A 50 m cut-off depth and 2 km distance exclusion have been widely used in the Canterbury region to determine whether a stream depletion assessment is required, based on the assumption that deep wells (> 50 m) and distant wells (>2 km from a stream) do not cause significant depletion. Installation of wells to depth slightly greater than 50 m became commonplace to avoid stream depletion assessment requirements and potential water take restrictions during low stream flows. Environment Canterbury engaged KSL to test the 50 m depth and 2 km distance assumptions to support robust water resource management decisions.
We developed a simple numerical model via the FeFlow Python (FePy) interface to assess the relationship between stream depletion rates, well depth, pumping rate and separation distance for a broad suite of statistically-derived aquifer properties. We then used the numerical simulation results as input data for development of two meta models.
Metal model 1 used a nonlinear (sigmoid) function to predict 150-day stream depletion as a function of both depth and distance and included a subset of models for each vertical hydraulic conductivity quartile (i.e. 25th, 50th, and 75th percentiles). This facilitated rapid assessment of the potential stream depletion after 150 days for all consented takes.
Meta model 2 used a high order polynomial to predict stream depletion recovery as a function of time after the cessation of pumping (e.g. due to implementation of flow restrictions) at a stated depth and distance. The use of high order polynomials to predict non-linear effects is commonplace and sufficiently robust assuming the metamodel is not used to extrapolate beyond the ‘measured’ data
The results showed considerable potential for deeper and more distant wells to cause significant stream depletion effects with potentially significant implications for future water resource management in the region.