Abstract
The authors have developed a novel approach to the extraction of cloud-base height (CBH) from pairs of whole-sky images. The core problem is to spatially register cloud fields from widely separated whole-sky imaging (WSI) cameras; this complete triangulation then provides the CBH measurements. The wide camera separation (necessary to cover the desired observation area), occluded regions, and the self-similarity of clouds defeats standard matching algorithms when applied to static views of the sky. The authors address this with an approach that is based on optical flow methods, exploiting the fact that modern WSIs provide time-ordered sequences of images. The authors will describe the algorithm for CBH determination, a confidence metric, as well as a method to correct for the severe projective effects on cloud shape induced by the WSI camera. Finally, the authors present the performance as evaluated both on real data validated by ceilometer measurements and on a variety of simulated cases.
