Theory and observations have shown that the shear due to near-surface horizontal temperature gradients significantly modifies the boundary layer turbulence and hence the near-surface wind speed and direction. The magnitude of these modifications can exceed those due to the vertical temperature gradient in typical conditions. Previous studies of this effect over the ocean using conventional observations have been limited to a small number of sonde datasets in specific locations. The NASA Scatterometer (NSCAT) is a space-borne microwave radar that provides parallel 600 km wide swaths of 10 m neutral-equivalent wind vectors at 50 km resolution. Combining these data at synoptic times with ECMWF analyses of sea-level pressure, sea-surface temperature and temperature at 2m allows this baroclinic effect over the ocean to be examined over large regions. We restrict the analysis to the northern hemisphere because previous ECMWF/Scatterometer studies have demonstrated that this is where the error in ECMWF storm positioning is least. The high resolution of the NSCAT winds allows very accurate identification of storm centers and fronts. Errors in storm and front locations in the background data is a significant source of scatter in the analysis. A baroclinic signal in the NSCAT wind speed is compared to the previous observations and to the standard theories. Apparently the accuracy of the NSCAT wind direction is insufficient to to detect the baroclinic modification except in high wind conditions. The magnitude of the speed modification is in agreement with the theoretical predictions. However, the phasing of the peak modification as a function of the thermal wind direction has a phase shift relative to the theoretical predictions. The speed modification is asymmetric with thermal wind orientation consistent with the theoretical prediction of Foster and Levy (1998).