Regional Patterns of Erosion and Sediment and ... - Land and Water

Regional Patterns of Erosion and Sediment and Nutrient Transport in the Goulburn and Broken River Catchments, Victoria R.C. DeRose, I.P. Prosser, L.J. Wilkinson, A.O. Hughes and W.J. Young

CSIRO Land and Water, Canberra Technical Report 11/03, March 2003

C S I R O L A N D a n d W AT E R

Regional Patterns of Erosion and Sediment and Nutrient Transport in the Goulburn and Broken River Catchments, Victoria R.C. DeRose, I.P. Prosser, L.J. Wilkinson, A.O. Hughes and W.J. Young

CSIRO Land and Water, Canberra Technical Report 11/03, March 2003

Copyright ©2003 CSIRO Land and Water To the extent permitted by law, all rights are reserved and no part of this publication covered by copyright may be reproduced or copied in any form or by any means except with the written permission of CSIRO Land and Water.

Important Disclaimer To the extent permitted by law, CSIRO Land and Water (including its employees and consultants) excludes all liability to any person for any consequences, including but not limited to all losses, damages, costs, expenses and any other compensation, arising directly or indirectly from using this publication (in part or in whole) and any information or material contained in it. ISSN 1446-6163

Table of Contents Acknowledgments..................................................................................................................................................... 3 Abstract ......................................................................................................................................................................... 4 Main Research Report................................................................................................................................................... 5 Background ............................................................................................................................................................... 5 Project Objectives ..................................................................................................................................................... 8 Methods .................................................................................................................................................................... 8 Sediment Delivery through the River Network................................................................................................... 10 Contribution of Suspended Sediment to the Murray River ................................................................................. 12 Nutrient Delivery through the River Network .................................................................................................... 13 Model Inputs ........................................................................................................................................................... 13 River Hydrology and Channel Form................................................................................................................... 13 Hillslope Erosion ................................................................................................................................................ 15 Gully Erosion ...................................................................................................................................................... 16 River Bank Erosion............................................................................................................................................. 16 Nutrient Sources – Total P and N ....................................................................................................................... 21 Disaggregation of Mean Annual Loads to Daily Loads...................................................................................... 24 Results and Discussion............................................................................................................................................ 25 Hillslope Erosion Hazard .................................................................................................................................... 25 Gully Erosion Hazard.......................................................................................................................................... 25 Riverbank Erosion............................................................................................................................................... 26 Sediment Sources to the Stream Network........................................................................................................... 26 Nutrient Sources.................................................................................................................................................. 28 Sediment Delivery through the River Network................................................................................................... 28 River Suspended Loads....................................................................................................................................... 29 Bedload Deposition............................................................................................................................................. 29 Nutrient Budget................................................................................................................................................... 30 Contribution to Suspended Sediment Export to the Murray River ..................................................................... 41 Comparison of Suspended Sediment Loads........................................................................................................ 41 Comparison of Nutrient Loads............................................................................................................................ 44 Disaggregation of Annual to Daily Loads........................................................................................................... 47 Testing of Land Use Scenarios ........................................................................................................................... 47 Comparison with NLWRA Results..................................................................................................................... 49 Conclusions............................................................................................................................................................. 49 References............................................................................................................................................................... 50

1

List of Figures (abbreviated titles) Figure 1: Map of the Goulburn and Broken River catchments. .................................................................................. 7 Figure 2: Mean annual rainfall across the Goulburn and Broken River catchments................................................... 9 Figure 3: A river network showing links, nodes, Shreve magnitude of each link (Shreve, 1966) and internal catchment area of a magnitude one and a magnitude four link. ................................................................. 10 Figure 4: Conceptual diagram of the bedload sediment budget for a river link. STC is the sediment transport capacity of the river link, determined by Equation 1. ................................................................................ 11 Figure 5: Conceptual diagram for the suspended sediment budget of a river link.. .................................................. 12 Figure 6: Conceptual diagram for the nutrient budget of a river link........................................................................ 13 Figure 7: Distribution of average bank heights and channel widths in relation to upslope contributing area for 48 surveyed sites. ............................................................................................................................................ 15 Figure 8: Predicted hillslope erosion hazard in the Goulburn and Broken River catchments................................... 17 Figure 9: Average density of gully erosion for 10 x 10 km grid cells for the Goulburn and Broken River Catchments................................................................................................................................................. 18 Figure 10: Mapped amount of intact riparian vegetation. ......................................................................................... 19 Figure 11: Predicted bank erosion............................................................................................................................. 20 Figure 12: Pattern of dissolved N input to streams. .................................................................................................. 22 Figure 13: Pattern of dissolved P input to streams.................................................................................................... 23 Figure 14: Measured floodplain width. ..................................................................................................................... 31 Figure 15: Predicted depth of floodplain deposition. ................................................................................................ 32 Figure 16: Predicted suspended sediment load compared with loads measured at gauging stations within the Goulburn and Broken catchments. ............................................................................................................. 33 Figure 17: Predicted suspended sediment load expressed per unit area.................................................................... 34 Figure 18: Modelled sediment transport capacity. .................................................................................................... 35 Figure 19: Predicted bedload deposition................................................................................................................... 36 Figure 20: Predicted Total P load.............................................................................................................................. 37 Figure 21: Predicted proportion of dissolved P to Total P load. ............................................................................... 38 Figure 22: Predicted Total N load............................................................................................................................. 39 Figure 23: Predicted proportion of dissolved N to Total N load............................................................................... 40 Figure 24: Predicted contribution of suspended sediment to the Murray River for sub-catchments within the Goulburn and Broken River catchments. ................................................................................................... 42 Figure 25: Suspended sediment rating curve (left) and cumulative load distribution (right) for Delatite River at Tonga Bridge (405214).............................................................................................................................. 43 Figure 26: Comparison of average annual suspended sediment loads predicted from SedNet with estimated loads from gauging station (Figure 16) sediment rating curves (eg. Figure 25).................................................. 44 Figure 27: Variation in predicted mean daily loads from April to October 1993 for the Delatite River at Tonga Bridge......................................................................................................................................................... 46 Figure 28: Scenario testing in ArcMap.. ................................................................................................................... 48

List of Tables (abbreviated titles) Table 1: Average concentrations of N and P in runoff from selected land uses within the Goulburn and Broken River Catchments. ...................................................................................................................................... 24 Table 2: Components of the sediment budget for the Goulburn Broken River Basins. ............................................ 27 Table 3: Components of the nutrient budget for the Goulburn Broken River Basins. .............................................. 27 Table 4: Comparison of nutrient concentrations and loads to predicted loads of SedNet at selected gauging stations. ................................................................................................................................................................... 45

2

Acknowledgments This study forms part of the NLWRA Communications and Adoptions project for dissemination of NLWRA nutrient and sediment budget information to local catchment management agencies. It further tests the application of sediment modeling approaches to focus catchments such as the Goulburn and Broken Rivers through the use of regional data sources. As such we acknowledge the input from both the Goulburn-Broken Catchment Management Authority, Victoria Department of Natural Resources and Environment, and Goulburn-Murray Water who provided much of the necessary resource information required to complete this work. Theiss Environmental provided river cross-section information. In particular we thank Wayne Tennant and Pat Feehan who helped coordinate the collection of resource information. We also acknowledge the assistance of John Gallant of CSIRO for topographic analyses and Hua Lu who provided updated land cover factor and hillslope erosion assessments for the catchment. We gratefully acknowledge the funding and support provided by NLWRA that made this study possible. Goulburn Broken CMA also provided funding to assist with the collection and processing of resource information.

