How do Unmodeled Systematic Mean Sea Level Variations Affect Long-term Sea Level Trend Estimates from Tide Gauge Data?

Abstract

Although there are over 1,800 globally distributed tide gauge stations, only a few hundred of them are suitable for monitoring and analyzing global mean sea level (MSL) changes. This is because several tide gauge records span short periods of time and therefore their trend estimates are adversely affected by unmodeled systematic sea level changes such as seasonal, interannual, decadal variations. This limitation can be improved by using more elaborate models that account for systematic fluctuations in MSL for shorter time-series. In this study, analytic expressions were derived to analyze and quantify the epoch-by-epoch and lump-sum effects of these systematic changes to the local MSL trend estimates as a function of the time-series‘ lengths. The numerical results reveal that systematic MSL variations, particularly transient/episodic ones, if they are not properly modeled or omitted from the models, will bias the trend estimates for the tide gauge data series around the world by up to 0.6 mm/year for the 50-year time-series that are needed for more reliable inferences about global MSL. Random effects, which are not a factor in estimating MSL trends for the long-term (>50 years) time-series, need to be scrutinized together with the systematic errors for time-series shorter than 50 years.