cs.LGNov 16, 2026

Learning Calibrated and Samplable Predictive Distributions of Surface-Ocean ΔpΔpCO2_2 Using Tabular Diffusion Models

Authors: Gabriele Accarino, Mohammad M. Amirian, Thea H. Heimdal, Amanda R. Fay, Galen A. McKinley, Viviana Acquaviva

Organizations: Department of Earth and Environmental Engineering, Columbia University, New York, NY, USA · CUNY New York City College of Technology, Brooklyn, NY, USA · Lamont-Doherty Earth Observatory, Columbia University, Palisades, NY, USA

Abstract

The ocean absorbs almost 30% of anthropogenic CO2_2 emissions, but quantifying this sink requires estimating surface-ocean CO2_2 from observations that are sparse in space and time. Most existing machine-learning estimates of this field are deterministic, giving a single value per grid point, with uncertainty approximated post hoc via residuals, ensembles, or prescribed distributions. We instead learn a conditional predictive distribution for each observation: using Treeffuser, a conditional diffusion model with gradient-boosted-tree score functions, we estimate the full predictive distribution of ΔpΔpCO2_2 (the sea-air ppCO2_2 difference) from 13 environmental, spatial, and temporal predictors. On a held-out SOCAT test set, Treeffuser attains the lowest continuous ranked probability score among the uncertainty-quantification methods evaluated (8.229 μμatm) and the coverage closest to the nominal 68% / 95% levels (0.686 / 0.930), while retaining point-prediction skill comparable to a deterministic XGBoost baseline (RMSE 18.4 vs. 17.7 μμatm; R2R^2 0.77 vs. 0.79). These results show that conditional diffusion can provide well-calibrated, observation-level uncertainties while largely preserving point-prediction skill. Because Treeffuser is generative, its samples can support probabilistic analyses of ΔpΔpCO2_2; extending this to spatially coherent field reconstruction will require modeling dependence across grid cells and evaluation on spatially held-out data.

Figures & tables

Appendix figures & tables10 assets

Supplementary material from the paper’s appendix.

Appendix

Explore similar work

CardsList
  1. Amortized Probabilistic Retrieval of Atmospheric CO2 from OCO-2 Spectra Using Deep Learning with Laplace Approximations and Normalizing Flows

    Jun 16, 2026Alejandro Calle-Saldarriaga, Felix Jimenez, Jack Grosskreuz +3Amortized InferenceRemote Sensing

  2. TSCoNet: A Two-Stage Copula CNN-LSTM for Uncertainty-Aware Spatio-Temporal Forecasting

    Jul 11, 2026Jongwook Kim, Jong-Min KimMultivariate Time Series ForecastingCopula Models

  3. Proper Scoring Rule-based Diffusion for Probabilistic Weather Forecasting

    Sep 29, 2026Joonhyeong Park, Giung Nam, Hyungi Lee +3Probabilistic ForecastingProper Scoring Rules