Development and validation of a supervised machine learning radar Doppler spectra peak-finding algorithm

Kalesse, Heike; Vogl, Teresa; Paduraru, Cosmin; Luke, Edward

doi:https://doi.org/10.5194/amt-12-4591-2019

Articles | Volume 12, issue 8

https://doi.org/10.5194/amt-12-4591-2019

© Author(s) 2019. This work is distributed under
the Creative Commons Attribution 4.0 License.

https://doi.org/10.5194/amt-12-4591-2019

© Author(s) 2019. This work is distributed under
the Creative Commons Attribution 4.0 License.

Articles | Volume 12, issue 8

Research article

|

30 Aug 2019

Research article |

| 30 Aug 2019

Development and validation of a supervised machine learning radar Doppler spectra peak-finding algorithm

Heike Kalesse, Teresa Vogl, Cosmin Paduraru, and Edward Luke

Data sets

High Spectral Resolution Lidar (HSRL) B. Ermold, E. Eloranta, H. Michelsen, J. Garcia, J. Goldsmith, and R. Bambha https://doi.org/10.5439/1025200

Ka ARM Zenith Radar (KAZRGE) A. Matthews, B. Isom, D. Nelson, I. Lindenmaier, J. Hardin, K. Johnson, and N. Bharadwaj https://doi.org/10.5439/1025214

Ka ARM Zenith Radar (KAZRSPECCMASKGECOPOL) A. Matthews, B. Isom, D. Nelson, I. Lindenmaier, J. Hardin, K. Johnson, and N. Bharadwaj https://doi.org/10.5439/1025218

Short summary

In a cloud, different particles like liquid water droplets and ice particles can exist simultaneously. To study the evolution of cloud particles from cloud top to bottom one has to find out how many different types of particles with different fall velocities are present. This can be done by analyzing the number of peaks in upward-looking cloud radar Doppler spectra. A new machine-learning algorithm (named PEAKO) that determines the number of peaks is introduced and compared to existing methods.