Examining the spatial and temporal variation of groundwater inflows to a valley-to-floodplain river using 222Rn, geochemistry and river discharge: the Ovens River, southeast Australia

Radon ( 222Rn) and major ion geochemistry were used to define and quantify the catchment-scale groundwatersurface water interactions along the Ovens River in the southeast Murray–Darling Basin, Victoria, Australia, between September 2009 and October 2011. The Ovens River is characterized by the transition from a single channel within a mountain valley in the upper catchment to a multi-channel meandering river on flat alluvial plains in the lower catchment. Overall, the Ovens River is dominated by gaining reaches, receiving groundwater from both alluvial and basement aquifers. The distribution of gaining and losing reaches is governed by catchment morphology and lithology. In the upper catchment, rapid groundwater recharge through the permeable aquifers increases the water table. The rising water table, referred to as hydraulic loading, increases the hydraulic head gradient toward the river and hence causes high baseflow to the river during wet (high flow) periods. In the lower catchment, lower rainfall and finer-gained sediments reduce the magnitude and variability of hydraulic gradient between the aquifer and the river, producing lower but more constant groundwater inflows. The water table in the lower reaches has a shallow gradient, and small changes in river height or groundwater level can result in fluctuating gaining and losing behaviour. The middle catchment represents a transition in river-aquifer interactions from the upper to the lower catchment. High baseflow in some parts of the middle and lower catchments is caused by groundwater flowing over basement highs. Mass balance calculations based on 222Rn activities indicate that groundwater inflows are 2 to 17 % of total flow with higher inflows occurring during high flow periods. In comparison to 222Rn activities, estimates of groundwater inflows from Cl concentrations are higher by up to 2000 % in the upper and middle catchment but lower by 50 to 100 % in the lower catchment. The high baseflow estimates using Cl concentrations may be due to the lack of sufficient difference between groundwater and surface water Cl concentrations. Both hydrograph separation and differential flow gauging yield far higher baseflow fluxes than 222Rn activities and Cl concentrations, probably indicating the input of other sources to the river in additional to regional groundwater, such as bank return flows.


Introduction
Defining the relationship between rivers and adjacent groundwater systems is a crucial step in developing programs and policies for protecting riverine ecosystems and managing water resources.Rivers interact with various water stores, such as groundwater in local and regional aquifers, water in river banks, water in unsaturated zone, and soil water.Rivers can recharge groundwater (loosing streams) or receive groundwater as baseflow (gaining streams).These interactions can vary with topography along a course of river, for example rivers may be gaining in narrow valleys in the hills but losing when they flow across the broad plains.Furthermore, the direction and magnitude of water fluxes can change with times; a gaining stream, for instance, can become losing if the river rises above the water table during a storm event.The three main controls on catchment-scale groundwater-surface water (GW-SW) interactions are: (1) the basin morphology and the position of the river channel within landscape; (2) the hydraulic conductivities of the river channel and adjacent alluvial aquifer; (3) and the relation of river stage to water table level in the adjacent aquifer which is closely related to precipitation patterns (Sophocleous, 2002;Pritchard, 2005).Without a sound understand of GW-SW interactions in a catchment it is not possible to identify potential pathways for water contamination and to calculate water budgets for water allocation.The latter has become an important issue in Australia because of the growing demands from both the human and environment in a drought dominated continent.
GW-SW interactions can be investigated by several techniques.Hydrograph separation is a straightforward method for assessing baseflow at a catchment scale.However, it cannot be used for losing or highly-regulated systems, and the slowflow component isolated by the method may aggregate several water storages (such as bank return Figures

Back Close
Full flow or interflow) rather than representing only regional groundwater inflow (Griffiths and Clausen, 1997;Halford and Mayer, 2000;Evans and Neal, 2005).Geochemistry, such as major ion concentrations, stable isotopes and radiogenic isotopes may also be used in GW-SW studies (Brodie et al., 2007;Cook, 2012).The requirements for using geochemical tracers to study GW-SW interactions are that the concentration of the tracer in groundwater is significantly different to that in river water and that concentrations in groundwater are relatively homogeneous (or that any heterogeneities are known).Radon ( 222 Rn) is powerful tracer for examining GW-SW interactions from qualitative and quantitative perspectives (Ellins et al., 1990;Cook et al., 2006;Mullinger et al., 2007;Baskaran et al., 2009). 222Rn is a radiogenic isotope produced from the decay of 226 Ra in the uranium decay series.In surface water, the 222 Rn activity is usually low because of low dissolved 226 Ra activities, the relatively short half life of 222 Rn (3.825 days) and the rapid degassing of 222 Rn to the atmosphere.Groundwater has 222 Rn activities that are commonly two to three orders of magnitude higher those of surface water due to the near-ubiquitous presence of U-bearing minerals in the aquifer matrix (Ellins et al., 1990;Cook et al., 2003;Mullinger et al., 2007;Cartwright et al., 2011).Due to the short half-life, the activity of 222 Rn reaches secular equilibrium with 226 Ra over two to three weeks (Cecil and Green, 1999).Thus, 222 Rn may be used for detecting groundwater inflows into rivers, especially where the difference in major ion concentrations between groundwater and surface water is small, such as in many upper catchment streams.The change in 222 Rn activities in a gaining stream (d C r /dx) is governed by groundwater inflow, in-stream evaporation, hyporheic exchange, degassing, and radioactive decay as follows: (1) (Cook et al., 2006;Mullinger et al., 2007).In Eq. ( 1), Q is the stream discharge (m 3 day −1 ), C r is the 222 Rn activity within the stream (Bq m −3 ), x is distance in the direction of flow (m), I is the groundwater inflow rate per unit of stream length Introduction

