Last update 1 July 2020
As well as the tides, the sea level as measured by tide gauges is locally depressed or elevated by changes in the weight of the atmosphere above the sea, (local high or low air pressure, the Local Inverse Barometer (LIB) effect). To within approximately 0.5% accuracy, a local reduction in air pressure at sea level of 1 hPa (or mbar) is equivalent to 1 cm increase in sea level due to the LIB effect. For more technical details on the LIB effect, see Wunsch and Stammer (1997) or Pugh and Woodworth (2014; Cambridge: CUP, 407pp). Monthly-mean air pressure data are often needed to make LIB corrections to the monthly-mean sea-level data from the PSMSL. This page describes some of the sources of air pressure data (known as 'sea-level pressure') suitable for the corrections. Note that in many (or probably most) studies of mean sea-level data, it will be necessary to make use of gridded fields of sea-level pressure, rather than of station pressure time series, owing to the lack of station data near to gauges in many parts of the world.
There are many gridded atmospheric reanalysis datasets available, but we shortlist only a couple of the more suitable options for monthly-mean analysis, with global coverage. If a longer list is of interest, see this table on the Reanalyses.org website and consider a literature review of intercomparison of datasets. There is no single intercomparison study that covers all these reanalyses or focuses on the quality of sea-level pressure. What is more typical are studies that cover a subset, e.g. Slivinski et al. (2019) - their Figs. 13 and 14. For the shortlist, a brief literature review has been undertaken. Regional reanalyses, e.g. EMULATE - for the North Atlantic and Europe, are not covered in detail here, but may also be useful for studies with a particular regional focus.
ECMWF ERA5 (Hersbach et al., 2020) is recommended, because of its extensive user support and operational updates. Data is available from the CDS cloud server (https://cds.climate.copernicus.eu) . There is also online data documentation (https://confluence.ecmwf.int/display/CKB) that provides a detailed description of the various products and a list of all available geophysical parameters.
NOAA-CIRES-DOE 20CRv3 (Slivinski et al., 2019) is recommended, because it is the first ensemble of (sub‐daily) global atmospheric conditions spanning over 100 years. It provides a best estimate of the weather at any given place and time as well as an estimate of its confidence and uncertainty. It is further recommended that the ensemble (https://portal.nersc.gov) is used to get the uncertainty data, as certain times and places don't have useful data. There is some preliminary documentation available at https://vlab.ncep.noaa.gov/web/fv3gfs . Note that there are some issues with 20CRv2, especially involving atmospheric pressure bias, therefore v2 is best avoided for inverse barometer corrections.
People interested in comparing multiple full-input reanalyses as maps and time series may find WRIT (Smith et al., 2014) : https://psl.noaa.gov/data/writ/, useful. WRIT will also soon have tools for comparison of ocean reanalyses.
The PSMSL is grateful to Gil Compo, Rob Allan, Philip Brohan, Tara Ansell and Simon Josey for their advice on the information in this web page.
Note: In an earlier version of this web page, we suggested that the most suitable gridded air-pressure data set for sea level studies of the late-19th and 20th centuries was HADSLP2. This data set is still available from the Met Office's HADSLP2 page. However our advice now would be to use one of the above products instead as they are probably now more accurate.