The existence of temporal scaling in hydrological time series has been acknowledged for decades but seldom rigorously investigated in terms of the effect of various catchment-scale forcings on the scaling behaviour of local series. The aim of this work is to establish the relationship between these forcings, such as a catchment’s physical properties or scaling behaviour of local series, on the scaling behaviour of groundwater levels (GWL) in the catchment. The benefit from performing this exercise is that it sheds light onto the local processes governing the fluctuation of GWL.
The study site selected for this work is in Wallingford, United Kingdom. GWL and river stage – monitored at 1-minute intervals – and corresponding rainfall, temperature and other hydrometeorological series – monitored at 15-minute intervals – are available for a period of 4 years. The site is formed of a riparian aquifer that drains into River Thames and the GWL are highly responsive and fluctuate over a wide range of temporal sales.
To rigorously investigate local forcings on Wallingford’s GWL scaling behaviour, a physically-based coupled recharge-groundwater flow model is developed and calibrated using multi-objective evolutionary algorithms to simulate GWL scaling behaviour. The scaling behaviour of all series is analysed using a newly developed robust detrended fluctuation analysis procedure (r-DFAn) to reliably and objectively quantify and identify the different scaling regimes that may exist in hydrological variables (Habib, Sorensen et al. 2017).
By varying the various inputs and parameters of the model to study their effect on GWL scaling behaviour, it was found that only changes to rainfall scaling behaviour produced statistically significant changes to GWL scaling behaviour. Other inputs and parameters tested were the aquifer’s physical properties, parameters affecting the recharge process, potential evapotranspiration’s scaling properties, river stage’s scaling properties and distance of the GW borehole from the river. This implies that the fluctuation structure of GWLs at the Wallingford site is dictated by the fluctuation structure of the rainfall as opposed to local geology or parameters controlling the recharge process.
An additional finding is that the scaling behaviour of a simulated hydrological variable can be used as a metric to ensure that a hydrological model correctly replicates the variable’s fluctuation structure across all studied scales.