GeoPrior-Mamba: Structured Process Priors with Mamba for Fine-Resolution XCO2 Reconstruction

2026-10-07

AI Summary

This study addresses the challenge of learning heterogeneous source-sink relationships in high-resolution XCO2 reconstruction from sparse satellite observations by proposing the GeoPrior-Mamba framework. This framework pioneers the use of large language models to offline organize carbon cycle process knowledge, generating deterministic structured priors rather than direct numerical predictions. These priors are injected into a multi-directional Mamba backbone via lightweight adapters, enabling adaptive spatial instantiation of geo-ecological knowledge. Experimental results demonstrate that the proposed method achieves an RMSE of 0.81 ppm (R² = 0.93), reducing errors by 48.2% compared to CAMS. Furthermore, it exhibits significantly faster convergence than prior-free baselines and is validated against TCCON ground-based measurements, establishing a new paradigm for high-fidelity carbon field reconstruction.

Abstract

Reconstructing fine-resolution column-averaged dry-air CO2 (XCO2) fields from sparse satellite observations requires models to infer spatial structure that is only weakly constrained by direct measurements. Existing learning-based methods typically treat environmental covariates as ordinary numerical inputs and must therefore learn heterogeneous source-sink relationships largely from sparse supervision. We introduce GeoPrior-Mamba, a multi-directional Mamba framework augmented with offline language-model-induced structured process priors. Rather than using a language model to predict XCO2, we use it before training to organize relative process knowledge for biospheric uptake, ecosystem respiration, and anthropogenic emissions into deterministic prior tables. These priors are spatially instantiated using geographic, ecological, emission-related, and seasonal information and are adaptively injected into the reconstruction backbone through a lightweight knowledge adapter. Using OCO-2 observations from 2018-2020, GeoPrior-Mamba achieves an RMSE of 0.81 ppm and an R2 of 0.93 on held-out observations, reducing RMSE by 48.2% relative to CAMS background interpolation and by 3.1% relative to Trans-XCO2 under the same evaluation protocol. Ablation experiments show a measurable contribution from the knowledge-prior branch and substantially faster convergence than the knowledge-free Mamba backbone. Independent TCCON evaluation further supports the consistency of the reconstructed fields with ground-based column CO2 measurements. These results suggest that language models can provide a practical mechanism for constructing structured process priors when globally consistent process-response representations are difficult to obtain directly, while remaining outside the numerical prediction loop.

Problem

• XCO2 reconstruction
• sparse satellite observations
• fine-resolution
• source-sink relationships
• structured process priors

Innovation

• Mamba
• Structured Process Priors
• Large Language Model
• XCO2 Reconstruction
• Knowledge Adapter