Artificial neural network for the prediction of shoreline changes in Narrabeen, Australia

Abstract

Shorelines around the world are of great importance because of their role in people's lives. Therefore, understanding its behavior seems to be vital. Shoreline changes and its realignments are highly nonlinear and finding a method in such a way that it could model its patterns would be very useful. ANN's are popular methods, as they have been applied to numerous problems. In this study, recurrent ANNs such as NARNET and NARXNET are used to model the shoreline changes in Narrabeen Coast, Australia, between 1980 and 2014. Their outputs represent a reliable performance for predicting shoreline changes based on historical data. These results also are compared with other methods, including RBF, GRNN, and TDNN. The NARNET showed most accurate results with MAPE= 17.18%, and the NARXNET had the best correlation with CC=0.26. It has been indicated that the NARNET and NARXNET are better methods since they need less extra data, besides the shoreline position itself.