Groundwater-Level Forecasting: Evolution from Conventional to Quantum-Inspired Approaches

Groundwater is an essential source of fresh water that forms an important part of hydrological model. Ground Water Level (GWL) is dependent on many factors such as meteorological conditions, geographic conditions, and human interactions taking place for the purpose of agriculture, industrial uses and livestock. Forecasting ground water levels is a challenging aspect, due to the uneven, nonlinear behaviour of groundwater. Traditional monitoring methods and analytical methods, which are labour intensive and data-dependent struggle with data gaps and evolving climatic conditions. The review is conducted to identify the best modelling techniques followed and the challenges faced in predicting ground water level. The review compares the data used, its sources, data preprocessing techniques, feature extraction techniques, application of various machine learning techniques and optimization techniques used in GWL prediction. Machine learning and deep learning-based models have consistently performed well in not only accurately forecasting GWL variations, but also mapping groundwater potential, and detecting anomalies. It is also observed in the review that the wavelet transforms, ensemble models, and empirical mode decomposition (EMD) techniques have been successful in handling the complex nature of ground water data. The survey also highlights the significance of model tuning, time-step considerations, and methodological decisions. Future studies may benefit from the use of diverse and widespread data enabling researchers understand the complex interactions better. The study further encourages to integrate approaches that use physical principles with data driven techniques, hybrid models and nature inspired optimization techniques for increased model performance.