Disconnection of groundwater from surface water causes a fundamental change in hydrology in a forested catchment in south-western Australia

Journal of Hydrology 472–473 (2012) 14–24

Contents lists available at SciVerse ScienceDirect

Journal of Hydrology journal homepage: www.elsevier.com/locate/jhydrol

Disconnection of groundwater from surface water causes a fundamental change in hydrology in a forested catchment in south-western Australia J. Kinal a,⇑, G.L. Stoneman b a b

Science Division, Department of Environment and Conservation, Dwellingup, WA 6213, Australia Sustainable Forest Management Division, Department of Environment and Conservation, Crawley, WA 6009, Australia

a r t i c l e

i n f o

Article history: Received 21 May 2012 Received in revised form 21 August 2012 Accepted 4 September 2012 Available online 12 September 2012 This manuscript was handled by Peter K. Kitanidis, Editor-in-Chief, with the assistance of Nunzio Romano, Associate Editor Keywords: Connectivity Amplified streamflow Stream salinity Climate change Groundwater–surface water Jarrah forest

s u m m a r y In south-western Australia, significant declines in annual rainfall in recent decades have been accompanied by even greater declines in annual streamflow. The disproportionate decline in annual streamflow is perplexing and while there has been speculation about the causes, the mechanisms responsible for the disproportionate decline have not been demonstrated. This study seeks to clarify the role of groundwater in the disproportionate decline in annual streamflow in a small catchment in the jarrah forest by examining records of annual streamflow, stream salinity and groundwater which progressively declined from 1976 to 2011. The records span the transition from connected groundwater–surface water systems to disconnected. This provided a unique opportunity to differentiate the groundwater contribution to streamflow for two reasons. Firstly, because the change in streamflow following disconnection can be largely attributed to streamflow that was previously generated because groundwater was connected. Secondly, because groundwater was the main source of stream salinity and hence stream salinity was a natural tracer which indicated the presence and relative proportion of groundwater in streamflow. Disconnection occurred around 2001 and was signalled by a change in the annual stream salinity signature from moderately high and variable, to low and constant, and by the transition in piezometric levels at the catchment outlet from mostly above ground, to mostly below ground. Following disconnection, the average runoff coefficient which had been slowly declining, abruptly fell by more than half and subsequently remained relatively low and constant. This indicated that whilst groundwater was connected it played a key role in streamflow generation. The contribution by groundwater to streamflow generation was non-linear and was dominant at higher rainfalls. The annual stream salinity signature indicated that direct groundwater discharge to the stream was a relatively minor component, especially at higher rainfalls. Hence connected groundwater contributed to streamflow generation mostly indirectly, thus amplifying other streamflow-generating processes, by facilitating additional surface runoff and/or throughflow which were relatively fresh. As groundwater levels, and hence connectivity, declined, the amplifying effect of groundwater-facilitated streamflow generation also declined. We suggest that the large disparity between the rates of decline in annual rainfall and inflow into reservoirs elsewhere in the forests in south-western Australia mostly reflects the aggregate loss of connectivity in the catchments of streams contributing inflow into the reservoirs. Crown Copyright Ó 2012 Published by Elsevier B.V. All rights reserved.

1. Introduction Climate change has manifest as a significant decline in rainfall across much of southern Australia over the past five decades (Timbal et al., 2006; Gallant et al., 2007; Hope et al., 2009). Particularly strong rainfall declines occurred in south-western Australia from the late 1960s with further declines in the late 1990s, and in south-eastern Australia from the mid-1990s (Hope et al., 2009). The decline in rainfall has been accompanied by even greater ⇑ Corresponding author. Tel.: +61 8 9538 0028; fax: +61 3 9538 1206. E-mail address: [email protected] (J. Kinal).

declines in streamflow. For example, whereas average annual rainfall in south-western Australia declined by 10–15% since the mid 1970s (Indian Ocean Climate Initiative, 2002), average annual inflow into reservoirs declined by 70% in the same period (Water Corporation, 2008). Similarly in the Murray–Darling Basin in south-eastern Australia, average annual rainfall was about 4% lower over the period 1997–2006 compared with the long-term average whereas streamflow was 21% lower over the same period (Potter et al., 2008). Declining streamflow in south-western Australia has had major impacts on reservoirs which are at historically low levels (Petrone et al., 2010; Water Corporation, 2008). These reservoirs supply Perth and other large centres in south-western

0022-1694/$ - see front matter Crown Copyright Ó 2012 Published by Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.jhydrol.2012.09.013

J. Kinal, G.L. Stoneman / Journal of Hydrology 472–473 (2012) 14–24

Australia which together comprise over 75% of the population of Western Australia (Water Corporation, 2008). Declines in streamflow have also adversely affected agricultural production in the Murray–Darling Basin which contains about 65% of the irrigated land and nearly 40% of the value of agricultural production in Australia (Australian Bureau of Statistics, 2008). Demand for water is increasing (Bates et al., 2008), however, climate change is projected to result in further declines in average annual rainfall in southern Australia in ensuing decades (CSIRO and Bureau of Meteorology, 2007). As a result water scarcity is a major issue for many communities in these regions. The disproportionately larger decline in average annual streamflow than in average annual rainfall in south-western and southeastern Australia is perplexing. In south-eastern Australia there have been many studies investigating the causes and impacts of climate change and climate variability across the region (e.g. http://www.seaci.org/). The studies have shown that whereas rainfall is a key factor controlling streamflow (Potter et al., 2011) the observed decline in annual and/or seasonal rainfall totals have not totally explained the exaggerated decline in annual and/or seasonal streamflow totals in many areas (Kiem and Verdon-Kidd, 2010). Other factors considered to contribute to the disproportionate decline in streamflow include rising temperatures (Cai et al., 2009; Potter et al., 2011) and declining groundwater levels (Petheram et al., 2011). Petheram et al. (2011) proposed that declining groundwater levels may have contributed to streamflow decline in south-eastern Australia by reducing connectivity between groundwater and surface water. They suggested that, in predrought conditions, relatively high groundwater levels may have amplified overland flow by reducing the storage capacity of the unsaturated zone and by facilitating organised patterns of drainage and the connection of source areas of runoff as the soil wetted up during a rainfall event. Consequently a falling watertable would result in a loss of amplification of overland flow. In contrast to southeastern Australia, there has been relatively little research into the mechanisms underlying the exaggerated decline in average annual streamflow in south-western Australia. Petrone et al. (2010) speculated that a new hydrological regime may have resulted in many catchments in south-western Australia because of falling groundwater levels. Hughes et al. (2012) examined the linkages between rainfall, groundwater storage and streamflow in forested catchments in south-western Australia. They concluded that where groundwater levels were close to the soil surface in the riparian zone, streamflow was closely correlated with groundwater storage. However, whereas these researchers speculated about the causes of the disproportionate decline in annual streamflow, they have not demonstrated the mechanism. In this paper we further explore the links between groundwater levels and streamflow generation to provide additional insights into the role of groundwater in streamflow decline in the context of a drying climate. We do this by examining a hydrological data set for a small catchment in the jarrah (Eucalyptus marginata) forest in south-western Australia in which groundwater levels, streamflow and stream salinity have been progressively declining from 1976. This data set is unique because it spans the transition from connected to disconnected groundwater in a catchment where the salinity of groundwater provided a natural tracer to indicate the presence and relative proportion of groundwater in streamflow. Disconnection of groundwater from surface water provided an exceptional opportunity to better understand the role of connected groundwater in streamflow generation because the contribution of groundwater could be differentiated from other streamflow generation mechanisms by comparing the streamflow responses in the periods before and after groundwater disconnected. We hypothesised that (1) groundwater was a key factor in amplifying annual streamflow responses and, consequently,

15

(2) when groundwater disconnected from surface water, it ceased amplifying streamflow generation and thus caused a fundamental change in the annual streamflow regime.

