Integrating GIS and RDBMS technologies during construction of a regional groundwater model

Abstract

Finite difference groundwater flow models like MODFLOW require cell-by-cell averages for a myriad of parameters. In reality, the data a modeller uses comes from many sources in a variety of formats. Point measurements from boreholes are a critical dataset and can be combined with lines (eg water level and structural contours), polygons (eg surface geology and land use maps) and rasters (eg Landsat imagery). Firstly, the modeller needs a working environment to store, integrate and analyse these datasets and to derive the cell-by-cell model input. Secondly, the model output needs to be compared with the original source data that describes the real world. Borehole information is typically stored in a relational database management system (RDBMS) and geographic information systems (GIS) are designed for managing spatial information. These technologies have been used as the working environment for the Lower Darling model, which is a large regional groundwater flow model within the Murray Geological Basin, southeast Australia. Different strategies were developed to manipulate the available data into MODFLOW input files and also for the modelled heads and flows to be compared with field observations. Some of these strategies are specific to the Lower Darling model, but others are generic and can be easily applied in the data manipulation and calibration of groundwater models for regional aquifer systems.

Introduction

The Lower Darling groundwater flow model is one of five regional groundwater flow models subdividing the Murray Geological Basin, in southeastern Australia (Fig. 1). The other regional models cover the Lachlan Fan (Kellett, 1997), Lower Murrumbidgee (Punthakey et al., 1994), Southern Riverine Plain (Gutteridge, Haskins and Davey Pty Ltd, 1992) and the South Australian/Victoria Mallee (Sinclair Knight Merz Pty Ltd, 1995) areas. These models have been constructed to predict the long term groundwater response to natural resource management strategies within the Basin. The imperative is a rising watertable that has caused waterlogging and land salinisation as well as increased salt loads to the Murray River. This watertable rise is a consequence of increased recharge due to the clearing of deep-rooted, water-efficient native vegetation for the development of pasture, cropping and irrigation. Land and water resource managers require predictions on how the watertable and salt loads to the rivers would change due to practices such as further vegetation clearing, irrigation expansion or changes in river regulation.

The Lower Darling groundwater flow model covers nearly 80,000 km² of the northwest quadrant of the Murray Geological Basin(Fig. 1). The model area is typical of the Murray Mallee region, with a semi-arid climate, aeolian landforms and mallee eucalypt vegetation. The model contains the Darling River downstream of Wilcannia, the Menindee Lakes storages, Lake Victoria and the Murray River between Mildura and Morgan (Fig. 2).

At this stage, the Lower Darling model is a steady state, three-dimensional model simulating groundwater conditions observed in 1988 and uses MODFLOW code (McDonald and Harbaugh, 1988). It consists of five layers, corresponding to the layered aquifers and aquitards within the Murray Basin Cainozoic sediments. The underlying Lower Cretaceous sediments are indirectly represented in the model.

The Cainozoic sediments in the model area range from fluvial sands and silts, marginal marine clays and platform limestones (Brown and Stephenson, 1991). The structure of the underlying basement defines the pattern of sedimentation, with a series of linear northeast trending sub basins being centres of deposition. These coalesce into the deepest part of the Basin, the Renmark Trough, near the southern margin of the model, where over 600 metres of Cainozoic fill has accumulated. This progressive thickening of the Cainozoic sequence to the southwest is evident in the cross section of Fig. 3. In the northeast, fluvial conditions dominated during the Cainozoic with clays, silts and sands deposited in rivers, deltas and coastal swamps (Fig. 3). Seawards, towards the centre of the model, clays, marls and silts indicate shallow marine and lagoonal settings. More persistent marine conditions in the southwest are indicated by a series of platform carbonates.

The regional groundwater flow in the Cainozoic aquifers is from the Basin margins in the north and west, with the Basin structure of troughs and ridges directing flow towards the Murray River in the southwest (Fig. 4). Hydraulic gradients are typically low, around 5–10 cm/km. Along the Basin margins, the vertical hydraulic gradient is downward (Fig. 3) and leakage from the Darling River and the Menindee Lakes has caused a dilution aureole in the shallow aquifers. Hence, the freshest groundwater (600–900 mg/L) is found in a 2–10 km corridor along the Darling River channel (Fig. 4).

Groundwater salinity increases rapidly down gradient and can exceed sea water concentration of 35,000 mg/L (Fig. 4). In the southwest, groundwater discharge conditions prevail, manifested as active salinas in local depressions and saline groundwater accessions to the Murray River. Upwards groundwater flow is driven by high heads in the deeper aquifers, due to thinning of the aquifer and hydrostatic loading from overlying marine clays containing dense saline groundwaters (Fig. 3).

The combination of upwards leakage and high salinities in the southwest contributes significantly to the salt loads entering the Murray River between Mildura and Morgan. This is particularly true for the Woolpunda Reach, where 193 tonnes of salt is estimated to enter the river every day (Smith and Watkins, 1994). This has required the construction of a major groundwater interception scheme. In addition, irrigation development along the Murray River corridor has resulted in groundwater mounds laterally displacing saline groundwater into the river.