SUSTAINABLE YIELD OF GROUNDWATER OF THE GNANGARA MOUND, PERTH, WESTERN AUSTRALIA

Transcription

1 SUSTAINABLE YIELD OF GROUNDWATER OF THE GNANGARA MOUND, PERTH, WESTERN AUSTRALIA Salama, R.B., Bekele, E., Hatton, T.J., Pollock, D.W. and Lee-Steere, N. CSIRO Land and Water, Floreat Laboratories, Private Bag 5, WEMBLEY WA 6913 ph (8) , fax (8) Abstract: The traditional notion of defining the sustainable yield of an aquifer based on the average annual recharge rate or some derivative of the recharge component of the water balance is now obsolete. Groundwater technocrats are now aware that ecologically sustainable development of groundwater resources requires the identification of different users including the environment, the determination of the water requirements of each user, and securing the additional groundwater recharge or discharge which is essential to balance the legitimate use by one or more user. Alternatively, the provision of water at rates less than these requirements will mean a loss of environmental value or restricted human use. The Gnangara groundwater Mound located within the Perth Basin is a typical example of an area where a balance is actively being sought to meet the growing demand for groundwater by various users while securing water for the environment. The Gnangara Mound is a major source of water for market gardens, private and metropolitan water supply, as well as sustaining major pine plantations, native Banksia woodland and numerous wetlands and lakes. Due to increasing consumption by multiple users, water levels are declining at a constant rate. This is an indication that current groundwater discharge is being removed from aquifer storage and from reductions in the aquifer outflow component. It is anticipated that water levels will most likely continue to decline unless preventative measures are taken to increase the recharge or decrease groundwater ouflow to the ocean. The following measures are strongly advocated to abate and reverse current water level trends: (1) the removal of pine plantations to increase recharge; (2) the relocation of high volume abstraction to the discharge margins of the Gnangara Mound; and (3) the injection of wastewater in a line of wells parallel to the ocean west of the supply wells to reduce the outflow component and prevent seawater intrusion. Applying all these measures will only reduce the environmental impacts and not eliminate them, because the moment an aquifer is developed as a water source, even taking into consideration all the environmental water requirements, several ecologically sensitive areas will be affected. The delicate balance between the groundwater supply and demand by different users cannot be achieved without securing additional recharge for the utilised overdrawn parts of the aquifer or reducing groundwater discharge (aquifer outflow). Nevertheless, it is very difficult to achieve this balance, especially with changing weather patterns, increasing water demands and the relatively low infrastructure cost of groundwater abstraction that encourages further development to satisfy demand during drought periods. Key Words: Gnangara Mound, groundwater budget, recharge, safe yield, sustainable yield. BACKGROUND Hydrogeology Perth groundwater resources are obtained from three major aquifers that occur beneath the Perth metropolitan area: the Superficial aquifer, Leederville aquifer and Yarragadee aquifer. The Superficial aquifer is unconfined and composed of Quaternary-Tertiary sediments of the Swan Coastal Plain. The plain consists of a series of geomorphic entities, which are sub-parallel to the present coastline. The generalised surface geology of these entities is largely reflected in the distribution of soils, as shown in Figure 1, and consists of the Guildford clay in the east, the Bassendean Dune System and the Spearwood Dune System in the middle, with the Quindalup Dune System fringing the present coastline (McArthur & Bettenay 196). The aquifer consists mainly of sand, silt and clay in varying proportions and ranges in thickness up to 1 m. Radiocarbon age dating reveals that the groundwater in the Superficial aquifer varies from modern at the water table to 2, years at the base of the aquifer (Thorpe & Davidson 1991). The Leederville aquifer is composed of interbedded lenses of sandstone, siltstone and shale. It is confined above by shale and siltstone of the Osborne Formation or Kings Park Formation, and below by the South Perth Shale, Parmelia Formation or Cockleshell Gully Formation. It is in direct connection with the Superficial aquifer where the Osborne Formation has been eroded. Groundwater in the Leederville aquifer ranges in age from about 1,9 years to more than 36,8 years, but is generally less than 36, years (Thorpe & Davidson 1991). The Yarragadee aquifer consists of interbedded sandstone, siltstone and shale. The South Perth Shale or shale beds within the Parmelia Formation confine the aquifer above. It is in hydraulic connection with the Leederville aquifer where the South Perth Shale or Parmelia Formation is absent. Groundwater within the Yarragadee aquifer ranges in age from about 6 years in the recharge area to more than 37,7 years, but it is generally older than 36, years (Thorpe & Davidson 1991). 1

2 Soil and Logger Locations Farms # Strawberry Farm # Turf Farm # Loggers Cadastre Soil Bassendean Dunes Ja Jas J G Ws P Wp Wy DL Spearwood Dunes Ky Kg Kls Sp B Quindalup Dunes Q1 Q2 Q3 Q4 Qu Qp Qs Alluvial Terrain Ya GG Water beach lake river-tce swamp Swamp # JP14 # JP # PM33 PM34# # JP13 (O) # # NR4C GD17 # WM4# NR9C # WM9 # WM8 # # L35C TPL1C # MS6 # # # # JB1 # L26C WM16 L2C # WM18 WM4 # # # L13C # WM23 L17C # WF9 WM33 # # JB6 # PB2 # WF1 L1C # WM28 L8C ## L9C # WF7 # # WH31 (O) MM9 # # # # # WF12 # MM MM16 MM6 GD12 MM14 # # GN1 MM Figure 1: Site map of the Gnangara Mound showing monitoring wells installed with water level recorders Landuse and Recharge The three largest water users (Table 1) in the Gnangara Mound are (Figure 2): pine forest plantations (23,8 ha), modified remnant native vegetation (86,1 ha) and market gardens (1,9 ha). All three are major water users and they either decrease the recharge to the aquifer through interception losses (forests and reserves ) or take water from the aquifer through evapotranspiration (pines and market gardens). Other minor land uses are remnant vegetation (8 ha), drainage lines (2,27 ha) and water bodies and swamps (1,26 ha). 2

