Analog ensemble data assimilation and a method for constructing analogs with variational autoencoders

TL;DR

This paper introduces a novel approach combining analog ensemble data assimilation with variational autoencoders to improve ensemble filtering methods, demonstrating comparable or superior performance in a Lorenz-`96 model.

Contribution

It proposes a new method for constructing analogs using VAEs and integrates this into ensemble data assimilation, showing improved performance over traditional methods.

Findings

Analog methods with catalog analogs improve EnOI performance.

Constructed analogs via VAEs perform as well as full ensemble filters.

The proposed method is robust across various tuning parameters.

Abstract

It is proposed to use analogs of the forecast mean to generate an ensemble of perturbations for use in ensemble optimal interpolation (EnOI) or ensemble variational (EnVar) methods. A new method of constructing analogs using variational autoencoders (VAEs; a machine learning method) is proposed. The resulting analog methods using analogs from a catalog (AnEnOI), and using constructed analogs (cAnEnOI), are tested in the context of a multiscale Lorenz-`96 model, with standard EnOI and an ensemble square root filter for comparison. The use of analogs from a modestly-sized catalog is shown to improve the performance of EnOI, with limited marginal improvements resulting from increases in the catalog size. The method using constructed analogs (cAnEnOI) is found to perform as well as a full ensemble square root filter, and to be robust over a wide range of tuning parameters.