Artificial Intelligence-assisted decadal scale beach change forecasting

点击翻译按钮获取中文摘要

Annual UK cost of flood and coastal erosion damage is estimated at £540 million. Sea level rise and extreme weathers associated with climate change rapidly increases coastal erosion and flooding, and economic impacts. The UK Government's Resilient Nation prosperity outcome recognises the need to implement climate-adaptive, safe, and sustainable coastal management, focusing on natural processes. Planning and implementation of sustainable adaptation requires new, efficient tools to predict decadal-scale coastal change. The uncertainty that is inherent in coastal behaviour means that these must be built on advanced technologies such as Artificial Intelligence (AI) to allow efficient and robust testing against multiple scenarios. This project will make a major contribution to tackle this challenge and train and validate an efficient AI tool to emulate time-dependent, decadal-scale coastal morphodynamic change by adopting a sequential learning AI emulation framework. A well-validated high-fidelity process-based coastal computational model, driven by multiple decadal-scale continuous stochastic time series of sea states, generated using a Monte Carlo approach, will provide synthetic beach change data for training and validation of the AI emulator and address the widely accepted data-scarcity issue. The emulator will be a powerful surrogate to computationally costly conventional modelling approaches for predicting decadal-scale beach change at local and global scale, and a tool to discover nature-based, climate-resistant coastal management interventions.