Conclusions References
Tables Figures

Back Close
Full (m 3 m −1 day −1 ), C i is the 222 Rn activity in the inflowing groundwater, F h is the flux of 222 Rn from hyporheic zone (Bq m −1 day −1 ), w is the width of the river surface (m), E is the evaporation rate (m day −1 ), k is the gas transfer coefficient (day −1 ), d is the mean stream depth (m) and λ is the radioactive decay constant (0.181 day −1 ).Groundwater inflow can be calculated by rearranging this equation.Equation (1) can also be used for other tracers.For major ions, such as Na or Cl, that do not degas to the atmosphere or decay, the last two terms on the right-hand side are redundant.This study uses 222 Rn activities and major ion geochemistry in conjunction with physical hydrological data to determine the GW-SW relationships and the contribution of baseflow along the Ovens River (Fig. 1), from its upper catchment to its discharge point at the Murray River.The study covers a period of 26 months that include the end of the 2000s Australian drought and the 2010 Victorian floods.From hydraulic heads and river heights, CSIRO (2008) indicated that the Ovens River is gaining in the upper catchment, alternately gaining and losing in the middle catchment and mainly losing in the lower catchment.However, the precise distribution of gaining and losing reaches, the temporal of GW-SW exchange and the quantity of baseflow to the river remains unknown.The results will provide an important background for future GW-SW studies in this and other catchments in the Murray-Darling Basin, and elsewhere., 2011).The river in the upper and middle regions is unregulated, but the flow downstream is partially regulated due to the storages on the Buffalo and King tributaries.

Groundwater along the Ovens River
The stratigraphy of the Ovens Catchment comprises Palaeozoic basement overlain by Tertiary-Recent fluviatile sediments (Lawrence, 1988;van den Berg and Morand, 1997).The basement in the upper and middle catchments is generally 10-50 m below the surface while the basement is up to 170m below the surface in the lower catchment.
Several basement highs and local outcrops exist at Myrtleford in the middle catchment and between Killawarra and Peechelba in the lower catchment.The basement predominantly consists of metamorphosed Ordovician turbidites intruded by Silurian and Devonian granites that form a fractured-rock aquifer with a hydraulic conductivity of 0.3-10 m day −1 and a transmissivity of < 10 m 2 day −1 (Slater and Shugg, 1987).The overlying sediment consists of, from the base to top, the Calivil Formation, the Shepparton Formation and the Coonambigal Formation.The sedimentary cover has the maximum thickness in the lower catchment, and thins and pinches out over basement Introduction

Conclusions References
Tables Figures

Back Close
Full  (Shugg, 1987;Cheng and Reid, 2006).The Quaternary fluvio-lacustrine Shepparton Formation has similar composition.The alluvial deposits of the Holocene Coonambnidgal Formation in the river valleys are contiguous with and undistinguishable from those of the underlying Shepparton Formation.
The Shepparton Formation and Coonambnidgal Formation together are up to 170 m thick and comprise a heterogeneous mixture of clay and silt, and "shoestring lenses" of sand and gravel (Tickell, 1978).As a result, they form a complex semi-confined to unconfined aquifer with varying degrees of interconnectivity.The alluvial sediments transits from unsorted cobbles and coarse gravels with fragment of basement rocks and minerals upstream to mature fine weathered sand and silt downstream.The hydraulic conductivity of the Shepparton and Coonambigal Formations is 0.1-10 m day −1 with an average of 0.2-5 m day −1 (Tickell, 1978).The Ovens River is within the Coonambidgal Formation, except for several locations upstream where it incises into the basement, for example, in Smoko, Bright and Myrtleford.The surface aquifers receive recharge mainly through direct infiltration on the valley/alluvial floors, and through exposed and weathered bedrocks at the margins of valley during the winter months.The head gradients in the Ovens catchment are usually downward, and the regional groundwater flow is down valley.The groundwater has a total dissolved solids (TDS) content of 100-500 mg L −1 which is higher than that of the Ovens River (TDS of 25-48 mg L −1 ) (Victorian Water Resource Data Warehouse, 2011).

Climate and land use
The   (22-34 km).This section includes a 2 km long and 2-4 m deep canyon in the basement followed by the beginning of the transition from the valley to the alluvial flood plains.River samples were collected from approximately 1 m above the riverbed using a collection beaker attached to a pole.In September 2009, groundwater was sampled from the Coonambigal and Shepparton Formations close to the Ovens River; some bores were re-sampled in the following rounds.The groundwater was sampled by using an impeller pump set at the screened Introduction

Conclusions References
Tables Figures

Back Close
Full interval, and at least 3 bore volumes of water were purged prior to sampling. 222Rn activities were measured using a portable in-air monitor (RAD-7, Durridge Co.) following methods described by Burnett and Dulaiova (2006)

Flow conditions
Between September 2009 and June 2010, the discharge of the Ovens River at Peechelba was between 160 and 4360 ML day

