Sediment Transport and Geomorphological Evolution in the Transgressive Ship Shoal, Louisiana: Insights from Geophysical Observation, Modeling, and Machine Learning Studies

Abstract

Ship Shoal has been a high-priority target sand resource for dredging activities to restore the eroding barrier islands in Louisiana, USA. Many studies have been performed in Ship Shoal in recent decades, but our knowledge of temporal and spatial variations of sediment transport and post-dredge morphology is still limited. The objectives of this dissertation are to explore the sediment infilling process, the spatial and temporal evolution of the morphology of dredge pits near and on Ship Shoal by using multiple methods. Scientific methods used in this study include (a) geophysical observation using bathymetry, side-scan and subbottom technique, (b) sediment transport hindcasts using Regional Ocean Modeling System, and (c) machine learning technique including random forest, classification tree, and logit lasso. There are a total of four primary findings in this dissertation. (1) Caminada dredge pit located on the east of Ship Shoal displays a unique mixing and redistribution of cohesive mud with noncohesive sand. Initial sediment infilling in this pit appears to be dictated by micro-topography on the pit bottom created by the dredging cuts. Raccoon Island dredge pit located north of Ship Shoal is in a mixed environment in which sand was capped by mud before dredging, and it was filled with mud after dredging. The sediment infilling rate and pit wall slope changing rate in Raccoon Island pit is greater than these in the Caminada pit, likely due to turbid ambient sediment concentration and trough-like morphology. (2) A sediment transport model shows a minimal amount of suspended mud from the Atchafalaya River that can be delivered to Ship Shoal in a one-year time scale, but suspended mud from nearby estuaries and inner shelf could be episodically transported to Ship Shoal to fill in dredge pits. Two hurricanes and one tropical storm during 2017-2018 changed the direction of sediment transport flux near Ship Shoal and contributed to the pit infilling (less than 10% for this specific period). (3) Geophysical and geomorphological features, including backscatter, roughness derived from bathymetry, rugosity derived from backscatter, and bathymetry, are identified as the most effective predictors of sediment texture in Caminada pit via multiple supervised machine learning methods. (4) High-quality sand from sandy shoals (such as Caminada pit) and mud-capped paleo-channels (such as Raccoon Island pit) can both be used as borrow areas for barrier system restoration in Louisiana coast. However, Caminada and Raccoon Island pits are not considered as renewable resources for future restoration