Soheil Ataei Hassankiadeh, Mehdi Adjami, Saeid Gharachelo,
Volume 20, Issue 44 (10-2024)
Abstract
With the advancement of machine learning and algorithms and the establishment of research sites in coastal areas, valuable information has become available that can be utilized in the development of coastal engineering. This study attempts to examine the coastal area of Narrabeen, Australia with a new approach using machine learning. One of the most important factors in understanding coastal behavior is its self-reconstruction under non-stormy and long-term conditions. Recognizing and describing the phenomena affecting the equilibrium performance of the coast is also of great significance. After sorting the initial data, the best behavioral pattern was examined using a regression decision tree algorithm by evaluating the combination of error and model complexity, and the most appropriate scenarios were selected to describe the influencing parameters on objective functions (shoreline changes and coastal platform geometry). Accordingly, to describe shoreline changes, ∆BW, Berm Slope, SLR, and ζ were considered with values of R2=82% and RMSE=3.489 meters; to describe changes in the elevation of the coastal platform, BC Height, ∆x Shoreline, ∆x BC, and P were considered with values of R2=48% and RMSE=0.397 meters; and to describe the horizontal position of the coastal platform crest, BW, Berm Slope, ∆y BC, BC Height, E, and SLR were considered with values of R2=67% and RMSE=9.807 meters. According to the results of the description and the impact of hydrodynamic and morphodynamic phenomena using the regression decision tree method and the obtained error values and coefficient of determination, it can be stated that this method is suitable and reliable for understanding the governing phenomena on coastal features; Based on this, in order to investigate the morphodynamic changes of the coastal platform and to understand its equilibrium behavior, its geometric shape and initial slope play a significant role.
