Regional Patterns of Erosion and Sediment and Nutrient Transport in the Mary River Catchment, Queensland

Transcription

1 Regional Patterns of Erosion and Sediment and Nutrient Transport in the Mary River Catchment, Queensland R.C. DeRose, I.P. Prosser, L.J. Wilkinson, A.O. Hughes and W.J. Young CSIRO Land and Water, Canberra Technical Report 37/02, August 2002 CSIRO LAND and WATER

2 Regional Patterns of Erosion and Sediment and Nutrient Transport in the Mary River Catchment, Queensland R.C. DeRose, I.P. Prosser, L.J. Wilkinson, A.O. Hughes and W.J. Young CSIRO Land and Water, Canberra Technical Report 37/02, August 2002

3 Copyright 2002 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

4 Table of Contents Acknowledgments... 3 Abstract... 4 Main Research Report... 5 Background... 5 Project Objectives... 7 Methods... 7 Hillslope Erosion Hazard... 7 Gully Erosion Hazard... 8 River Bank Erosion Hazard... 9 Sediment Delivery through the River Network... 9 Contribution of Suspended Sediment to the Coast River Nutrient Budget Model Hydrology Disaggregation of Mean Annual Loads to Daily Loads Results and Discussion Hillslope Erosion Hazard Gully Erosion Hazard Riverbank Erosion Sediment Sources to the Stream Network Nutrient Sources Sediment Delivery through the River Network Bedload Deposition River Suspended Loads Nutrient Budget Contribution to Suspended Sediment Export at the Coast Calibration of Suspended Sediment Load Disaggregation of Annual to Daily Loads Testing of Land Use Scenarios Comparison with NLWRA Results Conclusions References

5 List of Figures (abbreviated titles) Figure 1: A river network showing links, nodes, Shreve magnitude of each link (Shreve, 1966) Figure 2: Conceptual diagram of the bedload sediment budget for a river link Figure 3: Conceptual diagram for the suspended sediment budget of a river link Figure 4: Conceptual diagram for the nutrient budget of a river link. HSDR is hillslope sediment delivery ratio.. 15 Figure 5: Location map for the Mary River catchment Figure 6: Map showing mean annual rainfall for the Mary River catchment Figure 7: Land use map of the Mary River catchment Figure 8: Predicted hillslope erosion hazard in the Mary River catchment Figure 9: Density of gully erosion based on the Land Disturbance Survey of the Mary River Catchment Figure 10: Mapped amount of intact riparian vegetation Figure 11: Predicted bank erosion Figure 12: Pattern of dissolved N input Figure 13: Pattern of dissolved P input Figure 14: Measured bank height at gauging stations in relation to catchment area Figure 15: Measured channel width at gauging stations in relation to catchment area Figure 16: Regionalisation of the discharge term of streampower (ΣQ 1.4 ) with catchment area for the Mary catchment Figure 17: Predicted bedload deposition Figure 18: Predicted specific suspended sediment load Figure 19: Predicted proportion of dissolved N to Total N for the Mary catchment Figure 20: Predicted contribution of suspended sediment to the coast for sub-catchments within the Mary River catchment Figure 21: Suspended sediment rating curve for the Mary River Figure 22: Comparison of average annual suspended sediment loads predicted from SedNet with loads estimated from gauging stations Figure 23: Variation in predicted mean daily loads from August 1999 to February 2000 for the Mary River Figure 24: Scenario testing in ArcMap List of Tables (abbreviated titles) Table 1: Gully erosion rankings for the Mary Catchment... 8 Table 2: Average concentrations of N and P in runoff from land uses within the Mary Catchment Table 3: Hillslope soil loss from land use categories in the Mary River catchment Table 4: Components of the sediment budget for the Mary River Basin Table 5: Components of the nutrient budget for the Mary River Basin

6 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 Mary River through the use of regional data sources. As such we acknowledge the input from both the Queensland Department of Natural Resources and Mines (NRM) and the Environmental Protection Agency (APA) who provided much of the necessary resource information required to complete this work. In particular we thank Heather Hunter, Andrew Moss, John Bennett, Mark Salloway, Brian Stockwell, Bill Macfarlane, Cia Musgrove and John Ruffini. We also acknowledge the assistance of John Gallant of CSIRO for topographic analyses. We gratefully acknowledge the funding and support provided by NLWRA that made this study possible. 3