Groundwater levels
The bore hydrographs of shallow bores (< 20 m deep) at Bright indicate that recharge occurred on the valley alluvial plain in June 2010-February 2011 and June-September 2011 (Fig. 3a).The annual hydraulic head variation at Bright in the upper catchment between 2009 and 2011 was 0.5-3.0m (Victorian Water Resource Data Warehouse, 2011).There was a high lateral head gradient of ∼ 0.007 between the edge of valley (B57144) and the river bank (B51747 & B51743) towards the river.There were several head reversals between the bores in the bank prior to June 2010.As with upper catchment, there was uniform recharge at Eurobin and Myrtleford in the middle catchment in the same period (Fig. 3b and c).However, the annual hydraulic head variation was only 0.5-1.0 m.Furthermore, the lateral head gradient toward the river in the middle catchment was 0.002-0.004.The head gradients were reversed in the river bank at Myrtleford during recharge periods (May 2009 andAugust 2010).No data is available for the groundwater level near the river in the lower catchment during the study period.However, the historical data at Peechelba indicates the hydraulic heads in the flood plains (B11308 & B11307) varies by only a few millimetres per year (Fig. 3d).In contrast, the hydraulic head in the river bank (B11306) shows greater a variation of up to 1.5 m.The lateral head gradient toward the river in the lower catchment is ∼ 0.0001.Introduction

Conclusions References
Tables Figures

Back Close
Full

Electrical conductivity
The EC values of the Ovens River increases from ∼ 30 µS cm −1 in the upper catchment to 37-55 µS cm −1 at Peechelba in the lower catchment (Fig. 4).There is always rapid increase in the EC values in the first 5 km river reach from Harrietville.However, most of the increase in EC values occurs from the middle catchment downstream.Higher EC values (35-55 µS cm −1 ) were recorded in March 2010, March 2011 and June 2011 at the end of summer or during low flow.The EC trend in the Bright-Porepunkah river section has a small peak (an increase of 2.8 µS cm −1 in March 2011, 1.2 µS cm −1 in June 2011) in the Canyon (at 28 km) followed by a progressive increase in EC values downstream towards Porepunkah.
The EC values of shallow groundwater (< 60 m) increases down catchment from 50-100 µS cm −1 in the upper catchment to 100-400 µS cm −1 in the middle catchment and to 520-1200 µS cm −1 in the lower catchment (Table 2).EC values for groundwater throughout the study period and after the 2010 Victorian floods remained similar.

Conclusions References
Tables Figures

Back Close
Full is the September 2010 sampling round where the Na/Cl ratio remained between 1.70 and 2.70 along the entire river.Other cations/Cl ratios have similar trends to the Na/Cl.Groundwater in the upper and middle catchment is dominantly of a mixed magnesium and sodium or potassium bicarbonate type.Na comprises 22-58 % of the total cations with 20-43 % Mg, 16-30 % Ca and 3-21 % K, and HCO 3 accounts for 64-95 % of anions with 5-18 % Cl, 1-20 % SO 4 and > 1-25 % NO 3 .Groundwater in the lower catchment is a sodium or potassium chloride type with relative cation concentrations of 38-83 % Na, 4-54 % Ca, 2-27 % Mg, and > 1-3 % K, and relative anion concentrations of 29-64 % Cl, 20-68 % HCO 3 , > 1-16 % and > 1-4 % NO 3 .Molar Na/Cl ratios of the low salinity (TDS < 100 mg L −1 ) groundwater from the upper and middle catchments are mainly between 1.0 and 3.9 with some up to 11, whereas the more saline groundwater from the lower catchment has Na/Cl ratios close to those of rainfall (0.8-1.5).

Radon activities
While the Ovens River at uppermost site (0 km) in Harrietville has consistently low 222  The 222 Rn activities of groundwater in the upper catchment are in the range of 30 000-110 000 Bq m −3 (Table 5).The groundwater 222 Rn activities are 20 000-42 000 Bq m −3 in the middle catchment and 10 000-20 000 Bq m −3 in the lower catchment.The decrease trend in the groundwater 222 Rn activities across the subcatchments reflects a change in lithology from immature sediments containing abundant fragments of granitic and metamorphic materials, which contain a significant amount of U-bearing minerals, in the alluvial valleys to more mature weathered sediments that are dominated by quartz and feldspar on the plains.There were no significant differences in groundwater 222 Rn activities between the sampling rounds even after the 2010 floods.

Discussion
The geochemistry of the Ovens River allows the major geochemical process to be defined and the distribution and magnitude of groundwater inflows to be calculated.There are no occurrences of halite in the Ovens Valley, and Cl in groundwater and surface water is derived from rainfall (Cartwright et al., 2006).Since the molar Na/Cl ratio of the rainfall in the region is 1.0-1.3(Blackburn and McLeod, 1983), the high river Na/Cl (Fig. 5c) and other cation/Cl ratios in the upper reaches of the Ovens River most probably reflect rock weathering by surface runoff in the upper catchment.The decrease in river Na/Cl ratios down the catchment is likely caused by influxes of both groundwater in the valley aquifers and surface runoff from the middle and lower catchments, both of which have relatively low Na/Cl ratios.Thus, Na and other cations are derived from Introduction

Conclusions References
Tables Figures

Back Close
Full rainfall and rock weathering.Overall, the high 222 Rn activities in the upper and middle catchments of the Ovens River (Fig. 6) together with the progressive downstream increase in EC values (Fig. 4) and major ion concentrations (Fig. 5a and b) suggest that the Ovens River receives groundwater inflows. 222Rn is used to identify gaining reaches and to calculate baseflow in this study because the difference of 222 Rn activities between groundwater and river water in the Ovens Catchment is 2-3 orders of magnitude, whereas the difference in the EC values and concentrations of major ions in the groundwater and river water are much smaller.For comparison, baseflow fluxes are also calculated using Cl, which is conservative in this catchment, and estimated from the hydrographs.Assessing other potential methods of estimating baseflow is valuable as river discharge and major ion data are far more extensive than the 222 Rn data.

