A lumped conceptual model to simulate groundwater level time-series

• Published in: Environmental modelling & software : with environment data news 2014-11, Vol.61, p.229-245
• Main Authors: Mackay, J.D., Jackson, C.R., Wang, L.
• Format: Article
• Language: English
• Subjects: Animal, plant and microbial ecology; AquiMod; Biological and medical sciences; Fundamental and applied biological sciences. Psychology; General aspects. Techniques; Groundwater level simulation; Lumped conceptual modelling; Methods and techniques (sampling, tagging, trapping, modelling...); Observation borehole hydrograph
• ISSN: 1364-8152; 1873-6726
• DOI: 10.1016/j.envsoft.2014.06.003

Lumped, conceptual groundwater models can be used to simulate groundwater level time-series quickly and efficiently without the need for comprehensive modelling expertise. A new model of this type, AquiMod, is presented for simulating groundwater level time-series in unconfined aquifers. Its modular design enables users to implement different model structures to gain understanding about controls on aquifer storage and discharge. Five model structures are evaluated for four contrasting aquifers in the United Kingdom. The ability of different model structures and parameterisations to replicate the observed hydrographs is examined. AquiMod simulates the quasi-sinusoidal hydrographs of the relatively uniform Chalk and Sandstone aquifers most efficiently. It is least efficient at capturing the flashy hydrograph of a heterogeneous, fractured Limestone aquifer. The majority of model parameters demonstrate sensitivity and can be related to available field data. The model structure experiments demonstrate the need to represent vertical aquifer heterogeneity to capture the storage–discharge dynamics efficiently. •We develop a lumped modelling framework to simulate groundwater level time-series.•The framework is flexible and can accommodate different structures.•We test different model structures in four contrasting aquifers in the UK.•Acceptable calibration and evaluation efficiencies are achieved for all sites.•The majority of model parameters can be related to known physical characteristics.