Align-DA: Align Score-based Atmospheric Data Assimilation with Multiple Preferences

TL;DR

Align-DA introduces a data-driven, alignment-based approach to atmospheric data assimilation, replacing manual tuning with reward-guided priors, leading to improved analysis quality in high-dimensional, observation-sparse settings.

Contribution

This paper presents a novel generative, score-based framework for data assimilation that uses reward signals for automatic background prior alignment, enhancing analysis accuracy without manual tuning.

Findings

Consistent improvement in analysis quality across metrics

Effective use of multiple reward signals for alignment

Automatic adaptation of complex background priors

Abstract

Data assimilation (DA) aims to estimate the full state of a dynamical system by combining partial and noisy observations with a prior model forecast, commonly referred to as the background. In atmospheric applications, this problem is fundamentally ill-posed due to the sparsity of observations relative to the high-dimensional state space. Traditional methods address this challenge by simplifying background priors to regularize the solution, which are empirical and require continual tuning for application. Inspired by alignment techniques in text-to-image diffusion models, we propose Align-DA, which formulates DA as a generative process and uses reward signals to guide background priors, replacing manual tuning with data-driven alignment. Specifically, we train a score-based model in the latent space to approximate the background-conditioned prior, and align it using three complementary reward signals for DA: (1) assimilation accuracy, (2) forecast skill initialized from the assimilated state, and (3) physical adherence of the analysis fields. Experiments with multiple reward signals demonstrate consistent improvements in analysis quality across different evaluation metrics and observation-guidance strategies. These results show that preference alignment, implemented as a soft constraint, can automatically adapt complex background priors tailored to DA, offering a promising new direction for advancing the field.