Arctic Sea Level Reconstruction

Peter Limkilde Svendsen

Research output: Book/ReportPh.D. thesis

358 Downloads (Pure)


Reconstruction of historical Arctic sea level is very difficult due to the limited coverage and quality of tide gauge and altimetry data in the area. This thesis addresses many of these issues, and discusses strategies to help achieve a stable and plausible reconstruction of Arctic sea level from 1950 to today.The primary record of historical sea level, on the order of several decades to a few centuries, is tide gauges. Tide gauge records from around the world are collected in the Permanent Service for Mean Sea Level (PSMSL) database, and includes data along the Arctic coasts. A reasonable amount of data is available along the Norwegian and Russian coasts since 1950, and most published research on Arctic sea level extends cautiously from these areas. Very little tide gauge data is available elsewhere in the Arctic, and records of a length of several decades,as generally recommended for sea-level reconstruction, are completely absent outside the Norwegian and Russian sectors. Since the early 1990s, altimetric satellite missions have provided more spatially complete observations of sea level. This allows extraction of the primary variation patterns, which can be used as calibration for a reconstruction method.For oceanographic purposes, the altimetric record over the Arctic Ocean is inferiorin quality to that of moderate latitudes, but nonetheless an invaluable set of observations. During this project, newly processed Arctic altimetry from the ERS-1/-2 and Envisat missions has become available, allowing analysis ingreater detail, though much early progress on the project was based on ocean model data.Like other published sea level reconstructions, this project is based on the combination of tide gauge records and altimetry patterns. It is found that while it is possible to reconstruct the timestep differences and cumulate these to obtain a reconstructed sea-level record, this approach may yield widely variable results and is difficult to stabilize due to the many gaps in the data. A more robust approach, as described by Ray and Douglas (2011), takes into account the entirety of each tide gauge record and makes the reconstruction much less prone to drifting away over time.Unfortunately, many of the Russian-sector tide gauge records end around 1990,leaving a fairly sparse record after this. This project examines the effect of introducing a subset of the altimetric dataset as “virtual tide gauges” to remedy this sparsity, and appears to further stabilize the reconstruction. As Arcticsea level changes are particularly concentrated in the Beaufort Gyre area, this also introduces observations in an important area. However, this approach to some extent relies on relatively stationary conditions before the altimetric era,though previous research indicates largely stationary amounts of freshwater until the 1980s. This project initially aimed to obtain a robust reconstruction through the use of alternative decompositions, rather than the commonly used empirical orthogonal functions (EOFs), for the calibration. While one alternative decomposition,maximum auto correlation factors (MAFs), is investigated, it is found that preprocessing and handling of gaps (through appropriate method choice) in the tide gauge record is the primary concern for obtaining robust sea level reconstructions in the Arctic area.The reconstructions obtained in this project concern the period 1950 to 2010 using monthly data. The spatial coverage is all ocean areas above 68°N, limited to the north depending on the calibration dataset used (90°N for Drakkarcalibrated reconstructions, 82°N for altimetry-based reconstructions).
Original languageEnglish
Place of PublicationKgs. Lyngby
PublisherDTU Space
Number of pages168
Publication statusPublished - 2015


Dive into the research topics of 'Arctic Sea Level Reconstruction'. Together they form a unique fingerprint.

Cite this