Baseflow fluxes calculation using 222 Rn activities
Groundwater influxes to the river for the sampling rounds were calculated by rearranging Eq. ( 1). 222Rn activities are from Table 4. Stream discharges were estimated by linear interpolation of data from the five gauging stations.River depths and widths were estimated in the field; river depths vary from 1.2-8.0m in winter and 0.3-6.7 m in summer, and river widths range from 15-100 m in winter and 7-90 m in summer.Evaporation rates were estimated as 0.05 m day −1 and 0.2 m day −1 for winter and summer months respectively (Australian Bureau of Meteorology, 2011).Based on the distribution of groundwater 222 Rn activities (Table 5), 222 Rn activities of 76 000 Bq m −3 , 32000 Bq m −3 and 19 000 Bq m −3 were assigned to the upper, middle and lower catchments respectively.Hyporheic exchange can also cause an elevation in 222 Rn activity in rivers where there is no or low groundwater input (or groundwater has low radon activities) (Cook et al., 2006;Lamontagne and Cook, 2007;Cartwright et al., 2011;Cook, 2012).This exchange needs to be accounted for in 222 Rn mass balance to prevent over-estimating groundwater inputs although it can be difficult to estimate for larger rivers.The Ovens River and groundwater have generally high 222 Rn activities.Introduction

Conclusions References
Tables Figures

Back Close
Full Thus errors associated with hyporheic exchange are likely to be small, and initially the F h term was omitted.Gas exchange coefficients (k) were estimated using the modified gas transfer models of O' Connor and Dobbins (1958) and Negulescu and Rojanski (1969) as described by Mullinger et al. (2007): (2) where v is the average stream velocity (m day −1 ) derived from the steam discharge, river depth and river width data.Equations ( 2

Baseflow fluxes calculation using Cl concentrations
Groundwater inputs to the river were also calculated by Cl concentrations (Table 3) via: groundwater inflow of 810 000 and 6 700 000 m 3 day −1 respectively.Overall, the cumulative groundwater inflows are 2-36 % of total flow.When the assigned groundwater Cl concentrations were adjusted by ±1 standard deviation (0.9, 17 and 120 mg L −1 for the upper, middle and low catchments) for their spatial variations, the groundwater fluxes for individual reach also vary accordingly: a 25-37 % decrease for the higher assumed groundwater Cl concentrations and a 62-150 % increase for the lower assumed groundwater Cl concentrations.Sensitivity analysis on both 222 Rn and Cl mass balance calculations shows that decreasing groundwater end-member concentration (C i , Cl i ) makes a greater change in groundwater fluxes compared with increasing C i or Cl i .As the difference between groundwater and river concentrations (C i -C r ) or (Cl i -Cl r ) get smaller by decreasing C i or Cl i , it is more likely to magnify any errors in the mass balance calculation.The greater sensitivity is more apparent in the Cl mass balance calculations because the difference between Cl i and Cl r is already low at the beginning.

Baseflow fluxes calculation via hydrograph separation
Hydrograph separation employs a low pass filtering technique to separate the slowflow component (assuming to be mainly baseflow) that shows a low frequency of variation from the high frequency signals associated with surface runoff and interflow.Recursive digital filters developed by Nathan and McMahon (1990) and Eckhardt (2005) were used in this study.The filter equation for Nathan and McMahon (1990) is: where y is the total stream discharge, f is the filtered quick flow, k is the time step number, and α is the recession constant.Discharge data used in the calculations are from the period of October 2000-October 2011 from the three gauging stations: Bright (22 km), Myrtleford (65 km) and Peechelba (187 km).α is the gradient of the falling limb Introduction

Conclusions References
Tables Figures

Back Close
Full of a hydrograph and was determined via linear regression following Eckhardt (2008): The calculated α values are 0.976 for Bright, 0.970 for Myrtleford and 0.967 for Peechelba.The filter was applied in three passes (forward, backward, forward) across the hydrograph as suggested by Nathan and McMahon (1990).The calculated percentages of baseflow in the May-October (wet) and November-April (dry) periods are 47 % and 83 % at 22 km, 51 % and 78 % at 65 km, 49 % and 79 % at 187 km respectively.The algorithm for the Eckhardt's filter is: (Eckhardt, 2005), where b is the filtered baseflow (bk ≤ yk), and BFI max is the maximum value of the baseflow index (BFI) that can be modelled by the algorithm.BFI max cannot be measured but is assigned based on the catchment lithology and river flow regime.Eckhardt (2005) proposes 0.8 for perennial streams with porous aquifers, 0.5 for ephemeral streams with porous aquifer and 0.25 for perennial steams with hard rock aquifers.Considering the change in lithology from the small and highly conductive aquifers and the large bedrock aquifer in the upper catchment to the less conductive aquifers in the lower catchment, an area weighted BFI max of 0.31 was assigned for the upper catchment, 0.36 for the middle catchment and 0.47 for the lower catchment.The filter was applied in a single pass across the hydrograph as suggested by Eckhardt (2005).In comparison to the Nathan and McMahon filter, the Eckhardt filter produces lower percentages of baseflow: 36 % and 52 % at 22 km, 43 % and 58 % at 65 km, and 54 % and 66 % at 187 km, in the May-October and November-April periods respectively.However, these values are still substantially higher than those estimated by 222 Rn activities: 3 % and 2 % at 22 km, 10 % and 9 % at 65 km, and 16 % and 12 % at 187 km.Introduction