2. Materials and methods 2.1. Geomorphology, climate and hydrology of the jarrah forest Geomorphology and climate are key determinants of the hydrological processes in the jarrah forest. The jarrah forest occurs on the Darling Plateau which forms the south-western part of the Great Plateau of Western Australia (Churchward and McArthur, 1980). The basement rock of the Darling Plateau is Archaean granite and gneiss which has weathered in situ to form a lateritic soil profile which may be as deep as 50 m (Dimmock et al., 1974). The laterite profile typically consists of a surface horizon of gravels, sands, and loams, at times including duricrust, that merges at depth into mottled and pallid clays. In the upper parts of the landscape, relatively more of the laterite profile remains intact whereas in the lower slopes and valleys, the profile has been eroded and overlaid with soils developed locally or transported from upslope (Churchward and Dimmock, 1989). The climate of the region is mediterranean with cool, wet winters and hot, dry summers. About 80% of the rainfall occurs in the 5 months from May to September (Hall et al., 1981). Annual rainfall is greatest near the western margin of the Darling Plateau and the south coast and follows a declining gradient inland (Gentilli, 1989). In the 1980s, the range in average annual rainfall was about 1300– 600 mm (Gentilli, 1989). Streamflow, groundwater flow, and salt storage are all strongly correlated with annual rainfall. To facilitate forest management, the jarrah forest was assigned to one of three zones which reflected the broad differences in hydrology and salt storage across the forest. The boundaries between the zones corresponded with isohyets based on the long-term average annual rainfall records until 1978 (Hayes and Garnaut, 1981), i.e., high (>1100 mm/y), intermediate (900–1100 mm/y), and low (<900 mm/y) rainfall zones. The subsurface hydrology of the region is characterised by two groundwater systems (Williamson et al., 1987; Schofield, 1988). The upper system is an ephemeral perched aquifer that forms in winter, typically within the sandy gravel topsoil that overlies duricrust or a clayey subsoil. The lower aquifer is permanent groundwater which located within the regolith overlying the bedrock. Shallow subsurface throughflow from the perched aquifer is considered to be the main source of streamflow (Stokes and Loh, 1982; Stokes, 1985; Turner et al., 1987; Williamson et al., 1987). Infiltration excess overland flow is considered to be a relatively small proportion of streamflow because the surface soils have high infiltration capacities which are rarely exceeded by rainfall events (Sharma et al., 1987). Permanent groundwater that is sufficiently close to the soil surface may contribute to streamflow generation both directly by discharging into the stream and indirectly by contributing to the formation of variable source areas. The proportion of groundwater discharge is relatively small. Stokes and Loh (1982) estimated that annual streamflow in 1980 in Salmon catchment in the high rainfall zone of the jarrah forest comprised about 2% overland flow, 7% groundwater, and 91% throughflow. Variable source areas are formed when perched or permanent groundwater tables intersect the ground surface. Saturation excess runoff from variable source areas is considered to be an important mechanism of streamflow generation in the jarrah forest (Stokes, 1985; Ruprecht and Schofield, 1989; Bari and Smettem, 2006). Ruprecht and Schofield (1989) found that, following clearing of Wights catchment in the high rainfall zone of the jarrah forest for agriculture, a significant

16

J. Kinal, G.L. Stoneman / Journal of Hydrology 472–473 (2012) 14–24

component of the increased streamflow was correlated with expansion of the groundwater-induced variable source area. Forest clearing resulted in increased recharge and a rise in groundwater levels. This in turn resulted in an expanding source area which led to increased throughflow, overland flow, and groundwater contributions to streamflow. However, little is known about the magnitude of the indirect contribution by groundwater to streamflow generation in relatively undisturbed jarrah forest. Over the last 3 or 4 decades, groundwater has frequently been sufficiently close to the soil surface in the high rainfall zone to contribute to streamflow generation whereas in the low rainfall zone, groundwater is too deep and streamflow is generated entirely by throughflow and overland flow. In the intermediate rainfall zone, groundwater has often been sufficiently close to the soil surface to contribute to streamflow generation. During this period the proportion of annual rainfall that discharged as streamflow ranged from about 20% in the high rainfall zone to less than 1% in the low rainfall zone (Borg et al., 1987). Most rainfall is evaporated back to the atmosphere, principally as transpiration. A lesser amount is intercepted by foliage and subsequently evaporated. Groundwater recharge is greater in lower than in upper topographic positions but overall it is a small proportion of annual rainfall. A significant proportion of winter rainfall is stored in the soil profile. The soil–water store is depleted in the dry season when the loss by transpiration exceeds the addition by rainfall. Streamflow of lower-order streams is usually intermittent, commencing sometime after winter rains have started and seldom continuing into the ensuing dry season unless groundwater discharges to the stream. Average soil salt storage increases exponentially with decreasing average annual rainfall from about 0.1 kg/m3 in the high rainfall areas to about 5 kg/m3 in the low rainfall areas (Stokes et al., 1980). Soil salt storage is highly variable locally but typically increases from hilltops to valley floors. The vertical distribution of soil salt storage typically takes the form of a monotonically increasing profile in the upper landscape and a bulge profile in the valleys and lower slopes (Johnston et al., 1980). Hence, in either case, the sandy gravel surface soils have relatively low salt storage compared with the underlying clays. The source of the salt is oceanic spray which is transferred by rainfall and dry fallout (Hingston and Gailitis, 1976). The salt is concentrated in the soil because of the high proportion of rainfall evaporated by transpiration. Stream salinity is determined by the relative proportions of groundwater, which mobilises and is the main source of salt, and throughflow and overland flow which are relatively fresh. For example, Stokes and Loh (1982) found that in Salmon catchment, about 60% of the salt was contributed by groundwater, 38%, by throughflow and 2% by overland flow. Over the past 3 or 4 decades, annual stream salinity has been highest in the intermediate rainfall zone where moderately saline groundwater discharge is partially diluted by relatively fresh throughflow and overland flow. The lowest annual stream salinity has been in the low rainfall zone where the highly saline groundwater is too deep to discharge to streams. In the high rainfall zone annual stream salinities are low because groundwater salinity is low. 2.2. Study site Yarragil 4X catchment has an area of 273 ha and is located in the intermediate rainfall zone of the jarrah forest (Fig. 1). Average annual potential evaporation (1500 mm, (Jeffrey et al., 2001) is substantially greater than average annual rainfall (Fig 4) which indicates the catchment is in a water-limited environment. The maximum relief is about 100 m and the maximum slope is less than 10°. The forest is uneven-aged and has been selectively harvested and silviculturally treated several times. Most of the catch-

