Groundwater Prediction in the Thames Basin, London, Using Temporal Fusion Transformer Models

Addressing Thames Basin aquifer complex dynamics in England, this study uses a Temporal Fusion Transformer (TFT) for groundwater level prediction. Our research combines extensive hydrological data with advanced machine learning suited to Thames Basin, where a complex network of rivers and streams substantially affects groundwater dynamics. Unlike previous studies, this research focuses on long-term forecasting with deep learning, offering a long prediction horizon. To rigorously examine the model performance and robustness on new, unseen data, we applied the walk-forward validation method and other matrices such as RMSE and R2 coupled with the Holdout technique. Our approach contrasts traditional Long-Short Term Memory (LSTM), Attention-based LSTM, and TFT, focusing on the basin’s aquifers, Chalk, Oolitic Limestone, and Lower Greensand. Whilst both LSTM models were optimised using the Bayesian technique, TFT was applied for its inherent capability in complex time series. Our methodology processed historical groundwater and rainfall data from 2001-2023, accounting for the potential lag in aquifer response to the proximity of the river system. The dataset served as training, validation, and holdout for each model, focusing on capturing the dynamic temporal fluctuation. The results clearly showed the superiority of the TFT model in all aquifer types compared to other models across all horizons 7, 30, and 60 days. In the 60 days, the best results were observed in the Chalk aquifer with RMSE of 0.04 and R2 of 0.97 in holdout validation. However, in Limestone and Lower greensand aquifers, the TFT showed RMSEs of 0.12 and 0.016 and R2s of 0.65 and 0.32, respectively. Traditional LSTM models demonstrated limited predictive power, with negative values in all aquifers, while Attention-based LSTM slightly improved the efficacy. This study highlights the potential of sophisticated machine learning in managing complex aquifers and predicting water tables.