Rain event detection in commercial microwave link attenuation data using convolutional neural networks

Polz, Julius; Chwala, Christian; Graf, Maximilian; Kunstmann, Harald

doi:https://doi.org/10.5194/amt-13-3835-2020

Articles | Volume 13, issue 7

https://doi.org/10.5194/amt-13-3835-2020

© Author(s) 2020. This work is distributed under
the Creative Commons Attribution 4.0 License.

https://doi.org/10.5194/amt-13-3835-2020

© Author(s) 2020. This work is distributed under
the Creative Commons Attribution 4.0 License.

Articles | Volume 13, issue 7

Research article

|

15 Jul 2020

Research article |

| 15 Jul 2020

Rain event detection in commercial microwave link attenuation data using convolutional neural networks

Julius Polz, Christian Chwala, Maximilian Graf, and Harald Kunstmann

Download

Final revised paper (published on 15 Jul 2020)
Preprint (discussion started on 19 Dec 2019)

Interactive discussion

Status: closed

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

- Printer-friendly version

- Supplement

RC1: 'Review', Anonymous Referee #1, 23 Jan 2020
- AC2: 'Final Response', Julius Polz, 13 Mar 2020
RC2: 'Review of "Rain event detection in commercial microwave link attenuation data using convolutional neural networks"', Andreas Scheidegger, 03 Feb 2020
- AC1: 'Authors response to general comments by Andreas Scheidegger', Julius Polz, 13 Feb 2020
- AC2: 'Final Response', Julius Polz, 13 Mar 2020

Peer-review completion

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

AR by Julius Polz on behalf of the Authors (04 May 2020) Author's response Manuscript

ED: Referee Nomination & Report Request started (14 May 2020) by Gianfranco Vulpiani

RR by Andreas Scheidegger (18 May 2020)

ED: Publish as is (25 May 2020) by Gianfranco Vulpiani

AR by Julius Polz on behalf of the Authors (26 May 2020)

Short summary

Commercial microwave link (CML) networks can be used to estimate path-averaged rain rates. This study evaluates the ability of convolutional neural networks to distinguish between wet and dry periods in CML time series data and the ability to transfer this detection skill to sensors not used for training. Our data set consists of several months of data from 3904 CMLs covering all of Germany. Compared to a previously used detection method, we could show a significant increase in performance.