Conclusions References
Tables Figures

Back Close
Full

Spatial variations in baseflow
The baseflow fluxes derived from 222 Rn activates indicate high groundwater inflows in the upper catchment.The high groundwater influx is likely caused by the catchment morphology and aquifer lithology.In the upper catchment, the narrow valley creates a high hydraulic gradient of ∼ 0.007 between the alluvial aquifers and the river (Fig. 3a).Furthermore, the high rainfall and high rates of recharge (120-180 mm yr −1 ) through the coarse grained sands and gravels in the upper catchment (Cartwright and Morgenstem, 2012) maintains the high hydraulic gradients.High groundwater inflow commonly occurs in the first few river sections (0-11 km) of the river (Fig. 7).These river reaches are located at the edge of the alluvial valley, rather in the centre of the valley.It is likely that groundwater is discharged to the river at these break of slopes as a result of the sudden change in topography.Moderate to high groundwater inflows also occur between 31 and 34 km at Porepunkah as indicated by the progressively increase in EC values and 222 Rn activities (Figs.4b and 6b).This location is coincident with springs and a spring fed stream.The river reach at 27.8-29.5m is in a moderately steep canyon.As the flow leaves the canyon, it cuts a channel deep through shallow sediments on the alluvial valley plains at this location.As the water table follows the topography, it is likely to intercept with the river channel to produce springs on the river plains and to recharge the river as baseflow.Groundwater inflows in the upper and middle catchments are also derived directly from the basement aquifer as evidenced by the presence of 222 Rn and EC peaks in the canyon (28.4-28.7 km) (Figs.4b and  6b).The magnitude of groundwater influx from the basement aquifer may be large and seasonal (up to 16 m 3 m −1 day −1 in March 2011).Since fractured bedrock aquifers often have very limited storativity, they deplete very quickly, discharging less groundwater to the river toward the end of summer as in June 2011.
Groundwater influxes in the middle catchment are generally equal to or lower than those in the upper catchment (Fig. 7).The middle catchment represents a transition Introduction

Conclusions References
Tables Figures

Back Close
Full from a narrow valley to open alluvial flood plains with the river in the middle of the valley.Therefore, the lateral head gradient toward the river is lower than that in the upper catchment (Fig. 3b and c).The aquifer sediments also have lower hydraulic conductivities, and both these factors cause a reduction of groundwater influxes to the river.The difference between the river stage and the water table is also small and thus, any small changes in river height or groundwater level can result in the observed fluctuating gaining and losing behaviour along the river.
Groundwater inflows are further reduced in the lower catchment.This is the result of the shallow hydraulic gradient developed between the river and the groundwater in the open and flat alluvial flood plains in a semi-arid environment (Fig. 3d).Furthermore, groundwater inflows are likely to be restricted by the less conductive alluvial sediments.
Despite the widening of alluvial plains, several locations (between 65 and 72 km and between 166 and 188 km in the middle and lower catchments) receive significant baseflow (up to 24 m 3 m −1 day −1 ) (Fig. 7).This gaining behaviour is probably caused by basement highs that deflect groundwater flow and induce upward head gradients.
Between 65 and 72 km several large outcrops of bedrocks are found near the river, while the river meanders closely to the Warby Ranges between 166 and 188 km (Fig. 1) (van den Berg and Morand, 1997).

Temporal variations in baseflow
Groundwater inflows in the upper catchment increase during high flow periods (Figs.7a     and 8a).The increased rainfall over autumn, winter and occasionally summer produces high surface runoff and also recharges the groundwater.Due to the coarse sediments, the hydraulic heads in the upper catchment can increase up to 3 m throughout a year (Fig. 3a).The rapid rising heads within the narrow river valley induces hydraulic loading, causing more groundwater to flow toward the river and resulting a greater amount of groundwater inflows.However, the magnitude of groundwater inflows do not always increase proportionally with river flows.For instance, the discharge in December 2010 was greater than that in September 2009, and yet the December 2010 around had a Introduction

Conclusions References
Tables Figures

Back Close
Full lower cumulative groundwater influx than the September 2009 around (Fig. 9a).The lower baseflow fluxes may be caused by the high river stage as a result of multiple floods in the previous winter/spring months which reduces the hydraulic gradient between the river and the adjacent water table.In contrast, the river was relatively dry in September 2009 after a period of drought, allowing a greater hydraulic gradient to be developed during the recharge period and thus producing a greater amount of baseflow (Fig. 2).The groundwater influxes in the upper catchment can be low during low flow periods.The coarse aquifer sediments enable relatively quick drainage of groundwater into the river during winter and spring months.The water table near the river is likely to drop to the river height, and thus there is less groundwater influxes to the river over summer months.The baseflow fluxes in the lower catchment are similar at both high and lower flow conditions.The constant baseflow fluxes are probably caused by the limited fluctuation in the water table.The water table in this region is relatively constant, ranging 0.5-1.5 m near the river and less than millimetres away from the river (Fig. 3d).The constant water table is due to low recharge rate of 30-40 mm yr −1 on the floodplains (Cartwright and Morgenstem, 2012) which is the result of reduced rainfall, flat topography and low conductivity of alluvial sediments.Since the water table near the river does fluctuate, it is possible for the river to recharge the adjacent aquifers and river banks during high flow conditions.

