**3. Study approach**

In order to investigate varying climatic conditions and the impact of frequent lower than average annual rainfall on observed groundwater levels the long-term climate data are analysed and compared for both study areas. In the long-term, rainfall, minimum and maximum temperatures are related to climate variability. The climate data obtained were analysed and statistically interpreted. Long-term data are from the South African Weather Service (Station: Cape Town Observatory/Airport) and Bureau of Meteorology (BOM with station in Laverton near Werribee).

The groundwater databases of the Department of Primary Industries (DPI) and Department of Sustainability and Environment (DSE) Groundwater Management System (GMS) were examined to select representative bores tapping the Werribee Delta aquifer. Also, from the National Groundwater Database (NGDB) managed by DWA, a few bores screened in the

Changes in Groundwater Level Dynamics in Aquifer Systems –

**4. Description of the study area** 

information are available.

Implications for Resource Management in a Semi-Arid Climate 81

Access to several unpublished reports has also yielded valuable information. A general overview of the study area is presented with the description of geology and hydrogeological

The vegetable growing Werribee Irrigation District (WID) lies on Melbourne's rapidlydeveloping western urban fringe underlain by shallow Delta aquifer. The name of the management area for the Werribee Delta aquifer is the Deutgam Water Supply Protection Area (WSPA). The aquifer is linked to both Port Phillip Bay and the tidal extent of the Werribee River (SRWA, 2009). Deutgam WSPA is located around the Werribee South irrigation area (Figure 1). On the other hand, the fresh fruits and vegetable farm area in Cape Town is located on the Cape Flats, especially the Phillipi-Mitchells Plain Irrigation area. A large portion of the area around Cape Town is the sand-covered coastal plain (Cape Flats) shown in figure 2. The City of Cape Town Management Area (CMA) is largely surrounded by the Atlantic Ocean to the west and south with the most prominent landmass being the Cape Peninsula, attached to the mainland by the sandy plain of the Cape Flats (Schalke, 1973; Theron et al., 1992). The greater portion of the entire sand cover of the Western Cape are been considered in this study, particularly the south-western part of the City of Cape Town and the north-western end (Atlantis), where basic data and bore

Fig. 1. Location of the Werribee Plains and Deutgam WSPA, western fringe of Melbourne.

Inset: Deutgam WSPA (red spot) in Victoria (grey shade) within map of Australia

settings in order to first understand the groundwater system in both areas.

Cape Flats aquifer were selected. These bores were investigated by evaluating the groundwater levels and salinity (specifically the electrical conductivity) within shallow aquifers. The criteria for selection were continuous groundwater level record (minimum of 10 years record, with minimal interruptions or errors) and screened within the respective aquifers under this study.

The time-series groundwater data at selected locations within Cape Town area were compared with those of bore network data in the Werribee Plain. The analysis of this data was undertaken using Hydrograph Analysis: Rainfall and Time Trends (HARTT), a statistical tool that analyses groundwater data using the effect of long-term rainfall patterns, determined by accumulative residual techniques (Ferdowsian et al., 2001). This method can differentiate between the effect of rainfall fluctuations and the underlying trend of groundwater level over time. Rainfall is represented as an accumulation of deviations from average rainfall, and the lag between rainfall and its impact on groundwater is explicitly represented. HARTT produces a fitted curve through the groundwater level readings.

According to Ferdowsian et al. (2001), two variables are used to produce this line:

Rainfall variable (X1); accumulative monthly residual rainfall (AMRR, mm), or accumulative annual residual rainfall (AARR, mm).

Time trend (X2) (1,2,3 days…from first reading)

At any point along the fitted curve, the following equation holds:

$$\mathbf{Y} = \mathbf{c} + \mathbf{a}\mathbf{X}\_1 \text{(rainfall)} + \mathbf{b}\mathbf{X}\_2 \text{ (time trend)}\tag{1}$$

Where:

c is the intercept.

a and b are coefficients calculated in the multiple regression analysis.

Y is the water level depth at a point along the fitted curve.

So, to calculate the effect of rainfall, the following equation is used:

$$\mathbf{Y}' = \mathbf{a} \mathbf{X}\_1 (\text{rainfall}) \tag{2}$$

And to calculate the underlying trend, the following equation is used:

$$\mathbf{Y}^{\prime\prime} = \mathbf{c} + \mathbf{b}\mathbf{X}\_2 \text{(time trend)}\tag{3}$$

The R2 value (the coefficient of determination) is the degree of fit of the calculated curve compared to the recorded water levels (a value of 1 is a perfect fit; the degree of fit becomes less with decreasing values below 1). The p-value indicates the level of significance of each variable. If the p-value is less than 0.05, then the variable is significant. If it is less than 0.01, then it is highly significant. If the trend is not significant (as determined by R2) then the rate of rise or fall is not reliable. And if the rainfall variable is not significant then the reliability of the effect and the delay period (in the hydrograph response to effective rainfall) is low (Ferdowsian et al., 2001).

The method improves the estimation of time trends and allows for better interpretation of treatment effects on groundwater levels. The advantage and limitation of this method over other techniques of hydrograph analyses have been highlighted in Cheng et al. (2011).

Access to several unpublished reports has also yielded valuable information. A general overview of the study area is presented with the description of geology and hydrogeological settings in order to first understand the groundwater system in both areas.