3 Table 1: Daily water use by the different land uses in the Gnanagara Mound Vegetation Area (m 2 ) Rate of Water Use (mm/day) Daily Water Use (m 3 /day) Modified Remnant 861E E+5 Remnant Vegetation 8 E E+3 Pines 238 E E+5 Urban 31 E E+4 Market Garden 19 E E+4 Other 135 E+6.. E+ Daily total water use 1.17 E+6 Total Annual Water Use 4. E Landuse 647 Landuse Urban Market Garden Modified Remnant Vegetation Remnant Vegetation Pines Water Drainage Lines Unknown Figure 2: Site map of the Gnangara Mound showing landuse 3

4 Land use influences groundwater recharge to the Gnangara Mound and affects the natural water balance. Clearing deep-rooted native vegetation for agriculture decreases summer transpiration and interception water loss, thereby increasing groundwater recharge. Similarly, much of the urban storm water drainage is disposed of in soak pits near residential properties and to wetlands from the storm drains enhancing recharge. On the other hand, pine plantations decrease the recharge rate because of evapotranspiration and interception losses from the forest canopy and litter and water uptake from the soil water where the groundwater is deep, and directly from the aquifer system where the water levels are high. In a previous study, Carbon et al. (1982) showed that pine plantations transpired more water than the native forest they replaced. The pines depleted the soil water faster and to a greater degree. Deep drainage below 6 m was much less in the unsaturated zone below the plantations, indicating a decrease in recharge to the shallow aquifers. Estimates of recharge under pines vary from zero beneath dense mature pines to about 3 % of precipitation beneath eight-year plantations (Sharma et al. 1983). Estimates of recharge under Banksia differ greatly depending on the methods used. Farrington & Bartle (1989) estimated the rates of recharge to be about 2 % of annual rainfall using the water balance method. These values are much lower than the 34 % estimated by Carbon et al. (1982) and 28 3 % by Sharma et al. (1983). Average rainfall in the Gn angara Mound ranges between 7 and 8 mm and interception loss ranges between and 2 % (Farrington & Bartle 1989). Several studies provided rainfall recharge estimates using different methods. Bestow (1971) and Allen (1976) used water balance methods and obtained recharge estimates between 7.3 and 8.5 % of rainfall. Sharma & Pionke (1984) estimated that recharge to the Superficial aquifer is 12 % of rainfall over native bushlands, using bromide and water balance methods. Based on a flow net analysis, Davidson (1987) estimated that 14 % of rainfall recharges the aquifer, which Davidson (1995) later revised to 11 % of the rainfall in north Gnangara and % in south Gnangara using a modified flow net. Thorpe (1989) found that 21 % of rainfall recharges the crest of the Mound, using naturally occurring tritium to estimate recharge. Farrington & Bartle (1989) obtained recharge estimates between 2 to 22 % for Banksia woodland area by the chloride balance method. Sharma et al. (1991) showed that approximately 4 % of the rainfall over market gardens recharges the aquifer. In this sandy coastal aquifer, it is expected that recharge rates vary widely especially as in most of those areas there is no surface runoff and most of the rainfall is lost by evapotranspiration and the remaining part infiltrates as recharge. Recharge is controlled by the accumulation of water in depressions and water is lost by evapotranspiration. Historical Water Level Trends and Water Use Historical water level trends in the Gnangara Mound indicate that groundwater inflow and outflow to the aquifer reached a steady state with water levels rising during the wet winter months and falling during the dry summer months. Although the water level rises seasonally, depending on the rainfall rates, the water level recession also varied, resulting in virtually no net accession to aquifer storage. This is mainly due to the high transmissivity of the aquifer, which ranges up to a maximum of 1, m 2 d 1 along the coast where most of the groundwater discharge occurs. This equilibrium in the hydrologic budget was achieved through a balance between inputs from various components of recharge and outputs from evapotranspiration from the native vegetation and groundwater discharge to the rivers and the Indian Ocean. Two events of the 197s have severely disturbed this balance: the introduction of the pine plantations, which have high evapotranspiration rates, and water abstraction from the aquifer primarily for horticultural activity and later for the metropolitan water supply. Market gardens are the largest private user of groundwater from the Gnangara Mound. The irrigation requirements for crops are estimated at, to 18, m 3 /ha/yr. All irrigation supplies required by market gardens are abstracted from groundwater wells. Similarly, groundwater abstraction from the shallow aquifers is occurring from more than 1, private wells on private residential properties. Currently, Perth obtains more than one-third of its water supply as groundwater from the Gnangara Mound. Within the urban area, the total groundwater abstraction from the Superficial aquifer had been estimated to be equivalent to about 2 % of rainfall over the area (Davidson 1995). Several detailed groundwater studies have been conducted on the Gnangara Mound, starting with the "Perth Urban Water Balance Study" by Cargeeg et al. (1987) and more recently, the "Hydrogeology and Groundwater Resources of the Perth Region" by Davidson (1995). These studies tried to predict water resources of the Gnangara Mound based on water balance principles. The aim of this study is to investigate the systematic drawdown pattern in the areas adjacent to the well fields in the Gnangara Mound and to elucidate the reasons for the decline in water levels. DATA SOURCES AND METHODS One of the objectives of this study was to examine the water level patterns under different land use and management scenarios for the major soil types (Bassendean, Spearwood and Quindalup Sand Systems) in the southern Gnangara Mound. Fifty sites were selected in the Gnangara Mound for water level monitoring, soil and water sampling, and estimation of soil hydraulic parameter. The 5 sites were also instrumented with automatic water level recorders to monitor groundwater fluctuations under different management scenarios for the major soil types. Estimates of recharge were based on hourly measurements of water level fluctuations from 5 wells. Water level fluctuations were studied at two temporal scales: (a) short term (diurnal fluctuations); and (b) intermediate (seasonal fluctuations). Short term monitoring was conducted to determine the mechanical nature of the material overlying the aquifer and the extent of connectivity between the aquifer, the atmosphere and the degree of confinement of the aquifers. Seasonal monitoring was conducted to obtain data for estimating recharge and discharge rates for the different aquifers. To study the long-term water level patterns in the unconfined and confined aquifers, water level records from monitoring wells were obtained from the Water and River Commission database (SWRIS). 4

