Atmospheric Measurement Techniques
Atmospheric Measurement Techniques
AMT
Articles | Volume 13, issue 5
Articles | Volume 13, issue 5
https://doi.org/10.5194/amt-13-2257-2020
© Author(s) 2020. This work is distributed under
the Creative Commons Attribution 4.0 License.
https://doi.org/10.5194/amt-13-2257-2020
© Author(s) 2020. This work is distributed under
the Creative Commons Attribution 4.0 License.
Articles | Volume 13, issue 5
Research article
 | Highlight paper
 | 
11 May 2020
Research article | Highlight paper |  | 11 May 2020

A machine-learning-based cloud detection and thermodynamic-phase classification algorithm using passive spectral observations

Chenxi Wang, Steven Platnick, Kerry Meyer, Zhibo Zhang, and Yaping Zhou

Data sets

MODIS/Aqua Clouds 5-Min L2 Swath 1km and 5km S. Platnick, M. King, G. Wind, S. Ackerman, P. Menzel, and R. Frey https://doi.org/10.5067/MODIS/MYD06_L2.061

MYD35_L2 - MODIS/Aqua Cloud Mask and Spectral Test Results 5-Min L2 Swath 250 m and 1 km S. Ackerman, P. Menzel, R. Frey, and B. Baum https://doi.org/10.5067/MODIS/MYD35_L2.061

Continuity MODIS/Aqua Level-2 (L2) Cloud Mask Product S. Ackerman and R. Frey https://doi.org/10.5067/MODIS/CLDMSK_L2_MODIS_Aqua.001

Continuity VIIRS/SNPP Level-2 (L2) Cloud Mask Product S. Ackerman and R. Frey https://doi.org/10.5067/VIIRS/CLDMSK_L2_VIIRS_SNPP.001

Continuity MODIS/Aqua Level-2 (L2) Cloud Properties Product S. Platnick, K. Meyer, G. Wind, T. Arnold, N. Amrasinghe, B. Marchant, C. Wang, S. Ackerman, A. Heidinger, B. Holtz, Y. Li, and R. Frey https://doi.org/10.5067/MODIS/CLDPROP_L2_MODIS_Aqua.011

Continuity VIIRS/SNPP Level-2 (L2) Cloud Properties Product S. Platnick, K. Meyer, G. Wind, T. Arnold, N. Amrasinghe, B. Marchant, C. Wang, S. Ackerman, A. Heidinger, B. Holtz, Y. Li, and R. Frey https://doi.org/10.5067/VIIRS/CLDPROP_L2_VIIRS_SNPP.011

Download
Short summary
A machine-learning (ML)-based approach that can be used for cloud mask and phase detection is developed. An all-day model that uses infrared (IR) observations and a daytime model that uses shortwave and IR observations from a passive instrument are trained separately for different surface types. The training datasets are selected by using reference pixel types from collocated space lidar. The ML approach is validated carefully and the overall performance is better than traditional methods.
Read more