7 ABSTRACT This project was carried out to identify the major processes involved in the delivery of sediment and nutrients to rivers within the Mary River catchment. The loss of sediment and nutrients from the land can have impacts downstream on the river and the marine 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 Mary River, 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 a sediment and nutrient budget for the catchment. A sediment budget is an account of the major sources, stores and fluxes of sediment in the 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 sediment 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 catchment. The model results suggest that sheetwash and rill erosion contributes 5% of the total predicted sediment supply to channels. Furthermore, while it can vary by 3 orders of magnitude, only 1.3% of the catchment has high surface erosion potential. Much of this is restricted to steeper slopes on grazing land or to areas of cropping. Cropping land, although only occupying around 7% of the catchment area produces 26% of the total erosion on hillslopes. Erosion rates from grazing lands are not substantially higher than forested land as the latter occur predominantly on steeper slopes under areas of higher rainfall where natural erosion rates would be expected to be greater. Gully erosion is also a significant process contributing approximately 8% of the total predicted sediment supply. Gully erosion is a dominant sediment source in the drier western sub-catchments where grazing of native pastures is the main land use. Riverbank erosion makes the largest overall contribution of sediment to channels with 87% of the total predicted supply being derived from this source. This is a consequence of increased channel capacity and stream power through large historical channel incision and widening. Rapid accumulation of sand and gravel on the bed of rivers can degrade aquatic habitat, but we find that this is not a major concern in the Mary River catchment. Most river reaches are capable of transporting increases in supply of bedload from gully and riverbank erosion. Only in the lower third of the Mary River do we predict significant accumulation of sand and gravel. The sediment budget predicts that 95% of suspended sediment and <1% of bedload delivered to the river network in any year is exported from the river mouth. The large contribution of suspended sediment is due to limited opportunity for storage on floodplains or in reservoirs. This is in contrast to large river systems where much of the sediment remains stored for hundreds to thousands of years. We predict that the mean annual export of suspended sediment to the coast is 450 kt y -1. This figure lies within ± 20% of measured sediment loads based on samples collected at gauging stations along the Mary River. This represents good agreement. The small difference could be accounted for by a higher sediment delivery ratio from hillslopes than was applied in the model (ie., >5%) and a number of minor erosion processes, which are not presently modeled for, including landsliding and channel scour (bed incision). 4

8 Tributary creeks and streams entering the Mary River are predicted to supply 202 kt y -1 of suspended sediment to the main channel. Therefore, of the 445 kt y -1 of sediment exported to the coast, 55% is being derived from erosion within basin areas that adjoin the main river channel. Hence this region of the catchment contributes a disproportionately higher amount of sediment than other regions in the catchment with much of it being derived from bank erosion. Each of the sediment sources described above, together with dissolved contributions from hillslope runoff, deliver nutrients to the network of streams and rivers in the Mary River basin. We predict that 344 t y -1 and 1541 t y -1 of Total P and Total N, respectively, are exported to the coast. This represents 95% of Total P and 74% of Total N supplied to streams. The lower proportion of Total N export is caused largely by instream denitrification. While much of the Total P is transported attached to sediment, a higher proportion of Total N load (53%) is derived as dissolved contribution from hillslope runoff. The patterns of nutrient supply are thus not identical. Relative to Total P, a higher proportion of Total N is being derived as dissolved load from intensively managed agricultural land (cropping, improved grazing land) or where the sediment supply is low (forested land). 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 Mary River 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, 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 of achieving an 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 low input farming systems such as extensive beef grazing, 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 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 is derived from, where it is 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 sediment 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 5

9 management or even undertaking remediation measures we need to determine its significance and the 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 chiefly from three types of processes: 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; and 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. The suspended sediment is also deposited in low energy environments. The main depositional environment for suspended sediment along the Queensland coast is the in-shore marine environment of shallow waters, in-shore reefs, and tidal flats. Accelerated deposition in these areas can smother aquatic habitats and can increase turbidity through resuspension of the sediment. Not all suspended sediment delivered to streams is exported to the coast. Much of it is deposited along the way 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 substrate for benthic organisms (Jeffers, 1998). A reconnaissance level sediment budget for the Mary River catchment 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 sub-catchments 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

