Temporal analysis of shoreline recession and accretion
The precision with which estimates of the shoreline
rates-of-change reflect actual changes and predict future
changes is dependent on: (1) the accuracy achieved in
collecting shoreline position data, (2) the temporal variability
of the shoreline movement, (3) the number of measurements
used in the computation, (4) the proximity of the observations
to actual changes in the trend (sample bias), (5) the
period of time between measurements, (6) the total time
span of the record, and (7) the method used to calculate
the rate. Over 75% of the data we have assembled in a
comprehensive United States shoreline information system
(CEIS) were computed using the end point rate (EPR) method.
The EPR utilizes only two shoreline positions to calculate
rate-of-change values. Methods used less frequently include
the average of rates (AOR), linear regression (LR), and
jackknife (JK) methods. All of these computation methods
fit a linear model to shoreline response. For coastal
areas with constant rates of shoreline change through
time, the results of all the methods are identical. For
a coastline with a non-linear response, a linear estimation
method can only approximate the average rate-of-change.
As the response of the shoreline becomes more non-linear,
the differences among the rate-of-change estimates give
by the various methods increase.
Using data from a 65 km long section of the Outer Bank
of North Carolina, we demonstrate the differences in computational
methods for estimating shoreline changes and we show how
the potential sources of error can bias the final statistics.
Dolan, R., Fenster, M.S., and Holme
Journal of Coastal Research, v. 7, no. 3, p. 723-744