Reconstruction of past decades sea level using thermosteric sea level, tide gauge, satellite altimetry and ocean reanalysis data

This study investigates past sea level reconstruction (over 1950\textendash2003) based on tide gauge records and EOF spatial patterns from different 2-D fields. In a first step, we test the influence on the reconstructed signal of the 2-D fields temporal coverage. For that purpose we use global grids of thermosteric sea level data, available over 1950\textendash2003. Different time spans (in the range 10\textendash50~yr) for the EOF spatial patterns, and different geographical distributions for the 1-D thermosteric sea level time series (interpolated at specific locations from the 2-D grids), are successively used to reconstruct the 54-year long thermosteric sea level signal. In each case we compare the reconstructed trend map with the reference. The simulation indicates that the longer the time span covered by the spatial EOFs, the closer to the reference the reconstructed thermosteric sea level trends. In a second step, we apply the method to reconstructing 2-D sea level data over 1950\textendash2003, combining sparse tide gauge records available since 1950, with EOF spatial patterns from different sources: (1) thermosteric sea level grids over 1955\textendash2003, (2) sea level grids from Topex/Poseidon satellite altimetry over 1993\textendash2003, and (3) dynamic height grids from the SODA reanalysis over 1958\textendash2001. The reconstructed global mean sea level trend based on thermosteric EOFs (case 1) is significantly lower than the observed trend, while the interannual/decadal sea level fluctuations are well reproduced. Case 2 (Topex/Poseidon EOFs over 1993\textendash2003) leads to a global mean sea level trend over the 54-year time interval very close to the observed trend. But the spatial trends of the reconstruction over 1950\textendash2003 are significantly different from those obtained with thermosteric EOFs. Case 3 (SODA EOFs over 1958\textendash2001) provides a reconstruction trend map over 1950\textendash2003 that differs significantly from the previous two cases. We discuss possible causes for such differences. For the three cases, on the other hand, reconstructed spatial trends over 1993\textendash2003 agree well with the regional sea level trends observed by Topex/Poseidon.