3

ABSTRACT This project was carried out to identify the major processes involved in the delivery of sediment and nutrients to rivers within the combined Goulburn and Broken River catchments. The loss of sediment and nutrients from the land can have impacts downstream on the larger rivers such as the Murray River and the estuarine environments that receive this material. An essential part of minimizing the impact of sediment is to reduce losses from the landscape. In regional catchments, such as the Goulburn and Broken Rivers, there are a wide range of environments only some of which will contribute significant amounts of sediment to streams. There are also many opportunities for deposition of sediment in the catchment so that not all areas of erosion result in export of sediment from the catchment. The project looks at patterns of sediment transport through the river network, identifying which reaches may be impacted by deposition of sand on river beds, and which sub-catchments contribute the most to suspended sediment loads and export from the river basin. We address these issues by constructing sediment and nutrient budgets for the catchment. A budget is an account of the major sources, stores and fluxes of material in a catchment. Spatial modelling is the only practical method to assess the patterns of sediment and nutrient transport in a large complex catchment as there are only limited measurements of material transport rates. Modelling can be used to interpolate these measurements and combine them with a basic understanding of transport processes and geographical information on controlling factors. This includes mapping of soils, vegetation cover, geology, terrain, climate and measurements of river discharge. We produce maps and summary statistics of predicted surface wash erosion, gully erosion, riverbank erosion and bedload and suspended load transport across the catchments. The model results suggest that gully erosion is the dominant erosion process contributing approximately 57% of the total predicted sediment supply. Gully erosion is the dominant sediment source in a SW – NE trending zone through the middle of the catchments. Riverbank erosion also makes a significant contribution with 36% of the total predicted supply to streams being derived from this source. Reaches, where there is a combination of poor riparian vegetation cover coupled with high stream power, produce the bulk of this sediment. Sheetwash and rill erosion contributes 7% of the total predicted sediment supply to channels. While it varies by 3 orders of magnitude, only 11% of the catchment has moderate to high surface erosion potential. Much of this is restricted to steeper slopes on grazing land or to areas of cropping. Rapid accumulation of sand and gravel on the bed of rivers can degrade aquatic habitat. This is a significant concern in some areas of the Goulburn and Broken catchments. Accumulation of sand occurs along reaches of streams that have relatively gentle and wide channels and above which the upstream catchment area has relatively high erosion rates and sediment transport capacities to mobilize and move coarser sediments. Results presented here are considered to underestimate the extent of streams affected by sand accumulation. The sediment budget predicts that 42% of suspended sediment and about 1% of bedload delivered to the river network in any year is exported from the river mouth. Lakes, reservoirs and floodplains predominantly along the lower sections of the Goulburn and Broken rivers, provide the greatest opportunity for deposition of suspended sediment. We predict that the mean annual export of suspended -1 sediment to the Murray River is 132 kt y . This figure lies within ± 50% of measured sediment loads based on samples collected at gauging stations along the Goulburn River. This represents a reasonable agreement given that there are significant errors involved in the estimation of mean annual loads from the generally non-event based sediment samples and the limitations of the model. The difference between predicted and observed loads could relate to underestimation of the amount of floodplain deposition or that rates of gully erosion have declined in recent decades relative to the long term average. Each of the sediment sources described above, together with dissolved contributions from hillslope runoff and point sources, deliver nutrients to the network of streams and rivers in the Goulburn and Broken -1 -1 River basins. We predict that 287 t y of Total P and 2326 t y of Total N, are exported from the 4

catchment. This represents 59% of Total P and 67% of Total N supplied to streams. The higher proportion of Total N export partly reflects the higher proportion of N in a dissolved form (60 - 80%) compared with P in a dissolved form (20 – 40%) transported by the rivers. As a greater proportion of P is transported attached to sediment there are greater opportunities for deposition before reaching the river mouths and hence the proportion of Total P exported is less than that of Total N. The spatial patterns of nutrient supply and transport in the catchments differ and this has important implications for catchment management. The pattern of Total P loads in rivers is dominated by sedimentbound sources from areas of gully and riverbank erosion. The proportion of dissolved P in rivers remains low relative to sediment-bound P despite significant inflows of dissolved P from agricultural land. In contrast, the greater levels of dissolved N supplied by runoff from agricultural areas, and in particular irrigation in lowland areas, helps to maintain a high proportion of dissolved N in rivers despite significant inputs of sediment-bound N and denitrification occurring. Hence, while reducing the supply of sediment to streams will go a long way to reducing Total P supply, major reduction in the supply of Total N will not be achieved without attention being payed to dissolved sources as well. Results from the sediment and nutrient budget for the Goulburn and Broken Rivers have strong potential for guiding further investigation, identifying areas for improved management and setting targets for catchment restoration. The results predict that each erosion process (surface wash, gully erosion, and riverbank erosion) is significant and that these processes are highly focused, with much of the sediment and attached nutrients being generated from relatively small areas. If future efforts at minimising soil loss are targeted towards these hotspots, using management guidelines appropriate to the type of process, then a large benefit in reduced sediment and nutrient loads downstream can be achieved with comparatively less effort. As part of this project, a catchment scenario testing tool has been developed to assist extension providers and natural resource management agencies to investigate the relative effectiveness of different management strategies on long-term sediment loads and yields from river networks. In the ArcGIS environment, users can select an individual or group of watersheds and interactively change the attributes which effect sediment supply to streams. By rerunning the sediment model for this new land use scenario, the user can then see how these modifications will alter the sediment loads further downstream and finally sediment export to the coast. This scenario tool will help to maximise use of the limited resources that might be available at obtaining the best result for minimising sediment and nutrient export from the catchment.

MAIN RESEARCH REPORT Background A significant aspect to achieving ecologically sustainable land management is to ensure that the downstream impacts of land uses on streams are minimised. An essential part of minimising impact is to reduce the delivery of sediments and nutrients from land to streams. For many catchments with low input farming systems, the bulk of the nutrient load is transported attached to sediment so that sediment and nutrient transport are intimately linked. To put a particular land use or sub-catchment in context with the regional catchments in which it occurs requires us to conceptualise the critical sources, transport pathways and sinks of sediment and nutrient in a catchment. We need to identify where sediment and nutrient are derived from, where they are stored within the catchment, and how much is delivered downstream to rivers and the sea. To quantify sources, stores and delivery is to construct a material budget for a catchment or any part of a catchment. This is a critical step to conceptualise the context of land use in a large regional catchment and to focus more detailed studies on the areas of greatest potential impact. To date only a few regional studies of sediment and nutrient budgets have been undertaken. Most catchments are complex systems, often with considerable variation in land use pressures, and diverse topography, soils, rainfall and vegetation cover. Thus before changing any particular management or even undertaking remediation measures we need to determine its significance and the 5

spatial pattern that land uses impact for sediment and nutrient transport. We also need to put the more detailed investigations of other parts of this project in a broader regional environmental context for the results to be applicable across wider areas. Some parts of the landscape are inherently more at risk of increased erosion and sediment and nutrient transport than others. It is important to identify these areas for these will be the sites that require the most careful management to ensure a sustainable future. For example, some landscapes have inherently poor soils where grass cover is susceptible to dramatic and long-lasting decline when subjected to grazing pressure or drought. Other factors that contribute to inherent risk of sediment and nutrient delivery to streams include steep slopes, high channel density, and high rainfall erosivity. Sediment and nutrients are derived from a number of processes which include: • Runoff on the land, termed surface wash and rill erosion or alternatively hillslope erosion; • Erosion of gullies formed as a result of land clearing or grazing; • Erosion of the banks of streams and rivers; • Diffuse dissolved losses of nutrients; • Point sources for nutrients such as towns and industry. In many cases one process dominates the other in terms of delivering sediments and nutrients to streams, and the predominant process can vary from one part of a large catchment to another. Management aimed at reducing sediment and nutrient transport will target each process quite differently. For example, stream bank and gully erosion is best targeted by managing stock access to streams, protecting vegetation cover in areas prone to future gully erosion, revegetating bare banks and reducing sub-surface seepage in areas with erodible sub-soils. Surface wash erosion is best managed by promoting consistent groundcover, maintaining soil structure, promoting nutrient uptake and promoting deposition of eroded sediment before it reaches the stream. Consequently it is quite important to identify the predominant sediment and nutrient delivery process before undertaking catchment remediation or making recommendations for changed grazing practice. Sediment delivered to streams has several potential downstream impacts. High loads of suspended sediment, the silts and clays that are carried in the flow, degrade water quality in streams, reservoirs and estuaries. This is a result of both the sediment itself and the nutrients that the sediment carries. High concentrations of suspended sediment reduce stream clarity; inhibit respiration and feeding of stream biota; diminish light needed for plant photosynthesis; make water unsuitable for irrigation and require treatment of water for human use. Much of the sediment and nutrient is deposited on floodplains, providing fertile alluvial soils, or it is deposited in reservoirs. The extent of this deposition is highly variable from one river reach to another. Deposition potential must be considered when trying to relate catchment land use to downstream loads of sediment. The formation of gullies and accelerated erosion of stream banks can supply large amounts of sand and gravel to streams. These are transported as bedload, being rolled and bounced along the bed of streams. Where streams are unable to transmit the load of sand and gravel downstream, it is deposited, burying the bed, and in extreme examples forming sheets of sand referred to as sand slugs (Rutherfurd, 2000). Sand slugs are poor habitat. They can prevent fish passage, they fill pools and other refugia, and are unstable substrates for benthic organisms (Jeffers, 1998). A reconnaissance level sediment budget for the Goulburn and Broken River catchments will provide an understanding of the critical processes of sediment and nutrient transport that can lead to downstream impact. It will place the major land uses within a regional context. The budget will also identify subcatchments with the greatest potential for downstream impact on aquatic ecosystems. These are the first steps toward better targeting of remedial and land conservation measures.

