Articles | Volume 16, issue 5
https://doi.org/10.5194/amt-16-1195-2023
https://doi.org/10.5194/amt-16-1195-2023
Research article
 | 
08 Mar 2023
Research article |  | 08 Mar 2023

Optimizing cloud motion estimation on the edge with phase correlation and optical flow

Bhupendra A. Raut, Paytsar Muradyan, Rajesh Sankaran, Robert C. Jackson, Seongha Park, Sean A. Shahkarami, Dario Dematties, Yongho Kim, Joseph Swantek, Neal Conrad, Wolfgang Gerlach, Sergey Shemyakin, Pete Beckman, Nicola J. Ferrier, and Scott M. Collis

Data sets

Total Sky Imager (TSISKYIMAGE) D. Flynn and V. Morris https://doi.org/10.5439/1025309

ceil V. Morris, D. Zhang, and B. Ermold https://doi.org/10.5439/1181954

915-MHz Radar Wind Profiler/RASS (RWP915): wind consensus data P. Muradyan and R. Coulter https://doi.org/10.5439/1025135

Model code and software

RBhupi/plugin-cmv-fftpc: check previouse release for fftpc S. Shahkarami, dariodematties, and B. Raut https://doi.org/10.5281/zenodo.7676068

Download
Short summary
We studied the stability of a blockwise phase correlation (PC) method to estimate cloud motion using a total sky imager (TSI). Shorter frame intervals and larger block sizes improve stability, while image resolution and color channels have minor effects. Raindrop contamination can be identified by the rotational motion of the TSI mirror. The correlations of cloud motion vectors (CMVs) from the PC method with wind data vary from 0.38 to 0.59. Optical flow vectors are more stable than PC vectors.