State-observation augmented diffusion model for nonlinear assimilation with unknown dynamics

2024/07/31 by Zhuoyuan Li, Bin Dong, Li, Zhuoyuan +3 · 3 citations
Earth and Planetary Sciences · Environmental Science · #49N45 #60J60 #62F15 #68T20 #Climate variability and models #FOS: Computer and information sciences #Geophysics and Gravity Measurements #Machine Learning (cs.LG) #Machine Learning (stat.ML) #Meteorological Phenomena and Simulations

paper · pdf · doi:10.48550/arxiv.2407.21314

openalex publication_date 2024/07/31 · openalex created_date 2024/08/04 · openalex updated_date 2026/07/28

Abstract

Data assimilation has become a key technique for combining physical models with observational data to estimate state variables. However, classical assimilation algorithms often struggle with the high nonlinearity present in both physical and observational models. To address this challenge, a novel generative model, termed the State-Observation Augmented Diffusion (SOAD) model is proposed for data-driven assimilation. The marginal posterior associated with SOAD has been derived and then proved to match the true posterior distribution under mild assumptions, suggesting its theoretical advantages over previous score-based approaches. Experimental results also indicate that SOAD may offer improved performance compared to existing data-driven methods.

Cited by

• Reconstructing Multi-Scale Physical Fields from Extremely Sparse Measurements with an Autoencoder-Diffusion Cascade
• Dynamical Low-Rank Approximations for Kalman Filtering
• Divergence-kernel method for linear responses of densities and generative models

Related

• FlowDAS: A Stochastic Interpolant-based Framework for Data Assimilation
• Using Diffusion Models to do Data Assimilation
• PhyDA: Physics-Guided Diffusion Models for Data Assimilation in Atmospheric Systems
• Generating observation guided ensembles for data assimilation with denoising diffusion probabilistic model
• Score-based Data Assimilation