5 Water samples were collected from 85 production wells in six main well fields and analysed for the major cations (Na, K, Ca and Mg) and anions (Cl, HCO 3, SO 4 and NO 3 ), and for the stable isotopes deuterium and oxygen 18 to investigate the relationship between the Superficial, Leederville and Yarragadee aquifers in the Gnangara Mound. Chemical characterisation of waters in these aquifers also provided insight into the hydraulic connectivity between the aquifers caused by pumping. RESULTS Groundwater levels in the Gnangara Mound are a smoothed replica of the topography of the area (Figure 3). The water level forms a mound at the highest part of the area which discharge eastward and southward into the Swan River, northward toward Gingin and westward to the ocean. At low points in the landscape, the water table frequently intersects the land surface to form lakes and swamps. Under natural conditions, the major outflows from the aquifers are due to evaporation from open water surfaces and transpiration by native vegetation and pine plantations. Other losses occur through the groundwater flow to the ocean and rivers and recharge to the deeper aquifers at some localised points. The rate of movement of groundwater flow from north to south on the Gnangara Mound is about 3 m d 1 beneath the Bassendean Dunes and about 9 m d 1 east to west beneath the coastal strip. Based on the estimated rate of groundwater movement and longest flow direction north to south on the Gnangara Mound, the oldest groundwater in the aquifer is about 1,5 years old (Davidson 1995). Figure 3: Water level map of the Gnangara Mound showing the well fields, land uses and pathlines of flow Depth to Water Depth to groundwater ranges from 2 m at the Mound crest and decreases steadily moving eastwards towards the Swan Vally and westward towards the coast (Figure 4). Groundwater level is at the surface or in contact with surface water in several discharge areas to the east. The water table also intersects several lakes, which are at the boundary between the Bassendean and Spearwood Dune Systems. The depth to water increases again in the Quindalup Dune System due to the variable height of the dunes and can vary between 5 m, but near the coast the depth to groundwater varies from 1 3 m. Seasonal Water Level Trends in the Unconfined Aquifer Water levels in the Bassendean unit show very rapid responses to rainfall, and rising patterns (Figure 5). The water level rises slowly during the early part of the wet season and rapidly in August, and reaches a peak in October/November. The slow rise in the early part of winter is related to soil water deficit in the unsaturated zone and vegetation water use. When the water deficit is replenished and vegetation water use decreases in the winter months, the water levels start rising sharply due to the high conductivity of the sands. The annual water level rise is usually between.45 to 5.2 m (Table 2). 5

6 The water level patterns in the Spearwood unit (Figure 6) exhibited very high variability in depth to water and response to rainfall due to the variable depth to limestone. The effect of land management was more pronounced in this unit than in the Quindalup. The annual water level rise is usually between.1 and.75 m. The pattern of water level rise in the Spearwood unit is different from the Bassendean. The water level rose slowly throughout the rainy season without the characteristic sharp rise noticed in the Bassendean. In addition, the water level peak was not as prominent or as prolonged. The water level pattern in the coastal area was a subdued replica of the Spearwood dune topography. Water levels under the Quindalup unit (Figure 7) in coastal areas are generally deep due to the high sand dunes. The response to rainfall is minimal due to the high transmissivity of the limestone, and water level fluctuations were similar to the Spearwood with maximum rise of.3 m. The water level pattern for the eastern area (Figure 8), which is discharging into the Swan valley, showed a very rapid response to rainfall that is attributed to the shallow depth to water and the lower storage capacity (specific yield). Hydrographs from the area were also characterised by sharp recessions. The annual water level rise is usually between.4 to 1.7 m. Figure 4: Depth to water map of the Gnangara Mound 6

7 Water levels in the cleared and urban areas of the Bassendean Sands increased gradually at the start of winter (June and July), and then rose sharply from about the middle to the end of August, eventually reaching a maximum level by the end of September or early October. The peak continued to the end of October, when water levels slowly receded. The rate of rise varied from.6 to 5. m (Table 2), mainly because of the high conductivity of the Bassendean Sands and the additional recharge in the urban areas. The hydrographs under Banksia and other native vegetation follow the same trend, but the amount of water level rise was usually between m. Similar water level patterns were observed under the pines plantations. In some pine areas, water levels increased by.45.9 m, while elsewhere water levels increased by.6.8 m. The water level pattern for the eastern area (Figure 8), which discharges in to the Swan valley, showed a very rapid response to rainfall, attributed to the shallow depth to water. The area was also characterised by sharp recessions after major rainfall events. Table 2: Seasonal water level fluctuations and recharge in selected monitoring wells in the Gnangara Mound for 1998 to 1999 Site Total rise (m) Total recession (m) Recharge Soils Landuse (mm/yr) l9c Bass Bb mm Bass Bb wm Bass Bb L13c Bass Bb pb Bass Bb nr9c Bass Bb jb1p Bass Bb gn Bass Bb wm Bass C mm Bass C jb Bass Mg nr4c Bass Mg mm Bass Mg wm Bass Mg mm Bass Mg wm Bass P l1c Bass P L2c Bass P wm Bass P/C l8c Bass U wf Qu U wf Qu U wf Qu U wf Qu U jp Sp Bb jp Sp Bb/P wm Sp Bb jp Sp Bb Am2a.75.5 Sp Bb wm Sp Mg pm Sp Mg mm Sp Mg pm Sp Mg Wh Sp Mg/Bb ms Sp Mg/Bb wm Sp D1 l35c Sw All gd Sw All mm Sw Dl gd Sw V Bass: Bassendean, Sp: Spearwood, Qu: Quindalup, Sw: Swan alluvial, U: Urban, Mg: Market Garden, P: Pines, Bb: Banksia Bush, All: Alluvial, C: cleared land, Dl: drainage line, V: Vineyard (Specific yield.3 for Qu and Bass and.2 for Sp ) 7

