Search papers, labs, and topics across Lattice.
This study introduces a generative state-space model that leverages sparse and noisy observations to enhance ocean modeling, overcoming the limitations of traditional complete reanalysis datasets. By employing a hidden Markov model framework with neural networks for both initial-state and state-transition modules, the authors optimize the model using an expectation-maximization algorithm to reconstruct high-fidelity ocean fields. Experimental results on CMIP6 and FY-3D satellite data reveal that the model significantly improves the representation of ocean-state dynamics, demonstrating the potential of incomplete observations to boost evolutionary performance in ocean modeling.
Sparse observations can enhance ocean modeling performance, challenging the reliance on complete datasets that limit model capabilities.
Data-driven methods have revolutionized ocean modeling, yet current approaches rely heavily on complete reanalysis datasets, imposing computational constraints and limiting model performance to that of the training data. Here, we present a generative state-space model and an optimization framework that enable learning directly from sparse and noisy observations. The model is essentially a hidden Markov model with a continuous state space, where oceanic physical quantities are treated as hidden states and measurements as observations, enabling a unified representation of ocean fields and observational data. Both the initial-state and state-transition modules are implemented as neural networks to capture the complexity and temporal evolution of ocean states, while the emission module is formulated as a masked Gaussian distribution. To train the model from sparse observations, we derive an optimization framework based on the expectation-maximization (EM) algorithm. The framework alternately reconstructs high-fidelity ocean fields via Langevin dynamics and optimizes deep neural networks to capture temporal evolution. Theoretical analysis shows that the framework maximizes the likelihood of observations under the generative model. For efficiency, we assume that ocean-state evolution follows a stationary, ergodic, and Markovian stochastic process and adopt only length-two state sequences during optimization. Experiments on CMIP6 simulation data and FY-3D satellite data demonstrate high-fidelity reconstruction and accurate prediction, showing that sparse observations can directly improve the model's representation of ocean-state dynamics. This work offers a scalable pathway for next-generation Earth system models to learn directly from sparse, incomplete real-world observations.