Comparing Rn with Cl concentrations and hydrograph separation
In the upper and to some extent, the middle catchment, the baseflows estimated from Cl concentrations are often greater than those based on 222 Rn activities by 30-2000 % (Fig. 9).If 222 Rn activities do better indicate the variability in groundwater inflows, there must be sources of error in using Cl concentrations to estimate baseflows.Underestimating evaporation is unlikely since evaporation is a minor process in the catchment (Cartwright and Morgenstem, 2012).Another possible error is ignoring potential saline groundwater inputs.However, if the assigned groundwater Cl concentration was 5247 Introduction

Conclusions References
Tables Figures

Back Close
Full increased to 5 mg L −1 which is the highest Cl concentration in the upper catchment, the groundwater influxes would only decrease by 20-50 %.Thus, saline groundwater input is probably not the sole reason for the overestimation.The error is likely the result of the similarity between groundwater and river Cl concentrations in the upper and middle catchment.If the difference in the two end-member concentrations is small, it requires a significant input of groundwater in order to detect a rise in river Cl concentrations associated with the influx of groundwater.It also magnifies any calculation and measurement errors in mass balance calculations, particularly groundwater end-member concentration (Cook, 2012).The large increases in some of the river Cl concentrations (and thus large estimation of groundwater influxes) may be due to accumulation of Cl over several reaches or may come from other sources, such as water in the unsaturated zone or in pools on riverine plain.
In the lower catchments, the Cl-derived baseflow fluxes are usually lower than 222 Rnderived ones by 50-100 % (Fig. 9).If the assigned groundwater Cl concentrations in the calculations are reduced, the amount of baseflows would progressively increase, matching ones derived from the 222 Rn activities.This may be interrupted as the amount of saline regional groundwater contributing to the river probably being low.Rather, the majority of baseflows in this area probably come from the less saline water in the midchannel bars and river banks.Using regional groundwater concentration in the mass balance calculations would under-estimate the total groundwater discharge to the river but correctly identify the amount of regional groundwater discharge if the groundwater discharge comprises of both bank storage and regional groundwater (McCallum et al., 2010).The infilling of these mid-channel bars and river banks is supported by the larger variable hydraulic heads near the river (Fig. 3d).While these water stores are being replenished during high flow period, regional groundwater is likely to be the only source of groundwater inflows as evident by the agreement of the Cl-and 222 Rn-derived baseflow influxes during the rising limb of a high flow event in December 2010.Simultaneously river bank refilling and regional groundwater discharge are also suggested on the Riverine Plain (Cartwright et al., 2011).Introduction

Conclusions References
Tables Figures

Back Close
Full In comparison to 222 Rn mass balance, hydrograph separation yields larger estimates of accumulative baseflow fluxes across the catchments.However, some assumptions behind this method may not hold for the Ovens River as some reaches in the middle and lower catchment are losing and partly regulated.For Eckhardt's filter, the calculations are sensitive to the value of BFI max which can only be assigned subjectively.The cumulative baseflow flux could increase by 2-40 % if the BFI max was increased by 0.1.Finally, hydrograph separation cannot isolate the individual components of the slowflow component; this might include groundwater, bank returns and interflows and draining of pools on the floodplains.The baseflow fluxes calculated using hydrograph separation are likely to include these slowflow components rather than just regional groundwater.
The idea of these slowflow components contributing to the river is also suggested by the Cl concentrations data above.

Conclusions
The Ovens River is dominated by gaining reaches.The river and aquifer interactions are mainly controlled by the catchment geomorphology, aquifer sediments and rainfall distribution.In the upper catchment, the reaches are mostly gaining.Rapid groundwater recharge through the coarse sediments in the narrow valley creates hydraulic loading and causes higher baseflow flux during high flow periods.In the lower catchment, the open plains, fine sediments and reduced rainfall ensure low water table variability, leading to relatively lower and constant groundwater influx.The reaches also experience fluctuating gaining and losing conditions due to the similarity between the water table and river height on the open alluvial plains.The middle catchment represents a transition in river-aquifer interactions from upper to lower catchment.Basement highs in the middle and lower catchment also play role in inducing groundwater influx to the river.Groundwater from alluvial and basement aquifers contributes 4-22 % of total river discharge with higher baseflow during high flow periods while water from unsaturated zone and river bank is likely to play a greater contribution to the river flow.Introduction

