Study of the Red River plume in the Gulf of Tonkin from cluster analysis and ensemble simulations

– Duy Tung Nguyen – Doctorant, MEAE –

Résumé :

This study aims to better understand the variability of the Red River plume in the Gulf of  Tonkin (GOT) in the area near the mouth and further offshore, using numerical modeling. Understanding plume variability and the fate of delta waters is of paramount importance for a better understanding and better predictive ability of ocean circulation and hydrology in the GOT, as well as for better management of coastal waters. and monitoring of coastal ecosystems.

In the first part of the thesis, a configuration of the SYMPHONIE model is implemented with realistic forcings and a variable high-resolution grid, based on the configuration of V. Piton (2019). A simulation over a period of 6 years (2011-2016) is carried out to study the daily to interannual variability of the plume of the Red River and of three rivers whose mouth is nearby. The simulation is then compared to several sources of observations. Then, the plume is identified using passive tracers injected into the simulation. Using an unsupervised machine learning algorithm (K-means), the main plume regimes and their evolution over time are classified and analyzed according to four clusters and then related to different environmental conditions. In winter, the plume is narrow and mostly stays along the coast due to the coastal current and northeast wind. In early summer, the southwest monsoon wind blows the plume offshore. The plume reaches its greatest coverage in September, after the peak in flow. On the vertical, the thickness of the plume also shows seasonal variations. In winter, the plume is mixed throughout the water column, while in summer, the plume can be detached both from the bottom and from the coast. The plume may deepen offshore in summer, due to strong winds (in May, June) or specifically due to a recurring eddy occurring near 19°N (in August). This first part was the subject of a publication in 2021.

The cluster analysis above shows that, whatever the cluster, the plume is strongly affected by the wind. Therefore, in the second part of this thesis, I use a set of simulations to evaluate the response of the model to disturbances added to the forcing wind. The sensitivity of the simulation presented in the first part is statistically evaluated by calculating the dispersion and the distribution of the variables of interest from a set of 50 members. Due to computing and memory constraints, this study is carried out over a short period, from June to August 2015, corresponding to the high flow season. First, the error on the forcing wind is estimated by comparison with a satellite product. Then, its impact on the model is evaluated for surface and subsurface variables. For surface temperature and salinity, the uncertainty is higher near the Vietnamese coast and the Red River Delta. On the vertical, the uncertainty is greatest at the surface for salinity and at the sub-surface for temperature. I then analyze the sensitivity of the river plume. The dispersion of the plume surface is greatest in August, which is also the period when the plume surface is greatest. The cluster analysis shows some cluster changes between different members of the ensemble, but the cluster most likely to occur is always that of the reference simulation (with the wind undisturbed). These limited changes suggest that the Part I results are indeed robust to wind forcing errors. Finally, the set is verified using the available sets of observations.

Jury :

Lien Zoom :

https://cnrs.zoom.us/j/94859082386?pwd=QjdIeFFtMmZGV0FYT0tSb1NqcEVCUT09

ID de réunion : 948 5908 2386

Code secret :  d91L2x