A method for estimating surface layer similarity parameters from a satellite SAR over the ocean was presented in Young et al. (2000). Mean surface winds (and hence u*) are estimated using a scatterometer model function (CMOD-4) applied to the image with an assumed wind direction. The along-wind velocity variance is calculated from these wind images. In the convective limit zi/L is related to sigma_u and u*. The inversion height, zi, may be estimated from the scale of boundary layer convection which in turn can be estimated from the sea surface patterns imaged by SAR. Hence the Obukhov length may be retrieved from SAR in convective conditions. With an independent SST measurement the surface buoyancy flux can also be retrieved. We test this method against a large number of SAR images from the Tropical Pacific ocean over the TAO array and off the east coast of the USA over NOAA NDBC buoys. Surface truth is provided by both in situ flux measurements and by the COARE algorithm applied to the mean buoy data. Sensitivity to pixel size, sub-image size, stratification and wind direction are explored.