10 Project Objectives This report constitutes the final phase of the National Land and Water Resources Audit (NLWRA) 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 for the Mary River catchment. 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. Develop the ability to examine future management scenarios and their affect on sediment and nutrient export from catchments. Methods The only practical framework to assess the patterns of sediment and nutrient transport across a large complex area such as the Mary River catchment is a spatial modelling framework. 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 sediment 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. The methods used in each aspect of the spatial model are outlined below in brief. They were developed concurrently with a National Land and Water Resources Audit project on sediment budgets and reference is made to supporting technical documentation which contains details of the approach. Hillslope Erosion Hazard 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 annual soil loss (Y, tonnes ha -1 y -1 ) 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): Y = RKLSCP (1) The factors included in the RUSLE vary strongly across a diverse catchment such as the Mary, providing a method for estimating the spatial patterns of erosion using available spatial information for each factor. The precise form of each factor is based on soil loss measurements on hillslope plots, mainly in the USA. Limited local calibration of the RUSLE factors, particularly the C factor, have been undertaken in some catchments using plot scale measurements of erosion (McIvor et al., 1995; Scanlan et al., 1996). Mean annual values for rainfall erosivity and the cover factor are often used in direct application of Equation (1) to calculate mean annual hillslope erosion. This neglects often important seasonal patterns of rainfall erosivity and cover. High intensity rains for example may be associated with seasons of low ground cover. To incorporate these effects we used the product of mean monthly cover (C m ) and the proportion of annual rainfall erosivity for each month (R m /R). The monthly values of C m R m /R were then summed to give mean annual soil loss. It can be shown that incorporation of seasonal effects reduces 7

11 predicted mean annual soil loss in Australia's tropics by a factor of 1.5. The modifications of Equation (1) discussed above yield Y = KLS m = 12 Cm m= 1 R m R. (2) Twenty years ( ) of daily rainfall data mapped across Australia and 13 years ( ) of satellite vegetation data were used to apply Equation (2). Details of the use of this data are given in Lu et al. (2001). The soil erodibility factor (K) was derived from the Australian Soil Resources Information System (as detailed in Lu et al., 2001). The length and slope factors (L, S) were derived directly from the high resolution 25 m DEM of the Mary River catchment. Land use data was also used in the calculation of mean monthly cover factors. For the Mary catchment the land use cover map was derived from the Land Use, Vegetation Cover, and Land Disturbance Survey of the Mary River Catchment (Pointon, 1998). Land use codes in this survey were reassigned to one of 20 groups (Lu et al., 2001) for assessment of subfactors. The delivery of sediment to streams from sheet and rill erosion on hillslopes is modified by the hillslope sediment delivery ratio (HSDR). The HSDR is usually determined by calibration of hillslope erosion from runoff plots against stream sediment yields. It was found in the NLWRA that an average value of 5% was typical of hillslopes across the region covered by the Mary catchment and this was applied to all stream links and watersheds in the present study. Gully Erosion Hazard The spatial pattern of gullies in the Mary catchment was derived from the erosion survey undertaken as part of the Land Use, Vegetation Cover, and Land Disturbance Survey of the Mary River Catchment (Pointon, 1998). In this survey areas of land (termed UMA s), typically greater than 5 ha in areal extent, were assigned a ranking for intensity of gully erosion as shown in Table 1. Most gully erosion fell into the minor category, with fewer areas of moderate erosion and only 3 very small areas having severe erosion. These rankings were converted to gully densities (km of eroded gully per km 2 of area) by comparison with predicted densities obtained for a national gully density survey (Hughes et al., 2001) over the area of the Mary catchment. The predictions were based on correlation between measured gully densities and environmental variables such as geology, soil texture, mean annual rainfall and terrain-based attributes such as elevation, slope and hillslope length. These reflect the range of expected densities for landform types typical of the Mary catchment. Finally, a gully density map with improved accuracy compared to the NLWRA assessment was produced by assigning these densities to erosion codes in the mapped coverage of the spatial extent of gully erosion. It was assumed, within the errors of the broad catchment sediment budget, that all sediment derived from gully erosion was delivered to the river network. Table 1: Gully erosion rankings for the Mary Catchment. Code Code description Gully density (km km -2 ) no apparent erosion minor occasional areas of erosion (<10% of UMA eroded) moderate common areas of erosion (10-50% of UMA eroded) severe many eroded areas (> 50% of UMA)

