Ground Based Stereo Imaging as a Source for Collecting Measurements of Clouds

Abstract:

Satellite and ground-based cameras working at different spectral bands help to retrieve various features of atmosphere and clouds at an ever-improving accuracy with the advancements in imaging instruments and data processing capability. Ground based monitoring has the potential to continuously offer high temporal and spatial resolution for a particular region. With the installment of a stereo camera pair spaced approximately one kilometer apart, ground-based stereo can provide 3D coordinates of cloud features in a region as far as tens of kilometers away. With such a setup at the central facility of Southern Great Plains ARM site in Oklahoma, we developed and implemented an automatic stereo matching and reconstruction algorithm to retrieve and record cloud feature coordinates visible in the field of view of both cameras. Our cameras synchronously record image frames at a 30-second time interval during the day time. The algorithm identifies regions of clouds, matches high texture cloud features across synchronized frames and retrieves 3D positions of the matched features. In a three month campaign from mid April to mid July, 35 million cloud feature points are collected along with their 3D positions and the results are validated against Raman lidar measurements. Stereo imaging is able to make the distinction between clouds and rain or aerosols where Raman lidar suffers from confusion between these in the boundary layer. Stereo imaging also enables the estimation of vertical and horizontal velocities of cloud layers by tracking the cloud features in successive image frames in time. Overall, the analysis shows that ground based stereo imaging can successfully complement the current atmospheric measurement instruments by providing reliable, continuous, and high-resolution data over clouds in a localized region.