Accurate, short-term (0-2 hour) forecasts of convective initiation provide critical information about weather that has a major impact on aviation safety and system capacity. The Terminal Convective Weather Forecast (TCWF) algorithm is a key component of the FAA's operational Integrated Terminal Weather System (ITWS). Convective forecasts rely, in part, upon detection of convergence zones in the boundary layer. Detection of convergence requires accurate, high-resolution wind estimates, which may be based on measurements from many sources, including Terminal Doppler Weather Radar (TDWR), Next Generation Weather Radar (NEXRAD), Automatic Weather Observation System/Automatic Surface Observation System (AWOS/ASOS), aircraft (via the Meteorological Data Collection and Reporting System, MDCRS) and Low Level Wind Shear Alert System (LLWAS). These data may be directly analyzed, combined with satellite and sounding data or ingested into physical models that estimate winds and produce short term forecasts. We compare two windfield estimation techniques: Terminal Winds (TWINDS) [Cole et. al., 2000], an optimal estimation algorithm developed at Lincoln Laboratory that is deployed operationally in ITWS, and Variational Doppler Radar Analysis System (VDRAS) [Sun and Crook, 2001], a 4DVAR algorithm developed and fielded by the Research Applications Program (RAP) at NCAR. These techniques differ markedly in their use of physical models: TWINDS applies no physical constraints to its analysis, while VDRAS uses a 4DVAR technique to fit the data with a boundary layer model as a strong constraint. The techniques also differ in their computational requirements: TWINDS requires substantially less computational power than VDRAS. We were able to run TWINDS at higher horizontal resolution and update rate (1km grid spacing, 5 minute update) than VDRAS (2km and 12 minutes).