F. Bouttier
ECMWF, Shinfield Park, Reading, Berks RG2 9AX, GB

H. Van den Dool
Climate Prediction Center, 5200 auth Road, Camps Spring, MD20746

The GRACE satellite gravity mission will resolve the time-varying gravity field with sufficient accuracy to constrain surface hydrology and glacial ice variations to $\leq$1 cm of water-equivalent mass at scales of a few hundred km. To detect changes in water mass at this accuracy, it will be necessary to remove atmospheric mass contributions from the satellite measurements. A 1 mbar error in the pressure correction would introduce an error in the gravitational field equivalent to 1 cm of water, and would make uncertainties in pressure the limiting error source for using GRACE in hydrological applications at scales $>$ 200 km. We compare sea level pressure from NCEP/NCAR's Reanalysis model and from ECMWF with pressure measurements in the U.S. and the Mediterranean. In low-lying areas, the spatially-averaged RMS difference between models and observations is generally $<$1 mbar, but at high elevations the RMS can be up to 2 mbar. The discrepancy may be related either to elevation-dependent model errors %high-elevation modeling problems or to inconsistencies in how the stations and the models extrapolate to sea level. We are presently addressing this issue by comparing surface pressure values. Finally we will show that it may be possible to construct surface pressure fields accurate to 0.5 mbar using only barometric measurements.