Conclusions References
Tables Figures

Back Close
Full  Full  Full  Full Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | south-east margin of the 1 061 469 km 2 Murray-Darling Basin, the Ovens Catchment (Fig. 1) occupies just 7813 km 2 but contributes 6-14 % of the total flow of the Murray River (CSIRO, 2008).The Ovens River is the main river in the Ovens Catchment with a length of approximately 202 km that originates in the northern flank of the Victorian Alps and flows north-westwards.The catchment is characterised by multiple narrow V-shaped mountain valleys in the upper catchment and broad flat alluvial flood 5229 Discussion Paper | Discussion Paper | Discussion Paper | plains in the lower catchment.In the upper catchment, the river is 5-10 m wide and 1-2 m deep.It has small rapids with a steep channel gradient of around 6.5 m km −1 (Victorian Government Department of Sustainability and Environment, 2010a).Downstream of Porepunkah, the valley broadens and transits into open alluvial flood plains.The river in the lower catchment has a low gradient of less than 1 m km −1 (Victorian Government Department of Sustainability and Environment, 2010a) and develops a network of meandering and anastomosing channels downstream of Everton.In its lower reaches, it is 40-50 m wide and up to 8 m deep.It flows pass the Warby Ranges before discharging to the Murray River at Bundalong.The Ovens River is perennial and receives water from three main tributaries: the Buckland, Buffalo and King Rivers.The monthly flow at the gauging station located at Peechelba toward the discharge point varies between 200 and 30 200 ML day −1 with high flow occurring in Australian winter months (June-September) (Victorian Water Resource Data Warehouse Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | highs and in the valleys toward the highlands.The terrestrial Tertiary Calivil Formation has a thickness of up to 45 m.It does not crop out and occurs between 20 and 100 m below ground surface.It comprises consolidated gravel, sand slit, clay and rubbles with a hydraulic conductivity of 5-50 m day −1 climate of Ovens catchment is mainly controlled by the topography.The average rainfall decreases from 1127 mm in the alpine region in Bright to 636 mm on the alluvial plains in Wangaratta with most rainfall occurring in winter months (Bureau of Meteorology, 2011).In the 2000s Australian drought (particularly 2006-2009), rainfall in the Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | −1 with several moderately high flow events up to 11 420 ML day −1 during the 2009 winter (Fig. 2) (Victorian Water Resource Data Warehouse, 2011).Multiple extremely high flow events of up to 93 570 ML day −1 occurred between August 2010 and March 2011.The unusual wet periods during the spring and summer (November-March) were contributed by the 2010-2011 La Ni ña event.The river flow returned to 1910-3800 ML day −1 in the period of March-June 2011, followed by multiple moderately high flow events of up to 25850 ML day −1 Discussion Paper | Discussion Paper | Discussion Paper | between July and September 2011.The September 2009, September 2010, December 2010, and March 2011 sampling rounds all took place during high flow conditions with the September 2009 (10 178 ML day −1 ) and December 2010 (18 520 ML day −1 ) rounds on the rising limb of a flow event, and the September 2010 (6635 ML day −1 ) and March 2011 (4894 ML day −1 ) rounds on the receding limb of a flow event.The discharge in the March 2010, June 2010, June 2011 and October 2011 sampling rounds were 995 ML day −1 , 1114 ML day −1 , 2292 ML day −1 and 2606 ML day −1 respectively, and these sampling rounds were conducted during low flow periods.
Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper |

