30 Apr 2015
With this year's update of our page showing the trends in the tide gauge data set, including the new data obtained over 2014, we have introduced a new method of determining the trends. The new method takes into account that the variability in the tide gauge time series may not be independently random, but may be correlated over different time spans. This correlated variability can decrease our ability to determine if a long-term trend in sea level is occurring at the tide gauge, or in other words, increase our uncertainty in the estimated trend.
The new methods page gives the details of the analysis. However, there are several changes that you may notice. Firstly, the pop-up boxes for each tide gauge shown on the trends map now include an uncertainty. Both the estimated trend and the uncertainty will change as one changes the time span chosen by moving the sliders. Secondly, in order to calculate these results, we had to switch from using the annual means to using the monthly means. Thus, the standard deviation of the residuals listed in the long-term trends file with respect to a combined trend, annual and semi-annual periodic terms have noticeably increased, by a factor approaching the square root of 12. Finally, that file has a slightly different format to previous years, which we hope will not be an inconvenience to you.