Groundwater surface mapping informs sources of catchment baseflow

Hydrol. Earth Syst. Sci., 19, 1599–1613, 2015 www.hydrol-earth-syst-sci.net/19/1599/2015/ doi:10.5194/hess-19-1599-2015 © Author(s) 2015. CC Attribution 3.0 License. Groundwater surface mapping informs sources of catchment baseflow J. F. Costelloe1 , T. J. Peterson1 , K. Halbert2 , A. W. Western1 , and J. J. McDonnell3,4 1 Department of Infrastructure Engineering, University of Melbourne, Melbourne, Australia 2 Ecole Centrale de Nantes, Nantes, France 3 Global Institute For Water Security, University of Saskatchewan, Saskatoon, Canada 4 School of Geosciences, University of Aberdeen, Aberdeen, Scotland Correspondence to: J. F. Costelloe () Received: 13 October 2014 – Published in Hydrol. Earth Syst. Sci. Discuss.: 5 November 2014 Revised: 9 March 2015 – Accepted: 9 March 2015 – Published: 7 April 2015 Abstract. Groundwater discharge is a major contributor to stream baseflow. Quantifying this flux is difficult, despite its considerable importance to water resource management and evaluation of the effects of groundwater extraction on streamflow. It is important to be able to differentiate between contributions to streamflow from regional groundwater discharge (more susceptible to groundwater extraction) compared to interflow processes (arguably less susceptible to groundwater extraction). Here we explore the use of groundwater surface mapping as an independent data set to constrain estimates of groundwater discharge to streamflow using traditional digital filter and tracer techniques. We developed groundwater surfaces from 88 monitoring bores using Kriging with external drift and for a subset of 33 bores with shallow screen depths. Baseflow estimates at the catchment outlet were made using the Eckhardt digital filter approach and tracer data mixing analysis using major ion signatures. Our groundwater mapping approach yielded two measures (percentage area intersecting the land surface and monthly change in saturated volume) that indicated that digital filter-derived baseflow significantly exceeded probable groundwater discharge during most months. Tracer analysis was not able to resolve contributions from ungauged tributary flows (sourced from either shallow flow paths, i.e. interflow and perched aquifer discharge, or regional groundwater discharge) and regional groundwater. Groundwater mapping was able to identify ungauged sub-catchments where regional groundwater discharge was too deep to contribute to tributary flow and thus where shallow flow paths dominated the tributary flow. Our results suggest that kriged groundwa- ter surfaces provide a useful, empirical and independent data set for investigating sources of fluxes contributing to baseflow and identifying periods where baseflow analysis may overestimate groundwater discharge to streamflow. 1 Introduction Groundwater discharge is a major contributor to stream baseflow. Quantifying this flux is of considerable importance to water resource management (Woessner, 2000; Sophocleous, 2002; Cartwright et al., 2014). In recent decades there have been dramatic increases in the extraction of groundwater for agricultural use, driven by factors such as expansion of irrigated agriculture in southern Asia (Llamas and MartínezSantos, 2005; Perrin et al., 2011) and long-term drought in southeastern Australia (Leblanc et al., 2012; van Dijk et al., 2013). It has been long recognised that over-extraction from aquifers may result in significant long-term declines in groundwater levels, resulting in decreases in baseflow in rivers (Bredehoeft et al., 1982). As a result, the switch to groundwater as a source of irrigation supply has the potential to exacerbate decreases in baseflow in rivers already experiencing reductions in flow from drought or instream water use. Whilst these generalities of groundwater extraction and stream baseflow reduction are clear, the particularities for any given catchment are complex and difficult to quantify. The separation of baseflow contributions from regional groundwater (i.e. where aquifers are unconfined in the vicinity of streams) from other shallower sources, like interflow, Published by Copernicus Publications on behalf of the European Geosciences Union. 1600 J. F. Costelloe et al.: Groundwater surface mapping informs sources of catchment baseflow bank storage return and perched aquifer discharge, is technically difficult to quantify. Nevertheless, this is fundamentally important for quantifying how regional groundwater extraction may affect baseflow in rivers (Wittenberg, 1999). Despite decades of work (e.g. Nathan and McMahon, 1990; Tallaksen, 1995; Wittenberg, 1999; Eckhardt, 2005), methods to quantify and discriminate between “slow flow” (itself a poorly defined term) contributions to the stream using only streamflow data are approximate at best. From a physical perspective, the baseflow component of streamflow is the sum of the slow flow pathways into the river (Ward and Robinson, 2000). Regional, unconfined groundwater (often termed “deep groundwater”) can discharge into the river via the valley floor or through more shallow, lateral flow paths, such as discharge into tributaries draining the valley slopes. Rain event driven interflow pathways can also contribute to tributary streamflow and recent work has shown a continuum between groundwater and interflow processes (sometimes referred to as “shallow groundwater” in hilly terrains) along the stream reach (Jencso et al., 2009; Jencso and McGlynn, 2011). In terms of water resource extraction (e.g. for urban supplies or irrigation on the valley floor), groundwater pumping typically targets the deep groundwater, and often in alluvial valley locations where the depth to groundwater is at a minimum. Thus, it is important to be able to differentiate between contributions to streamflow from deep groundwater discharge (more susceptible to groundwater extraction) compared to shallower interflow processes (arguably less susceptible to groundwater extraction). But how can the baseflow components be identified? Digital recursive filters are the most common method of separating baseflow from streamflow but do not discriminate between the different components of baseflow, and the estimate is integrated over the entire catchment area upstream of the gauging station. The technique rests on the assumption that baseflow is comprised of linear or non-linear outflow from an aquifer (e.g. Nathan and McMahon, 1990; Wittenberg, 1999; Eckhardt, 2005). All of the filter approaches require calibration of 1–3 parameters based on subjective criteria (e.g. recession curve analysis, typical values, etc.). Calibration of these parameters against synthetic baseflow derived from a numerical model has shown that optimal values vary considerably with catchment and climatic characteristics, many of which are not known or not possible to know a priori for natural catchments (Li et al., 2014). There is typically significant variability in recession curves from a give (...truncated)