ment was harvested before 1920 and then again in the 1930s. It was silviculturally treated in the late 1930s, and about 20% of the catchment was thinned in the 1950s. The most recent harvest and silvicultural treatment occurred in summer of 2000/2001 when stand density was reduced by about 27% from an average basal area of 33 m2/ha to 24 m2/ha (Kinal and Stoneman, 2011). The overstorey is dominated by jarrah and Corymbia calophylla (marri) on the middle and upper slopes, and by marri, Eucalyptus patens and Eucalyptus rudis in the streamzone. The middle storey is predominantly Banksia grandis, Allocasuarina fraseriana, Persoonia longifolia, Xanthorrhoea preissii and juvenile stages of overstorey trees. The shrub layer comprises a diverse range of predominantly perennial woody species (Bell and Heddle, 1989). 2.3. Data Streamflow, stream conductivity, groundwater levels and rainfall were monitored in Yarragil 4X from 1976 to 2011. The stream gauging station is located on a second-order stream at an elevation of 254 m ASL (Fig. 2). Stream flow rate was estimated from continuous records of stage height in a sharp-crested V-notch weir. From 1976 to 1990, grab samples of stream water were taken periodically and the conductivity measured and converted to total dissolved salts (TDS, mg/L, Borg et al., 1987). Flow-weighted annual stream salinity was estimated by summing the TDS weighted by flow rate at the time of sampling (Borg et al., 1987). Stream conductivity was measured continuously from 1991 to 2011. Rainfall was recorded near the catchment outlet by a storage rain gauge read weekly from 1976 to 1983, and by a pluviometer from 1984 to 2011. Annual rainfall for the period 1934–1975 was estimated from a regression with annual rainfall at Dwellingup, 20 km north-east, where records began in 1934 (y = 0.81x 50.47, n = 36, R2 = 0.87, p < 0.0001). Groundwater levels were monitored in the piezometers shown in Fig. 2. The method of piezometer installation was described by Herbert et al. (1978). The depth of the piezometers and the duration of records are shown in Table 1. Piezometer water levels were measured approximately monthly although there were gaps in the records for some piezometers. Groundwater samples were taken biannually in autumn and spring from 2003 and the conductivity measured and converted to TDS. 2.4. Data analysis Double-mass analysis (Searcy and Hardison, 1960) was used to evaluate the relationship between annual streamflow and annual rainfall at Yarragil 4X. A change in the direction of the double-mass curve indicates a change in the relative proportionality between the two variables, which infers that factors other than rainfall may be affecting streamflow. Where a change point was identified, a best-fit regression line or curve was fitted either side of the change point. The slope of the regression line is the ratio of average annual streamflow to average annual rainfall and hence is an estimate of the average runoff coefficient for the period defined by the line. Where the regression is curvilinear, the average runoff coefficient at a point on the curve is estimated from the slope of the tangent at the point. The time series of annual rainfall at Yarragil 4X was tested for presence of a monotonic trend by using both Mann–Kendall (non-parametric), and least-squares linear regression (parametric), to test the slope of the trend. The presence of a change point in the time series of annual rainfall was tested using a cusum non-parametric test. Analysis of covariance was used to test for significant temporal changes in the relationship between the following pairs of variables over successive time periods (treatment variable):

J. Kinal, G.L. Stoneman / Journal of Hydrology 472–473 (2012) 14–24

17

N

Fig. 1. Location of Yarragil 4X catchment in relation to native forest on publicly-owned land (shaded) and average annual rainfall isohyets based on long term records obtained prior to 1978 (Hayes and Garnaut, 1981) in south-western Australia.

1. Annual streamflow (dependent variable) and annual rainfall (covariate). 2. Annual stream salinity (dependent) and annual streamflow (covariate). 3. Annual stream salinity (dependent) and annual rainfall (covariate). Data were log-transformed to render them normal and homoscedastic and preliminary checks were conducted to ensure homogeneity of regression slopes. Data for the low or zero-flow years 2001, 2006 and 2010 were omitted from the analysis.

3. Results 3.1. Rainfall Average annual rainfall in Yarragil 4X was 897 mm for the period 1976–2011, a decline of 12% from 1017 mm, the average for the period 1934–1975 (Fig. 3). During the period 1976–2011, the 10-

year moving-average annual rainfall declined by 1% from 896 in 1985 to 892 in 2011. A Mann–Kendall test found no evidence of a significant trend during the period 1976–2011. A linear regression of annual rainfall from 1976 to 2011 had a negative slope, however, the slope was not significantly different from zero (slope = 0.58, R2 = 0.001, p = 0.83). A cusum test of annual rainfall for the period 1976–2011 found no significant change points. 3.2. Groundwater Except for seasonal fluctuations, permanent groundwater levels in the catchment have been monotonically declining since at least the mid 1970s. Piezometer 61418409 is located close to the stream about 600 m upstream from the catchment outlet. The water level in this piezometer was within a metre of the ground surface in 1976 but declined over 11 m in the ensuing 35 y at an average rate of 0.33 m/y (Fig. 5, Table 1). Water levels in six other piezometers installed in 1985 declined in the ensuing 26 y by between 6.2 and 8.9 m, at an average rate of 0.24–0.34 m/y (Table 1). The average rate of groundwater decline was marginally lower in the valleys

18

J. Kinal, G.L. Stoneman / Journal of Hydrology 472–473 (2012) 14–24

Fig. 2. Topographic map of Yarragil 4X catchment showing location of piezometers, stream gauging station and pluviometer. Hydrographs for the labelled piezometers are shown in Figs. 4 and 5.

Table 1 Change in groundwater levels from 1976 (piezometer G61418409) or 1985 (remaining piezometers) until 2011, and average groundwater salinity from six-monthly measurements from 2003 to 2011 in seven piezometers in Yarragil 4X catchment.

G61418409 G61420218 G61420219 G61420220 G61420221 G61420222 G61420225

Piezometer depth (m)

Landscape position

28.3 30.5 8.3 23.7 29.9 30 33

Valley Midslope Valley Valley Midslope Midslope Midslope

Annual rainfall (mm)

1400

Groundwater level Period (y)

Average change (m/y)

Average

Standard error

11.6 7.2 6.3 8.4 8.8 7.5 8.6

35 26 26 26 26 26 26

0.33 0.29 0.24 0.32 0.34 0.29 0.33

897 581 978 299 1154 605 290

4 5 4 17 14 9 2

1934-1975 1976-2011 10-y moving-average

1200 1000 800 600 400 200 0 1976

1980

1984

1988

1992

1996

2000

2004

Salinity (mg/L)

Decline (m)

2008

Year

Water level below ground (m)

Piezometer

0 2 4 6 8 10 12 14 16 1975

1979

1984

1989

1994

1999

2004

2009

Year Fig. 3. Annual rainfall from 1976 to 2011, estimated average annual rainfall from 1934 to 1975, average annual rainfall from 1976 to 2011, and 10-year movingaverage annual rainfall from 1976 to 2011 in Yarragil 4X.

