The accuracy of a diagnostic wind model, such as ARL's 3-Dimensional Wind Field (3DWF) model, is closely related to its initial data accuracy and spatial density. Since 2003, a Doppler Wind Lidar has been flown on a Navy Twin Otter in a series of studies of the atmospheric boundary layer. The wind profiles were retrieved from a downward conical scan with 30 degree azimuth intervals using a volume velocity processing method. Each profile was taken within a 30 second time window yielding a complete u,v,w profile each 1.5 km from the surface to 2500 meters with a vertical resolution of 50 meters. We have developed an objective analysis algorithm to grid the airborne Doppler wind lidar observed wind profiles in a complex terrain. The gridded data is used to initialize an improved version of 3DWF model. A sensitivity analysis was carried out to determine the least number of wind profiles required to give a reasonably accurate wind field for the testing site.