6

BARMAH LAKE NATHALIA

!

Broken R iver

´

KATAMATITE

!

LOCH GARRY SHEPPARTON

! TATURA !

GREEN LAKE LAKE COOPER WARANGA BASIN !MURCHISON REEDY LAKE LAKE NAGAMBIE !

LAKE MOKOAN BENALLA

! EUROA

!

LAKE NILLAHCOOTIE SEYMOUR

!

Goul b

ur n R

MANSFIELD

YEA

! KILMORE

LAKE EILDON

iver

!

! EILDON ! ALEXANDRA

JAMIESON

!

!

BUXTON

!

Legend

!

Localities Roads Streams

Lakes/Reservoirs Goulburn - Broken Catchments

0

25

Kilometres 100

50

Figure 1: Map of the Goulburn and Broken River catchments.

7

Project Objectives This report constitutes the final phase of the National Land and Water Resources Audit (NLWRA; 2001) Communications and Adoptions Project on extending nutrient and sediment budgets to state agencies, regional resource managers and community groups. Specific objectives are: • To use the best available regional data sources and improve techniques to get a more accurate result than the NLWRA national results for the Goulburn and Broken River catchments. • To compare results with monitored sediment and nutrient loads. • To examine the time sequence of loads and demonstrate how mean annual loads can be disaggregated into daily loads. • To develop the ability to examine future management scenarios and their affect on sediment and nutrient export from catchments.

Methods 2

The Goulburn and Broken catchments together cover some 22 800 km (Figure 1). Main outflows to the Murray River occur at the mouths of the Goulburn River and Broken Creek. Streams from an area of 2 approximately 900 km to the west of Waranga Basin terminate in a number of small lakes. In addition, 2 streams from a 1000 km portion of the Shepparton Irrigation District discharge flow into the neighbouring Campaspe Catchment via Waranga Channel before flowing into the Murray. Therefore about 8% of the catchment does not contribute to the sediment and nutrient loads of the Goulburn River or Broken Creek. The only practical framework to assess the patterns of sediment and nutrient transport across a large complex area such as the Goulburn and Broken River catchments (Figure 1) is a spatial modelling framework. There is a large range of climate (eg., Figure 2), topography, and land use which can strongly affect erosion and sediment transport. There are few direct measurements of sediment transport in regional catchments, and it is unrealistic to initiate sampling programs of the processes now and expect results within a decade. Furthermore, collation and integration of existing data has to be put within an overall assessment framework, and a large-scale spatial model of material transport is the most effective use of that data. The assessment of sediment and nutrient transport is divided into four aspects: hillslope erosion as a source of sediment and attached nutrients; hillslopes as a source of dissolved nutrients; gully erosion as a source of sediment and nutrients; and river links as a further source, receiver and propagator of the sediment and nutrients. To calculate the supply of sediment, its deposition and its delivery downstream is to construct a river sediment budget. We calculated budgets for two types of sediment: suspended sediment and bedload. TM A suite of ArcInfo programs were used to define river networks and their sub-catchments; import required data; implement the model; and compile the results. The programs are referred to collectively as the SedNet model: the Sediment River Network model. The SedNet model calculates, among other things: • The mean annual suspended sediment output from each river link; • The depth of sediment accumulated on the river bed in historical times; • The relative supply of sediment from surface wash, gully and bank erosion processes; • The mean annual rate of sediment accumulation in reservoirs; • The mean annual export of sediment to the Murray River; • The contribution of each sub-catchment to that export.

8

NATHALIA

´

KATAMATITE

SHEPPARTON TATURA BENALLA MURCHISON EUROA

NAGAMBIE

SEYMOUR

MANSFIELD

YEA

ALEXANDRA EILDON JAMIESON

KILMORE BUXTON

Legend Average Annual Rainfall (mm) 394 - 500 500 - 750 750 - 1,000 1,000 - 1,250 1,250 - 1,500 1,500 - 1,750 1,750 - 1,858

0

25

Kilometers 100

50

Figure 2: Mean annual rainfall across the Goulburn and Broken River catchments.

9

Details of the model and its application to regional catchments in Australia are given in Prosser et al., (2001). The methods used in the construction and implementation of the SedNet programmes are described in detail in a number of CSIRO technical reports which are available at http://www.clw.csiro.au/publications/technical2002/ or through contacting the authors. Consequently only a brief overview of the model is included here and in particular where this regional application differs from the previous NLWRA work.

Sediment Delivery through the River Network The basic unit of calculation for constructing the sediment budgets is a link in a river network. A link is the stretch of river between any two stream junctions (or nodes; Figure 3). Each link has an internal subcatchment, from which sediment is delivered to the river network by hillslope and gully erosion processes. The internal catchment area is the catchment area added to the link between its upper and lower nodes (Figure 3). For the purpose of the model, the internal catchment area of first order streams is the entire catchment area of the river link. Additional sediment is supplied from bank erosion along the link and from any tributaries to the link.

4

1 3 2

1

1

1

Figure 3: A river network showing links, nodes, Shreve magnitude of each link (Shreve, 1966) and internal catchment area of a magnitude one and a magnitude four link. Sediment is processed sequentially through the river network beginning with first order links and terminating at the basin outlet (commonly the ocean or a major river such as the Murray River). The sediment load (yield at the outlet) for each link is calculated from the supply of sediment from tributary links and the local watershed, less losses through floodplain deposition (fine sediment), bed deposition (coarse sediment), and reservoir deposition (coarse and fine sediment). The branching network of streams for the Goulburn and Broken catchments (Figure 1) was built from a 20 m DEM (digital elevation model) (source: DNRE). The DEM was found to contain significant errors which were particularly evident in the gently sloping lowlands and caused significant misalignments of streams. Consequently a technique was developed to edit DEM elevations where errors occurred in order to generate the correct river network. The river network was defined as beginning at a catchment 2 area of 20 km . This area was selected to limit the number of links across the assessment area, while providing a good representation of the river network. The physical stream network extends upstream of the limit in most areas and these areas are treated as part of the internal catchment area contributing material to the river link. 10

Two anabranches occur along the Broken River from which flow is diverted down streams into the Broken Creek which flows separately into the Murray River. These anabranches are edited into the river network. For each anabranch a distributary ratio is assigned between 0 and 1 according to the portion of mean annual flow that is diverted down the anabranch. This ratio is determined from gauging records of daily flow. The sediment and nutrients loads entering the anabranching node are also split according to this ratio. The coarse sediment (bedload) budget is illustrated in Figure 4. The main aim of the bedload budget is to predict the formation of sand slugs. These are predicted to occur when there is an excess of sediment supply to a river link beyond the capacity of the link to transport bedload. This is known as the sediment transport capacity (STC) and is based on Yang’s (1973) relationship to unit stream power (Equation 1).

STC x =

86S x

1.3

∑Q

1.4 x

(1)

ωwx 0.4 Riverbank erosion (t/y)

Tributary supply (t/y)

Gully erosion (t/y)

STC (t/y) Downstream yield (t/y)

If loading < capacity

If loading > capacity

capacitycpacitycapacity • no deposition

• deposit excess • yield = capacity

• yield = loading

Figure 4: Conceptual diagram of the bedload sediment budget for a river link. STC is the sediment transport capacity of the river link, determined by Equation 1.

Sediment transport capacity is a function of the river width (wx), slope (Sx), discharge (Qx) settling velocity 1.4 of the bedload particles (ω) and hydraulic roughness of the river. ΣQx represents mean annual sum of 1.4 -1 daily flows raised to a power of 1.4 (Ml y ). Using Yang's (1973) equation, and an average value for -1 Mannings roughness coefficient of 0.025, enabled prediction of STC in a river link (t y ). The value of ω was determined for particles with a mean diameter of 2 mm, being the average size observed for sediment slug deposits (Rutherfurd 1996). All coarse sediment entering reservoirs and lakes is deposited. The suspended sediment budget is illustrated in Figure 5. The main aim of this budget is to predict the export of suspended sediment after loss of sediment on floodplains and in lakes and reservoirs. Sediment deposition in reservoirs is calculated in the model as a function of the mean annual inflow into the reservoir and its total storage capacity (Heinemann, 1981). For the Goulburn and Broken catchments the extent of floodplain alongside rivers was determined from a detailed floodplain map equivalent to a 1 in 25 year flood event (FLOODWAY25, source DNRE and Goulburn-Broken CMA). A relatively simple model of floodplain deposition is implemented in SedNet. Floodplain deposition in this case is simply the proportion of sediment that goes overbank and settles out during a typical flood. It is calculated as the ratio of the median overbank flow above bank full discharge multiplied by the proportion of sediment that would be expected to settle out during overbank flow (see Figure 5). Particle settling is a function of the residence time of water on the floodplain. The longer that water covers the floodplain the greater the proportion of the suspended load that is deposited. The residence time of water on 11

floodplains increases with floodplain area and decreases with floodplain discharge. This simple model of floodplain deposition assumes a uniform sediment concentration and that the majority of suspended sediment is transported at times of high river flow.

