Composing a surrogate observation operator for sequential data assimilation
This addresses a specific challenge in data assimilation for fields like meteorology or oceanography, but it is incremental as it builds on existing neural network techniques.
The study tackled the problem of state estimation in data assimilation when the observation operator is unknown by proposing a method to compose a surrogate operator using a neural network, which iteratively improves to reduce differences from observations, and a twin experiment showed it outperforms methods using a fixed operator.
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.