Fig. 4. Hydrograph of valley piezometer G61418409 from 1976 to 2011.

(0.30 m/y) than in the hillslopes (0.31 m/y), however, there was no statistical relationship between rate of decline and depth to groundwater. Several other piezometers in middle and upper slope positions have dried since installation.

Valley piezometer 61420219 is situated near the stream at the catchment outlet (Fig. 2). The piezometer recorded a hydraulic head 3 m above ground when installed in 1985 but water levels have since fallen progressively. Before the mid 1990s, piezometer

19

50

-4

1976-1988

-2

1989-2000

40

0 2 4 6 1984

1988

1992

1996

2000

2004

2008

Year Fig. 5. Hydrograph of valley piezometer G61420219 from 1985 to 2011. Negative values indicate hydraulic head above ground. Gaps between lines indicate a discontinuity in the sampling record.

water levels were permanently above ground (Fig. 5). However, as water levels fell, the magnitude and duration of hydraulic head above ground each year also decreased. In early 2001, piezometer water levels fell below ground and remained below ground until late 2002. In the following 4 years, water levels were above ground for a relatively short period during the latter part of the wet season. Since 2006, water levels have remained below ground. Salinity of groundwater remained relatively constant in all piezometers throughout the period 2003–2011 (Table 1). Groundwater salinity in the piezometers ranged from about three to ten times stream salinity (Table 1, Fig. 9). 3.3. Streamflow The average and variance of annual streamflow in Yarragil 4X generally declined over the period 1976–2011 whereas the coefficient of variation remained relatively constant (Fig. 6). Ten-year moving-average annual streamflow decreased by 70% from 15.6 mm in 1985 to 4.7 mm in 2011 (Fig. 6). Average annual streamflow also declined relative to average annual rainfall. The average annual rainfall–streamflow relationship is shown in Fig. 7 approximated by a power function in each of the successive periods 1976–1988, 1989–2000, and 2002–2011. Analysis of covariance indicates a highly significant negative shift in the relationship between the successive periods, F(3, 29) = 22.78, p < 0.0001. Whereas the power curve in Fig. 7 shifted to the right between the 1970s and 1990s, the characteristic non-linear form of the curve remained the same, particularly the steeply rising gradient at relatively higher annual rainfalls. However, from around the early 2000s, the power curve shifted further to the right and also changed in form, becoming more linear and without the steeply rising gradient. 50

Streamflow (mm)

Annual streamflow

10-y moving-average

40 30 20 10 0 1976

1980

1984

1988

1992

1996

2000

2004

2008

Year Fig. 6. Annual streamflow (bars) and 10-year moving-average annual streamflow (line) from 1976 to 2011. In 2001, 2006, and 2010 streamflow was either zero or very low.

Annual streamflow (mm)

Water level below ground (m)

J. Kinal, G.L. Stoneman / Journal of Hydrology 472–473 (2012) 14–24

2002-2011

y = 4.12*10-14 x 4.89 2 R = 0.69 y = 2.21*10-16 x5.61. R2 = 0.82

30

2000

20

y = 1.57*10-12 x 4.18 R2 = 0.16

10

2002

0 400

600

800

1000

1200

Annual rainfall (mm) Fig. 7. Annual streamflow in relation to annual rainfall, fitted with power curves for three successive periods from 1976 to 2011. The power curves were determined by fitting a linear regression to log-transformed data. The zero or low-flow years 2001, 2006, and 2010 were omitted from curve fitting.

The effect, on the runoff coefficient, of the change in the average annual rainfall–streamflow relationship from a power function with a steep rising gradient to a relatively more linear function after 2000/2001 was greatest at high annual rainfalls. For example, the runoff coefficient was estimated from the curves in Fig. 7 for average annual rainfalls of 800 and 1100 mm in the periods 1989–2000 and 2002–2011. The runoff coefficient declined by 31% between the periods 1989–2000 and 2002–2011 for an average annual rainfall of 800 mm but by 56% for an average annual rainfall of 1100 mm (Table 2). The changing relationship between annual rainfall and streamflow between 1976 and 2011 is also shown in the double mass curve in Fig. 8. The best-fit regression model to the curvilinear first leg of the double mass curve from 1976 until around 2000 is a second-order polynomial equation. The coefficient of the quadratic term is negative indicating a continually decreasing slope of the polynomial curve. This reflects the progressively declining average annual streamflow relative to average annual rainfall. Sometime between 2000 and 2001, an abrupt change occurred in the double-mass curve indicating a distinct change in the proportionality between annual streamflow and annual rainfall compared with the period before 2000/2001 (Fig. 8). There was very little streamflow in 2001 hence the data point is not shown in Fig. 7. However, the data points for 2000 and 2002 are arrowed in Fig 7 to illustrate the relatively large change in the annual rainfall–streamflow relationship for the period to 2000 compared with the period from 2002. There was no corresponding change point in the time series of annual rainfall, indicating that the abrupt change in the annual rainfall–streamflow relationship was not due to a step change in annual rainfall. Several changes to the double-mass curve were evident after 2000/2001 (Fig. 8). Firstly, the slope of the curve was substantially lower. The average runoff coefficient, which is estimated by the slope of the double-mass curve, declined fairly uniformly by about 0.0037 per decade from the 1970s to the 1990s. However, within the space of 1–2 years, between the late 1990s and early 2000s, the average runoff coefficient dropped by more than half from 0.0119 to 0.0056 (Table 3). Secondly, there was less variation of data points around the curve compared to the period before 2000/2001. This corresponds with the relatively more linear

20

J. Kinal, G.L. Stoneman / Journal of Hydrology 472–473 (2012) 14–24

Table 2 Average runoff coefficient for a given annual rainfall estimated from the power curves in Fig. 7 and decline in average runoff coefficient in the successive periods. Average annual rainfall (mm)

Average runoff coefficient

800 1100

Decline in average runoff coefficient (%)

1976–1988

1989–2000

2002–2011

A

B

C

A?B

B?C

0.008 0.027

0.005 0.021

0.003 0.009

38 21

31 56

1976-1988 600

y = 0.0056x + 228.08 R2 = 0.98 300

2001 200

y = -2*10-7x 2 + 0.021x - 22.5 2 R = 0.99

100

Annual stream salinity (mg/L)

Cumulative annual streamf low (mm).

700 400

1989-2000 y = -0.74x + 1018.1 R2 = 0.54

500 400

y = -0.38x + 532.65 2 R = 0.37

300 200 100

0 0

10000

20000

0 500

30000

Cumulative annual rainf all (mm)

700 600 500 400 300 200 100 0 1976

1980

1984

1988

1992

1996

2000

2004

2008

Year Fig. 9. Time series of annual stream salinity from 1976 to 2011. There were no data available for 1990. In 2001, 2006, and 2010 annual streamflow was either zero or very low.

Table 3 Average runoff coefficient estimated at intervals along the polynomial curve and the line fitted to the double mass curve shown in Fig. 8. Year

Runoff coefficient

1979 1989 1999 2001 2009

0.0193 0.0156 0.0119 0.0056 0.0056

y = 0.07x + 46.06 R2 = 0.17 700

900

1100

1300

Annual rainfall (mm)

Fig. 8. Double mass curve of annual streamflow versus annual rainfall fitted with a second-order polynomial equation for 1976–2000 and a line for 2001–2011.

