Development and Applications of a Large Eddy Simulation based Synthetic Total Sky Imaging System

Recently, an algorithm has been developed to retrieve cloud size from Total Sky Imaging, which allows us to continuously monitor the cloud size distribution of shallow cumulus convection. In this study we compare the retrieved distribution with LASSO-based Large Eddy Simulations. To do so, we have developed a TSI Simulator using the Blender ray tracing engine. Using the simulator, we perform the following studies: 1) Validating the TSI projection algorithm via cross referencing with the cloud field from the LES, with a specific emphasis on the error introduced by cloud sides and background clouds for larger zenith angles; 2) Comparison of LES and TSI observations of cloud size distributions using the simulator; 3) Training of a machine learning system to automatically filter cloud base from cloud sides to increase the quality of the cloud sizes retrieved from this system.