During the duration of the project, the conducted research aimed at understanding dune stability mechanisms, dune equilibrium prediction, an investigation of in-situ deep water dune dynamics in the Queen Charlotte Sounds in New Zealand along with improving a numerical model. The tasks that were carried out during the Marie Curie GF-DUNAMICS project were as follows : 1) Setup and preparation of the flume experimentations (sieve analysis, flume calibration, instrumentation, testing), 2) PIV experimentation for five scenarios : observation of dune growth, equilibrium, acquisition of series of images at high frequency & resolution with the PIV camera, 3) Numerical modelling study with DUNE2D of dune saturation for seven configurations and explanation of dune saturation mechanisms, 4) Machine learning study using genetic programming and derivation of prediction equations for dune dimensions at equilibrium, 5) Environmental data analysis for the deep water dune fields in the Queen Charlotte Sounds, 6) Implementation of the tidal numerical model and simulations across several tidal cycles for analysis of in situ dune migration and saturation, 7)Testing the MIKE numerical model and implementing morphological solvers.
A detailed analysis of the physical mechanisms setting the position of the maximum of the bed shear stress over the dune profile was carried out. The findings gave a deeper insight into mechanisms controlling finite amplitude dune saturation for several flow conditions. It showed that four modes of dune saturation existed depending on the rescaled mean grain diameter, gravity, and the sediment flux inertia length. These findings were published in the journal Earth Surface Processes and Landforms (Doré and Coco, 2025). Machine Learning, particularly genetic programming, was utilized to develop highly accurate dune predictors that outperform current models (Doré and Coco, 2025b).
The flume study spanned over several months enabling the collection of valuable data essential for future analysis. A comprehensive sediment grain size analysis was conducted that served as the basis for the implementation of various experimental scenarios designed to replicate the formation of hydraulic dunes. However, the laboratory did not have a powerful enough LED system. An original methodology was therefore implemented using a powerful LED projector and lens system to focus beams for data acquisition on dune evolution. Five scenarios were successfully captured following the methodology.
The study of tidal sand dunes in the Queen Charlotte Sounds included the numerical modeling of representative dune profiles. The results showed a high frequency oscillation of dune crests during tides. For the first time, the results revealed the mechanisms of saturation of submarine dunes during tidal cycles. The analysis was carried out based on the previous findings that served to interpret the results. An article has been authored and is currently undergoing peer review at Marine Geology journal (Doré, Mac Donald and Coco, 2025).
The MIKE numerical model has undergone extensive testing, revealing that the existing morphological scheme struggles to accurately predict dune profiles, particularly at the crests where significant flux divergences occur. To address this issue, two new numerical schemes have been implemented within MIKE, designed to enhance stability and accuracy.
Doré A ., G. Coco, 2025. Numerical modelling investigation of finite amplitude subaqueous dune saturation, Earth Surface and Planetary Landforms , doi : 10.1002/esp.70046
Doré A ., G. Coco, 2025b. A hybrid, genetic programming and physically-based predictor of dune geometry, Geomorphology , doi : 10.1016/j.geomorph.2024.109495
Doré, A., MacDonald, H., Coco, G., 2025. Morphodynamics of Eddy-Driven Submarine Dunes in the Queen Charlotte Sound/Totaranui, New Zealand, Marine Geology, Under review.