8 Long Term Water Level Trends in the Unconfined Aquifer Water levels in most of the monitoring wells in the central area of the Gnangara Mound have declined systematically over the last 2 years. A typical hydrograph shows seasonal variations with water levels increasing in winter and decreasing in summer. The water table usually reaches a maximum elevation in July/August and a minimum in March/April. This seasonal trend continues; however, it seems that the maximum water level achieved during the accession period is slowly declining. A typical example is the hydrograph of Station WM1. Although water levels vary seasonally, the long-term trend line shows a decline in the water level (Figure 9). The rate of decline in water levels in wells in the Gnangara Mound is variable and depends mainly on location and land use. WM /4/98 19/1/98 19/4/99 19/1/ JB /4/98 19/1/98 19/4/99 19/1/ WH31 19/4/98 19/1/98 19/4/99 19/1/ Wm18 19/4/98 19/1/98 19/4/9919/1/99 MM /4/9819/1/9819/4/9919/1/ WM4 19/4/98 19/1/98 19/4/99 19/1/ PM /4/9819/1/9819/4/99 19/1/ MM /4/98 19/1/98 19/4/9919/1/99 WM /4/98 19/1/98 19/4/99 19/1/ MM /4/98 19/1/98 19/4/99 19/1/ MS6 19/4/98 19/1/98 19/4/99 19/1/ WM28 2/6/98 2/12/98 2/6/99 2/12/99 Figure 5: Seasonal trends of water levels in Bassendean Dune System Figure 6: Seasonal trends of water levels in Spearwood Dune System WF /4/98 19/1/98 19/4/99 19/1/99 WF /4/98 19/1/98 19/4/99 19/1/99 TPL1C 1.2 WF /4/98 19/1/98 19/4/99 19/1/99 WF /4/98 19/1/98 19/4/99 19/1/ GD12 2/5/98 2/9/98 2/1/99 NR4C 19/4/98 19/1/98 19/4/99 19/1/ L17C 21/7/98 21/1/99 21/7/99 L8C 19/4/98 19/1/98 19/4/99 19/1/99 Figure 8: Seasonal trends of water levels in the eastern part of the Gnangara Mound (Swan Valley) /4/98 19/1/98 19/4/99 19/1/99 Figure 7: Seasonal trends of water levels in Quindalup Dune System 8

9 Level (mahd) /6/74 2/3/79 2/12/83 2/9/88 2/6/93 2/3/98 Figure 9: Water level trend in WM1 Monitoring bores not affected by pumping from production bores are generally located in Banksia woodlands. Monitoring bores in Banksia woodlands located mainly in the eastern part of the study area also indicate the highest annual net gain in water levels (rising limb of the hydrograph) (Figure 1). On the other hand, water levels under Banksia affected by production (Figure 11) indicate that rates of rising water levels during the accession period are either greater than or equal to rates of falling water levels during recession. Water levels monitored in pine plantations that are affected by production generally show a falling trend in water levels (Figure 12), in comparison to water levels in areas not affected by production (Figure 13). In pine forests located far away from production bores, water levels have the largest range of annual variation in rate of rise and fall compared to the other land use areas. Since groundwater withdrawal from production bores does not occur in these areas, the dramatic changes in water levels observed in the monitoring bores in pine forests are most likely caused by water use by the pines. In comparison, monitoring bores in cleared areas have the lowest change in annual water level..2. Rise v Fall y = 1.763x - 6E-5 R 2 =.9388 Fall Rise Figure 1: Rise versus fall in water levels in monitoring bores in Banksia not affected by production bores Fall Rise v Fall y = 1.34x +.1 R 2 = Rise Figure 11: Rise versus fall in water levels in monitoring bores in Banksia affected by production bores 9

10 .2 Rise v Fall y = 1.613x + 5E-5 R 2 =.913. Fall Rise Figure 12: Rise versus fall in water levels in monitoring bores in Pines affected by production bores. Rise v Fall y =.8959x +.2 R 2 =.9383 Fall Rise Figure 13: Rise versus fall in water levels in monitoring bores in Pines not affected by production bores In summary, water levels in Banksia not affected by production have the lowest variability (Figure 14). In contrast, annual water level trends under Banksia affected by production have the highest variability (Figure ). Overall, pine plantations, regardless of production, tend to impact water levels the most, but especially within production areas (Figures 16 and 17). W/L (mahd) /6/68 2/12/73 /5/79 14/11/84 7/5/9 28/1/95 19/4/1 1/1/6 Figure 14: Water levels of monitoring bores in Banksia not affected by production W/L (mahd) /1/71 27/9/73 23/6/76 2/3/79 14/12/81 9/9/84 6/6/87 Figure : Water levels of monitoring bores in Banksia affected by production 1

