Preprints
https://doi.org/10.5194/gmd-2021-353
https://doi.org/10.5194/gmd-2021-353

Submitted as: development and technical paper 15 Nov 2021

Submitted as: development and technical paper | 15 Nov 2021

Review status: this preprint is currently under review for the journal GMD.

DINCAE 2: multivariate convolutional neural network with error estimates to reconstruct sea surface temperature satellite and altimetry observations

Alexander Barth, Aida Alvera-Azcárate, Charles Troupin, and Jean-Marie Beckers Alexander Barth et al.
  • GHER, University of Liège, Liège, Belgium

Abstract. DINCAE (Data INterpolating Convolutional Auto-Encoder) is a neural network to reconstruct missing data (e.g. obscured by clouds or gaps between tracks) in satellite data. Contrary to standard image reconstruction (in-painting) with neural networks, this application requires a method to handle missing data (or data with variable accuracy) already in the training phase. Instead of using a standard L2 (or L1) cost function, the neural network (U-Net type of network) is optimized by minimizing the negative log likelihood assuming a Gaussian distribution (characterized by a mean and a variance). As a consequence, the neural network also provides an expected error variance of the reconstructed field (per pixel and per time instance).

In this updated version DINCAE 2.0, the code was rewritten in Julia and a new type of skip connection has been implemented which showed superior performance with respect to the previous version. The method has also been extended to handle multivariate data (an example will be shown with sea-surface temperature, chlorophyll concentration and wind fields). The improvement of this network is demonstrated in the Adriatic Sea.

Convolutional networks work usually with gridded data as input. This is however a limitation for some data types used in oceanography and in Earth Sciences in general, where observations are often irregularly sampled. The first layer of the neural network and the cost function have been modified so that unstructured data can also be used as inputs to obtain gridded fields as output. To demonstrate this, the neural network is applied to along-track altimetry data in the Mediterranean Sea. Results from a 20-year reconstruction are presented and validated. Hyperparameters are determined using Bayesian optimization and minimizing the error relative to a development dataset.

Alexander Barth et al.

Status: open (until 10 Jan 2022)

Comment types: AC – author | RC – referee | CC – community | EC – editor | CEC – chief editor | : Report abuse

Alexander Barth et al.

Alexander Barth et al.

Viewed

Total article views: 179 (including HTML, PDF, and XML)
HTML PDF XML Total BibTeX EndNote
137 40 2 179 3 2
  • HTML: 137
  • PDF: 40
  • XML: 2
  • Total: 179
  • BibTeX: 3
  • EndNote: 2
Views and downloads (calculated since 15 Nov 2021)
Cumulative views and downloads (calculated since 15 Nov 2021)

Viewed (geographical distribution)

Total article views: 177 (including HTML, PDF, and XML) Thereof 177 with geography defined and 0 with unknown origin.
Country # Views %
  • 1
1
 
 
 
 
Latest update: 29 Nov 2021
Download
Short summary
Earth-observing satellites provide routine measurement of several ocean parameters. However, these dataset have a significant amount of missing data due to the presence of clouds or other limitations of the employed sensors. This paper describes a method to infer the value of missing data in satellite data based on a convolutional autoencoder (a specific type of neural network architecture). The technique also provides a reliable error estimate of the interpolated value.