Time trends and persistence in US sea level data: an investigation using fractional integration methods

Abstract

This paper analyses US sea level data using long memory and fractional integration methods. Specifcally, monthly data for 41 US stations covering the period from January 1950 to December 2018 are examined. Fractional integration methods suggest that all series exhibit orders of integration in the interval (0, 1), which implies long-range dependence with positive values of the diferencing parameter; further, signifcant positive time trends are found in the case of 29 stations located on the East Coast and the Gulf of Mexico, and negative ones in the case of four stations on the North West Coast, but none for the remaining 8 on the West Coast. The highest degree of persistence is found for the West Coast stations and the lowest for the East Coast ones. Thus, in the event of shocks, more decisive action is required in the case of West Coast stations for the series to revert to their original trend