11 W/L (mahd) /6/68 2/12/73 /5/79 14/11/84 7/5/9 28/1/95 19/4/1 1/1/6 Figure 16: Water levels of monitoring bores in Pines not affected by production W/L (mahd) /1/72 27/9/74 23/6/77 19/3/8 14/12/82 9/9/85 5/6/88 2/3/91 Figure17: Water levels of monitoring bores in Pines affected by production Regional Long Term Water Level Trends in the Unconfined Aquifer Water levels started falling in the unconfined aquifers of the Gnangara Mound since 1975, and by 198, water levels dropped by 2 m in the central area (Figure 18). In 1985, water levels in the same area dropped by another 1 m. Additional drawdown of 3 m occurred in the shallow aquifer in the central area by 199, followed by another metre by 1995 and again by The water level map for 1998 shows the central area dropped by 6 m in the northern area, 2-3 m in the southern area and 1 2 m in the eastern and western parts of the study area. The overall drop in water levels since 1975 at the crest of the Mound is about 5 m (Figure 19). Water Level Trends in the Confined Aquifers Water levels are declining in the Leederville (Figure 2) and the Yarragadee (Figure 21) aquifers at much higher rates than in the Superficial aquifer. Water level declines are mainly due to abstraction. Water level declines in the Leederville ranges from 5 m to m, while declines in the Yarragadee range from 5 m to m. Recent analysis of water levels showed that in some areas the rate of drawdown since 1994 nearly doubled. Estimates of recharge One of the difficulties of estimating recharge using water level fluctuations is the estimation of specific yield (Sy). This is mainly because the estimate of Sy requires extensive and long-term aquifer tests. Fortunately, due to the importance of the Gnangara Mound as a water source for Perth, extensive hydrogeological studies had been conducted and several measurements of specific yield are available in the literature and the open file reports of the Water and Rivers Commission. Davidson (1995) assigned specific yield values of.3 to Tamala Limestone,.2 for the Bassendean and Gnangara Sands and.5 for the Guildford Clay. From a detailed soil properties analysis carried out during a parallel study (Salama et al. 21), the average porosity of the Bassendean Sands was estimated to range from. to.35, the specific retention from.2.7, while the specific yield was estimated to be.2.3. From the delayed response of the water level to rainfall events, it has also been estimated that the wetting front needs from 7 to 11 days to reach the water table. According to short-term (seasonal) water level fluctuations for the season, recharge in the Bassendean Sands under Banksia (Table 2) ranged between 11 to 22 mm yr 1, whereas under pines recharge ranged from 9 to 18 mm yr 1 and under urban areas it reached 1, mm yr 1. For the Spearwood sands, recharge under Banksia ranged from 1 to mm yr 1. Net recharge under the market gardens ranged from 16 to 138 mm yr 1, which is much lower than previous estimates of 3 mm yr 1 by Sharma et al. (1991). Recharge in the urban areas of the Quindalup varied from 75 to 9 mm yr 1. The much lower than expected recharge for urban areas is mainly due to the use of the same specific yield value as the Spearwood. Due to the very high transmissivity and consequently high specific yield of the Quindalup, recharge under urban areas could possibly be more than 1, mm yr 1. 11

12 Waterlevel Change from 1975 to 198 Waterlevel Change from 198 to Waterlevel Change from 1985 to 199 Waterlevel Change from 199 to Waterlevel Change from 1995 to 1998 Waterlevel Change from 1975 to 1998 Figure 18: Fall in water levels in the Gnangara Mound for the years 198, 1985, 199, 1995, 1998 and for Figure 19: Map of study section indicating areas of watertable fall,