12 River Bank Erosion Hazard The supply of sediment from riverbank erosion was calculated from the results of a global review of river bank migration data (Rutherfurd, 2000). The best predictor of bank erosion rate (BE) was found to be bankfull discharge (Q 1.58 ) equivalent to a 1.58 recurrence interval flow. This was modified in the project to account for the condition of riparian vegetation (Pr). BE (1 PR) Q = in m per year (3) The condition of riparian vegetation was taken from the Mary State of the Rivers Report (SOR) (Johnson, 1997). Although locally more accurate surveys exist for selected reaches of the catchment the SOR survey is at present the only basin wide survey. In the SOR survey a riparian condition rating of between 0 and 100 was assigned according to riparian vegetation width and structure. A rating of 81 to 100 for example represents very good condition while 0 to 20 represents very poor condition. Of the 164 subsections across the catchment, three were found to have no assigned rating. These were subsequently assigned a rating according to those of neighbouring streams and evaluation of vegetation cover maps (Pointon and Collins, 2000). Bankfull discharge for the Mary River is on average equivalent to a 1 in 20 year recurrence interval flow owing to the large channel capacity and high banks throughout most of the main reaches. This differs from the rivers typical of those reviewed by Rutherfurd (2000). It was subsequently found that Equation 3 produced unreasonably high levels of sediment supply from bank erosion when compared with measured sediment loads at river gauging stations. Consequently, the coefficient in Equation 3 was reduced from to by calibration against the measured loads and to give more realistic results averaged over the 100 years to which the bank erosion rule is applied. Riverbank erosion, by definition, delivers sediment to the river network. Sediment Delivery through the River Network Hillslope, gully, and stream bank erosion, together supply sediment to the stream network (the network of creeks and rivers in a catchment). The sediment supplied to a reach of river is then either deposited within the river, and its surrounding floodplain, or is transmitted to the next reach downstream. There also may be substantial deposition in reservoirs. 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. A suite of ArcInfo TM 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. Details of the model and its application to regional catchments in Australia are given in Prosser et al., (2001). That document describes all the equations and input data used. Here we give a brief descriptive summary of the approach. 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 coast; and the contribution of each sub-catchment to that export. 9

13 For this project, suspended sediment is characterised as fine textured sediment carried at relatively uniform concentration through the water column during large flows. The main process for net deposition of suspended sediment is overbank deposition on floodplains (eg., Walling et al., 1992). The amount of deposition depends upon the residence time of water on the floodplain and the sediment concentration of flood flows. The residence time of suspended sediment in streams is low, so there is negligible transient deposition of suspended sediment. Suspended sediment is sourced from surface wash erosion of hillslopes, gully erosion and riverbank erosion. The sediment budget is reported as mean annual values for current land use. Bedload is sediment transported in greatest proportions near the bed of a river. It may be transported by rolling, saltation, or for short periods of time, by suspension. Transport occurs during periods of high flow, over distances of hundreds to thousands of metres (Nicholas et al., 1995). Residence times of coarse sediment in river networks are relatively long so there is transient deposition on the bed as the sediment works its way through the river network. In addition to that deposition, an increase in sediment supply from accelerated post-european erosion can cause the total supply of sediment in historical times to exceed the capacity of a river reach to transport sediment downstream. The excess sediment will be stored on the bed and the river will have aggraded over historical times (Trimble, 1981; Meade, 1982). There has been a significant increase in supply of sand and fine gravel to rivers in historical times and deposition of this bedload has formed sand slugs: extensive, flat sheets of sand deposited over previously diverse benthic habitat (Nicholas et al., 1995; Rutherfurd, 1996). The bedload budget aims to predict the formation of these sand slugs. 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 1). 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 1). 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 Figure 1: 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