Salinity (mg/L)

2002-2011

annual rainfall–streamflow relationship after 2000/2001 and hence relatively more constant proportionality between annual streamflow and annual rainfall (Fig. 7). By definition, the data

Fig. 10. Annual stream salinity in relation to annual rainfall, fitted with linear regressions for three successive periods from 1976 to 2011. The zero or low-flow years 2001, 2006, and 2010 were omitted from curve fitting.

points will fall exactly on the double-mass curve, provided the proportionality between the two variables remains constant. In contrast, in the period before 2000/2001, the relationship between annual rainfall and annual streamflow was distinctly non-linear and was approximated well by a power function (Fig. 7). Consequently an incremental increase in annual rainfall resulted in a much greater increase in annual streamflow. Accordingly, before 2000/2001, the data points varied about the double-mass curve depending on the magnitude of the annual rainfall. Thirdly, the double-mass curve was relatively linear after 2000/2001 suggesting that annual streamflow was no longer declining relative to annual rainfall. 3.4. Stream salinity Flow-weighted annual stream salinity (hereafter called annual stream salinity), generally declined and became less variable in the period from 1976 until around the late 1990s (Fig. 9). In 1976, annual stream salinity was 620 mg/L but from around 1999 it remained relatively low and constant and, except for 2001, ranged between 91 and 131 mg/L and averaged 113 mg/L. In 2001 annual stream salinity was 41 mg/L and corresponded with an exceptionally low annual streamflow. There was no salinity record for 2010 because there was no streamflow. Average annual stream salinity also declined relative to average annual rainfall. From 1976 until the late 1990s/early 2000s, annual stream salinity was an inverse linear function of annual rainfall (Fig. 10). Hence, when annual rainfall was high, annual stream salinity was low. However, this relationship progressively weakened and since the late 1990s/early 2000s annual stream salinity remained low and relatively constant irrespective of the annual rainfall. Analysis of covariance indicates a highly significant nega-

J. Kinal, G.L. Stoneman / Journal of Hydrology 472–473 (2012) 14–24

700

1976-1988

Annual stream salinity (mg/L)

600

1989-2000 2002-2011

500

400

y = 810.48x-0.39 R2 = 0.68

300

200

y = 335.54x-0.31 2 R = 0.52

100

y = -1.16x + 121.98 R2 = 0.09

0 0

10

20

30

40

50

Annual streamflow (mm) Fig. 11. Flow-weighted annual stream salinity in relation to annual streamflow in three successive periods from 1976 to 2011. Power curves are fitted to data for 1976–1988, and 1989–2000 by linear regression of log-transformed data. A linear regression is fitted to data for 2002–2011. The zero or low-flow years 2001, 2006, and 2010 were omitted from curve fitting.

tive shift in the relationship between the successive periods 1976– 1988, 1989–2000, and 2002–2011, (F(3, 28) = 32.04, p < 0.0001). Similarly from 1976 until the late 1990s/early 2000s, annual stream salinity was a negative power function of annual streamflow (Fig. 11). Hence at high annual streamflows annual stream salinity was low but at progressively lower annual streamflows annual stream salinity increased steeply. However, the strongly nonlinear relationship gradually weakened until around the late 1990s/early 2000s when annual stream salinity remained relatively low and constant even at low annual streamflows. Analysis of covariance indicates a highly significant negative shift in the relationship between the successive periods 1976–1988, 1989– 2000, and 2002–2011, (F(3, 28) = 47.15, p < 0.0001). 4. Discussion In 1976 in Yarragil 4X catchment, groundwater was connected to surface water, and contributed to streamflow generation and stream salinity by discharging saline groundwater into the stream. This is evident from the moderately high annual stream salinity. In the ensuing years, until the late 1990s, average annual streamflow declined progressively but more rapidly than average annual rainfall. The average runoff coefficient decreased by about 0.0037 per decade. Annual stream salinity also decreased despite declining average annual rainfall and streamflow whereas, on the basis of the general form of the relationship between these variables, the opposite would have been expected (Figs 10 and 11). The key hypothesis of this paper is that whilst groundwater was connected to surface water it amplified streamflow generation processes. As groundwater levels progressively declined, connectivity diminished accordingly. This resulted in a progressively declining contribution of saline groundwater to streamflow generation. Hence, as connected groundwater declined, it mediated the decline in annual streamflow and annual stream salinity. When groundwater levels declined sufficiently to disconnect from surface water at the catchment outlet, average annual streamflow dropped abruptly in the absence of the amplifying effect of groundwater on streamflow generation, and annual stream salinity became relatively low and constant. This is further discussed below.

21

In the early 2000s, probably in 2001, groundwater disconnected from surface water at the catchment outlet. This is evident from the annual stream salinity response which changed from being highly variable and dependent on annual rainfall and streamflow before disconnection, to relatively constant and unresponsive to variation in annual rainfall or streamflow after disconnection. Following disconnection, saline groundwater no longer discharged into the stream as evidenced by the low stream salinity for the period 2001–2011. Average annual stream salinity in Yarragil 4X, after disconnection, was 113 mg/L which is consistent with stream salinity in lower rainfall areas of the jarrah forest where saline groundwater is too deep to contribute to streamflow, and salt load is derived entirely from surface runoff and the perched groundwater system (Loh et al., 1983; Williamson et al., 1987). Disconnection was also evident from the height of the water level in piezometer G61420219 relative to the ground surface. G61420219 is located close to the catchment outlet hence piezometric levels should reasonably approximate the upward hydraulic pressure of groundwater beneath the stream channel nearby. As groundwater levels declined, the transition from piezometric levels that were entirely or mostly above ground to mostly or entirely below ground occurred around 2001. Disconnection caused profound changes to the annual streamflow regime. The annual rainfall–streamflow relationship changed from non-linear with a steeply increasing gradient before disconnection, to an almost linear function with a low gradient after disconnection. This resulted in an abrupt and large decrease in the average runoff coefficient. Furthermore the average runoff coefficient, which had been declining by about 0.0037 per decade before disconnection, appeared to stop declining and remained at a relatively low and constant 0.0056 after disconnection. The dramatic changes in the annual streamflow regime, following disconnection, occurred largely because groundwater ceased contributing to streamflow generation. This indicates that groundwater played a key role in streamflow generation. The transition from connected to disconnected groundwater in Yarragil 4X provides a unique opportunity to estimate the extent to which connected groundwater augments streamflow generation. This is because the reduction in average annual streamflow following disconnection can be largely attributed to streamflow that was generated by connected groundwater (Fig. 7). The magnitude of the reduction in annual streamflow following disconnection varied according to annual rainfall. For example, for an average annual rainfall of 800 mm, average annual streamflow was about 31% lower in the 2000s, after disconnection, than in the 1990s, before disconnection, whereas for an average annual rainfall of 1100 mm, average annual streamflow was about 56% lower in the 2000s than in the 1990s (Table 2). This indicates that connected groundwater was a key factor controlling streamflow generation and, at higher annual rainfalls, dominated streamflow generation. Connected groundwater contributed to streamflow generation in Yarragil 4X both directly as groundwater discharge, and indirectly by facilitating additional throughflow or surface runoff. The relative proportions of groundwater discharge and groundwater-facilitated streamflow varied according to annual rainfall. However, at higher annual rainfalls, direct discharge was a relatively minor component of annual streamflow and it was primarily the indirect contribution by groundwater that amplified streamflow generation. Since groundwater discharge is relatively saline and facilitated streamflow is fresh, the relative proportions of groundwater discharge and facilitated streamflow are reflected in the annual stream salinity levels. Annual stream salinity was inversely related to annual streamflow (Fig. 11) whereas annual streamflow was directly related to annual rainfall (Fig. 7). This indicates that, at progressively higher annual rainfalls, a proportionately larger component of annual streamflow was relatively fresh, thus diluting the