Hillslope erosion (t/y) Tributary supply (t/y)

HSDR

Riverbank erosion (t/y)

Floodplain

Gully erosion (t/y)

Af

Downstream yield (t/y)

 vA  −  fx  Qfx  1 − e  Qfx  Dx = I x Qtx   

Figure 5: Conceptual diagram for the suspended sediment budget of a river link. HSDR is hillslope sediment delivery ratio. The equation is for the amount of sediment deposited on the floodplain (t/y), where Ix is the sediment load input to the link, Qfx/Qtx is the proportion of flow that goes overbank, Afx/Qfx is the ratio of floodplain area to floodplain discharge and ν is the sediment settling velocity.

Contribution of Suspended Sediment to the Murray River The differentiation of sub-catchments which contribute strongly to total river sediment export is an important aspect of catchment management as this enables catchment managers to target areas for rehabilitation. It is not always possible, or sensible, to implement erosion control works effectively across large areas. Not all suspended sediment delivered to rivers is exported from the basin as there are extensive opportunities for floodplain deposition along river courses. There are usually strong spatial patterns in sediment delivery to basin outlets because some tributaries are confined in narrow valleys with little opportunity for deposition, while others may have extensive open floodplains. There will also be strong, but different patterns in sediment delivery to streams. Thus a map of contribution to export may be very different to a map of erosion. The contribution that each sub-catchment makes to sediment export can only be calculated once the mean annual suspended sediment export is known. The sub-catchments are the link internal areas described in Figure 3. The method tracks back upstream calculating from where the sediment load in each link is derived. The calculation takes a probabilistic approach to sediment delivery through each river link encountered on the route from source to sea. Each internal link catchment area delivers a mean annual load of suspended sediment (LFx) to the river network. This is the sum of gully, hillslope and riverbank erosion delivered from that sub-catchment. The sub-catchment delivery and tributary loads constitute the load of suspended sediment (TIFx) received by each river link. Each link yields some fraction of that load (YFx). The rest is deposited. The ratio of YFx/TIFx is the proportion of suspended sediment that passes through each link. It can also be viewed as the probability of any individual grain of suspended sediment passing through the link. The suspended load delivered from each sub-catchment will pass through a number of links on route to the catchment mouth. The amount delivered to the mouth is the product of the loading LFx from the sub-catchment and the probability of passing through each river link on the way:

12

CO x = LFx x

YFx YFx +1 YF x x......x n TIFx TIFx +1 TIFn

(2)

where n is the number of links on the route to the outlet. Dividing this by the internal catchment area expresses contribution to outlet export (COx) as an erosion rate (t/ha/y). The proportion of suspended sediment passing through each river link is ≤ 1. A consequence of Equation 2 is that all other factors being equal, the further a sub-catchment is from the mouth, the lower the probability of sediment reaching the mouth. This behaviour is modified though by differences in source erosion rate and deposition intensity between links.

Nutrient Delivery through the River Network The nutrient budget model (Annual Network Nutrient Export - ANNEX) predicts the average annual loads of phosphorous and nitrogen in each link in a river network in a similar way to SedNet, with which it is run in conjunction (see Young et al., 2001 for model details). The model considers only the physical (not biological) stores of nutrients in the river system, and is also primarily concerned with the physical nutrient transport processes. It does, however, consider denitrification - a biological process resulting in loss of N to the atmosphere, and phosphorous adsorption-desorption, a physical process influenced by biological activity. The main source terms are hillslope erosion, gully erosion, riverbank erosion, dissolved loads in runoff water and point sources (Figure 6). As with SedNet, the model then routes nutrient loads through the river network estimating the losses associated with floodplain and reservoir deposition and in stream denitrification.

Land use map

Hillslope erosion (t/y)

Soil nutrient conc (kg/t)

HSDR, nutrient enrichment

Riverbank erosion (t/y)

Nutrient concentration (kg/ML)

Gully erosion (t/y)

Bank nutrient conc. (kg/t)

Runoff volume (ML/y) Point sources

Tributary yield (t/y)

Deposition, P equilibration, denitrification Downstream yield (t/y)

Figure 6: Conceptual diagram for the nutrient budget of a river link. HSDR is hillslope sediment delivery ratio.

Model Inputs River Hydrology and Channel Form SedNet combines a number of hydrological parameters into the calculation of river sediment budgets. As such, the correct representation of river hydrology is important for routing sediment and nutrients through 13

the river network. The parameters need to be predicted (interpolated) for each river link across the river basin. The variables used are: • The mean annual flow (Qa); • The median daily flow (Qmd) used in the nutrient budget; • The mean annual sum of Q

1.4

for calculating mean annual sediment transport capacity;

• The bank full discharge (Qbf); • A representative flood discharge for floodplain deposition (in this case median overbank flow – Qob). Values for these variables were derived from the time series of daily flows for 26 gauging stations (source: Victorian Water Resources Data Warehouse) within the Goulburn Broken catchments. In general, only gauging stations with reasonably long discharge records (ie., > 20 years) were selected for analysis. Each hydrological variable was regionalized across the catchments by developing a simple 2 empirical rule with catchment area (A in km ) and mean annual rainfall (R in mm) using standard linear least-squares fits on log transformed data points. Gauging stations lying on streams with unregulated flow conditions (eg., no major extractions, or reservoirs occur upslope) were used to determine the regionalizations for unregulated flow as below:

Qa = 3.56 × 10 −5 × A1.087 × R 2.205

∑Q

1.4

R2 = 0.97

(< Qbf ) = 1.531 × 10 −5 × A1.515 × R 2.346 R2 = 0.97

Qbf = 8.65 × A0.964 × R 0.112

R2 = 0.92

Qob = 6.44 × A0.872

R2 = 0.80

Qmd = 1.413 × 10 −14 × A1.125 × R 4.296

R2 = 0.89

It was necessary to also predict regulated flow parameters for river links below Lake Eildon (Figure 1) and Lake Nagambie as both have a major impact on river flows. This was undertaken by back-calculating the effective catchment area that would be required to give the observed flow parameter at each gauging station using the regionalizations for unregulated flow. The average effective reduction in catchment area is then calculated for reaches with similar flow regulation prior to interpolation of the respective flow parameter within SedNet. For example, the effective reduction in catchment area required to produce the 2 observed (regulated) mean annual flows (Qa) along the Goulburn River decreased from 1050 km below 2 2 Eildon, to 7420 km below Nagambie, to 10240 km from Broken Confluence to the main outlet on the Murray River. Gauging records were also used to determine distributary ratios along the Broken River. In this way approximately 10% of the mean annual flow of the Broken River above Casey Weir (near to Lake Mokoan) is diverted down the Broken Creek channel. A second distributary occurs further down Broken River at Gowangardie Weir where flows enter Pine Lodge Creek. No flow data was available for this distributary and as such a default value of 1% was assumed. The calculations of bank full discharge and median overbank flow are based on the average flood recurrence interval determined from the time series of daily flows recorded at rated gauging stations and existing bank heights. A total of 11 cross-sections were examined across the catchment and these indicated a bank full discharge ranging from 0.5 to 20 years and averaging 5 years.

14

80

Channel width (m)

70 y = 6.0367x 0.2591 R2 = 0.6605

60 50 40 30 20 10 0 1

10

100

1000

10000

100000

100

1000

10000

100000

12

Bank height (m)

10

y = 1.3258x 0.1521 R2 = 0.3646

8 6 4 2 0 1

10

Contributing area (km2)

Figure 7: Distribution of average bank heights and channel widths in relation to upslope contributing area for 48 surveyed sites.

The channel characteristics of width and bank height are used in SedNet for a number of calculations (eg., bank erosion, sediment transport capacity, flood frequency). It is therefore necessary to estimate channel form for all river links. This is done by regionalizing point measurements of bank height and channel width on the basis of upslope contributing area. This procedure produces an average estimate which is applied equally to all stream links on the basis of contributing area. Coefficients of variation of 71 and 81% for bank height and channel width, respectively, demonstrate how spatially variable channel form is. For the Goulburn and Broken Rivers, surveyed cross-sections were used to assess bank height and channel width at points along the river network (Figure 7). These were obtained from surveys taken at 19 of the river gauging stations (Theiss Environmental) together with a further 29 very detailed channel surveys undertaken by Goulburn Broken CMA. In the case of the more detailed surveys bank heights and channel widths represent the average of measurements taken over the length of the channel surveyed.

