Composing a surrogate observation operator for sequential data assimilation

TL;DR

This paper introduces a neural network-based method for constructing surrogate observation operators in data assimilation when the true operator is unknown, improving state estimation accuracy.

Contribution

It presents a novel iterative neural network approach to compose surrogate operators, outperforming fixed-operator methods in data assimilation tasks.

Findings

Outperforms fixed-operator approaches in twin experiments

Neural network surrogate reduces observation-model discrepancy

Improves state estimation accuracy in data assimilation

Abstract

In data assimilation, state estimation is not straightforward when the observation operator is unknown. This study proposes a method for composing a surrogate operator when the true operator is unknown. A neural network is used to improve the surrogate model iteratively to decrease the difference between the observations and the results of the surrogate model. A twin experiment suggests that the proposed method outperforms approaches that tentatively use a specific operator throughout the data assimilation process.