Regional Sea Level Variability and Trends, 1960–2007: A Comparison of Sea Level Reconstructions and Ocean Syntheses

Abstract

Several existing statistical and dynamical reconstructions of past regional sea level variability and trends are compared with each other and with tide gauges over the 48 year period 1960–2007, partially predating the satellite altimetry era. Evaluated statistical reconstructions were built from tide‐gauge data (TGR), and dynamical reconstructions from ocean data assimilation (ODA) approaches. Although most of the TGRs yield global‐mean time series of sea level with trends deviating within only ±0.1 mm yr−1, the spatial anomalies of the trends deviate substantially between the reconstructions over the period predating altimetry. In contrast, TGRs match observed regional trend patterns fairly well during the satellite altimetry era. TGRs match tide‐gauge data better than ODA results; however, they exhibit less variability in the open ocean compared to altimetric data. Over the prealtimetry period, all reconstructed regional sea level trend patterns deviate substantially from each other. In terms of detrended correlations in this earlier period, the reconstructions match tide gauges, and each other, much better in the Pacific than in the Atlantic. An ensemble of all TGR and ODA estimates provides some improvements in correlations and trends to both tide gauges and altimetry. Nevertheless, a lack of independent open ocean sea surface height data predating altimetry makes impossible the validation of the ensemble for prealtimetry open ocean sea level trends and variability. Estimating regional sea level changes prior to altimetry therefore remains an unsolved challenge.

Initialization and Ensemble Generation for Decadal Climate Predictions: A Comparison of Different Methods

“…This may suggest that modes of variability of the reanalysis are not exactly compatible with the modes from the prediction system or that they are not yet sufficiently sampled by the available data used to construct the EOFs. In this respect, FAI might benefit from attempting to capture better the variability modes in the North Atlantic using a larger EOF basis, using a different weighting method to determine better the structure of the modes in this region, or using regional (per‐basin) EOFs, which have been shown to perform significantly better in, for example, reconstructing Atlantic sea level variability (Meyssignac et al, ) in comparison to using global EOFs (Carson et al, ). Apart from this, the FAI hindcasts show the highest skill for the ENSO region at interannual timescales which is associated with the improved zonal momentum balance in the equatorial Pacific as compared to Preop‐LR (not shown) and appear to have a substantial potential for further improvements in the future.Given that MODINI only uses the wind stress data for initialization, it compares reasonably well with the other initialization methods that are tested here.…”

Steric Sea Level Changes from Ocean Reanalyses at Global and Regional Scales

“…The dynamical consistency of ocean reanalyses may be further exploited in centennial sea level reconstructions, where reanalyses provide the training data through which long records of tide gauge data are used to reconstruct the spatial patterns of sea level variability [33]. With the systematic production of multi-model ocean reanalyses, this application of reanalysis is expected to further develop and replace early attempts of using climate model simulations to draw error statistics for climate reconstructions [118].…”

“…Another observation-based approach of sea level estimation is the tide gauge-based sea-level reconstruction [32,33], where tide gauge data are used to estimate large-scale patterns of variability, generally based on statistical relations (e.g., empirical orthogonal functions, EOFs) computed over well-observed training periods (e.g., altimetry). Such reconstructions have the crucial merit to offer a long multi-decadal reconstruction owing to the availability of long tide gauge data records; on the other hand, different reconstructions show significant discrepancy at regional scales [33].…”

“…They showed that the local mismatch between reanalyses and tide gauges is mostly due to meteorological forcing. However, during the pre-altimetry period (i.e., before 1993), small consistency is found between reanalyses and tide gauge-based sea level reconstructions, the latter generally outperforming the former in reproducing the inter-annual variability measured by tide gauges [33]. This might be partly attributed to the fact that some tide gauge data records are not representative of large-scale dynamics.…”

Towards Comprehensive Observing and Modeling Systems for Monitoring and Predicting Regional to Coastal Sea Level

“…An important part of projected SL trends on a regional scale arises from the dynamical and thermal and haline adjustment of the ocean related to changes in the circulation. On timescales up to decades, model improvements are needed to better capture the interannual variability of SL associated with climate modes discussed in section "Causes of Coastal Sea Level Variability" (e.g., Frankcombe et al, 2013;Carson et al, 2017). Simulations of such variability by climate models require further validation with emerging longer data sets of SL, mass or density changes.…”