The predictive accuracy of shoreline change rate methods in Point Pelee, Canada

Abstract

Shoreline retreat has significant consequences for Point Pelee National Park's (PPNP) ecological and economic systems. Using airphoto-based data, three methods for shoreline position prediction are evaluated for predicting observed shoreline positions: end point rate (EPR), linear regression (LR) and lake level predictor (LLP).The triangular cuspate foreland park has two sides facing the lake from east and west. On both sides, short-term predictions were more accurate than the longer-term. For eastern and western PPNP, the LR and EPR methods performed best, respectively. The LLP method performed better for the western side, underscoring the relationship between water level and shoreline position. For all methods, the highest errors in prediction were for the northeast PPNP, an area influenced by artificial structures adjacent to the park.This study proposes site-specific method testing before predicting shoreline positions to quantify the errors associated with each method. The LR method performs best whenever there is a strong long-term trend for shoreline position changes. The performance of the EPR method depends largely on the selection of the two points used in its calculations. Human alterations of the sediment budget likely lead to high uncertainty in shoreline position predictions for affected shores.