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

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.

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

References in corpus (5)

• Very Deep Convolutional Networks for Large-Scale Image Recognition
• Sequence to Sequence Learning with Neural Networks
• Striving for Simplicity: The All Convolutional Net
• Pruning Filters for Efficient ConvNets
• Multi-scale and Context-adaptive Entropy Model for Image Compression

Cited by in corpus (1)

• Data Assimilation in the Latent Space of a Neural Network