22

J. Kinal, G.L. Stoneman / Journal of Hydrology 472–473 (2012) 14–24

saline groundwater component. However, at progressively higher annual rainfalls, annual streamflow was increasingly dominated by the indirect contribution that groundwater makes to streamflow generation (Table 2, Fig 7). This indicates that, at progressively higher annual rainfalls, a correspondingly greater proportion of relatively fresh annual streamflow was generated by groundwater. The indirect contribution by groundwater to streamflow generation by facilitating throughflow or surface runoff in relatively undisturbed jarrah forest has not previously been demonstrated. This reflects the difficulty in differentiating throughflow or surface runoff that has been facilitated by groundwater from that generated by other mechanisms. In contrast, groundwater discharge has been quantified using salt as a natural tracer. Stokes and Loh (1982) used chloride ion and a combined hydrograph-separation and mass-balance approach to estimate that annual streamflow in 1980 in Salmon catchment in the high rainfall zone of the jarrah forest consisted of about 7% of groundwater discharge. This is relatively small compared with hydrological systems globally where groundwater is typically a major component of streamflow. For example, Alley et al. (1999) suggest that the proportion of groundwater in streamflow, although highly variable, averages between 40% and 50% of streamflow in small and medium-sized streams. The dramatic changes in the streamflow and stream salinity regimes when groundwater disconnected from surface water in Yarragil 4X are similar, but occurred in reverse sequence, to the dramatic hydrological changes in Lemon catchment when the progressively rising groundwater table intersected the ground surface (Ruprecht and Schofield, 1991). Lemon is a first-order catchment in the low rainfall zone of the jarrah forest. The lower half of Lemon was cleared for agriculture to study the impact of forest clearing on stream salinity. Before deforestation, the highly saline groundwater table was 15–20 m below the stream channel. Streamflow was derived entirely from relatively fresh surface runoff and throughflow and consequently annual stream salinity was low. Following deforestation in 1976, groundwater levels rose progressively. Annual stream salinity remained low initially whilst groundwater levels were below the ground surface but began to rise in 1987 when groundwater was close to the ground surface. Around 1989, the groundwater table intersected the ground surface in the valley floor, creating a streamzone saturated area. Subsequently there was a dramatic increase in annual streamflow and annual stream salinity whereby average annual streamflow increased by 67% and annual stream salinity increased by an order of magnitude. Although there are similarities in the magnitude of the streamflow responses following connection between the groundwater and surface water systems in Lemon, and disconnection in Yarragil 4X, there are differences in the streamflow generation processes between the two catchments. In Lemon, the abrupt change in annual streamflow occurred when groundwater intersected the ground surface, creating a saturated area. Ruprecht and Schofield (1991) attributed the large increase in annual streamflow mainly to the indirect contribution by groundwater which facilitated throughflow and saturation excess runoff from the groundwater seep. The seep functioned as a variable source area, expanding as groundwater levels continued to rise, thus generating additional throughflow and saturation excess runoff. Before connection occurred, streamflow was intermittent and was generated mostly by throughflow from an ephemeral perched aquifer. Following connection, the newly emergent saline groundwater seep provided an ongoing source of baseflow during the dry summer season hence streamflow became perennial. In contrast to Lemon catchment after connection, streamflow in Yarragil 4X was intermittent before disconnection occurred. During the wet season, streamflow in Yarragil 4X comprised surface runoff, throughflow, and groundwater discharge. During the dry

summer period there was no streamflow, indicating that the water table was below ground even though the potentiometric level near the catchment outlet was above ground throughout the year until the early 1990s, or for most of the year in the mid-late 1990s (Fig. 5). It is unclear how groundwater contributed to streamflow in Yarragil 4X in this context, but the process may be by way of the capillary fringe groundwater ridging mechanism of streamflow generation (Sklash and Farvolden, 1979). According to this model, upward hydraulic pressure by saline groundwater would maintain soil water at matric potentials just below saturation in the nearstream zone. In the wet season, infiltrating rainfall would rapidly change the moisture status of the soil in the near-stream zone from unsaturated to saturated, creating a groundwater-induced saturated variable source area. The source area would discharge saline groundwater and facilitate relatively fresh throughflow and saturation excess runoff. In the dry season, the near-stream zone soil moisture would revert to an unsaturated state and there would be no groundwater discharge or baseflow. As groundwater levels in the upper parts of the catchment progressively declined from the mid 1970s, the upward hydraulic pressure near the streamzone declined, the zone of connectivity retreated downstream and the near-stream zone area that was conducive to groundwater ridging would have progressively diminished. Around 2001, the potentiometric level in the vicinity of the catchment outlet declined below the level that would sustain a capillary fringe sufficiently close to the soil surface to enable groundwater ridging, and the groundwater and surface water systems in the catchment disconnected permanently. Streamflow was subsequently derived entirely from relatively fresh surface runoff and throughflow from an ephemeral perched aquifer. Whilst groundwater was connected to surface water in Yarragil 4X it controlled stream salinity and amplified streamflow generation processes. Hence, as groundwater levels declined, groundwater mediated the declines in annual stream salinity and streamflow, relative to annual rainfall. This is apparent as the downward shift of the annual rainfall–salinity relationship in Fig 10, the shift of the annual rainfall–streamflow relationship to the right in Fig. 7, and the convex curvature of the first leg of the double mass curve of annual streamflow versus annual rainfall in Fig. 8. Following disconnection, the sudden loss of the amplifying effect of groundwater in streamflow generation resulted in an abrupt and large decline in the runoff coefficient. Subsequently, groundwater levels continued to decline but seem to have had little influence on the runoff coefficient which appears to have stabilised (second leg of the double mass curve of annual streamflow versus annual rainfall in Fig. 8). A literature search has found no comparable occurrence of a progressive shift in streamflow regime in response to declining groundwater levels or an abrupt change in hydrological regime in response to loss of connectivity between groundwater and surface water systems. This may be because of the severity of climate change in south-western Australia and the unique opportunity presented by the fortuitous monitoring of a small headwater stream for several decades and which coincidentally spanned disconnection of groundwater from surface water. The decline in average annual streamflow in Yarragil 4X since the mid 1970s was disproportionately greater than the reduction in average annual rainfall in the same period. More broadly across the jarrah forest, average annual streamflow has also declined at a disproportionately greater rate than average annual rainfall. When groundwater disconnected from surface water in Yarragil 4X, the average runoff coefficient dropped by more than half (Table 3) in the absence of the amplifying influence of the groundwater contribution on streamflow generation. The sharp drop in runoff coefficient further exaggerated the difference between the long-term rate of decline in annual streamflow and annual rainfall. It is likely that the large disparity between the rates of decline in annual