14 A branching network of river links joined by nodes was defined from the 25 m DEM of the Mary Catchment. The river network was defined as beginning at a catchment area of 10 km 2. This area was selected to limit the number of links across the assessment area, while providing a good representation of the channel 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. Short links, where the catchment area had reached less than 20 km 2 by the downstream node, were removed. Links were further separated by nodes at the entry to and exit from reservoirs and lakes. Internal catchment areas for each link were determined from the 25 m DEM. A sediment budget for bedload was calculated for each river link (x) in the network, working from the top of the basin to the sea (Figure 2). The aim was to define those links subject to net deposition because the historical supply of bedload has exceeded sediment transport capacity. The total mean annual load supplied to the outlet of the link at any time is compared with the mean annual sediment transport capacity at that point. If the load is in excess of capacity, the excess is deposited and the yield to the link immediately downstream equals the sediment transport capacity. If the loading to the outlet is less than the sediment transport capacity there is no net deposition and the yield downstream equals the loading to the outlet. Riverbank erosion (t/y) Gully erosion (t/y) Tributary supply (t/y) STC (t/y) Downstream yield (t/y) If loading < capacity no deposition yield = loading If loading > capacity deposit excess yield = capacity Figure 2: Conceptual diagram of the bedload sediment budget for a river link. STC is the sediment transport capacity of the river link, determined by Equation 3. Bedload is supplied to a river link from tributary links and from gully and riverbank erosion in the internal catchment area of the link. Half the sediment derived from riverbank and gully erosion contributed to the bedload budget and the other half contributed to the suspended load budget. This reflects observed sediment budgets (eg., Dietrich and Dunne, 1978) and the particle size of bank materials. Gully density was converted to a mean annual mass of sediment derived from gully erosion by assuming development of gullies over 100 years and a mean gully cross-sectional area of 10 m 2. Similarly, bank retreat was converted to a mean annual mass of sediment supplied by bank erosion by multiplying Equation 3 by bank height, channel length, and a dry bulk density of 1.5 t m -3. Once calculated, the total supply of bedload to a river link is compared to sediment transport capacity (STC x ). Sediment transport capacity is a function of the river width (w x ), slope (S x ), discharge (Q x ), particle size of sediment and hydraulic roughness of the channel. Yang (1973) found strong relationships between unit stream power and STC. Using Yang's (1973) equation, and average value for Mannings roughness coefficient of 0.025, we predicted sediment transport capacity in a river link (t y -1 ) from: 11

15 STC x 86S 1.3 x ωw 0.4 x Q 1.4 x = (4) where ω is the settling velocity of the bedload particles (m s ), and ΣQ x represents mean annual sum of daily flows raised to a power of 1.4 (Ml 1.4 y -1 ). This represents the disproportionate increase in sediment transport capacity with increasing discharge. 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). The suspended sediment loads of Australian rivers, and rivers in general, are supply limited (Olive and Walker, 1982; Williams, 1989). That is, rivers have a very high capacity to transport suspended sediment and sediment yields are limited by the amount of sediment delivered to the streams, not discharge of the river itself. Consequently, if sediment delivery increases, sediment yields increase proportionally. Deposition is still a significant process, however, and previous work has shown that only a small proportion of supplied sediment leaves a river network (Wasson, 1994). Suspended sediment is supplied to a river link from four sources: river bank erosion, gully erosion, hillslope erosion and tributary suspended sediment yield (Figure 3). Prediction of surface wash and rill erosion was described above but only a small proportion of sediment moving on hillslopes is delivered to streams. The difference occurs for two reasons. First the RUSLE is calibrated against hillslope plots considerably smaller than the scale of hillslopes. Much of the sediment recorded in the trough of the plots may only travel a short distance (less than the plot length and much less than the hillslope length) so that plot results cannot be easily scaled up to hillslope predictions. Second, there are features of hillslopes, not represented by erosion plots, which may trap a large proportion of sediment. These include farm dams, contour banks, depressions, fences, and riparian zones. The most common way of representing the difference between plot and hillslope sediment yields is to apply a hillslope sediment delivery ratio (HSDR) to the RUSLE results (eg., Williams, 1977; Van Dijk and Kwaad, 1998). This ratio represents the proportion of sediment moving on hillslopes that reaches the stream and is generally determined by comparing the results of hillslopes plots against sediment yields from tributary streams. The main location for deposition of suspended sediment is on floodplains. A relatively simple conceptualisation of floodplain deposition is to consider that the proportion of suspended sediment load that is available for deposition is equal to the fraction of total discharge that goes overbank. This assumes uniform concentration of suspended sediment with depth. The actual deposition of material that goes overbank can be predicted as a function of the residence time of water on the floodplain. The longer that water sits on the floodplain the greater the proportion of the suspended load that is deposited. The residence time of water on floodplains increases with floodplain area and decreases with floodplain discharge. Floodplain area was mapped from the 25 m DEM of the Mary catchment using a topographic lowness index developed by Gallant (pers comm.). This index can be matched against terrace and valley floor extents to provide a reasonable map of floodplain extent. An increase in supply of suspended sediment from upstream results in a concomitant increase in mean sediment concentration and mean annual suspended sediment yield. Thus increases to suspended sediment supply have relatively strong downstream influences on suspended sediment loads. Sediment deposition in reservoirs is incorporated in the model as a function of the mean annual inflow into the reservoir and its total storage capacity (Heinemann, 1981). The procedures above were applied in sequence to each river link from the top of the basin to the sea, adding suspended load and predicting its loss through deposition along the way. The final calculation is of mean annual suspended sediment export to the sea. 12

