A neural network approach to estimating a posteriori distributions  of Bayesian retrieval problems

Pfreundschuh, Simon; Eriksson, Patrick; Duncan, David; Rydberg, Bengt; Håkansson, Nina; Thoss, Anke

doi:https://doi.org/10.5194/amt-11-4627-2018

Articles | Volume 11, issue 8

https://doi.org/10.5194/amt-11-4627-2018

© Author(s) 2018. This work is distributed under
the Creative Commons Attribution 4.0 License.

https://doi.org/10.5194/amt-11-4627-2018

© Author(s) 2018. This work is distributed under
the Creative Commons Attribution 4.0 License.

Articles | Volume 11, issue 8

Research article

|

09 Aug 2018

Research article |

| 09 Aug 2018

A neural network approach to estimating a posteriori distributions of Bayesian retrieval problems

Simon Pfreundschuh, Patrick Eriksson, David Duncan, Bengt Rydberg, Nina Håkansson, and Anke Thoss

Download

Final revised paper (published on 09 Aug 2018)
Preprint (discussion started on 29 Mar 2018)

Interactive discussion

Status: closed

AC: Author comment | RC: Referee comment | SC: Short comment | EC: Editor comment

- Printer-friendly version

- Supplement

RC1: 'Neural Network Retrievals', Christian Kummerow, 07 May 2018
- AC1: 'Response to Christian Kummerow', Simon Pfreundschuh, 26 Jun 2018
RC2: 'AMT-2018-91 Review Document', Anonymous Referee #2, 15 Jun 2018
- AC2: 'Response to Anonymous Referee #2', Simon Pfreundschuh, 26 Jun 2018

Peer-review completion

AR: Author's response | RR: Referee report | ED: Editor decision

AR by Simon Pfreundschuh on behalf of the Authors (26 Jun 2018) Author's response Manuscript

ED: Publish subject to technical corrections (28 Jun 2018) by Andrew Sayer

AR by Simon Pfreundschuh on behalf of the Authors (27 Jul 2018) Manuscript

Short summary

A novel neural-network-based retrieval method is proposed that combines the flexibility and computational efficiency of machine learning retrievals with the consistent treatment of uncertainties of Bayesian methods. Numerical experiments are presented that show the consistency of the proposed method with the Bayesian formulation as well as its ability to represent non-Gaussian retrieval errors. With this, the proposed method overcomes important limitations of traditional methods.