📄 earth_science

How Uncertain Are Estimates of Marine Carbon Dioxide Removal? Insights from a Simplified Model

Using a simplified one-dimensional model and Monte Carlo framework, this study reveals that uncertainties in marine carbon dioxide removal estimates are driven by mixed-layer properties in the short term and vertical mixing or deployment depth in the long term, creating persistent decades-long uncertainties that critically impact monitoring and carbon crediting frameworks.

Original authors: Ruth Musgrave

Published 2026-06-25

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Ruth Musgrave

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 or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine the ocean as a giant, multi-story sponge. Scientists are trying to figure out how well this sponge can soak up carbon dioxide (CO2) from the air if we add a special "cleaning agent" (like adding baking soda to the water, known as Ocean Alkalinity Enhancement). The goal is to pull CO2 out of the atmosphere to fight climate change.

This paper asks a simple but tricky question: How sure can we be about our calculations of how much carbon this sponge actually soaks up?

The author, Ruth Musgrave, built a simplified computer model (a "one-dimensional" model) to act like a vertical slice of this ocean sponge. She ran over 1,000 different simulations, tweaking the rules of the game slightly each time to see how much the results changed. Here is what she found, explained through everyday analogies:

1. The "First Few Years" are a Rollercoaster

In the beginning, right after we add the cleaning agent, the results are very shaky. It's like trying to guess how fast a cup of hot coffee will cool down when you don't know the exact temperature of the room, how windy it is, or how thick the cup is.

The Analogy: Imagine you pour a drop of dye into a cup of water. If you don't know exactly how fast the water is swirling (wind), how deep the cup is (mixed layer depth), or how sticky the water is (gas exchange), your guess about how fast the dye spreads will be all over the place.
The Finding: In the first few years, small changes in these "cup conditions" can make our estimate of carbon removal swing wildly—sometimes by as much as double or half of what we expected.

2. The "Deep Ocean" is the Long Game

After the first few years, the water starts to mix deeper. This is where the real uncertainty hides for decades. It depends on how fast the "cleaning agent" sinks from the top layer into the deep ocean and how fast the deep water comes back up.

The Analogy: Think of the ocean as a two-story house. The top floor is the "mixed layer" (where the wind blows), and the basement is the "deep ocean."
- If the stairs between the floors are wide and open (high mixing), the cleaning agent sinks fast.
- If the stairs are narrow and clogged (low mixing), the agent stays on the top floor longer.
The Finding: The biggest uncertainty comes from not knowing exactly how "wide" those stairs are (a property called pycnocline diffusivity). If the stairs are narrow, the agent stays at the top, and the ocean sucks up carbon quickly. If the stairs are wide, the agent sinks too fast before it can do its job, and it takes decades for the deep water to come back up and finish the job. This uncertainty can last for 50 years or more.

3. The "Wrong Floor" Problem

The paper also looked at where we put the cleaning agent.

The Analogy: Imagine you are trying to clean a room, but you accidentally pour the cleaning solution into the basement instead of the living room. It will take a very long time for that solution to work its way back up to where it's needed.
The Finding: If the intervention is deployed even a little bit too deep (below the top layer of water), the results are terrible. The carbon removal is much lower, and the uncertainty is huge. It's a waste of time and money.

4. The "Net Zero" Countdown

Finally, the paper looks at the "Time to Net Neutrality." This is the moment when the ocean has sucked up enough carbon to pay back the "carbon debt" created by making and shipping the cleaning agent.

The Analogy: Imagine you borrow $100 to buy a tool that earns you $100 back. But the tool costs $50 to build. You need to earn $50 to break even.
The Finding: Because our calculations are so uncertain, we don't know when we break even.
- In the "best-case" scenarios, we might break even in less than a year.
- In the "worst-case" scenarios (where the agent sinks too fast or the water doesn't mix right), it could take 50 years to break even.
- About 4% of the simulations showed it would take longer than 50 years to become "carbon negative."

The Bottom Line

The paper concludes that while we know the theory of how this works, our predictions are very uncertain because we don't fully understand the physics of the ocean (how fast it mixes, how fast gas moves, etc.).

Why does this matter?
If we want to sell "carbon credits" (certificates saying we removed carbon), we need to be sure the carbon is actually gone. This study warns that right now, we might be selling credits based on guesses that could be wrong by a factor of two, or that might take decades to actually happen. We need better measurements of the ocean's "mixing" and "wind" to know if these plans will work on the timescales we need.

Technical Summary: How Uncertain Are Estimates of Marine Carbon Dioxide Removal? Insights from a Simplified Model

Problem Statement
Marine Carbon Dioxide Removal (mCDR) techniques, specifically Ocean Alkalinity Enhancement (OAE) and Direct Ocean Removal (DOR), rely on reducing surface ocean partial pressure of CO2 to draw down atmospheric CO2. While the theoretical maximum efficiency of these interventions is known (e.g., ~0.8 mol CO2 per mol alkalinity for OAE), the time-evolution of carbon removal is highly sensitive to physical and biogeochemical processes. Current estimates of mCDR efficacy depend heavily on ocean models, yet the uncertainties inherent in these models—particularly regarding vertical mixing, air-sea gas exchange, and watermass properties—are poorly understood. Existing global models often employ fixed parameterizations that may underestimate natural variability, and their complexity makes it difficult to isolate the specific mechanisms driving uncertainty in CDR calculations. This lack of quantified uncertainty poses a fundamental challenge for Monitoring, Reporting, and Verification (MRV) frameworks and the timing of net carbon neutrality for mCDR projects.