J. Kinal, G.L. Stoneman / Journal of Hydrology 472–473 (2012) 14–24

rainfall and inflow into reservoirs in the northern jarrah forest in south-western Australia (Water Corporation, 2008) has also resulted in large part from the progressive disconnection of groundwater as it retreats from higher parts of the landscape in areas of the intermediate and high rainfall zones where groundwater was previously connected to surface water. In south-eastern Australia, the much greater than expected reductions in average annual streamflow relative to the decline in average annual rainfall experienced in recent decades may be similarly caused by a combination of diminishing connectivity and hence decreasing amplification of streamflow generation by groundwater as proposed by Petheram et al. (2011), together with the progressive disconnection of groundwater from surface water which has further exaggerated streamflow decline as has occurred in Yarragil 4X. Climate change is projected to result in further declines in average annual rainfall in south-western Australia in ensuing decades (Bates et al., 2008). This is likely to result in continued declines in groundwater levels in forested areas. This in turn will result in disproportionately greater declines in streamflow in areas where further disconnection of groundwater from surface water may occur. The potential for further disconnection is greatest in the high rainfall zone of the northern part of the forest where groundwater levels have been highest in the past and hence where the areal extent of connectivity has been greatest, and in the southern part of the forest where the rate of decline in groundwater levels has been relatively slower than in the north and hence where groundwater levels remain relatively high (Conservation Commission of Western Australia, 2012). The results from this study, if built into hydrologic models, will increase confidence in simulations of streamflow responses to future climate change scenarios in forested catchments in southwestern Australia. In the past, models have not anticipated the magnitude of the amplifying effect of groundwater on streamflow generation. Model calibration in some catchments has overestimated the initial years of streamflow records and underestimated the more recent years of records because they have not provided for the substantial change in streamflow generation that accompanies disconnection of a declining groundwater table (CSIRO, 2009). Consequently past simulations may potentially have underestimated the magnitude of the decline in streamflow in response to further rainfall decline (CSIRO, 2009). Forest thinning has been proposed as a strategy to arrest and reverse declining inflows into reservoirs in catchments in south-western Australia (Water Corporation, 2005). Experimental thinning treatments of jarrah forest in the past have resulted in increased streamflow which subsequently declined to former levels as the forest regenerated (Ruprecht et al., 1991; Ruprecht and Stoneman, 1993; Stoneman, 1993). The results from this study indicate that thinning of catchments is likely to be most effective where groundwater is connected to surface water. Where groundwater is disconnected, thinning will be most effective if the treatments ensure a sufficient rise in groundwater levels to enable, and to sustain, the connection between groundwater and surface water systems.

5. Conclusion This study showed that whilst groundwater was connected to surface water it was a key factor that contributed to, and at higher annual rainfalls dominated, streamflow generation in Yarragil 4X catchment. This is evident from the abrupt and highly significant drop in runoff coefficient that coincided with disconnection. The results also indicate that connected groundwater contributed to streamflow both directly by discharging into the stream and indirectly by facilitating generation of surface runoff and/or throughflow. However, direct discharge was a relatively small

23

component of streamflow, particularly at higher annual rainfalls. Instead, the groundwater-facilitated component of streamflow generation dominated at higher annual rainfalls, thus amplifying annual streamflow. This is apparent from the annual stream salinity signature. When groundwater disconnected from surface water it ceased amplifying streamflow, consequently the average runoff coefficient dropped abruptly and steeply and remained relatively constant, thus fundamentally changing the hydrological regime. Acknowledgments We thank Lachie McCaw, Matt Williams, and three anonymous reviewers for helpful comments on the manuscript. References Alley, W.M., Reilly, T.E., Franke, O.L., 1999. U.S. Geological Survey circular 1186. U.S. Government Printing Office, Denver CO. Australian Bureau of Statistics, 2008. Water and the Murray–Darling Basin – A Statistical Profile, 2000-01 to 2005-06, Canberra, Australia. Bari, M.A., Smettem, K.R.J., 2006. A conceptual model of daily water balance following partial clearing from forest to pasture. Hydrol. Earth Sys. Sci. 10, 321– 337. Bates, B.C., Hope, P., Ryan, B., Smith, I., Charles, S., 2008. Key findings from the Indian Ocean Climate Initiative and their impact on policy development in Australia. Clim. Change 89, 339–354. Bell, D.T., Heddle, E.M., 1989. Floristic, morphologic and vegetational diversity. In: Dell, B., Havel, J.J., Malajczuk, N. (Eds.), The Jarrah Forest: A Complex Mediterranean Ecosystem. Kluwer Academic Publishers, Dordrecht, pp. 53–66. Borg, H., King, P.D., Loh, I.C., 1987. Stream and Groundwater Response to Logging and Subsequent Regeneration in the Southern Forest of Western Australia. Interim Results from Paired Catchment Studies. Water Authority of Western Australia, Report No. WH34, Perth, Australia, 163p. Cai, W., Cowan, T., Briggs, P., Raupach, M., 2009. Rising temperature depletes soil moisture and exacerbates severe drought conditions across southeast Australia. Geophys. Res. Lett. 36, L21709. http://dx.doi.org/10.1029/2009GL040334. Churchward, H.M., Dimmock, G.M., 1989. The soils and landforms of the jarrah forest. In: Dell, B., Havel, J.J., Malajczuk, N. (Eds.), The Jarrah Forest: A Complex Mediterranean Ecosystem. Kluwer Academic Publishers, Dordrecht, pp. 13–21. Churchward, H.M., McArthur, W.M., 1980. Landforms and soils of the Darling System, Western Australia. In: Department of Conservation and Environment (Ed.), Atlas of Natural Resources, Darling System, Western Australia. Department of Conservation and Environment, Perth, Australia, pp. 25–33. Conservation Commission of Western Australia, 2012. Forest Management Plan 2004–2013 End-of-term Audit of Performance Report. CSIRO, 2009. Surface Water Yields in South-west Western Australia. A Report to the Australian Government from the CSIRO South-West Western Australia Sustainable Yields Project. CSIRO Water for a Healthy Country Flagship, Australia. CSIRO and Bureau of Meteorology, 2007. Climate change in Australia. Technical Report, Canberra, Australia. Dimmock, G.M., Bettenay, E., Mulcahy, M.J., 1974. Salt content of lateritic profiles in the Darling Range, Western Australia. Aust. J. Soil Res. 12, 63–69. Gallant, A.J.E., Hennessy, K.J., Risbey, J., 2007. Trends in rainfall indices for six Australian regions: 1910–2005. Aust. Met. Mag. 56, 223–239. Gentilli, J., 1989. Climate of the jarrah forest. In: Dell, B., Havel, J.J., Malajczuk, N. (Eds.), The Jarrah Forest: A Complex Mediterranean Ecosystem. Kluwer Academic Publishers, Dordrecht, pp. 23–40. Hall, N., Wainwright, R.W., Wolf, L.J., 1981. Summary of Meteorological Data in Australia. CSIRO Division of Forest Research. Divisional Report No.6, Canberra, Australia. 117p. Hayes, R.J., Garnaut, G., 1981. Annual Rainfall Characteristics of the Darling Plateau and the Swan Coastal Plain. Public Works Department of Western Australia. Water Resources Branch, Unpublished Report No. WRB3, Perth, Australia. 23p. Herbert, E.J., Shea, S.R., Hatch, A.B., 1978. Salt Content of Lateritic Profiles in the Yarragil Catchment, Western Australia. Forests Department of Western Australia. Research Paper 32. Hingston, F.J., Gailitis, V., 1976. The geographic variation of salt precipitated over Western Australia. Aust. J. Soil Res. 14, 319–335. Hope, P., Timbal, B., Fawcett, R., 2009. Associations between rainfall variability in the southwest and southeast of Australia and their evolution through time. Int. J. Climatol. http://dx.doi.org/10.1002/joc.1964. (accessed March 2012). Hughes, J.D., Petrone, K.C., Silberstein, R.P., 2012. Drought, groundwater storage and stream flow decline. Geophys. Res. Lett. 39, L03408. http://dx.doi.org/10.1029/ 2011GL050797. Indian Ocean Climate Initiative, 2002. Climate Variability and Change in South West Western Australia. Department of Environment, Perth. Jeffrey, S.J., Carter, J.O., Moodie, K.M., Beswick, A.R., 2001. Using spatial interpolation to construct a comprehensive archive of Australian climate data. Environ. Modell. Softw. 16 (4), 309–330.