13 Hydraulic Connectivity between the Superficial Aquifer and the Deep Confined Aquifers Geochemical and isotopic studies were critical in defining the hydraulic connectivity and degree of mixing of water between the different aquifers. Isotope results were compared with the Perth meteoric water line given by δd = * δ 18 O Groundwater recharge to the Leederville aquifer takes place where the confining Osborne Formation shale is absent, and the Leederville is in direct hydraulic connection with the Superficial formations and where the downward hydraulic gradient is significant (Davidson 1995). Similarly, groundwater recharge to the Yarragadee aquifer is by downward leakage of groundwater from the Leederville aquifer where the South Perth Shale is absent and downward hydraulic head prevails (Davidson 1995). The Superficial aquifer of Gwelup well field (Figure 22) contains mainly NaCl NaHCO 3 -type water, while the Leederville aquifer contains mainly NaHCO 3 -type water. Groundwater in the shallow aquifer contains less total dissolved solids than the Leederville aquifer. The isotopic composition of the groundwater from the Gwelup well field is similar to the mean composition of rainwater (Figure 23). Groundwater in the Leederville aquifer is more depleted in oxygen 18 and deuterium, but in the shallow aquifer, the groundwater is more enriched and quite distinctive according to the isotopic composition diagram (Figure 23). These results indicate that groundwater in the Leederville water was recharged during different climatic conditions, possibly cooler times. Water quality in the Wanneroo well field (Figure 24) is similar to the Gwelup well field with the Superficial aquifer containing less soluble salts than the Leederville aquifer. The isotopic composition of groundwater from the three aquifers occupy three distinct areas along the Perth meteoric water line. The Yarragadee groundwater is the most depleted in deuterium and oxygen 18, but the Leederville contains groundwater that is slightly more enriched. The isotopic composition of water from the shallow aquifer most closely resembles rainwater compared to the other aquifers. Two wells that are screened in the Leederville aquifer (P and P65) plot in the Superficial aquifer zone, indicating that these two production wells are pumping from the shallow aquifer. Groundwater from the Lexia well field (Figure ) falls above the isotopic composition given by the Perth meteoric water line. This indicates a slightly different type of water compared to both the shallow and Leederville waters in Gwelup and Wanneroo well fields. Similar results are also observed in the Mirrabooka well field (Figure 26) where most of the shallow groundwater plots above the meteoric water line. In this case, although the Leederville groundwater also plots above the line, they are also less depleted than the shallow aquifer. All samples from the shallow groundwater in Neerabup well field (Figure 27) plot below the meteoric water line with several areas plotting along an evaporation trend. This indicates that the recharging water was subjected to some degree of fractionation before infiltration. Groundwater from the Leederville aquifer in this area plots above the meteoric water line, with two wells showing mixing with the shallow aquifer. The stable isotope results for shallow groundwater from all of the well fields (Figure 28) indicate a regional trend given by the line: δd = * δ 18 O The isotopic trend for shallow groundwater from Neerbaup well field indicates a meteoric source and enrichment due to evaporation. The isotopic signature of groundwater from the Leederville and Yarragadee aquifers follows a different line (δd = * δ 18 O ), which is above and parallel to the meteoric water line. DISCUSSION Effects of Declining Rainfall Prior to major abstraction from the Gnangara Mound, the aquifer had nearly reached steady state. Water levels varied seasonally, increasing during winter and falling during summer with no net change in water level. Groundwater recharge was essentially equivalent to groundwater discharge. Water levels in several areas in the Gnangara Mound started to decrease after Although falling water levels coincided with a decline in rainfall, there are several points to consider. First, rainfall has not been declining at a constant rate, yet the water levels have been consistently falling. Second, the decrease in the water table could be produced by a combination of other factors (i.e. the start of abstraction from private wells, the development of the market gardens and the pine plantations). The poor correlation between the long-term rainfall trend and the rise and fall in water levels in most of the wells in the Gnangara Mound suggests that the declining water levels in the Mound are mainly due to the cumulative effect of discharge from the aquifer system and, to a smaller degree, to the decline in rainfall (Green et al. 1997). Effects of Abstraction The semi steady state of the aquifer prior to 1975 indicates that the recharge and discharge (outflow + evapotranspiration) components of the aquifer were balanced. Prior to the establishment of agricultural areas and the development of the Urban Water Scheme, water level patterns reached a semi steady state, indicating recharge is approximately equal to discharge. Estimates of recharge to the Superficial aquifer are between 1 % of rainfall; assuming the semi equilibrium status of the aquifer, then groundwater discharge is occurring roughly at the same rate. If water levels after the recharge period during the winter do not rebound to their original level, then discharge exceeds recharge. Combined outflow (discharge) from the Gnangara Mound has been steadily increasing since The decline in the water level is an indication that the abstraction is withdrawing water from storage and that the natural and artificial discharge is exceeding the recharge even allowing for increased recharge in urban areas. 13

14 The fact that the water levels are declining in the central part of the study area (Figure 19) and not in the western or eastern areas where no pumping is taking place, is an indication that the drawdown in the aquifer is mainly caused by pumping and water uptake by pines. There are many production bores in the Leederville aquifer within the central Perth area. Since abstraction is currently exceeding groundwater through-flow and recharge, the potentiometric heads within the aquifer are gradually declining (Davidson 1995). Recent records show that this rate of decline is doubling. AM18 (Leederville) AM19 (Leederville) /8/76 18/2/82 11/8/87 31/1/93 24/7/ /5/79 14/11/84 7/5/9 28/1/95 AM27A (Leederville) AM28 (Leederville) /3/71 28/8/76 18/2/82 11/8/87 31/1/93 24/7/ /1/8 29/3/86 19/9/91 11/3/97 AM3A (Leederville) AM34(Leederville) /3/71 28/8/76 18/2/82 11/8/87 31/1/93 24/7/98 level(mahd) /1/78 18/2/82 29/3/86 7/5/9 /6/94 24/7/98 AM35 (Leederville) AM36A(Leederville) /1/8 29/3/86 19/9/91 11/3/ /3/71 28/8/76 18/2/82 11/8/87 31/1/93 24/7/98 Figure 2: Leederville Water Levels 14

15 AM27 (Yarragadee) AM29 (Yarragadee) /3/71 /5/79 11/8/87 28/1/ /2/82 11/8/87 31/1/93 24/7/98 AM3 (Yarragadee) AM32(Yarragadee) /3/71 28/8/76 18/2/82 11/8/87 31/1/93 24/7/98 level(mahd) /1/78 18/2/82 29/3/86 7/5/9 /6/94 24/7/98 AM33A (Yarragadee) AM34A (Yaragadee) level(mahd) /2/82 11/8/87 31/1/93 24/7/98 level(mahd) /2/82 29/3/86 7/5/9 /6/94 24/7/98 AM35A (Yarragadee) AM36 (Yarragadee) /2/82 11/8/87 31/1/93 24/7/98 level(mahd) /2/82 29/3/86 7/5/9 /6/94 24/7/98 Figure 21: Yaragadee Water Levels

16 Figure 22: Gnangara Mound Well Fields Perth Meteoric Water Line δ 2 H = 7.3 δ 18 O Gwelup Borefield Shallow Leederville -5-1 δ 2 H G12 G11 G13 G2 G16 G12 - G65 G G1 G1 G135-2 δ 18 O - Figure 23: Gwelup Well Field 16

