Evaluation of a Wavelet-Based Optical Flow Algorithm through the Use of Large Eddy Simulation

Recent progress in the ability to observe two-component wind fields in the atmospheric surface layer by horizontally scanning aerosol lidar and motion estimation algorithms elevates interest in issues such as accuracy, precision, and resolution.  Toward this, we have started using large eddy simulations (LES) to advect and disperse a passive tracer in the model domain that mimics atmospheric aerosol concentration.  A wavelet-based optical flow algorithm, named Typhoon, was applied to horizontal fields of the passive tracer near the surface of the model domain and dense motion vectors were derived for all grid points on those slices.  Each motion vector can be compared with the wind velocity at the corresponding grid point.  Moreover, large eddy simulations can be conducted for a wide variety of boundary layer states and environments, thereby allowing us to infer how the motion estimation may perform in various situations.