_steric}$

Data sources included:

Total Sea Level: Delayed-mode Level-4 satellite altimetry (CMEMS).
Ocean Mass: JPL GRACE/FO Release 06.3 mascon solutions.
Upper-Ocean Steric: Ensemble mean of five gridded Argo products (BOA, IAP, GDCSM, JAMSTEC, SIO).
Validation Data: Direct hydrographic measurements from BATS, HOT, RAPID arrays, and MOVE arrays.
Benchmarking: The Estimating the Circulation and Climate of the Ocean (ECCO) V4r4 state estimate, which is dynamically consistent and closes the sea-level budget locally, was used to diagnose discrepancies.

The study analyzed interannual variability and linear trends, employing Monte Carlo simulations to quantify uncertainties and confidence intervals.

Key Results

Limited Detection Capability: The residual method fails to reliably reproduce the magnitude of observed deep-ocean steric variability. At all seven sites, the standard deviation of the residual time series (8.2–19.3 mm) was substantially larger than the directly observed DSSLa (2.7–5.8 mm), indicating the residual is dominated by noise, instrumental drifts, or errors in the budget terms.
Correlation Issues: While raw correlations between residuals and observations were moderate to strong at some sites (e.g., MOVE3), detrending revealed that much of this agreement was driven by shared long-term trends rather than interannual variability. After removing trends, correlations dropped significantly (e.g., from 0.67 to 0.08 at RAPIDeast), suggesting that similar linear trends may arise from unrelated processes.
Trend Discrepancies: In most cases, the residual trends were substantially larger than the observed DSSLa trends, and their 95% confidence intervals did not overlap. For instance, at RAPIDeast, the residual trend was 3.61 ± 0.28 mm/year, nearly eight times larger than the observed trend of 0.41 ± 0.08 mm/year.
Diagnosis of Systematic Error: The study identified that the large misclosure in the Northeast Atlantic (specifically at RAPIDeast) is not caused by Argo salinity drift, as previously hypothesized by other studies. Instead, the discrepancy is dominated by a non-oceanographic gravity signal in the GRACE/FO data.
- Spatial analysis showed a basin-scale negative mass trend in the Northeast Atlantic in GRACE/FO data that is absent in ECCO and inconsistent with physical oceanography.
- If the GRACE/FO mass trend were real, it would imply a massive strengthening of the East-West pressure gradient and a significant trend in the lower limb of the Atlantic Meridional Overturning Circulation (AMOC). However, RAPID observations and ECCO do not support such a trend.
- Comparisons with ECCO confirmed that the mismatch resides in the mass/OBP term, not in total sea level or steric components.

Significance and Claims
The paper concludes that detecting regional deep-ocean steric changes using sea-level budget analyses remains challenging, even with decade-long records. The signal-to-noise ratio of the residual approach is insufficient for robust detection in most regions.

Crucially, the authors identify a significant, previously unstudied issue with GRACE/FO data in the Northeast Atlantic. They argue that the apparent mass loss observed by GRACE/FO in this region is likely a non-oceanographic gravity signal originating from the solid Earth, rather than a true oceanographic mass change or a simple data processing error. This finding challenges previous interpretations that attributed sea-level budget misclosures in the North Atlantic primarily to Argo salinity drift. The authors caution that until the solid Earth process causing this gravity signal is isolated and modeled, users of GRACE/FO data for oceanographic studies in the Northeast Atlantic should exercise caution. The study underscores the necessity of independent validation (such as long-term moorings and dynamically consistent state estimates) when interpreting residual sea-level budgets.

The "Leftover" Trick

The Problem: The "Leftover" is Noisy

The Mystery of the Northeast Atlantic

The Bottom Line

More like this