The primary motivation of this study is to analyze the 1D-2DH hydrodynamic model of the San Francisco Bay and Sacramento-San Joaquin Delta (SFBD) outlined in Nederhoff et al. (2021). I compared model water level data to 70 tidal records from the National Oceanic and Atmospheric Association (NOAA), the United States Geological Survey (USGS), the California Data Exchange Center (CDEC), and from local municipalities throughout the Bay Area to investigate how the model captures water levels and tidal constituent amplitudes. While the Nederhoff et al (2017) model analyzed an extended time period from 1950-2019, I analyzed M2 amplitude and tidal water levels for the water year of 2017 (WY2017) with a larger dataset that extended into the Sacramento-San Joaquin Delta. Because WY2017 was a high river flow year for the Sacramento Delta, the model was able to be evaluated throughout a large range of flow regimes. I used harmonic analysis through the MATLAB package UTide (Codiga et al. 2011) to assess the model’s ability to replicate M2 amplitudes. I assessed the error for these M2 values as well as for tidal water levels. The average RMSE for M2 amplitude is 0.111 m across the entire model domain during WY2017, performing fairly consistent throughout the model. The one exception being the shallow and complex Grizzly Bay, which performed significantly worse, with RMSE values around 0.5 m. The model better replicated water levels in the 2DH grid representation of the San Francisco Bay ( Attempts to improve the model were mostly unsuccessful. I tried to increase the grid resolution at the Carquinez Strait to improve tidal propagation upstream, but altering the grid caused the coupling between the 2DH grid and 1D network to detach. This prevented the propagation of water flow in either direction at the coupling near Collinsville. The software required to fix this coupling was non-standard and unavailable for my usage, so I was unable to resolve the issue. I also attempted to create a new wind forcing file using in-situ data rather than the ERA5 reanalysis. This new wind forcing made negligible difference in water level and M2 model skill. An experiment in removing river flow showed that riverine impacts on elevating extreme water levels only have effects (>0.05 m) east of the Carquinez Strait. Extreme water levels west of this point in the San Pablo, Central, and South Bays are dominated by tides, storm surge, and to a lesser extent local wind. A decrease in tidal amplitude by river flow potentially decreases flood risk in some parts of the Bay during times of high outflow from the Sacramento-San Joaquin Delta. I also investigated maximum equilibrium effects of constant wind in the two prevailing wind directions (southerly and westerly) of the San Francisco Bay. The wind setup effect become more prominent (>0.05 m) at and above a steady 10 m/s in both directions. This study also showed that wind likely exerts a small influence on tidal properties, especially for winds greater than 10 m/s.

Les submersions marines d’origine météorologique sont des catastrophes naturelles majeures, responsables chaque année de milliers de morts et de milliards d’euros de dégâts. La partie centrale du Golfe de Gascogne est un territoire particulièrement vulnérable à cet aléa, comme nous l’a rappelé la forte submersion engendrée par la tempête Xynthia en 2010. L’objectif de ce travail est d’améliorer la compréhension des surcotes et des submersions marines dans cette zone de France dans une approche pluridisciplinaire mêlant la géomorphologie, l’océanographie et l’analyse d’archives historiques. Afin de juger du caractère exceptionnel de Xynthia, une recherche de l’ensemble des submersions marines qui ont affecté la région d’étude depuis 500 ans a été menée. La modélisation numérique des surcotes des 5 tempêtes engendrant des submersions au 20ème siècle, révélées par ces recherches, montre que des conditions météo-marines variées ont induit des niveaux d’eau et des submersions comparables à ceux provoqués par Xynthia. Ce constat est en désaccord avec les estimations de périodes de retour de niveau marin extrêmes basées sur l’analyse statistique de mesures marégraphiques et met en avant l’apport de l’approche historique dans de telles problématiques. Devant la forte vulnérabilité des Pertuis Charentais aux submersions marines, la modélisation statique de la submersion marine, méthode simple mais néanmoins fréquemment utilisée pour estimer l’extension des zones inondées, a été évaluée. Cette méthode fournit de bonnes estimations de l’extension de l’inondation dans les zones de faibles altitudes caractérisées par une faible distance entre le trait de côte et la limite continentale de la zone inondable, mais mauvaises lorsque cette distance est grande. En effet, lorsque l’inondation se propage loin du trait de côte, la dynamique de l’écoulement ne peut plus être négligée sur ces grandes distances. Afin d’anticiper de futures submersions, deux configurations des digues ont été testées par modélisation numérique au travers de l’exemple de l’estuaire de la Charente. Les hauteurs d’eau et l’inondation de Xynthia sont simulées en augmentant la hauteur des digues de l’estuaire, empêchant toute inondation des zones basses adjacentes ; puis en abaissant les digues bordant l’estuaire au niveau des plus hautes marées astronomiques et en créant une seconde rangée de digues protégeant les zones habitées. Cette seconde configuration permet l’inondation des zones non habitées mais empêche l’inondation des zones à enjeux importants, comme la ville de Rochefort. Il est montré que la rehausse de l’ensemble des digues entraîne des niveaux d’eau supérieurs de 1.2 m à Rochefort par rapport à la simulation sans modification de digues, alors que l’abaissement de celles-ci et la protection des zones à forts enjeux ne modifient pas la hauteur d’eau dans l’estuaire. Ainsi, la rehausse des digues côtières n’est pas une solution systématique car la protection contre l’inondation de toutes les zones côtières peut augmenter la vulnérabilité des zones à forts enjeux. / Storm-induced coastal flooding are major natural disasters, responsible for thousands of deaths and billions of euros of damages each year. The central part of the Bay of Biscay is vulnerable to this hazard, as recently shown by the strong flooding induced by Xynthia in 2010. This study aims to improve the understanding of storm surges and coastal floods in this area of France, using several methods such as numerical or static modeling. To assess the uniqueness of Xynthia, historical researches of coastal floods affecting the study area for 500 years was conducted. Numerical modeling of the storm surges related to 5 storms of the 20th century revealed by these researches shows that various meteo-oceanic settings induced water levels and coastal floods comparable to those caused by Xynthia. This finding challenges return periods estimations of extreme sea levels based on statistical analysis of tide gauges measurements and highlights the contribution of the historical approach to such issues. Given the high vulnerability on coastal floods of Pertuis Charentais, static modeling, a simple but frequently used method to estimate the extension of flooded areas is evaluated. This analysis shows that this method provides good estimations of flood extents in low-lying areas characterized by a small distance between the shoreline and the continental limits of the lower area, but bad estimations when this distance is large. These poor performances when floods spread away from the coastline are explained by the dynamics of the flow, which can no longer be ignored. Two coastal defenses strategies are investigated in the Charente-river Estuary by numerical modeling. Water levels and coastal floods induced by Xynthia are simulated with increased dikes height, preventing flooding of adjacent low-lying areas, and then with dikes lowered to highest high spring tide height and with a second rank of dikes preventing flooding of important issues areas, such as the town of Rochefort. It is shown that raising dikes leads to higher levels of 1.2 m in Rochefort compared to the simulation without changing dikes, while protecting issues do not affect the water level in the estuary. Thus, it is demonstrated that the systematic raising of dikes is not a solution because it can increase the vulnerability of important issues areas.