Estimating groundwater recharge and its associated uncertainty: Use of regression kriging and the chloride mass balance method

摘要

The chloride mass balance method of estimating groundwater recharge is generally applied at a point scale but water resources management generally requires information at a regional scale. To estimate groundwater recharge regionally previously used upscaling methods have ranged from simple averaging to empirical relationships and geostatistical interpolation. This study combined the best components of these methods by using regression kriging: in data-sparse areas the recharge is upscaled using global regression equations with gridded rainfall and surface geology as covariates, while in data-dense areas the kriging of the regression equation residuals ensures that the upscaled recharge estimates respect the point estimates of recharge. The uncertainty in the recharge estimates was quantified using 1000 stochastic replicates of the chloride deposition of rainfall, the chloride exported in runoff, the chloride concentration of the groundwater and the regression equations used to perform the upscaling. This study focused on the coal bearing Sydney, Gunnedah, Gloucester and Surat Basins of Eastern New South Wales (Australia). Historically, groundwater recharge to the Permian units of these basins has received little attention due to their low yields of poor quality water. The increased potential extraction of regional groundwater due to coal development has demanded a greater understanding of the water balance. We found that recharge is highest in the younger productive aquifers (up to 20% of long-term mean annual rainfall) and lowest in the Permian units (-1% of long-term mean annual rainfall). The magnitude of the uncertainty is often close to the magnitude of the median recharge estimate. The method developed here for upscaling the point estimates of recharge using the chloride mass balance provides robust estimates of recharge and its associated uncertainty. The method is applicable to any regional study with variable density input data. Crown Copyright (C) 2017 Pub