An objective method to generate high resolution wind maps from the irregularly spaced scatterometer data was developed and tested on synthetic data for the northeast Pacific Ocean. Current assimilation methods for atmospheric global circulation models (GCMs) reproduce only the large-scale features (>600 km) of the wind field. In contrast, this method attempts to retain the high spatial resolution of the scatterometer where there are adequate measurements. To estimate the covariance function and to compute realistic error estimates for the method, a synthetic data set was constructed using wind fields from the European Centre for Medium Range Weather Forecasts. The wave number spectrum of the synthetic data was based on an analysis of Seasat scatterometer data. The patchiness of the scatterometer data caused discontinuities in the objective estimates, which would give unacceptable wind stress curl fields; thus the initial estimates were selectively smoothed using overdetermined biharmonic splines. A constraint on the horizontal divergence of the wind, applied only in regions of large estimated errors, improved the estimates. For the NASA scatterometer this method would produce a wind map every 12 hours with spatial resolution that preserves the small-scale features of the original data over about half the mapped region and with somewhat decreased resolution and accuracy over the rest of the region. ¿ American Geophysical Union 1990