16 Hillslope erosion (t/y) Tributary supply (t/y) HSDR Riverbank erosion (t/y) Gully erosion (t/y) Floodplain Q fx D = x I x 1 e Q tx A f va Q fx fx Downstream yield (t/y) Figure 3: 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 I x is the sediment load input to the link, Q fx /Q tx is the proportion of flow that goes overbank, A fx /Q fx is the ratio of floodplain area to floodplain discharge and ν is the sediment settling velocity. Contribution of Suspended Sediment to the Coast One of the strongest interests in suspended sediment transport at present is the potential river export to estuaries and the coast. Because of the extensive opportunities for floodplain deposition along the way, not all suspended sediment delivered to rivers is exported to the coast. There will be strong spatial patterns in sediment delivery to the coast 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. Differentiation of sub-catchments which contribute strongly to coastal sediment loads is important because of the very large catchments involved in Australia. It is not possible, or sensible, to implement erosion control works effectively across such large areas. The contribution of each sub-catchment to the mean annual suspended sediment export from the river basin was calculated. The sub-catchments are the link internal areas described in Figure 1. The calculations were made once the mean annual suspended sediment export was calculated. 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 (LF x ) 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 (TIF x ) received by each river link. Each link yields some fraction of that load (YF x ). The rest is deposited. The ratio of YF x /TIF x 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 LF x from the sub-catchment and the probability of passing through each river link on the way: YF YF YF CO LF x x x... x x x+ 1 n x = x (5) TIFx TIFx+ 1 TIFn 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 (CO x ) as an erosion rate (t/ha/y). The proportion of suspended sediment passing through each river link is 1. A consequence of Equation 5 is that all other factors 13

17 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. River Nutrient Budget Model The nutrient budget model (Annual Network Nutrient Export - ANNEX) is a steady state model that 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. For each link, ANNEX requires values for the sediment-bound and dissolved nutrient inputs from the immediate catchment of the river link (Figure 4). The model then routes nutrient loads through the river network estimating the losses associated with floodplain and reservoir deposition and instream denitrification. Whilst the sediment-bound and dissolved nitrogen budgets are calculated separately, for phosphorous, the exchanges between the sediment bound and dissolved phases during transport are modeled. ANNEX has been calibrated for current conditions at a national scale using nutrient load estimates from flow and water quality measurement at 93 stations (Young et al., 2001). As with SedNet, the nutrient budgets calculated by ANNEX are average annual budgets. 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. ANNEX therefore assumes that at the annual time scale the changes in biological nutrient stores within a river network link and the fluxes between river network links due to biological transport processes are small in comparison to the fluxes due to physical nutrient transport processes and the changes in physical nutrient stores. The main source terms are hillslope erosion, gully erosion, riverbank erosion, dissolved loads in runoff water (Figure 4), and point sources. Point sources were excluded from the present study as they were not considered to make a significant contribution to total nutrient loads due to the dominance of agricultural sources. 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 (C p ) in the hillslope soil (Equation. 6). For C p > HSDR, SC NC = (6) Cp In 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 6) 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 yields times the soil nutrient concentration, which for phosphorous was taken to be 0.25 g kg -1 and for nitrogen 1 g kg