Hillslope Erosion Hillslope erosion from sheet and rill erosion processes is estimated using the Revised Soil Loss Equation (RUSLE; Renard et al., 1997) as applied in the NLWRA (Lu et al., 2001). The RUSLE calculates mean -1 -1 annual soil loss (Y, tonnes ha y ) as a product of six factors: rainfall erosivity factor (R), soil erodibility factor (K), hillslope length factor (L), hillslope gradient factor (S), ground cover factor (C) and land use practice factor (P): (3)

Y = RKLSCP 15

For soil erodibility (K) and rainfall erosivity (R) we used the NLWRA data (Lu et al., 2001). The length and slope factors (L, S) across the Goulburn and Broken River catchments were derived directly from the high resolution 20 m digital elevation model (DEM). Improved regional land use data (area with annual rainfall > 650 mm only) was also used in the calculation of mean monthly cover factors. Land use codes in this survey were reassigned to one of 20 groups for assessment of subfactors before calculation of the resultant soil loss estimates (Figure 8). The delivery of sediment to streams from sheet and rill erosion on hillslopes is modified by the hillslope sediment delivery ratio (HSDR). HSDR is determined by calibration of hillslope erosion from runoff plots against stream sediment yields. A average value of 5% was found in the NLWRA to be typical of hillslopes across the region covered by the Goulburn Broken catchment and this was applied to all stream links and watersheds in the present study. All sediment produced by sheet and erosion is assumed to contribute to the suspended sediment load of rivers.

Gully Erosion The spatial pattern of gullies in the Goulburn and Broken catchments was derived directly from a gully map of Victoria produced by Lindsay Milton and others in the 1960’s (Ford et al., 1993). Measured gully -2 2 lengths were converted to average densities (km km ) for grid cells of 10 x 10 km (see Figure 9) before implementation in SedNet. The methods used in generating average gully densities from gully maps are described in Hughes and Prosser (2002). -1

In SedNet, gully densities are converted to a sediment supply (kt y ) by multiplying gully density by sub2 2 catchment area (km ), average gully cross-sectional area of gullies (10 m ), average dry bulk density of -3 eroding materials (1.5 t m ), and then dividing by the time over which gullies have been active. In this case the gullies are assumed to have been active since the mid 1850’s (ie., a period of 150 years) when many of the gullies developed as a consequence of land use development and forest clearance. Details of the assumptions used in the calculation of erosion rates can be found in NLWRA technical reports. Sediment generated from gullies contributes to both suspended and bedload sediment. Because of the coarse texture of many of the soil types where gullies occur in the Goulburn and Broken catchments (ie., granitic terrain) we have assumed that on average 70% of eroding sediment contributes to bedload and 30% contributes to the suspended sediment load.

River Bank Erosion River bank erosion is modeled as there are generally few direct measurements of bank erosion over the length of individual river reaches. A global review of river bank migration data (Rutherfurd, 2000) -1 suggested the best predictor of bank erosion rate (BE; my ) to be bank full discharge (Q1.58) equivalent to a 1.58 recurrence interval flow. It was, however, found that this model overestimated the amount of bank erosion along the lower reaches of the Goulburn River by calibration with river loads measured at gauging stations, and by comparison with descriptive reports. Rutherfurd also found a significant relationship between bank erosion and stream power ( ρgQ x S x ) -1

where p is the density of water, g is the acceleration due to gravity, Qx is the mean annual flow (ML y ) and Sx is the energy slope normally approximated to channel gradient. Subsequently a model based on stream power was used to estimate bank erosion for the Goulburn and Broken catchments (Equation 4).

BE = 0.00002 × ρ × g × Q x × S x (1 − PR )(1 − e −0.008 Fx )

16

(4)

NATHALIA

!

´

KATAMATITE

!

SHEPPARTON

! TATURA !

BENALLA

!

MURCHISON

!

EUROA

!

NAGAMBIE

!

SEYMOUR

!

MANSFIELD

! YEA

!

ALEXANDRA

!

EILDON

!

KILMORE

!

JAMIESON

!

BUXTON

! Legend Hillslope Erosion (t/ha/y) 0 - 0.1 0.1 - 0.5 0.5 - 1 1-5 5 - 10 > 10

.

0

25

Kilometers 100

50

Figure 8: Predicted hillslope erosion hazard in the Goulburn and Broken River catchments.

17

NATHALIA

´

KATAMATITE

SHEPPARTON TATURA BENALLA MURCHISON EUROA

NAGAMBIE

SEYMOUR

MANSFIELD

YEA

ALEXANDRA EILDON JAMIESON

KILMORE BUXTON

Legend Gully Density (km/km2) 0 - 0.01 0.01 - 0.1 0.1 - 0.5 0.5 - 1 >1

0

25

Kilometers 100

50

Figure 9: Average density of gully erosion for 10 x 10 km grid cells for the Goulburn and Broken River Catchments. 18

NATHALIA

!

´

KATAMATITE

!

SHEPPARTON

! TATURA !

BENALLA

!

MURCHISON

!

EUROA

!

NAGAMBIE

!

SEYMOUR

!

MANSFIELD

! YEA

!

ALEXANDRA

!

EILDON

!

KILMORE

!

JAMIESON

!

BUXTON

!

Legend % Riparian Vegetation 0 - 0.20 0.21 - 0.40 0.41 - 0.60 0.61 - 0.80 0.81 - 1.00

0

25

Kilometers 100

50

Figure 10: Mapped amount of intact riparian vegetation.

19

NATHALIA

!

´

KATAMATITE

!

SHEPPARTON

! TATURA !

BENALLA

!

MURCHISON

!

EUROA

!

NAGAMBIE

!

SEYMOUR

!

MANSFIELD

! YEA

!

ALEXANDRA

!

EILDON

!

KILMORE

!

JAMIESON

!

BUXTON

!

Legend Bank Erosion (m/yr) 0.000 - 0.001 0.002 - 0.01 0.01 - 0.05 0.05 - 0.1 > 0.1

0

25

Kilometers 100

50

Figure 11: Predicted bank erosion.

20

Bank erosion is modified by the proportion of riparian vegetation (PR) and floodplain width (Fx). Where there is 100% riparian vegetation cover then no bank erosion occurs. Similarly for very narrow or absent floodplains then the model produces little or no bank erosion. This is component of the model is necessary to take into consideration narrow or rocky gorges where there is often little bank erosion due to the predominance of resistant bedrock materials. The average proportion of riparian vegetation (Figure 10) was assessed from a combination of the Victorian In-Stream Condition (ISC) Index (data supplied by the Department of Natural Resources and Environment) and the land use cover map of the catchment which contained mapped areas of Crown Frontages and Parks and Reserves. In these areas riparian vegetation is considered to be intact (ie., 100%). -1

The predicted bank erosion rate (Figure 11) is converted into sediment supply (kt y ) by multiplying BE by -3 channel length (m), bank height (m), average particle density of bank materials (1.5 t m ) and dividing by a conversion factor of 1000. Sediment generated from bank erosion contributes to both suspended and bedload sediment supply. In view of the lack of spatial information about bank particle size distributions, a 50:50 split was assumed as per the NLWRA for contribution to the fine and coarse sediment budgets

Nutrient Sources – Total P and N The nutrient load from hillslope erosion is calculated as the product of the hillslope sediment yield (hillslope erosion x HSDR) multiplied by the nutrient concentration of this load (NC). The nutrient concentration of the sediment load is determined from the proportion of clay and nutrient concentration of the bulk soil (SC). ANNEX uses a two-part mixing model that assumes all nutrients are associated with the clay fraction. For internal catchment links where the percentage clay is greater than the HSDR, all sediment delivered to the channel is assumed to be clay. The nutrient concentration is then the bulk soil concentration divided by (‘enriched’) the proportion of clay (Cp) in the hillslope soil (Equation. 5). For Cp > HSDR,

NC =

SC Cp

(5)

In the unlikely cases where the proportion of clay is less than the HSDR, only a portion of the delivered sediment is clay and so the nutrient concentration (Equation 5) is reduced by the ratio of the proportion of clay to the HSDR. Data on soil clay proportions and nutrient concentrations for P and N were extracted from the Australian Soil Resource Information System (Bui et al., 2001). The loads from riverbank and gully erosion are calculated as the product of their respective sediment -1 yields times the soil nutrient concentration, which for phosphorous was taken to be 0.25 g kg and for -1 nitrogen 1 g kg . Estimation of dissolved loads due to runoff differs from that used in the NLWRA project. In this instance, the dissolved load of a sub-catchment is determined as the product of the mean nutrient concentration in runoff multiplied by the mean annual volume of runoff. The volume of runoff represents the increase in discharge between the inlet and outlet of each stream link. The nutrient concentration is the average of area weighted nutrient concentrations for the dominant land uses within the sub-catchment area (Figures 12 and 13). Since an average concentration in runoff is used in the model, no distinction is made between surface and subsurface runoff volumes. The concentration of soluble N and P was assessed from gauging station records (source: Victorian Water Resources Data Warehouse) for regions with dominant land uses (Table 1). Data for major drains in irrigation areas comes from a 5 year monitoring survey (source: Goulburn Murray Water, 2002). Nutrient loads in urban areas was assessed from storm water runoff (Wong et al., 2000).

