Lagrangian Flux Decomposition (LFD): Fast Probabilistic Marine Carbon Dioxide Removal Modeling via Stochastic Lagrangian Transport and Carbonate System Accounting

[Geoengineering News]

https://essopenarchive.org/doi/full/10.22541/essoar.15002617/v1

‎

Authors: Jordan Hood Miller, Trond Kristiansen, and Dr. Momme Butenschön 

30 April 2026

Primary Abstract

Marine carbon dioxide removal (mCDR) requires ocean modeling to quantify efficacy, environmental risk, and uncertainty across scales. Estimating air–sea carbon fluxes and associated impacts is crucial for project planning and for Measurement, Reporting, and Verification under emerging standards. Existing approaches, primarily Eulerian dynamic models, face scalability limits due to tradeoffs between resolution and domain size, restricting their ability to assess multiple sites, configurations, and future ocean scenarios. We introduce the Lagrangian Flux Decomposition framework, which quantifies the difference in air–sea carbon dioxide flux between an mCDR intervention and a counterfactual and attributes that difference to tracer particle contributions over time. The framework reuses ocean and atmospheric reanalysis or machine learning products as forcing fields within a stochastic ensemble based on Markov chain sampling of transport and flux perturbations. This reduces computational cost while enabling high resolution global analysis. Its modular design integrates Lagrangian transport, carbonate chemistry, and air–sea flux models to evaluate multiple scenarios and quantify uncertainty. The method produces probability distributions of changes in ocean conditions and air–sea flux. We apply the framework at four locations to simulate dispersal, efficiency, and carbonate system impacts. Results show it reproduces key physical and chemical processes with high efficiency and aligns with published models. The approach supports marine spatial planning and verification of carbon removal credits, while explicitly linking input and parameter uncertainty to outcome uncertainty and identifying priorities for further validation.

Source: ESS Open Archive