18 Hillslope erosion (t/y) Dissolved conc. (kg/ml) Soil nutrient conc (kg/t) Riverbank erosion (t/y) Gully erosion (t/y) Runoff depth (ML/y) HSDR, nutrient enrichment Bank nutrient conc. (kg/t) Area weighted land use Point sources Tributary yield (t/y) Deposition, P equilibration, denitrification Downstream yield (t/y) Figure 4: Conceptual diagram for the nutrient budget of a river link. HSDR is hillslope sediment delivery ratio. Estimation of dissolved loads due to surface and sub-surface runoff differs from that used in the NLWRA project. In this instance, the dissolved load is determined as the sum of the load from each of the major contributing land uses within a catchment link. The load for each land use is calculated as the product of the typical nutrient concentration in runoff (Table 2) multiplied by the mean annual volume of runoff. The volume of runoff was in turn calculated as the product of the increase in average annual discharge experienced between the inlet and outlet of the stream link multiplied by the proportionate area of the land use within the contributing watershed. In this way, for each catchment link, no differentiation is made in the mean annual runoff depth between land uses. The concentration of soluble N and P was assessed from gauging station records and small scale catchment nutrient studies for regions with dominant land uses (Bubb et al. 2000, 2002a, 2002b, Bramley and Roth 2002, Cosser 1989, Hunter 2002). These are summarized in Table 2. Since an average concentration in runoff is used in the model, no distinction is made between surface and subsurface runoff volumes. Table 2: Average concentrations of N and P in runoff from land uses within the Mary Catchment. Land use Percentage of catchment Soluble N concentration (ug L -1 ) Soluble P concentration (ug L -1 ) Comments Sugar cane Other crops Row, tree, turf, grain Grazing Primarily beef and dairy on improved and native pastures and in native forest Forests Pine plantations, Parks and public areas Other (including urban, rural residential, industrial) Assumed similar to native forest grazing, rural residential has higher N component 15

19 Hydrology The correct representation of river hydrology is important for routing sediment and nutrients through the river network. Several hydrological parameters are used in the river sediment budget methods. These need to be predicted for each river link across the river basin. The variables used are: the mean annual flow (Q a ) the median daily flow (Q md ) used in the nutrient budget the mean annual sum of Q 1.4 for calculating mean annual sediment transport capacity; the bankfull discharge (Q bf ); and a representative flood discharge for floodplain deposition (in this case median overbank flow Q ob ). Values for these variables were derived from the time series of daily flows generated from IQQM modeling procedures over a 109 year period at each of the 25 main river monitoring (gauging) stations together with two dummy nodes, one of which included the Mary River outflow. Flow parameters were calculated for unregulated conditions as total loss in flow caused by dams and irrigation draw off amounted to only 1% of the total discharge from the Mary River. Each hydrological variable was regionalized by developing a simple empirical rule with catchment area (A in km 2 ) and mean annual rainfall (R in mm) using standard linear least-squares fits on log transformed data points. This allows prediction of values in ungauged river links throughout the rest of the river network Q a = A R R 2 = Q ( < Qbf ) = A R R 2 = Q bf = A R R 2 = 0.94 Q ob = A R R 2 = 0.92 Q md = A R R 2 = 0.96 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 Mary catchment suspended sediment has been sampled at a number of the gauging stations since the early 1970 s. These were generally nonevent 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: Mary River at Miva; Mary at Fishermans Pond; Mary at Bellbird; Munna Creek; Wide Bay Creek; Amamoor Creek; and Six Mile Creek. A modified form of the standard rating curve (Equation 7) was adopted to determine the relationship between suspended sediment concentration (C) and daily flow (Q d ). C c + aq d b = (7) 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 16

20 flows from the regression as they caused significant bias and underestimation of the predicted suspended sediment concentration at high flows. Mean daily load (L d ) passing each gauging station is represented by Equation 8. L d d ( 1+ b) d = c Q + a Q (8) Since the first term in this equation contributes little to total load, then the mean annual load (L a ) can be represented as the sum of daily loads for the time series of daily flows (Equation 9). L 365 i = n (1+ b) a = aq di n i= 1 (9) 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 7), then the coefficient a can be calculated using Equation 10. 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. L a = i = i= n n a (10) Q (1+ b) di Routine monitoring of total N and total P has only been conducted since the mid-1990 s in the Mary Catchment and of these samples, few were taken at high river flows. Consequently trends between concentration and flow could not be established for nutrients and thus concentrations were treated as being constant for the range of river flows. Results and Discussion The major localities and roads in the Mary River catchment are shown in Figure 5 in relation to the river network. This is presented to help locate areas not shown on the maps of erosion and sediment transport. Similarly, average annual rainfall is illustrated across the catchment in Figure 6. Average annual rainfall shows a strong gradient away from parts of the catchment that are close to the coast. Thus catchments to the east, such as Munna and Wide Bay Creeks (near and to the north of Kilkivan, Figure 5), have lower rainfall than those to the west of the Mary River and this feature is carried through into patterns of erosion hazard, vegetation cover, and river discharge. Land use is also often mentioned in relation to erosion and sediment supply and Figure 7 shows the principal land uses (Pointon, 1998) within the Mary catchment. This provides comparison with subsequent illustrations of sediment budget items. Hillslope Erosion Hazard The pattern of hillslope erosion, as estimated from the Revised Universal Soil Loss Equation, is shown in Figure 8. It can be seen that most areas of the catchment have low soil loss rates of < 2.5 t ha -1 y -1. The fewer areas with soil loss rates > 2.5 t ha -1 y -1 relate to steeper slopes on cleared land. The few isolated patches of land with high soil loss rates > 10 t ha -1 y -1 relate to cropping land that occur predominantly in central areas of the catchment and about Maryborough. A number of pixels had very high erosion rates > 100 t ha -1 y -1 and these are considered to relate to minor errors in the grid analysis. The higher erosion rates that might have been expected in the higher rainfall areas towards the southwest did not occur due to good cover of forest and remnant vegetation on hillslopes in these areas. 17

