Atmospheric Radiation Measurement Climate Research Facility US Department of Energy

mplcmaskml > Micropulse Lidar Cloud Mask Machine Learning VAPVAP Type(s) > Evaluation

The Micropulse Lidar Cloud Mask Machine Learning (MPLCMASKML) value-added product is an alternative approach to traditional cloud detection algorithms. Taking advantage of machine learning capabilities and capitalizing on the trained eye as an interpreter of lidar images, a deep neural network is trained to recognize desired cloud features in MPL lidar data.

This approach has been applied to data from the ARM fast-switching polarized micropulse lidars to develop a machine learning model that can segment images semantically and produce pixel-to-pixel predictions of clouds in lidar images (Cromwell and Flynn 2019). In addition to these cloud predictions, the model provides a confidence rating of each pixel prediction. MPLCMASKML provides the cloud mask as well as the number of cloud layers and the cloud layer boundaries.

Reference: Cromwell E and D Flynn. 2019. “Lidar Cloud Detection With Fully Convolutional Networks.” In 2019 IEEE Winter Conference on Applications of Computer Vision (WACV), 619-627, doi:10.1109/WACV.2019.00071.

Primary Derived Measurements


  • Fixed
  • AMF1
  • AMF2
  • AMF3


Flynn D, E Cromwell, and D Zhang. 2021. Micropulse Lidar Cloud Mask Machine-Learning Value-Added Product Report. Ed. by Robert Stafford, ARM user facility. DOE/SC-ARM-TR-274. 10.2172/1824785.


Cromwell E and D Flynn. 2019. Lidar Cloud Detection with Fully Convolutional Networks. In Winter Conf. on Applications of Computer Vision, : IEEE.

View All Related Publications


Damao Zhang
Pacific Northwest National Laboratory

View All VAP Translators