Four-dimensional variational data assimilation using lidar data

Given the high resolution measurements of the atmospheric boundary layer by LIght-Detection-And-Ranging (lidar) technologies, a four-dimensional variational data assimilation (4D-Var) technique is employed to extract three-dimensional wind vector and temperature data not observed by lidar. Optimal control theory is applied in 4D-Var to drive a numerical model to obtain an atmospheric state that best fits observational data.; The identical twin experiments that utilize the 4D-Var's prediction model to generate synthetic observational data, are first applied to assess the applicability of this technique to microscale atmospheric boundary layer (ABL) flow. The sensitivity of retrieval quality to initial guesses, observational frequencies and observational errors are evaluated. Three new features are developed to improve retrieval of microscale turbulence structures. A surface flux model driven by observational data is implemented. A dynamic procedure for determining the weights of smoothness penalty constraints is proposed and evaluated. The profiles of eddy viscosity and diffusivity are treated as control variables and recovered from observational data by the 4D-Var.; The 4D-Var is finally applied to radial velocity data of the high resolution solid state Doppler lidar. The wind and temperature fields are retrieved in detail and they reveal some interesting flow structures, such as horizontal rolls and micro-fronts. The retrieved velocity field provides valuable information on the flow structures involved in the transport process in the atmospheric boundary layer. The success of retrieving wind vectors from lidar radial velocity data suggests that the 4D-Var technique and multiple lidar systems can be further integrated to reconstruct more realistic atmospheric states.