Equation‐Free Surrogate Modeling of Geophysical Flows at the Intersection of Machine Learning and Data Assimilation

Abstract There is a growing interest in developing data‐driven reduced‐order models for atmospheric and oceanic flows that are trained on data obtained either from high‐resolution simulations or satellite observations. The non‐intrusive nature offer significant computational savings compared to large‐scale numerical models. These low‐dimensional can be utilized reduce the burden of generating forecasts estimating model uncertainty without losing key information needed assimilation (DA) produce accurate state estimates. This paper aims at exploring an equation‐free surrogate modeling approach intersection machine learning DA Earth system modeling. With this objective, we introduce end‐to‐end (NIROM) framework equipped with contributions modal decomposition, time series prediction, optimal sensor placement, sequential DA. Specifically, use proper orthogonal decomposition (POD) identify dominant structures flow, long short‐term memory network dynamics POD modes. NIROM integrated within deterministic ensemble Kalman filter (DEnKF) incorporate sparse noisy observations locations through QR pivoting. feasibility benefit proposed demonstrated NOAA Optimum Interpolation Sea Surface Temperature (SST) V2 set. Our results indicate stable long‐term forecasting SST reasonable level accuracy. Furthermore, prediction accuracy gets improved by almost one order magnitude DEnKF algorithm.