Groundwater level time series modelling focuses on reproducing and interpreting variations in groundwater levels measured in monitoring boreholes over time. Rather than explicitly resolving spatial groundwater flow, these approaches use lumped or semi distributed representations, or machine learning methods, to relate external drivers - such as rainfall, evapotranspiration, recharge or pumping - to observed groundwater responses.
Numerous groundwater level time series modelling approaches are available, each offering different strengths depending on data availability, hydrogeological complexity and the questions being addressed. Two papers that provide an overview of these approaches are those of Collenteur et al. (2024) and Gilbert (2025).
Freely available groundwater level time-series modelling software
AquiMod
AquiMod, developed by the British Geological Survey, represents aquifer behaviour using a parsimonious, physically interpretable model structure that links climate inputs to groundwater levels through conceptual recharge and storage components. It has been successfully applied across a wide range of hydrogeological settings worldwide.
GARDENIA
Developed by the French geological survey, BRGM, GARDENIA simulates the water balance and groundwater level of a catchment using a lumped modelling approach.
HydroSight
HydroSight is a hydrogeological statistical software package comprising a groundwater hydrograph time-series modelling and simulation framework plus a data quality analysis module. The toolbox can be used from a stand-alone graphical user interface or from within Matlab. It uses a transfer function noise modelling approach to link the response of a groundwater system to stresses (e.g. rainfall, evaporation, pumping) via statistical response functions.
Pastas
Pastas is an open-source python package for simulating groundwater level times-series. It uses a transfer function noise modelling approach to simulate groundwater levels. An associated paper describes how Pastas models are constructed in seven steps: import Pastas, read the time series, create a model, specify the stresses and the types of response functions, estimate the model parameters, visualize output, and analyze the results.
Machine learning methods
Machine learning (ML) methods are increasingly being used to simulate groundwater level time-series. For example, Python codes implementing the artificial neural networks described by Wunsch et al.