17 δ 2 H Perth Meteoric Water Line δ 2 H = 7.3 δ 18 O Wanneroo Borefield Shallow Leederville Yarragadee -5-1 W7 W26 W3 P3 W3 W1 P4 W35 P14 W5 W85 W55 P17 P97W7 W W57 W12 P P δ 18 O Figure 24: Wanneroo Well Field Perth Meteoric Water Line δ 2 H = 7.3 δ 18 O Lexia Borefield Shallow Mirrabooka -5 δ 2 H -1 L41 L42 L31 L12 - L7-2 - δ 18 O Figure : Lexia Well Field 17

18 Perth Meteoric Water Line δ 2 H = 7.3 δ 18 O Mirrabooka Borefield Shallow Leederville Mirrabooka -5 δ 2 H -1 M1 M5 M41 M11 M31 M13M21 M33 M3 M182 M23 M M M35 δ 18 O - Figure 26: Mirrabooka Well Field Perth Meteoric Water Line δ 2 H = 7.3 δ 18 O WT1 WT4-5 Q3 Q2 Neerabup Borefield Shallow Leederville -1 δ 2 H WT8 Q4 Q18Q35 Q12Q14 WT6 Q - WT WT45 Q165-2 δ 18 O - Figure 27: Neerabup Well Field 18

19 Shallow Leederville Yarragadee Perth Meteoric Line δ 2 H = * δ 18 O δ 2 H Shallow δ 2 H = * δ 18 O (R 2 =.97671) -1 - Leederville and Yarragadee δ 2 H = * δ 18 O (R 2 =.77698) δ 18 O -2 - Figure 28: Stable isotopes for all wells The Safe Yield and Sustainable Yield Concepts: Which one is applied in the Gnangara Mound? (Bredehoeft 1997; Sophocleus 1997) Groundwater abstraction from the Superficial aquifer in the Gnangara Mound takes place from more than 12, private shallow wells (17 x 1 6 m 3 ). Groundwater abstraction by the Water Corporation and market gardens is roughly the same amount. The total abstraction from the Superficial aquifer is estimated to be 213 x 1 6 m 3 (Davidson 1995). Long-term water level monitoring in the Gnangara Mound indicates that the water levels in the areas of groundwater abstraction are continuously falling. The water levels will continue to decline as long as abstraction continues at the current rate. The ever increasing number of private bores and the increase in production by the Water Corporation and market gardens will create further water level declines in the aquifers. Current groundwater allocation practices for the Gnangara Mound are based on the water balance approach (Davidson 1995), which is based on the assumption that the effects of groundwater abstraction from the confined aquifers will eventually propagate to the water table. For example, groundwater abstraction from the Yarragadee aquifer will induce downward leakage from the Leederville aquifer which, in turn, will induce downward leakage from the Superficial aquifer, resulting in a slight lowering of the water table. It was also suggested (Davidson 1995) that to minimise water table drawdown and to spread the effects of abstraction over wide areas, groundwater should be pumped preferentially from the confined aquifers rather than the Superficial aquifer, particularly where the confined aquifers are in direct vertical hydraulic connection with the Superficial aquifer (Davidson 1995). Although this might succeed in areas where the confining layers between the aquifers are absent, the isotopic signature from the pumping well fields indicate that there are distinct water types in each of these aquifers and that although some leakage will take place due to pumping, it is not expected that this leakage will compensate for the amount of withdrawal. In addition, pumping from the confined aquifers in areas where the confining layer is absent will cause more drawdown in the shallow aquifer. The high pumping rates used in these aquifers will produce significant localised drawdown around well fields. The fact remains that a new state of dynamic equilibrium in the Gnangara Mound can be achieved only by an increase of recharge, a decrease in discharge or a combination of these two factors (Theis 194; Bredehoeft et al. 1982). Logically, the most viable way for increasing recharge in the Gnangara Mound is by decreasing evapotranspiration and water uptake from the aquifer by removing the pine plantations. As most of the existing and planned pumping schemes are within the plantations, it is essential that they should be systematically removed. In this case, the amount of water utilised by the plantations will be available for recharge and subsequently for abstraction from the Superficial aquifer. Otherwise, pumped water from the Gnangara aquifers will be supplied from both storage and from reduced natural discharge and not from recharge. Although a thinning programme has been suggested for the pine forest as a way to reduce the water uptake by evapotranspiration, it is doubtful whether this will succeed. While current competition for water between the trees will be reduced, the remaining trees will have more water for extra growth and are likely to 19

