Detecting and Understanding the Accelerated Sea Level Rise Along the Mid-Atlantic Coast during Recent Decades

Wednesday, 17 December 2014
Jessica Sarah Kenigson, University of Colorado at Boulder, Boulder, CO, United States and Weiqing Han, Univ Colorado, Boulder, CO, United States
A “hotspot” of accelerated sea level rise has recently been reported along the east coast of the United States, potentially linked with slowdown of the Atlantic Meridional Overturning Circulation in a warming climate. However, separating acceleration in the long-term trend from transient acceleration due to natural variability - particularly the approximately 60-year cycle associated with the Atlantic Multidecadal Oscillation - poses technical difficulties. The Empirical Mode Decomposition (EMD) and Ensemble EMD (EEMD) methods have been used to isolate the nonlinear trend from oscillations on various timescales, allowing robust acceleration estimates for the trend. Yet the accuracy of these methods in detecting accelerated sea level rise, particularly given the limited lengths of available tide gauge records, has not yet been fully justified. In this study, idealized sea level time series are constructed based upon interannual, decadal, and multidecadal oscillations obtained from tide gauge observations and prescribed trends with known accelerations. The idealized records are then analyzed with the EMD and EEMD methods and time-varying acceleration error estimates are obtained. Finally, the methods are applied to tide gauge observations from the Atlantic coast encompassing the hotspot region, and the acceleration estimates are interpreted in light of the results from the idealized data. Generally, EEMD provides more stable acceleration estimates that are less sensitive to the record start date than EMD. When the data record exceeds twice the ~67-year multidecadal oscillation period, the EEMD acceleration error falls to ~25% or less for idealized annual mean records with a quadratic trend. For records of intermediate length – between one and two multidecadal oscillation periods – the acceleration error is strongly time-varying. For data records shorter than ~67 years, the multidecadal oscillation is inadequately separated from the nonlinear trend and contributes significantly to the large acceleration detected in the hotspot region during recent decades. However, there is evidence of secular acceleration at Atlantic City, NJ and Philadelphia, PA. These findings have important implications for improved detection of regional sea level acceleration in a warming climate.