Modeling study demonstrates accurate large-scale shoreline change hindcasting across coastal sandy beaches, suggesting coastal orientation and latitude govern morphodynamic memory.
Key Points
Develop a hybrid, scale-adaptive convolutional framework that incorporates physical hydrodynamic forcing into data-driven shoreline change modeling over large coastal domains.
Formulated a framework combining convolution operations and optimized kernel functions over hydrodynamic forcing to capture beach response and morphodynamic memory.
Hindcast 40 years (1984–2024) of satellite-derived shoreline positions across 22 littoral cells covering ~530 km of coastline in Oregon and Washington, analyzing over 10,000 transects at 50-m resolution.
Achieved a median validation Root Mean Square Error of 14.1 m and a Mielke’s index of 0.51 across the regional hindcast domain.
Identified that geographical latitude dictates alongshore response timescales, whereas coastal orientation controls the magnitude of seasonal cross-shore shoreline shifts, intensifying on more west-facing beaches.