Relative sea-level variations revealed by tide-gauge records of long duration

Abstract

Trends in mean relative sea-level and pat.terns of occurrence of extreme sea levels are analyzed separately in the two parts of this thesis. In Part 1, twenty-eight of the world's longest tide-gauge records are examined for clues to the uncertain balance among factors contributing to relative sea-level (RSL) fluctuations. Obtaining these clues requires a description of t he data in terms of component functions, whether chosen for t heir special properties (regression analysis) or empirically determined (EOF analysis). Part 1 describes a regression model that allows for gradual changes of the RSL trend and for sudden changes in the level to which tide-gauge measurements are referred (gauge zero); also described is an EOF analysis procedure that offers certain advantages in the handling of missing observations. Although the rate of RSL rise exhibits significant gradual change over 60- to 120-year analysis intervals at many stations, no support is found for the idea of a gradual global acceleration of RSL rise. Results that seem to show RSL rise accelerating globally over the last century can be attributed instead to the changing geographical distribution of tide-gauge observations. Regional and local controls, which include vertical crustal movements and changes due to oceanographic or meteorological effects, must be responsible for the accelerations documented at many stations, and are dominant in controlling shorter-term departures from the RSL trend too. Preliminary results of EOF analysis reveal regionally coherent fluctuations of annual mean RSL in the Baltic Sea with r.m.s. amplitudes as large as 70 mm. Globally coherent fluctuations have smaller r.m.s. amplitudes: a tentative upper limit is 20 mm. In Part 2, 40 years of hourly sea-level records from two stations on the mid-Atlantic coast of the U.S. a re used in a compilation of monthly 'surge'-level exceedance counts. 'Surge' level is defined as observed sea level minus predicted tide level. The results are compared with previously published storm counts, and the annual cycle of 'surge'-level exceedance frequency is found to lead that of storm frequency by nearly two months. It is recommended that further work aimed at modelling 'surge'-level exceedances should include: (i) recognition that the tide record includes meteorological/ oceanographic as well as astronomical components, especially at the frequency of the solar annual tide, and (ii) quantification of the relative importance of tide and surge in the timing of extreme sea-level occurrences at different times and places.