English

Attention-based Convolutional Autoencoders for 3D-Variational Data Assimilation

Machine Learning 2021-01-07 v1 Computational Engineering, Finance, and Science

Abstract

We propose a new 'Bi-Reduced Space' approach to solving 3D Variational Data Assimilation using Convolutional Autoencoders. We prove that our approach has the same solution as previous methods but has significantly lower computational complexity; in other words, we reduce the computational cost without affecting the data assimilation accuracy. We tested the new method with data from a real-world application: a pollution model of a site in Elephant and Castle, London and found that we could reduce the size of the background covariance matrix representation by O(10^3) and, at the same time, increase our data assimilation accuracy with respect to existing reduced space methods.

Keywords

Cite

@article{arxiv.2101.02121,
  title  = {Attention-based Convolutional Autoencoders for 3D-Variational Data Assimilation},
  author = {Julian Mack and Rossella Arcucci and Miguel Molina-Solana and Yi-Ke Guo},
  journal= {arXiv preprint arXiv:2101.02121},
  year   = {2021}
}

Comments

Published in Computer Methods in Applied Mechanics and Engineering in Dec 2020

R2 v1 2026-06-23T21:50:45.107Z