Assessment of Tidal Range Changes in the North Sea From 1958 to 2014

Abstract

We document an exceptional large-spatial scale case of changes in tidal range in the North Sea, featuring pronounced trends between -2.3 mm/yr at tide gauges in the United Kingdom and up to 7 mm/yr in the German Bight between 1958 and 2014. These changes are spatially heterogeneous and driven by a superposition of local and large-scale processes within the basin. We use principal component analysis to separate large-scale signals appearing coherently over multiple stations from rather localized changes. We identify two leading principal components (PCs) that explain about 69% of tidal range changes in the entire North Sea including the divergent trend pattern along United Kingdom and German coastlines that reflects movement of the region's semidiurnal amphidromic areas. By applying numerical and statistical analyses, we can assign a baroclinic (PC1) and a barotropic large-scale signal (PC2), explaining a large part of the overall variance. A comparison between PC2 and tide gauge records along the European Atlantic coast, Iceland, and Canada shows significant correlations on time scales of less than 2 years, which points to an external and basin-wide forcing mechanism. By contrast, PC1 dominates in the southern North Sea and originates, at least in part, from stratification changes in nearby shallow waters. In particular, from an analysis of observed density profiles, we suggest that an increased strength and duration of the summer pycnocline has stabilized the water column against turbulent dissipation and allowed for higher tidal elevations at the coast.

Keywords: Kriging; North Sea; amphidromes; empirical orthogonal function (EOF); stratification; tidal range.

Figure 1

Time series of mean annual high and low tidal water levels for three exemplarily selected stations in the German Bight. For illustration purposes, all records are shown with different artificial vertical offsets. The increase in the tidal range is illustrated for the three sites as gray shaded areas between high and low water level time series.

Figure 2

Bathymetry of the North Sea (Becker et al., ; Schrottke & Heyer, 2013). Also shown are the locations of tide gauges (black dots) used in this study including their respective numbering (see also Table 1). The black propellers indicate the location of the three semidiurnal amphidromic areas (including the amphidromic points for the M₂ and S₂ constituent) and the black dotted lines indicate contours of equal mean tidal range (Sündermann & Pohlmann, 2011).

Figure 3

Changes in tidal range before (a) and after (b) applying ordinary kriging and removing the nodal cycle.

Figure 4

Linear trends of tidal range between 1958 and 2014. Trends at measured sites are shown as dots with a black edge. Dots in between stations are based on Kriging.

Figure 5

(a) Linear trends of the M₂ and (b) S₂ tidal constituents between 1958 and 2014 (significant trends outlined).

Figure 6

Results of the PCA. (a) Shown are time series of PC1, and PC2 and (b) their corresponding spatial patterns. Panels (c) and (d) map the correlations between observations and PC1 (c) and PC2 (d) for each site. PCA, principal component analysis.

Figure 7

Linear trends in tidal range with 95% significance intervals from measurements (blue) and the reconstruction (red) based on PC1 and PC2, with the respective difference shown in (b). (c) Spatial distribution of the linear trends from the reconstruction (significant trends outlined) and (d) explained variance of the two PCs as share of the total variance.

Figure 8

(a) Extended network of tide gauges with additional stations shown in red, (b) correlations of all tide gauges (except 5–7) with PC2 and (c) comparison between measured and reconstructed values of tidal range at the newly added tide Gauges 1–7. The reconstruction in (c) is based on PC2, and the numbers in parentheses indicate the respective correlation.

Figure 9

(a) Vertical profiles of potential density as averaged over all query points in (b) at depths from 0 to 35 m for the years 1958–2013 (black), the year 1995 (blue) and the years 1998 (red). The two selected years feature the greatest deviation from the mean density profile.

Figure 10

(a) Spatially averaged density profiles (0–35 m) from the query area in Figure 9b spanning the period 1958–2013. (b) Comparison between PC1 changes and the stability index (see main text), where both time series were scaled by their standard deviation and adjusted for long‐term trends.