Statistical modeling of historic shore erosion rates on the Chesapeake Bay in Maryland

Abstract

Few strong relationships exist along the Chesapeake Bay shoreline between the historic erosion rate and the distribution of any of several coastal parameters which were defined and tested using traditional regression and discriminant analysis procedures To develop a simple predictive equation for shore erosion that could be used by coastal managers, the entire Chesapeake Bay shoreline was partitioned into naturally occurring reaches 2–5 km in length, and the historic erosion rate on each reach was modelled as a function of five variables (a) shoreline type, (b) “100-year” storm surge height, (c) mean tide range, (d) wave climate, and (e) potential littoral drift rate The statistical analysis yielded a multiple correlation coefficient (r 2) of 30 8%, discriminant analysis showed only the first two variables listed above are useful predictors (i e, statistically significant) of historic erosion rates A 95-mile portion of the same bay shoreline in Queen Anne’s and Talbot counties was then partitioned into shorter reach lengths (1/2–2km) and more variables were included The multiple correlation coefficient (r 2) improved slightly to 32 9%, but only shoreline type and potential littoral drift rate were found to be useful predictors of historic erosion rates Curiously, the ability to model statistically the historic shore erosion rate is best on those reaches already substantially protected by structures For Queen Anne’s and Talbot counties, the multiple regression coefficient improved to 61 5% when only reaches 1/2–2km in length protected by structures were considered.