20 return toward the original Leaf Area Index of the unthinned plantation and thus approach the original rates of evapotranspiration. A new state of dynamic equilibrium cannot be achieved unless the natural discharge is decreased. Groundwater discharge of the Gnangara aquifer takes place along a very long stretch of the Indian Ocean and surrounding rivers. Decreasing the discharge by extensive pumping along a line parallel to the ocean could produce massive sea water intrusion problems. However, discharge of fresh water from the aquifer can be reduced through the injection of treated wastewater along a line of injection wells parallel to the ocean. Simultaneously, abstraction can take place by another parallel line of wells further inland. If these two suggestions are not followed, the aquifer will continue to be depleted and, consequently, water levels will continue to decline. It is strongly recommended that Government issue licenses and fees for all private wells and a monitoring system to be established in the urban areas, especially in the coastal suburbs, to manage the risk of sea water intrusion. Stokes et al. (1996) predict that between now and 21, groundwater will comprise approximately 85 % of proposed additional water source development for the Perth Metropolitan Area. By 21, groundwater will contribute about 5 % of all water supplied to the system (Stokes et al. 1996). Our results indicate that although the aquifer storage would be capable of sustaining the water requirements for a long time, the water levels in the aquifer will continue to decline at a greater rate and some private wells will not be operational due to their shallow depth. At the same time most of the wetlands in the region will be affected. CONCLUSIONS Preliminary analysis of seasonal water level trends indicated that geomorphology and geology control the water level patterns. Each aquifer has a unique water level pattern. Landuse is partly responsible for water level patterns and recharge; however, the hydraulic properties control aquifer response to a greater degree. Analysis of short and long-term water level data from the southern part of the Gnangara Mound reveals that the water levels prior to 1975 were in a semi steady state. After 1975, a combination of excessive water use by pine plantations, heavy pumping from private boreholes in market gardens, pumping from the shallow wells in private homes and intensive pumping in the Gnangara Mound for the Metropolitan water supply, caused water levels to decline at a constant rate, over and above the decline expected with reduced rainfall since that time. This is an indication that groundwater loss and discharge is taking place from aquifer storage. The decline of water levels is likely to continue unless measures are taken to increase the recharge (removal of pine plantations), reduce the outflow component and prevent seawater intrusion problems through the injection of wastewater in a line of wells parallel to the ocean, and relocate high volume abstraction to the discharge margins of the Gnangara Mound. The geochemistry and stable isotope signatures of groundwater sampled from three different aquifer systems revealed distinct water types that suggest very little hydraulic connection or mixing of waters between aquifers. Recharge to the shallow aquifer is decreasing due to the decrease in rainfall and the maturing of the pine plantations and recharge to the confined aquifers is negligible and will also decrease; therefore, declining water levels will continue unless strict measures are taken to increase recharge or reduce discharge. At the same time, it is important that the environmental criteria, which control sustainable yield of the aquifers, be sensibly based. ACKNOWLEDGEMENTS The authors would like to thank ACIAR (Project No 9454: Agrochemical Pollution of Water Resources under tropical intensive Agriculture Systems) for financial support; the Water and Rivers Commission and the Water Corporation of Western Australia for data and support. REFERENCES Allen AD (1976) Outline of the hydrogeology of the Superficial formations of the Swan Coastal Plain. Western Australia Geological Survey, Annual Report 1975: Bestow TT (1971) The water balance in the north Gnangara area. Western Australia Geological Survey, Annual Report 197: Bredehoeft J (1997) Safe yield and the water budget myth. Groundwater 35(6):929. Bredehoeft JD, Papadopulos SS, Cooper H (1982) Groundwater, The Budget Myth. Scientific Basis of Water-Resource Management. National Academy Press, Washington DC. Carbon BA, Roberts FJ, Farrington P, Beresford JD (1982) Deep drainage and water use of forests and pastures grown on deep sands in a Mediterranean environment. Journal of Hydrology 55: Cargeeg GC, Boughton GN, Townley LR, Smith GR, Appleyard SJ, Smith RA (1987) Perth Urban Water Balance Study Vol 1, Findings. Water Authority of Western Australia, Perth. Davidson WA (1987) Hydrogeology of the Lexia area Perth, Western Australia: Western Australia Geological Survey, Hydrogeology Report No 2796 (unpublished). Davidson WA (1995) Hydrogeology and groundwater resources of the Perth region, Western Australia. Geological Survey of Western Australia Bulletin 142. Farrington P, Bartle GA (1989) Water and chloride balance of Banksia woodlands on coastal deep sands of South Western Australia. In: Sharma ML (ed) Groundwater recharge. Balkema, Rotterdam, Green TR, Charles SP, Bates BC, Fleming PM (1997) Effects of simulated climate change on groundwater recharge in Western Australia. 24 th Hydrol. and Water Resour. Symp., Auckland, Ne w Zealand, November 1997,

21 McArthur WM, Bettenay E (196) The development and distribution of the soils of the Swan Coastal Plain, Western Australia. Soil Publication No 16, CSIRO, Melbourne. Salama RB, Pollock DW, Byrne JD, Bartle, GA (21) Geomorphology, soils and landuse in the Swan Coastal Plain in relation to contaminant leaching. In: Agrochemical Pollution of Water Resources, Proceedings of a Conference held on February 2, Hat Yai, Thailand, ACIAR Proceedings No 14: Sharma ML, Byrne JD, Herne, DE, and Kin PG (1991) Impact of horticulture on water and nutrients fluxes to a sandy aquifer: Australia. CSIRO Division of Water Resources, Report 91/33, December Sharma ML, Farrington P, Fernie M (1983) Localised groundwater recharge on the "Gnangara Mound", Western Australia. In: Papers of the International Conference on Groundwater and Man, Volume 1: The investigation and assessment of Groundwater Resources. Australian Water Resources Council, Conference Series No 8, Vol 1: Sharma ML, Pionke HB (1984) Estimating groundwater recharge from measurements of environmental tracers in the vadose zone. In: Nielsen, DM (ed) NWWA/US EPA Conference on characterisation and monitoring of the vadose (unsaturated) zone, December 1983, Las Vegas, Nevada, Proceedings, Sophocleus M (1997) Managing water resources systems: Why safe yield is not sustainable. Groundwater 35(4):561. Stokes RA, Martens AS, Ng YH (1996) The importance of groundwater to Perth s water consumer. In: Proceeding of Groundwater and Land-use Planning Conference September Perth. Centre for Groundwater Studies, Australia. SWRIS Database, Water and Rivers Commission, Western Australia Theis CV (194) The source of water derived from we lls: essential factors controlling the response of an aquifer to development, Civil Eng. 1: Thorpe PM (1989) Tritium as an indicator of groundwater recharge to the Gnangara Mound on the Swan Coastal Plain, Perth, Western Australia. In: Sharma ML (ed) Groundwater recharge. Balkema, Rotterdam, Thorpe PM, Davidson WA (1991) Groundwater age and hydrodynamics of the confined aquifers, Perth, Western Australia. In: Proceedings of the International Conference on Groundwater in Large Sedimentary Basins, Perth, Western Australia, 199: Australian Water Resources Council, Conference Series no 2, pp