Rethinking deep learning: linear regression remains a key benchmark in predicting terrestrial water storage
This work addresses hydrologists and machine learning practitioners by showing that incremental improvements in deep learning may not always justify complexity for certain hydrological predictions.
The study tackled the problem of predicting terrestrial water storage (TWS) by comparing linear regression against deep learning models like LSTM and Temporal Fusion Transformers, finding that linear regression outperformed these more complex models on the HydroGlobe dataset.
Recent advances in machine learning such as Long Short-Term Memory (LSTM) models and Transformers have been widely adopted in hydrological applications, demonstrating impressive performance amongst deep learning models and outperforming physical models in various tasks. However, their superiority in predicting land surface states such as terrestrial water storage (TWS) that are dominated by many factors such as natural variability and human driven modifications remains unclear. Here, using the open-access, globally representative HydroGlobe dataset - comprising a baseline version derived solely from a land surface model simulation and an advanced version incorporating multi-source remote sensing data assimilation - we show that linear regression is a robust benchmark, outperforming the more complex LSTM and Temporal Fusion Transformer for TWS prediction. Our findings highlight the importance of including traditional statistical models as benchmarks when developing and evaluating deep learning models. Additionally, we emphasize the critical need to establish globally representative benchmark datasets that capture the combined impact of natural variability and human interventions.