24

J. Kinal, G.L. Stoneman / Journal of Hydrology 472–473 (2012) 14–24

Johnston, C.D., McArthur, W.M., Peck, A.J., 1980. Distribution of Soluble Salts in Soils of the Manjimup Woodchip Licence Area, Western Australia. CSIRO Division of Land Resources Management. Technical Paper No. 5, Canberra, Australia, 29p. Kiem, A.S., Verdon-Kidd, D.C., 2010. Towards understanding hydroclimatic change in Victoria, Australia – preliminary insights into the ‘‘Big Dry’’. Hydrol. Earth Syst. Sci. 14, 433–445. Kinal, J., Stoneman, G.L., 2011. Hydrological impact of two intensities of timber harvest and associated silviculture in the jarrah forest in south-western Australia. J. Hydrol. 399, 108–120. Loh, I.C., Ventriss, H.B., Collins, P.D.K., 1983. Water resource quality in Western Australia. In: Water Quality – Its Significance in Western Australia. Water Research Foundation of Australia Seminar, Perth. Petheram, C., Potter, N., Vaze, J., Chiew, F., Zhang, L., 2011. Towards better understanding of changes in rainfall–runoff relationships during the recent drought in south-eastern Australia. In: Proceedings of the19th International Congress on Modelling and Simulation, Perth, Australia, 12–16 December, 2011. . Petrone, K.C., Hughes, J.D., Van Niel, T.G., Silberstein, R.P., 2010. Streamflow decline in southwestern Australia, 1950–2008. Geophys. Res. Lett. 37, L11401. http:// dx.doi.org/10.1029/2010GL043102. Potter, N.J., Chiew, F.H.S., Frost, A.J., Srikanthan, R., McMahon, T.A., Peel, M.C., Austin, J.M., 2008. Characterisation of recent rainfall and runoff in the Murray–Darling Basin. A Report to the Australian Government from the CSIRO Murray–Darling Basin Sustainable Yields Project. CSIRO, Australia, 40pp. Potter, N.J., Petheram, C., Zhang, L., 2011. Sensitivity of streamflow to rainfall and temperature in south-eastern Australia during the Millenium drought. In: Proceedings of the19th International Congress on Modelling and Simulation, Perth, Australia, 12–16 December, 2011 . Ruprecht, J.K., Schofield, N.J., 1989. Analysis of streamflow generation following deforestation in southwest Western Australia. J. Hydrol. 105, 1–17. Ruprecht, J.K., Schofield, N.J., 1991. Effects of partial deforestation on hydrology and salinity in high salt storage landscapes. I. Extensive block clearing. J. Hydrol. 129, 19–38. Ruprecht, J.K., Stoneman, G.L., 1993. Water yield issues in the jarrah forest of southwestern Australia. J. Hydrol. 150, 369–391.

Ruprecht, J.K., Schofield, N.J., Crombie, D.S., Vertessy, R.A., Stoneman, G.L., 1991. Early hydrological response to intense forest thinning in southwestern Australia. J. Hydrol. 127, 261–277. Schofield, N.J., 1988. Predicting the effects of land disturbances on stream salinity in south-west Western Australia. Aust. J. Soil Res. 26, 425–438. Searcy, J.K., Hardison, C.H.,1960. Double-Mass Curves. Geological Survey Water Supply Paper 1541-B. U.S. Government Printing Office, Washington, DC, 66p. Sharma, M.L., Baron, R.J.W., Fernie, M.S., 1987. Areal distribution of infiltration parameters and some soil physical properties in laterite catchments. J. Hydrol. 94, 109–127. Sklash, M.G., Farvolden, R.N., 1979. The role of groundwater in storm runoff. J. Hydrol. 43, 45–65. Stokes, R.A., 1985. Stream Water and Chloride Generation in a Small Forested Catchment in South Western Australia. Water Authority of Western Australia, Report No. WH7, Perth, Australia, 176p. Stokes, R.A., Loh, I.C., 1982. Streamflow and solute characteristics of a forested and deforested catchment pair in south-western Australia. In: O’Loughlin, E.M., Bren, L.J. (Eds.), Proceedings of the First national Symposium on Forest Hydrology, Melbourne, 11–13 May 1982. The Institution of Engineers Australia, National Conference Publication 82/6, Canberra, Australia, pp. 60–66. Stokes, R., Stone, K., Loh, I., 1980. Summary of Soil Salt Storage Characteristics in the Northern Darling Range. Public Works Department, Western Australia, Water Resources Technical Report No. 94. Perth, Australia, 20p. Stoneman, G.L., 1993. Hydrological response to thinning a small jarrah (Eucalyptus marginata) forest catchment. J. Hydrol. 150, 393–407. Timbal, B., Arblaster, J.M., Power, S., 2006. Attribution of the late-twentieth-century rainfall decline in southwest Australia. J. Clim. 19, 2046–2062. Turner, J.V., Macpherson, D.K., Stokes, R.A., 1987. The mechanisms of catchment flow processes using natural variations in deuterium and oxygen-18. J. Hydrol. 94, 143–162. Water Corporation, 2005. Wungong Catchment Environment and Water Management Project, Perth, Western Australia. Water Corporation, 2008. Water Forever: Options for Our Water Future, Perth, Western Australia. Williamson, D.R., Stokes, R.A., Ruprecht, J.K., 1987. Response of input and output of water and chloride to clearing for agriculture. J. Hydrol. 94, 1–28.