Search papers, labs, and topics across Lattice.
This paper introduces HLOBA, a hybrid-ensemble data assimilation method that operates in a latent space learned via an autoencoder to improve accuracy, efficiency, and uncertainty quantification in atmospheric data assimilation. HLOBA maps both model forecasts and observations into a shared latent space using an autoencoder and an observation-to-latent-space mapping network, fusing them through a Bayesian update weighted by time-lagged ensemble forecasts. Experiments demonstrate that HLOBA matches dynamically constrained four-dimensional DA methods in analysis and forecast skill, while achieving end-to-end inference-level efficiency and providing element-wise uncertainty estimates.
Achieve the accuracy of complex 4D data assimilation with the speed of end-to-end inference by operating in a learned latent space.
Data assimilation (DA) combines model forecasts and observations to estimate the optimal state of the atmosphere with its uncertainty, providing initial conditions for weather prediction and reanalyses for climate research. Yet, existing traditional and machine-learning DA methods struggle to achieve accuracy, efficiency and uncertainty quantification simultaneously. Here, we propose HLOBA (Hybrid-Ensemble Latent Observation-Background Assimilation), a three-dimensional hybrid-ensemble DA method that operates in an atmospheric latent space learned via an autoencoder (AE). HLOBA maps both model forecasts and observations into a shared latent space via the AE encoder and an end-to-end Observation-to-Latent-space mapping network (O2Lnet), respectively, and fuses them through a Bayesian update with weights inferred from time-lagged ensemble forecasts. Both idealized and real-observation experiments demonstrate that HLOBA matches dynamically constrained four-dimensional DA methods in both analysis and forecast skill, while achieving end-to-end inference-level efficiency and theoretical flexibility applies to any forecasting model. Moreover, by exploiting the error decorrelation property of latent variables, HLOBA enables element-wise uncertainty estimates for its latent analysis and propagates them to model space via the decoder. Idealized experiments show that this uncertainty highlights large-error regions and captures their seasonal variability.
