REACT: Physically and Chemically Consistent Reconstruction of Marine Active Tracers

2026-10-02

AI Summary

This study addresses the physicochemical inconsistencies in reconstructing ocean pH from sparse observations, where conventional AI models neglect carbonate chemistry closure and carbon conservation. To overcome this limitation, we propose REACT, a framework that introduces a "carbon-first" paradigm by embedding physical conservation laws and chemical equilibrium constraints into a deep learning architecture. Specifically, dissolved inorganic carbon serves as the foundational variable for conservative advection-diffusion modeling; following source-term corrections, it is mapped to pH via neural network decoding under hard carbonate equilibrium constraints. Experimental results demonstrate that REACT reduces the normalized root-mean-square error of pH by 14.7% and decreases chemical consistency errors by 24.0%, while exhibiting superior robustness across multiple temporal scales.

Abstract

Reconstructing global sea surface pH from sparse observations is critical for monitoring ocean acidification and understanding marine carbon cycling. Traditional assimilation and inverse models are physically grounded but costly for large-scale reconstruction. Recent black-box and physics-guided AI models improve efficiency, but are mainly designed for passive tracers, where the reconstructed variable is also the transported inventory. In contrast, pH is an active carbonate tracer: it is the prediction target, while dissolved inorganic carbon (DIC) is the conserved carbon inventory. This mismatch can produce low pH error while violating carbonate closure and source-free carbon conservation. To address this, we introduce \textbf{REACT}, a carbon-first reconstruction framework that decouples transport, active correction, and chemical decoding. REACT transports a latent carbonate state with a conservative advection--diffusion solver, captures non-conservative carbon-cycle variations with a source module, decodes the corrected state into pH, and constrains the output through carbonate equilibrium. This design keeps pH as the target while enforcing consistency on the underlying carbon state. On simulation data, REACT reduces pH NRMSE by (14.7%) and chemical consistency error by (24.0%) over the best baseline. Cross-temporal-scale evaluations show robustness against error accumulation from coarse to fine temporal scales, and ablation studies validate the effectiveness of each component.

Problem

• sea surface pH reconstruction
• active tracer
• carbonate chemistry consistency
• ocean acidification
• dissolved inorganic carbon

Innovation

• Active Tracer Reconstruction
• Physics-Informed AI
• Carbonate Equilibrium
• Advection-Diffusion Solver
• Decoupled Framework