Automatic calibration of hydrodynamic models for fringing reefs
As global sea levels rise and storm frequency and intensity shift due to climate change, tropical coral reef-lined coasts are becoming increasingly susceptible to wave-driven flooding. In addition, coral reefs, which are essential for coastal protection, are deteriorating due to ocean acidification...
| Autor: | |
|---|---|
| Tipo de recurso: | tesis de maestría |
| Fecha de publicación: | 2024 |
| País: | España |
| Institución: | Universidad de Cantabria (UC) |
| Repositorio: | UCrea Repositorio Abierto de la Universidad de Cantabria |
| Idioma: | inglés |
| OAI Identifier: | oai:repositorio.unican.es:10902/35933 |
| Acceso en línea: | https://hdl.handle.net/10902/35933 |
| Access Level: | acceso abierto |
| Palabra clave: | Nearshore hydrodynamics Hybrid modelling Fringing coral reefs Wave breaking coefficient Friction coefficient Hidrodinámica costera Modelado híbrido Arrecifes de coral de franja Coeficiente de rotura de olas Coeficiente de fricción |
| Sumario: | As global sea levels rise and storm frequency and intensity shift due to climate change, tropical coral reef-lined coasts are becoming increasingly susceptible to wave-driven flooding. In addition, coral reefs, which are essential for coastal protection, are deteriorating due to ocean acidification and other environmental stressors, thereby weakening their ability to dissipate wave energy. In this scenario, accurate downscaled predictions of nearshore wave processes are vital for reducing the susceptibilities of these environments and developing adequate adaptation strategies. To achieve these accurate predictions, traditionally, forecasting tools relied on modeling coastal dynamics using high-fidelity numerical models. While these models provide detailed simulations, they are costly in terms of computational resources when applied under a dynamic downscaling approach. To overcome this limitation, hybrid approaches, also known as metamodels, have been developed. These metamodels combine numerical models with statistical techniques, aiming to reduce computational costs by predicting wave behavior with fewer exhaustive simulations. However, the accuracy achieved by these approaches is highly dependent on the correct estimation of certain calibration coefficients of the numerical models. A calibration process commonly involves comparing real measurements with the numerical model outputs to identify the most accurate set of calibration coefficients. The longer the observations and the greater the number of coefficients to be calibrated, the more the number of required simulations increases exponentially, which can lead to significant computational effort and make it difficult to obtain accurate results. To address these challenges, this project introduces CHySwash, an advanced methodology that builds upon the foundation of its predecessor, HySwash (Ricondo et al., 2024). CHySwash is designed to streamline the calibration of numerical models, significantly reducing the time and computational resources typically required. It achieves this by integrating advanced techniques, including sampling, clustering, and interpolation, alongside an automatic calibration process powered by the Shuffled Complex Evolution optimization algorithm, renowned for its effectiveness in parameter optimization. The proposed methodology is applied in a monitored coral reef-lined coast, Molokai, Hawaii. Specifically, we aim to predict the optimal wave breaking and friction coefficients, which govern the wave breaking process and the dissipation of wave energy. |
|---|