21

NATHALIA

!

´

KATAMATITE

!

SHEPPARTON

! TATURA !

BENALLA

!

MURCHISON

!

EUROA

!

NAGAMBIE

!

SEYMOUR

!

MANSFIELD

! YEA

!

ALEXANDRA

!

EILDON

!

KILMORE

!

JAMIESON

!

BUXTON

! Legend Dissolved N input (ug/L) 0 - 100 100 - 500 500 - 1,000 1,000 - 1,500 > 1500

0

25

Kilometers 100

50

Figure 12: Pattern of dissolved N input to streams.

22

NATHALIA

!

´

KATAMATITE

!

SHEPPARTON

! TATURA !

BENALLA

!

MURCHISON

!

EUROA

!

NAGAMBIE

!

SEYMOUR

!

MANSFIELD

! YEA

!

ALEXANDRA

!

EILDON

!

KILMORE

!

JAMIESON

!

BUXTON

! Legend Dissolved P input (ug/L) 0 - 10 10 - 50 50 - 100 100 - 500 500 - 1,000

0

25

Kilometers 100

50

Figure 13: Pattern of dissolved P input to streams.

23

Table 1: Average concentrations of N and P in runoff from selected land uses within the Goulburn and Broken River Catchments. Land use

Percentage of catchment

Forests Grazing Rural/Residential Cropping Non-irrigated dairy Irrigated fruit/veges Irrigated crops Irrigated improved pasture Irrigated diary Irrigated annual pasture Urban

30 52 1 5.6 0.05 0.3 1.3 4 0.04 3.5 0.4

Soluble N concentration (ug L-1) 287 510 800 500 700 2250 1350 1125 600 1400 3450

Soluble P concentration (ug L-1) 4 8 8 22 210 250 320 500 512 750 605

Point sources can also produce significant inflows of nutrients to rivers. For this reason Total N and P in loads in the effluent discharge from sewage treatments plants needs to be taken into consideration. Within the Goulburn and Broken catchments a total of 13 significant points sources were identified (Goulburn-Broken Water Quality Working Group Report, 1995). The load from each point source was treated as an input to the stream link nearest the respective town locality. Measured loads for point -1 -1 sources ranged from 1500 to 96400 kg yr for Total N and from 430 to 22300 kg yr for Total P.

Disaggregation of Mean Annual Loads to Daily Loads Ecologists often consider daily sediment and nutrient loads to be of more importance than mean annual loads for assessing the impacts of water quality on rivers and estuaries. Estimates of mean annual loads produced by SedNet can be disaggregated into mean daily loads if the relationship (ie., rating curve) between concentration and flow is known for river links. Within the Goulburn and Broken catchments suspended sediment has been sampled at many of the gauging stations since the early 1970’s although sampling tends to me more comprehensive over the last decade. In most cases these are non-event based samples taken on an irregular basis. Only those sites having a reasonable sample size and coverage in terms of the range of flow conditions were subsequently selected for the development of rating curves. These included 14 on unregulated streams and a further 10 stations along regulated channels. A modified form of the standard rating curve (Equation 6) was adopted to determine the relationship between suspended sediment concentration (C) and daily flow (Qd).

C = c + aQd

b

(6)

where a, b, are coefficients determined from linear regression of log transformed data and c is a coefficient representing the mean concentration at low flow. The coefficients a and b were determined from samples taken at high flows only. It was found necessary to exclude concentrations taken at low flows from the regression as they caused significant bias and underestimation of the predicted suspended sediment concentration at high flows. Mean daily load (Ld) passing each gauging station is represented by Equation 7.

Ld = c × Qd + a × Qd

(1+ b )

(7)

Since the first term in this equation contributes little to total load, then the mean annual load (La) can be represented as the sum of daily loads for the time series of daily flows (Equation 8). 24

La =

365 i =n (1+ b ) aQdi ∑ n i =1

(8)

where n is the number of days for which flows have been recorded. Hence, in the case where the exponent b has been determined from rating curves (via Equation 6), then the coefficient a can be calculated using Equation 9. This equation enables the disaggregation of mean annual loads predicted by SedNet into mean estimated daily loads for all stream links in the river system.

a=

La × n i =1

365∑ Qdi

(9) (1+ b )

i =n

Routine monitoring of Total N and Total P has only been conducted since the mid-1990’s and of these too few samples have been taken at high enough river flows to construct reliable concentration flow rating curves. Present sampling records suggest there is no significant variation in Total N or P for the range of sampled flows at most gauging stations. Consequently, the concentrations of Total N and P were treated as being constant for the range of river flows. Sampling records at some gauging stations showed significant increases in nitrates and nitrites with increasing flow suggesting the possibility of developing rating curves through the separate treatment of dissolved and sediment attached components. This is, however, outside the scope of the present study.

Results and Discussion Hillslope Erosion Hazard The pattern of hillslope erosion (Figure 8) shows that the majority (89%) of the Goulburn and Broken -1 -1 -1 -1 catchment has low soil loss rates of < 0.5 t ha y . The areas with soil loss rates > 0.5 t ha y relate primarily to areas of high slope on agricultural lands. Comparatively few areas have high erosion rates -1 -1 exceeding 5 t ha y and these usually relate to steep slopes on areas mapped as grazing land. Some of the land lying to the east of Seymour and south of Euroa falls into this category. Elsewhere areas with high erosion rate relate to cropping land such as to the east of Shepparton and north of Benalla. Overall, however, cropping land is predicted to have relatively low sheet and rill erosion rates due to the low rainfall and negligible slope of land in areas where cropping is undertaken. A number of spots had very -1 -1 high erosion rates > 50 t ha y and these are considered to relate to minor errors in the grid analysis. 6

It was found that about 1.04 x 10 tonnes of soil is moved annually on hillslopes in the Goulburn and -1 -1 Broken catchments. This equates to an average soil erosion rate of 0.46 t ha y , about half the previous estimate of the National Land and Water Resources Audit (NLWRA) project. This reflects the improved quality and accuracy of input information to USLE factors such as the regional land use map, and improved slope angle and slope length measurements derived from the 20 m DEM. The values of hillslope erosion represent local movement of soil on hillslopes. By applying the 5% HSDR discussed earlier, the total input of sediment from sheet and rill erosion to streams is estimated to be -1 about 52 kt y .

Gully Erosion Hazard The spatial pattern of gully density (Figure 9) shows that extensive areas of the Goulburn and Broken catchments have undergone gully erosion in the past. Overall, 37% of the catchment area has an -2 average gully density exceeding 0.1 km km while 38% has little or no gully erosion. The highest density of gullies is in areas to the west and south west of Seymour and in areas trending northeast from Seymour through to the area near Benalla. In these areas, gully densities are typically between 0.1 and 1 -2 -2 km km . Gully densities rarely exceed 1 km km .

25

A typical gully is considered to have on average an eroded depth of 2 m and a width of 5 m giving a total 3 sediment yield of 10,000 m per kilometre of channel. Thus, areas with a moderate gully density (0.5 km -2 -1 -1 -1 km ) would have produced a total sediment yield of approximately 75 t ha , equivalent to 0.5 t ha y -3 assuming a sediment density of 1.5 t m and 150 years of activity. While this amount is similar to the average amount of sediment produced by hillslope erosion, the main difference is that all sediment produced by gullies, enters the river network. 5

Gully erosion removes 4.36 x 10 tonnes of soil annually from watershed areas in the Goulburn and Broken catchments. This is 75 % of the level of gully erosion estimated for the NLWRA.