Methodology
To systematically examine these uncertainties, the study employs a one-dimensional (1D), abiotic ocean model. This reduced-complexity approach isolates vertical processes (turbulent diffusion and air-sea gas exchange) while ignoring horizontal advection, subduction, and biological interactions. The model evolves Total Alkalinity (TA) and Dissolved Inorganic Carbon (DIC) over time, calculating surface CO2 flux based on air-sea gas exchange.

The study utilizes two primary analytical frameworks:

Sensitivity Simulations: Individual model parameters (e.g., mixed-layer depth, wind speed, gas transfer coefficient, pycnocline diffusivity, background TA/DIC, deployment depth) are perturbed by values consistent with observed natural variability. These are compared against a reference simulation to determine the "CDR sensitivity ratio" ( $r_{CDR}$ ), defined as the ratio of CDR in a sensitivity run to the reference run.
Monte Carlo Ensembles: Over 1,000 simulation pairs (baseline and intervention) are run simultaneously, with parameters drawn randomly from statistical distributions representing observed ocean variability. This approach assesses how combined uncertainties in physical processes and watermass properties manifest in CDR estimates.

The simulations are conducted for ocean depths of 200 m and 2,000 m over a 50-year period to evaluate the impact of depth on uncertainty persistence.

Key Results
The study identifies three distinct temporal regimes governing CDR uncertainty:

Initial Rapid-Equilibration ( $t \ll \tau_{mix}$ ): Within the first few years, CDR estimates are highly sensitive to parameters controlling the initial air-sea gas exchange timescale ( $\tau_{air-sea}$ ), including mixed-layer depth, wind speed, gas transfer coefficients, and background carbonate chemistry (TA, DIC, temperature). During this phase, CDR estimates can vary by a factor of two depending on parameter biases.
Alkalinity Loss from Mixed Layer ( $t \approx \tau_{mix}$ ): As vertical mixing removes alkalinity from the surface, uncertainties become dominated by the interplay between the air-sea exchange timescale and the mixing timescale ( $\tau_{mix}$ ). Uncertainties in pycnocline diffusivity ( $\kappa_{pyc}$ ) become critical, causing CDR estimates to peak in variability (up to 30%) 2–3 years post-deployment.
Deep Mixing ( $t \gg \tau_{mix}$ ): In the long term, uncertainties persist for decades if the timescale for mixing into the deep ocean is comparable to or faster than air-sea equilibration. The study finds that uncertainty in pycnocline diffusivity is the dominant factor for long-term CDR uncertainty, as it controls both the fraction of equilibrated alkalinity before deep subduction and the rate at which unequilibrated deep waters return to the surface.

Quantitative Findings:

Magnitude of Uncertainty: In the first few years, CDR estimates can differ by up to 100% (a factor of two) from the reference simulation. By 20–50 years, this narrows to 10–20%, though significant uncertainty remains for deep ocean deployments.
Deployment Depth: Deploying alkalinity below the mixed layer (even by 20–40 m) results in significantly lower efficiency and higher uncertainty, with some estimates remaining far below the reference simulation even after 50 years.
Net Neutrality Timing: The time required for an intervention to become net carbon-negative (accounting for Life Cycle Assessment costs) is highly variable. While a reference case suggests ~1.4 years, the Monte Carlo ensemble shows a wide distribution: ~30% of simulations reach net neutrality in under a year, 50% take longer than 2 years, and ~4% take longer than 50 years. This "long tail" is primarily associated with deep deployments or rapid mixing that prevents early equilibration.
Depth Dependence: Uncertainties persist longer in 2,000 m deep oceans compared to 200 m deep oceans, where the entire water column can equilibrate within the simulation timeframe.

Significance and Claims
The paper argues that uncertainties in mCDR estimates are not merely transient but can persist for decades due to the timescales of interior mixing relative to air-sea gas exchange. The authors claim these results identify a fundamental challenge for regional and global mCDR simulations, where governing physical processes are often poorly constrained.

Specifically, the study highlights that:

MRV Implications: The uncertainty in the time-evolution of carbon removal directly impacts the timing of net neutrality, which is critical for carbon crediting frameworks. Without well-quantified uncertainties, the awarding of carbon credits may be premature or inaccurate.
Model Limitations: Current ocean models often rely on parameterizations for turbulence and gas exchange that are subject to substantial spatial and temporal biases. The study suggests that validating these specific physical processes (mixing and gas transfer) is more critical for mCDR than for general climate projections.
Deployment Strategy: The results emphasize the necessity of ensuring alkalinity is deployed strictly within the surface mixed layer to maximize efficiency and minimize uncertainty.

The authors maintain a modest scope, noting that their 1D model excludes horizontal advection and biological feedbacks, which could introduce further uncertainties. However, they conclude that even in an abiotic system, uncertainties arising from physical processes alone are significant enough to warrant serious further investigation to ensure mCDR interventions are genuinely carbon-negative on decadal timescales.

1. The "First Few Years" are a Rollercoaster

2. The "Deep Ocean" is the Long Game

3. The "Wrong Floor" Problem

4. The "Net Zero" Countdown

The Bottom Line

More like this