CART MODEL FOR PREDICTING DUNE EROSION BASED ON STORM INTENSITY AND BEACH MORPHOLOGY

Abstract

Hurricane Sandy (2012) resulted in historic losses to the beach and dune system in many parts of New Jersey, leaving inland structures susceptible to wave and surge-induced damages. To defend against future threats, the state, in conjunction with the US Army Corps of Engineers, has pursued a statewide system of engineered beaches comprised of a dune paired with a beach nourishment. Currently in the state there is much debate over the degree of protection these projects currently offer, and how this may change in the future due to climate change. Process-based models can be used to try to answer these questions; however, as the number of test cases increase the computational costs may become prohibitive. Here, a data-driven modeling approach is used to assess the variability in the vulnerability of the beach-dune system in New Jersey to a range of storms. Inputs to the model include storm intensity as characterized by the Storm Erosion Index (SEI) (Miller and Livermont 2008) and parameterized pre-storm morphology. The resulting classification tree ensemble has been shown to accurately predict dune volume loss (as a percent) due to historical storms (Lemke and Miller, 2021). Here, the model is applied to further investigate how changes to beach-dune system and sea level rise affect the probability of dune impact.