Coastal storms and shoreline change: signal or noise?

A linear regression (studentized) residual analysis was used to identify potential shoreline position outliers and to investigate the effect of the outliers on shoreline rate-of-change values for transects along the Outer Banks, North Carolina. Results from this analysis showed that, over a 134 year period, storm-influenced data contribute statistically significant information to the long-term signal. Consequently, storm-influenced data points do not appear to be temporal outliers and thus, do not need to be excluded from a long-term analysis of shoreline changes. Furthermore, projections of the upper and lower confidence intervals (CIs) for the regression line to the year 2010 (24 year extrapolation) showed that including or excluding outliers had minimal effects on shoreline position predictions.