Modelling groundwater flow in a variably-connected aquifer-stream system

University of Technology, Sydney. Faculty of Engineering and Information Technology.Catchments with variably-connected surface and subsurface flow systems are not uncommon in Australia or through various parts around the globe. Management of the available groundwater and surface water resources in such generic types of catchments is critical to ensure the sustainability of these valuable assets. This requires a decision making to be based on quantitative estimates of available volumes in the various interconnected water bodies, usually derived via suitable modelling. Fully coupled modelling of such systems still faces several complications such as: proper choice of spatial scale that better represents the interconnected system; availability and ease of access to the data required; availability of capable software to perform the simulation; and, occasionally certain jurisdictional conflicts- where these water bodies cross the trans-boundaries between neighbouring regions. Under these challenges, pseudo coupling- a less data-intensive but still rigorous modelling of aquifer-stream system- can provide estimates of acceptable accuracy for management decisions about the resource. Therefore, the primary objectives of the present research were to investigate the effect of variable model cell dimensions on the resulting simulated aquifer-stream water balance estimates; and to develop a pseudo-coupled groundwater-surface water model of a representative catchment with an unregulated intermittent stream. The first objective is to prove the effect of grid resolution on the calculated water balance components in modelling an interconnected aquifer-stream regime. Such analysis is of high importance for resolving double accounting and related issues in management of water allocations. This objective has been accomplished by pseudo coupling of MODFLOW with RIV and STR algorithms and MODFLOW-SURFACT with RIV applied to a synthetic aquifer-stream. Six cell resolutions (lOxlO, 25x25, 50x50, lOOxlOO, 150x150, and 250x250 m) were developed for pumping and nonpumping scenarios. It was found that as the cell dimension increases, the difference in the exchanged fluxes between the river and the aquifer also increases and could be more than 100% of the base case (i.e. the 1 Ox 10 m mesh, which has cell dimensions equal to the river width). The outcomes of the grid variation experiment were applied to a three-dimensional flow model with grid resolution of 250 x 255 m for the entire aquifer-stream system in Zone 2 of Cox's Creek Catchment. The MODFLOW-SURFACT and RIV algorithm has been utilised in the pseudo-coupled simulation of the groundwater and surface water regimes over 24 years. The qualitative assessments and the quantitative calibration measures illustrated that the model could reproduce the observed groundwater level variations. The hydrographs support the observational inference that the lower aquifers are probably used for irrigation more than the upper one. The contribution of Cox's Creek to the total inflow recharge is about 2852 ML/yr, which is nearly 13.4% of the total feed to the aquifers, and is around three times that from rainfall. The aquifers recharge the Cox's Creek by approximately 111 ML/yr (0.5% of the total groundwater outflow). The Creek receives the least amount of its flow from the underlying aquifers, a finding which supports the work of other researchers. The simulated monthly average leakage and baseflow of the Creek were 7.76 and 0.26 ML respectively throughout the simulation. These values provide further evidence that the Creek is generally a losing stream. While the simulation model has been designed for Zone 2 in the Cox's Creek region, it has potential for application to other catchments with unregulated intermittent streams. Such merits should prove helpful to decision makers in managing water resources in regions of similar character