UNDERSTANDING THE RELATION BETWEEN WIND-DRIVEN AND PRESSURE-DRIVEN SEA-LEVEL VARIABILITY

Sea surface adjustment to combined wind and pressure forcing is examin ed using numerical solutions to the shallow water equations. The exper iments use coastal geometry and bottom topography representative of th e North Atlantic and are forced by realistic barometric pressure and w ind stress fields. The response to pressure is essentially static or c lose to the inverted barometer solution at periods longer than a few d ays and dominates the sea level variability, with wind-driven sea leve l signals being relatively small. With regard to the dynamic signals, wind-driven fluctuations dominate at long periods, as expected from qu asi-geostrophic theory. Pressure becomes more important than wind stre ss as a source of dynamic signals only at periods shorter than approxi mately three days. Wind- and pressure-driven sea level fluctuations ar e anticorrelated over most regions. Hence regressions of sea level on barometric pressure yield coefficients generally smaller than expected for the inverted barometer response known to be the case in the model . In the regions of significant wind-pressure correlation effects, to infer the correct pressure response using statistical methods, input f ields must include winds as well as pressure. Because of the nonlocal character of the wind response, multivariate statistical models with l ocal wind driving as input are not very successful. Inclusion of nonlo cal wind variability over extensive regions is necessary to extract th e correct pressure response. Implications of these results to the inte rpretation of sea level observations are discussed.