Coastal flooding is threatening the personal safety, property, and social development of the low-lying land around the coast worldwide. Stormsurge is one of the main sources of coastal flooding. Tide and surge models can provide timely water level forecasts for coastal management
...
Coastal flooding is threatening the personal safety, property, and social development of the low-lying land around the coast worldwide. Stormsurge is one of the main sources of coastal flooding. Tide and surge models can provide timely water level forecasts for coastal management with the early warning of flooding. Although a regional model can be used to study effects of climate change in a specific area, global water level modeling provides some advantages, such as the long-term response of the extreme sea level and coastal flooding due to global warming and comparison of global surge differences between regions. Global hydrodynamic modeling is becoming an increasingly important research topic. Nowadays, with ever increasing resolution, neglected physical processes and parameter uncertainties due to the inaccurate input or empirical values is becoming more and more dominating the model accuracy. At the same time, measurements like the satellite altimeter and the in-situ tide gauges are able to monitor the water level changes, which offers the possibility to estimate uncertain parameters. In this thesis, we develop a parameter estimation scheme and implement it to a global tide and surge model, and subsequently, apply to improve the water level forecast skill. Themain challenges for large-scale parameter assimilation for tide models are in assessing parameter uncertainties, large computational demand, large memory requirement and insufficient observations. In this thesis, we explore these challenges using an application to the Global Tide and Surge Model (GTSM). A computationally efficient and low memory usage iterative estimation scheme is designed and applied to GTSM for bathymetry and bottomfriction coefficient calibration. In addition, we study how to make the best use of spatial sparse distributed observations...@en