(
Rn activities (112-245 Bq m −3 ), other river reaches in the upper catchment have 222 Rn activities between 373 and 2903 Bq m −3 (Fig. 6a).The location at 4.8 km often records the highest 222 Rn activity, with the exception of September 2010 and March 2011. 222Rn activities in the upper catchment were highest in September 2009, June 2011 and October 2011 and lower in March 2011, September 2011 and December 2011.In the Bright-Porepunkah river section, there was a significant 222 Rn peak of 905 Bq m −3 (March 2010) and 817 Bq m −3 (June 2010) at 28 km in the canyon (Fig. 6b), which is the site where the small increase in EC values was observed.The 222 Rn activities in the last 1.5 km of this river section were 881-1243 Bq m −3 .A small stream and spring on the floodplain at Porepunkah had 222 Rn activities of 2663 Bq m −3 March 2010) and 8083 Bq m −3 (June 2010), and 10 488 Bq m −3 (March 2010) and 50 450 Bq m −3 (June 2010) respectively.In the middle catchment, 222 Rn activities generally decrease downstream from 601-2174 Bq m −3 to 231-440 Bq m −3 , with several 222 Rn peaks occurring between 46.6 and 62.6 km (Fig. 7a).High 222 Rn activities were Introduction Discussion Paper | Discussion Paper | Discussion Paper | recorded in September 2009 and June 2011, whereas activities were lowest in December 2010.River reaches in the lower catchment have the lowest 222 Rn activities, ranging between 80 and 754 Bq m −3 .The temporal variation in the 222 Rn activities in the lower catchment is minimal with a maximum difference of ∼ 200 Bq m −3 .Despite the low activities, locations between 140 and 187 km in September 2009 and March 2010 exhibited elevated 222 Rn activities (699 and 754 Bq m −3 respectively).
Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | ) and (3) give a maximum and minimum k value respectively, and these values were used to calculate maximum and minimum groundwater influxes.The k values for the winter months are generally 3.0-8.0day −1 in the upper catchment, decreasing to 0.2-1.0day −1 in the lower reaches.Lower values were obtained for the summer months, from 3.0-4.0day −1 in the upper catchment to 0.2-0.3day −1 in the lower catchment.High values of k in the upper catchment reflects the high velocities caused by the shallow river depth and steep channel gradient, while low values of k in the lower catchment is the result of lower velocities due to the greater rive depth and low channel gradient.Studies on low-gradient rivers indicated that the k values typically vary from 0.5 to 2.5 day −1(Raymond and Cole, 2001;Cook et al., 2003;Cartwright et al., 2011).k values for shallow and turbulent rivers have been estimated to be up to 34 day −1(Mullinger et al., 2007).Thus the calculated k values are within the range recorded in other studies.Each sub-catchment was assigned an average value of k from all individual reaches within the sub-catchment to calculate the maximum and minimum groundwater influxes.The 222 Rn dilution at the three confluences was calculated by combining the 222 Rn activity and the discharge at the sampling site Discussion Paper | Discussion Paper | Discussion Paper | downstream of the confluence with the 222 Rn activity (Table 4) and the discharge near the exit of the tributary.The calculations indicate most reaches are gaining (I > 0 m 3 m −1 day −1 ), except for one reach in the upper catchment (at 11 km, June 2011) and few reaches in the middle and lower catchments.Reaches with I < 0 m 3 m −1 day −1 are assigned as having zero baseflow flux.For maximum groundwater inputs (using values from Eq. 2), the baseflow fluxes are 0.2-9.0m 3 m −1 day −1 for the upper catchment, 0.2-24.4m 3 m −1 day −1 for the middle catchment and 0.1-24.1 m 3 m −1 day −1 for the lower catchment during the high flow periods (September 2009, September 2010, December 2010, March 2011) (Fig. 7a); and 0.1−1 for the middle catchment and 0.4-3.7 m 3 m −1 day −1 for the lower catchment during the low flow periods (March 2010, June 2010, June 2011and October 2011) (Fig.7b).Relatively higher groundwater inflows occur in the upper and middle catchment.In addition, there are high groundwater inputs of up to 24 m 3 m −1 day −1 at several locations (65-72 km and 166-188 km) in the middle and lower catchments.Furthermore, groundwater inputs, particularly in the upper catchment, often increase during the high flow periods.The increase in groundwater influxes during the high flow periods is also reflected by the higher cumulative groundwater influxes in September 2009 (2 400 000 m 3 day −1 ), December 2010 (740 000 m 3 day −1 ) and March 2011 (660 000 m 3 day −1 ) (Fig.8a) compared to the low flow periods in Fig.8b).The cumulative groundwater inflow for the catchment during the study period was 150 000-2 400 000 m 3 day −1 or 4-22 % of total flow.Repeating the calculations with values from Eq. (3) (minimum k) reduces groundwater influxes by 18-70 %.The large percentage changes often occur in the gaining reaches with a declining222 Rn activity and in the gaining reaches with a small increase in 222 Rn activity.Sensitivity to k has been found to increase with lower groundwater inflow rate (that is a small d C r /dx)(Cook et al., 2006;Cook, 2012).It also results in additional two losing reaches in the upper catchment and increases the number of losing Discussion Paper | Discussion Paper | Discussion Paper | pling round.The background river 222 Rn activities for each sub-catchment are derived from the lowest river 222 Rn activities in the upper and lower catchments: 220 Bq m −3 , 175 Bq m −3 and 130 Bq m −3 for upper, middle and lower catchments respectively.The revised groundwater influxes are 3-70 % lower.The large discrepancies usually occur in some reaches of the middle and lower catchments where these reaches have a relatively low 222 Rn activity.However, these reaches only contribute a small proportion of baseflow to the catchment, and these large discrepancies will only have a small effect on the catchment-scale groundwater inflow.The over-estimation on the cumulative groundwater inflow in September 2009 due to ignoring hyporheic flow is about 17 %Discussion Paper | Discussion Paper | Discussion Paper |

(
Cartwright et al., 2011), where Cl r and Cl i are Cl concentrations in the river and groundwater respectively.Based on the distribution of Cl concentrations in groundwater (Table2), three average Cl r concentrations were used for distance 0-65 km, 65-127 km and 127-202 km: 3.25, 45.0 and 275 mg L −1 respectively.Comparing with the 222 Rn mass balance calculations, Cl mass balance calculations indicate fewer gaining reaches in the upper, middle and lower catchments (Fig. 9).The locations of high groundwater inflow do not always mirror the ones predicted by the 222 Rn activities.The groundwater influxes for the upper and middle catchments based on Cl concentrations are higher than those based on 222 Rn activities.Conversely, Cl concentrations often yield lower groundwater influxes in the lower catchment than 222 Rn activities.Several reaches in the middle catchment have extremely high calculated baseflow of up to 1414 m 3 m −1 day −1 .The best match between 222 Rn and Cl derived groundwater influxes are the ones in the upper catchment in March 2010 and June 2010, and the ones in the lower catchment in December 2012.During the high flow periods, groundwater influxes of 5upper, middle and lower catchment respectively.For the low flow periods, groundwater influxes are lesser: 0.08-3.4m 3 m −1 day −1 for the upper catchment, 0.2-4.6 m 3 m −1 day −1 for the middle catchment and 0.01-0.2m 3 m −1 day −1 for the lower catchment.Except for the two high flow conditions (September and December 2010), the Cl mass balance calculations indicate lower cumulative groundwater influxes, for example 1 000 000 m 3 day −1 in September 2009, 17 000 m 3 day −1 in March 2010, and 140 000 m 3 day −1 in June 2011.September and December 2010 had a cumulative Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper |

Fig. 4 .Fig. 5 .Fig. 7 .Fig. 8 .Fig. 9 .
Fig. 4. (A) EC values along the course of the river (Table1).EC values gradually increase downstream.(B) EC values along the Bright-Porepunkah reach in March and June 2011.Distinct EC peaks at 28.5 km, followed by a gradual increase in EC values in both sampling rounds.

Table 1 .
EC values of the Ovens River.Notes: nm -not measured

Table 2 .
EC values and Cl concentrations of groundwater in the Ovens Catchment.Notes - * : if single value is reported, the value represents the depth of the middle of the bore screen below ground surface.

Table 3 .
Cl concentrations of the Ovens River.