Model analysis winds such as those from European Center for Medium-Range Weather Forecasts (ECMWF) and National Centers for Environmental Prediction/National Center for Atmospheric Research (NCEP/NCAR) reanalyses are widely used for climate studies. However, large errors and different climate change signals are found among different reanalysis systems over high latitude regions (Francis and Cermak, 2001, Bromwich et al. 2000). In addition, because of the complicated error structure resulted from assimilation of various types of observations and complicated physical parameterizations, it is difficult to identify many of the error sources in the reanalyses. This study attempts to derive a satellite-based, geostrophic-like wind field for climate studies over the middle and high latitude oceans. A technique that uses the satellite-based surface wind and temperature soundings for deriving the three-dimensional atmospheric wind fields is developed. In this technique, the thermal wind derived from the satellite soundings is added to the surface wind to obtain a first-guess, nonmass-conserved atmospheric wind profile. Then a Lagrange multiplier in a variational formalism is used to force the first-guess wind to conserve mass. Two mass conservation schemes are proposed. One is to use the mass flux conservation equation across a latitudinal wall as a constraint to derive the meridional wind first, and then the vertically-integrated mass conservation equation is used to infer the zonal wind. The zonal and meridional winds are obtained separately in this approach. The second scheme is to use the vertically-integrated mass conservation equation as a constraint to retrieve the zonal and meridional winds simultaneously from the first-guess field. Temperature soundings from the Television Infrared Observational Satellite (TIROS) Operational Vertical Sounder (TOVS) Pathfinder A dataset (Susskind et al. 1997) and a Special Sensor Microwave Imager (SSM/I) satellite-based surface wind field (Atlas et al. 1996) are used in deriving the wind fields. The two mass conservation schemes yield two different wind fields. They are compared with the ECMWF and NCEP/NCAR reanalyses and radiosonde observations over the Southern Ocean. It is found that the wind of the first scheme is closer to the reanalyses and radiosonde observations in most situations, while the second scheme yields a yearly-mean bias in both the zonal and meridional wind fields compared to the radiosonde observations. Differences between the two schemes and their differences from the reanalysis winds are discussed.