Riverbank Erosion The lower Goulburn River is known to have undergone significant lateral migration in historical times with -1 -1 locally high erosion rates of up to 0.5 m yr and an average rate for reaches of about 0.05 m yr (Erskine et al., 1993). The combined effect of reduction in flood peaks through construction of dams and water offtakes and improved riparian condition along some reaches of the river are consistent with lower rates of bank erosion observed in recent decades. Some sections of the river now show little or no bank erosion. One of the two main factors controlling riverbank erosion in the model is the extent of riparian vegetation as shown in Figure 10. In general the majority of the river network has a reasonable riparian condition with better than 40% riparian cover. Poorest riparian cover tends to occur in an area running from the southwest of Seymour through Euroa to the northeast of the catchment. Here average riparian cover along streams lies in the range 20 to 60%. Areas with good riparian cover (>60%) include forested watersheds in the south east and well established riparian margins along the main rivers, particularly below Shepparton. -1

Predicted bank erosion (m y ) is a function of stream power, riparian condition and floodplain width (Figure 11). Results of the model suggest that bank erosion rates are at their greatest along the section of the Goulburn River from Eildon to just below Seymour where a combination of high stream power and poor riparian cover along some reaches makes banks particularly susceptible to erosion. Rates between -1 0.2 and 0.5 m y are predicted for these reaches, consistent with maximum observed rates for the -1 Goulburn River (Erskine et al., 1993). These rates are equivalent to between 1500 and 3750 t y of sediment being input per kilometre of the river network. -1

-1

-1

Elsewhere bank erosion rates of 0.002 to 0.01 m y (9 - 45 t y km ) are predicted along tributary -1 streams where poor riparian cover occurs. Negligible bank erosion ( 100000

0

25

Kilometers 100

50

Figure 18: Modelled sediment transport capacity.

35

NATHALIA

!

´

KATAMATITE

!

SHEPPARTON

! TATURA !

BENALLA

!

MURCHISON

!

NAGAMBIE

!

EUROA

!

SEYMOUR

!

MANSFIELD

! YEA

!

ALEXANDRA ! EILDON

!

KILMORE

!

JAMIESON

!

BUXTON

!

Legend Bedload Accumulation (m/yr) 0 - 0.1 0.1 - 0.3 0.3 - 1 1-2 >2

0

25

Kilometres 100

50

Figure 19: Predicted bedload deposition.

36

NATHALIA

!

´

KATAMATITE

!

SHEPPARTON

! TATURA !

BENALLA

!

MURCHISON

!

EUROA

!

NAGAMBIE

!

SEYMOUR

!

MANSFIELD

! YEA

!

ALEXANDRA

!

EILDON

!

KILMORE

!

JAMIESON

!

BUXTON

! Legend Total P load (t/yr) 0.2 - 1 1 - 10 10 - 50 50 - 100 100 - 150 150 - 200 200 - 270

0

25

Kilometers 100

50

Figure 20: Predicted Total P load. 37

NATHALIA

!

´

KATAMATITE

!

SHEPPARTON

! TATURA !

BENALLA

!

MURCHISON

!

EUROA

!

NAGAMBIE

!

SEYMOUR

!

MANSFIELD

! YEA

!

ALEXANDRA

!

EILDON

!

KILMORE

!

JAMIESON

!

BUXTON

!

Legend Dissolved P:Total P 0 - 0.2 0.2 - 0.4 0.4 - 0.6 0.6 - 0.8 0.8 - 1

0

25

Kilometers 100

50

Figure 21: Predicted proportion of dissolved P to Total P load.

38

NATHALIA

!

´

KATAMATITE

!

SHEPPARTON

! TATURA !

BENALLA

!

MURCHISON

!

EUROA

!

NAGAMBIE

!

SEYMOUR

!

MANSFIELD

! YEA

!

ALEXANDRA

!

EILDON

!

KILMORE

!

JAMIESON

!

BUXTON

! Legend Total N load (t/yr) 0.02 - 10 10 - 100 100 - 250 250 - 500 500 - 1000 1000 - 2000 2000 - 2400

0

25

Kilometers 100

50

Figure 22: Predicted Total N load.

39

NATHALIA

!

´

KATAMATITE

!

SHEPPARTON

! TATURA !

BENALLA

!

MURCHISON

!

EUROA

!

NAGAMBIE

!

SEYMOUR

!

MANSFIELD

! YEA

!

ALEXANDRA

!

EILDON

!

KILMORE

!

JAMIESON

!

BUXTON

!

Legend Dissolved N:Total N 0 - 0.2 0.2 - 0.4 0.4 - 0.6 0.6 - 0.8 0.8 - 1

0

25

Kilometers 100

50

Figure 23: Predicted proportion of dissolved N to Total N load.

40

Contribution to Suspended Sediment Export to the Murray River The sediment budget predicts that 42% of suspended sediment and < 1% of bedload delivered to the river network in any year is exported from the river mouth. While bank erosion contributes evenly to the coarse and fine budgets and gully erosion is estimated to contribute 70% to the coarse sediment budget, by the catchment mouth, bedload makes up only 0.7% of the total sediment export, meeting common observations of suspended load dominance in large rivers (Richards, 1982). Given that around 18% of sediment delivered to streams within the Goulburn and Broken River basins is exported to the Murray River, it can be concluded that increased erosion upstream in a sub-catchment will not result in the same increase at the river mouth. In other words, within the Goulburn and Broken catchments there is a low degree of connectivity between upstream erosion in sub-catchments and contribution to the Murray. Sub-catchments which make a substantial contribution to the export at the Murray are those with high erosion and limited floodplain extent between the source and the Murray. Sub-catchments close to the Murray are more likely to contribute to sediment export because of limited possibilities for that sediment to be deposited along the way. This overall pattern of suspended sediment contribution to export (Figure 24) shows that the Goulburn and Broken catchments can be divided into 3 roughly contiguous regions of land. In lowland areas of the -1 -1 catchment from below Nagambie contribution to export is low (< 0.01 t ha y ) as a result of low erosion rates in this predominantly flat landscape. Above Lake Eildon and south of Yea sub-catchments are similarly making little contribution to export due to a combination of the long travel distance for sediment to the river mouth, low erosion rates under forest, and the high sediment trapping efficiency of the reservoir. The third region comprises a broad band of land trending from the south west to the north east in the direction of Seymour, Euroa and Benalla. Here, contribution to export is typically much higher (0.05 -1 -1 to 0.5 t ha y ) primarily due to much higher erosion rates but also the close proximity to the river mouths. A few sub-catchments along the main river channels where bank erosion rates are high, are -1 -1 making the greatest contribution to export (>0.5 t ha y ). If the goal is to reduce sediment delivery to the Murray River then remedial works can be focussed on particular sediment sources and the land uses and erosion processes found there. Obviously targeting the areas with a disproportionately high level of contribution should be a priority, such as areas of bank erosion. Targeting these areas will obviously have the greatest effect on reducing sediment export to the Murray. However, because the majority of the catchment is still making a significant contribution, it is unlikely that major reductions in suspended sediment loads to the Murray will be achieved, without more widespread rehabilitation of problem areas.

Comparison of Suspended Sediment Loads The preceding sections have concentrated on predictions of the river sediment and nutrient budgets and their implications with little discussion on tests of their accuracy. For the Goulburn Broken Rivers it is possible to directly compare the results from SedNet with load estimates based on suspended sediment samples taken at gauging stations and sediment deposition in Lake Eildon. Furthermore the relatively dense network of gauging stations allows a spatial analysis of uncertainties in the model. Suspended sediment loads were calculated for 24 gauging sites across the catchment by applying sediment rating curves to the maximum length of daily flows measured at the stations (Figure 16). Figure 25 shows one of the better sampling records of suspended sediment taken within the catchment. Equation 10 is the best fit to the spread of data points and provides a rating curve for suspended sediment concentration against daily flow at this station.

