Global shoreline forecasting with satellite imagery
Present forecasting of shoreline behavior is focused on local models. Locally means that the model is limited to forecasting shoreline behavior only locally. The (trained) model is designed and validated for purely local cases and not applicable to other cases. Next to that, the most frequently used forecasting models are process-driven and statistical forecast methods. Both process-driven as statistical methods are methods which are not actively including information on other locations worldwide. As this thesis will focus on sandy shorelines and since 31 % of the ice-free shorelines are sandy shorelines, there is a wide availability of shoreline data. Machine learning forecasting algorithms could benefit from this large set of time series data, indicating a shift from a local to a global model.
This study aims to investigate possibilities to improve shoreline forecasting position by utilizing information from other coastal areas. This will be done by creating a global model for shoreline forecasting. A global model will enhance the exchange of cross-time series information worldwide and therefore will contribute to learning from best practices.
