Historic beach profiles in Wales have only been collected for the shorter chainage length of the Mean Low Water Neaps datum. These profiles are essential for informing the impact of Climate Change and sea level rise but cannot be compared to modern profiles which collect data on the longer chainage length of Mean Low Water Springs. Utilising AI techniques, we would like to extend chainage measurements along the shoreline to Mean Low Water Springs, enhancing comparability with modern profiles. This research offers an opportunity to delve into coastal geomorphology, contributing to a deeper understanding of coastal evolution.

This report focuses on predicting Mean Low Water Springs (MLWS) for different regions using Long Short-Term Memory (LSTM) networks. The primary objective is to analyze and predict trends for individual regions based on the dataset provided. The study includes data preprocessing, model development, training, and testing, followed by the visualization of actual and predicted trends. The findings will contribute to a better understanding of regional MLWS variations, which are crucial for coastal management, navigation safety, and environmental studies. The research technique includes several critical processes, beginning with the collecting and preparation of historical tidal data from Barry Iceland Beach. In order to improve model performance, data preprocessing involves cleaning the dataset to eliminate inconsistencies, normalizing the data, and arranging the data into a time-series format that is appropriate for LSTM model training. The study also highlights the significance of region-specific analysis, acknowledging that there might be substantial regional variations in MLWS patterns. In order to capture the intricate temporal correlations, the LSTM model creation procedure involves designing a network architecture that is specific to the properties of tidal data and uses numerous LSTM layers. The model's performance is optimized through hyperparameter tweaking, and its accuracy and generalizability are guaranteed through thorough training and validation.