C = 3 + 0.031 × Qd 1.07 ,

2

R = 0.47

(10)

41

NATHALIA

´

KATAMATITE

SHEPPARTON TATURA BENALLA MURCHISON EUROA

NAGAMBIE

SEYMOUR

MANSFIELD

ALEXANDRA

YEA

EILDON JAMIESON

KILMORE BUXTON

Legend Contribution of suspended sediment to the Murray River (t/ha/y) 0 - 0.01 0.01 - 0.05 0.05 - 0.1 0.1 - 0.5 0.5 - 1 >1

0

25

Kilometers 100

50

Figure 24: Predicted contribution of suspended sediment to the Murray River for sub-catchments within the Goulburn and Broken River catchments. 42

The rating curve (Equation 10) when applied to the time series of daily flow from 1947 to 2001 results in a mean annual load of 5 kt estimated to flow past this gauging station. The same general trend as shown in Figure 25 was observed for the majority of gauging stations analysed in the Goulburn and Broken catchments. Principal differences were the mean suspended sediment concentration at low flows, which -1 were found to increase from 4 to 13 mg L with increasing catchment area, and the slope of the rating curve (equal to the exponent in Equation 10) which tended to vary between 0.3 and 1.2. While there appeared to be no systematic variation in the exponent in relation to catchment area, as stations lacking sampling at high flows tended to have the lower values, it was found that regulated channels such as the Goulburn river tended to have lower values and therefore, flatter rating curves. 100 90 Cummulative percent load

Suspended sediment concentration (mg/L)

1000

100

10

80 70 60 50 40 30 20 10 0

1 10

100

1000

10

10000

Daily flow (ML)

100

1000

10000

Daily flow (ML)

Figure 25: Suspended sediment rating curve (left) and cumulative load distribution (right) for Delatite River at Tonga Bridge (405214). Care needs to be taken in the interpretation of gauging station samples as there are significant sources of -1 error in the calculation of river loads. For example, the maximum sampled flow (2700 ML day ) in Figure 25 is less than a 1 in 1 year flow event for this site. Furthermore, only 8 samples were taken at flows -1 above 1000 ML day , and no samples were taken at flows above 3000 ML when 47% of the river load is estimated to be transported based on extrapolation of the rating curve to high flows. The poor replication of event based flows, typical of most gauging stations, produces relatively large errors in the estimation of -1 river loads. In this example, the standard error (53 mg L ) in the mean suspended sediment -1 concentration measured at flows above 1000 ML (97 mg L ) suggests an error of at least ± 50%. The standard error of the estimate for the regression curve was not used in this instance because of the large bias caused by the bulk of samples occurring at low flows. Despite these limitations, the predicted load from SedNet for the Delatite River near to Tonga Bridge of 4 -1 kt y compares favourably with the measured load. Figure 26 compares the predicted load plotted against measured load at the 24 stations. Overall, while there is some variation for individual stations which can be partly attributed to errors in measured loads, there is a good agreement between the loads predicted by SedNet and measured loads. There is however a systematic trend evident in Figure 27 with -1 loads being overestimated along the main channels (eg., loads > 10 kt y ) and underestimated along the smaller tributary channels. -1

SedNet suggests a decline in sediment load along the Goulburn River from 143 to 124 kt y from below the junction with the Broken River to the outlet at the Murray River. The same trend is apparent in -1 -1 measured loads which decrease from 100 kt y at Shepparton to 78 kt y at McCoy Bridge. The loss in sediment is largely attributed to overbank deposition and the lack of any significant sediment inflows along the lower section of the River. For the main reach of the Goulburn River below Seymour, SedNet results are on average about 50% higher than measured loads. While this is similar to the ± 50% approximate error in estimation of river loads, all measured loads are below those predicted by SedNet, suggesting a systematic difference. Therefore, either rates of erosion from the three main sources have been over-predicted in the modelling or there has been insufficient flood deposition modelled along the lower floodplain area. Given that the 43

biggest differences occur for sub-catchments dominated by gully erosion (eg., Sugarloaf River and the Goulburn River at Seymour) this suggests there has been a decline in the rate of gully erosion in recent decades. It is important to keep in mind that SedNet predicts long term average sediment loads while measured loads apply to much shorter periods in time. Some of the difference, therefore, can be attributed to temporal variation in the activity of sediment sources. 1000

SedNet modelled load (kt/yr)

100

10

1 1.3

y = 0.5x 2

R = 0.8 0

0 0

0

1

10

100

1000

Measured load (kt/yr)

Figure 26: Comparison of average annual suspended sediment loads predicted from SedNet with estimated loads from gauging station (Figure 16) sediment rating curves (eg. Figure 25). Dotted line is the 1:1 relationship while the solid line is the least squares regression fit. Under prediction of loads by SedNet for the smaller tributary channels could be due to either underestimation of erosion inputs or overestimation of floodplain deposition. Sites showing the greatest difference all occur in sub-catchments dominated by forest. As these all have very limited floodplain extents then underestimation of erosion inputs is the likely cause. At present SedNet does not consider bank erosion in forested watersheds. The implication of the results is that significant levels of bank erosion are being maintained in forested catchments, particularly those in wetter and steeper catchments where stream power would be expected to be higher. Rates of Sediment deposition in Lake Eildon also provides a useful comparison with results from SedNet -1 -1 which predicts that 67 kt y of coarse sediment (100% of load) and 50 kt y of suspended sediment (88% of load) are being stored in the reservoir. This is approximately half the measured deposition (Abrahams, 3 -1 -1 1972) in the reservoir of 211 800 m y , which is equivalent to 212 kt y if an approximate dry bulk -3 density of 1 t m is assumed for lake sediments that are subject to periodic drying. As the predicted -1 suspended sediment export from the reservoir (6.8 kt y ) is close to the measured load based on sediment rating data at the Eildon gauging station (Figure 16), then this suggests that the sediment loads of streams entering the lake have been under predicted, consistent with the comparison of measured loads from gauging stations.

Comparison of Nutrient Loads The estimation of nutrient loads is generally more problematic than for sediment loads as nutrients are mostly sampled at times of low flow and rarely at high flows, let alone during flood events. Relationships between nutrient concentrations and daily flow are usually poor, if not non-existent. Approximate load 44

estimates can be derived from the limited sampling data available at gauging stations by assuming average concentrations of both sediment-bound nutrients and soluble nutrients and then applying these to the estimated mean annual sediment loads and discharges, respectively. In the Goulburn and Broken catchments measurement of nutrients at gauging stations has included total Kjeldahl N, total nitrates and nitrites (NOx), soluble P, and Total P. Sediment-bound P can be determined from the difference between soluble and Total P. The sum of Kjeldahl N and NOx is equal to the Total N concentration. Ammonium concentrations are required to calculate sediment-bound N, and since these have generally not been measured, then only an approximate measure of sediment-bound N can be obtained from the relationship between Kjeldahl N and suspended sediment concentrations. Three gauging sites were selected to compare measure loads against predicted loads (Table 4) for different catchment contributing areas ranging from the larger rivers (Goulburn River at McCoy Bridge) to smaller tributaries (Delatite River at Tonga Bridge). The three sites suggest a similar trend of over prediction of Total N and under prediction of Total P, although differences are probably within the errors involved in calculating nutrient loads from sampling data.

Table 4: Comparison of nutrient concentrations and loads to predicted loads of SedNet at selected gauging stations. Sed = sediment-bound nutrient, Diss = dissolved nutrient. Site

McCoy - 405323

Murchison - 405200

Delatite - 405214

Sampled

Predicted

Sampled

Predicted

Sampled

Predicted

SS Load (kt y-1)

78

130

94

108

5

4

Sed P conc. (g kg-1)

3.2

1.8

2.8

Sed N conc. (g kg-1)

7

8

5

Sed P load (t y-1)

250

180

170

110

14

2.6

Sed N Load (t y-1)

560

490

710

410

25

11

1600000

1430000

923000

920000

125000

105000

Discharge (ML) Diss P conc. (mg L-1)

0.020

0.0046

0.0044

Diss N conc. (mg L-1)

0.7

0.53

0.26

Diss P load (t y-1)

30

50

4

20

0.55

1.8

Diss N load (t y-1)

1140

1740

490

1200

33

43

Total P load (t y-1)

280

230

174

130

15

4.4

Total N load (t y-1)

1700

2330

1200

1610

58

54

Diss P: Total P ratio

0.12

0.22

0.03

0.18

0.04

0.41

Diss N: Total N ratio

0.67

0.75

0.40

0.75

0.57

0.80

When comparing dissolved and sediment-bound loads there is systematic under prediction of sedimentbound nutrient loads and over prediction of dissolved nutrient loads for both N and P resulting in larger predicted ratios of dissolved to Total N or P (Table 4). This suggests that the dissolved concentrations used in nutrient grids (Figures 12 and 13) have been overestimated and that the bulk soil concentrations of eroding sediment may be greater on average in the Goulburn and Broken catchments that the default values (see Page 21) assumed in SedNet.

45

Alternately, the systematic trend could reflect inadequate sampling of the range of river flows, with the average sediment-bound P and N concentrations and dissolved P and N concentrations not being representative of high flow conditions when daily loads are greatest (Figure 27). If increases in dissolved nutrient and decreases in sediment-bound nutrient concentrations with increasing flow were to be assumed (as indicated by the analysis of some, but not all, gauging records), then better agreement would have resulted between the sampled and predicted loads shown in Table 4. There are also a number of other parameters in the model which affect the nutrient concentrations in streams. For example, overestimation of dissolved N could be due to under prediction of the level of denitrification occurring in the Goulburn and Broken Rivers. To meet the observed dissolved N level along the lower Goulburn River would require about twice the level of denitrification to occur than is presently predicted.

Daily Flow (ML)

5000 4000

Daily flow

(