Air pressure effects on sea level changes during the twentieth century

Interpretation of tide gauge data in terms sea level (η) and ocean dynamics requires estimates of air pressure (p_a) to determine the ocean's isostatic response—the inverted barometer effect ((Formula presented.)). Three gridded p_a products (HadSLP2, NOAA-20CRv2, and ERA-20C) are used alongside meteorological station p_a and tide gauge η records to evaluate the contribution of (Formula presented.) to η changes over the twentieth century. Agreement between gridded products is better during more recent periods and over regions with good historical data coverage, whereas it is worse for earlier time periods or in ocean areas with poor observational data coverage. Comparison against station data reveals the presence of systematic errors in the gridded products, for example, such that uncertainties estimated through differencing the gridded products underestimate the true errors by roughly 40% on interannual and decadal time scales. Notwithstanding such correlated errors, gridded products are still useful for interpretation of tide gauge data. Removing gridded estimates of (Formula presented.) from η records reduces spatial variance in centennial trends across tide gauges by 10–30%, formal errors in centennial trends from individual gauges by ∼5%, and the temporal variance in detrended records by 10–15% on average (depending on choice of gridded product). Results here advocate for making the (Formula presented.) correction to tide gauge records in studies of ocean circulation and global η over long, multidecadal, and centennial time scales using an ensemble mean taken across several gridded (Formula presented.) products.