SPATIAL CHANGES IN SEA LEVEL ALONG THE EAST COAST OF THE UNITED STATES

Sea level trend estimation has been a topic of discussion for many years. Studies reporting accelerating rates of sea level rise along the Northeast of the United States have increased research efforts to better understand the processes driving this trend. This study analyzes long-term (1940-2013), monthly water levels from the Permanent Service for Mean Sea Level (PSMSL) to investigate spatial changes in sea level trends along the east coast of the United States. Different polynomial fits to the data with and without seasonal cycle were considered. The goodness of fit to a number of statistical models has been evaluated using the Bayesian Information Criterion (BIC). A model combining a linear trend and seasonal cycle yielded the best fit as evaluated by the BIC. Water levels from stations in the Northeast are not highly correlated with water levels in the Southeast (R²~0.7-1.0 for pairs of stations in the same region; R²<0.6 for pairs in different regions). Water levels along the East Coast correlated with ENSO with small (R~0.07-0.15) but significant values Water level correlations with the North Atlantic Oscillation were of the same order as ENSO’s in the Southeast, but intensified in the Northeast (up to 0.3). Using the Multivariate Adaptive Regression Splines (MARS) methodology, a breakpoint (separating the time series into regimes) for most of the stations was identified around 2002. However, 1998 was the calculated breakpoint using the Gaussian Mixture Model (GMM), which separates spatially consistent regimes with distinct trends. The sea level trends after 1998 were significantly higher for stations in the Northeast, and they ranged from 5-7 mm/year. The results suggest accelerating sea level rise mostly in the Northeast stations of the United States during approximately the last 15 years, with no evidence of such behavior prior to that.