21 ! MARYBOROUGH! KILKIVAN GYMPIE! Legend! Localities Major Roads Streams KENILWORTH! MALENY! Dams Mary Catchment Kilometres Figure 5: Location map for the Mary River catchment. 18

22 MARYBOROUGH KILKIVAN GYMPIE Legend Average Annual Rainfall (mm) Streams KENILWORTH MALENY Kilometers Figure 6: Map showing mean annual rainfall for the Mary River catchment. 19

23 MARYBOROUGH KILKIVAN GYMPIE Legend Land Use Urban Open forest Production forest National parks Sugar cane Other field crops Improved pastures Native pastures KENILWORTH MALENY Streams Kilometers Figure 7: Land use map of the Mary River catchment. 20

24 MARYBOROUGH KILKIVAN GYMPIE Legend Annual Hillslope Erosion (t/ha/yr) Very low Low Medium 5-10 High >10 Very High Streams KENILWORTH MALENY Kilometers Figure 8: Predicted hillslope erosion hazard in the Mary River catchment. 21

25 It was found that about 8.8 x 10 5 tonnes of soil is moved annually on hillslopes in the Mary catchment. This equates to an average soil erosion rate of 0.93 t ha -1 y -1, significantly below the national average (4 t ha -1 y -1 ) determined by applying the same method of assessing hillslope erosion to the Australian continent (as part of a National Land and Water Resources Audit project). The present estimate of hillslope erosion for the Mary catchment is also below that estimated for this catchment area in the NLWRA (5.4 t ha -1 y -1 ). This reflects the improved quality and accuracy of input information to USLE estimates such as the regional land use map and improved slope angle and slope length when derived from the 25 m DEM. Soil erosion potential varies by three orders of magnitude in the catchment. Furthermore, the distribution of erosion is highly skewed in relation to catchment area. Regions having a low erosion rate (< 2.5 t ha -1 y -1 ) occupy 96% of the catchment while regions with a high erosion rate (> 10 t ha -1 y -1 ) occupy only 1.3% of the catchment area. Overall, 16% of the area is predicted to erode at a rate greater than the catchment average rate of 0.93 t ha -1 y -1. This demonstrated the considerable value in targeting erosion control to problem areas, rather than uniform soil erosion control or that purely responsive to local perception. The values of hillslope erosion represent local movement of soil on hillslopes. It is important to realise that hillslope erosion values overestimate sediment delivery to streams as much of the sediment that is moving may be deposited before reaching the stream. For instance, material eroded on a ridge slope might end up being deposited in colluvial fans on flatter valley bottoms or on river frontage areas before reaching streams. Overall only about 5% of sediment moving on hillslopes finds its way to streams. Table 3 divides hillslope erosion into land use classes. There is relatively little difference in the predicted erosion rate between land use categories, although cropping lands tend to have the highest rates. The lack of difference in erosion rate is because of the association of land use with other factors that influence erosion. The forested areas of the catchment tend to be the steepest and wettest parts and thus have an inherently high erosion risk. The total soil loss is dominated by both native pastures and forested land simply because they occupy such a large proportion of the total catchment area. Cropping land, although only occupying around 7% of the catchment area produces 26% of the total erosion. Table 3: Hillslope soil loss from land use categories in the Mary River catchment. Land use category Area (km 2 ) Proportion of total area (%) Predicted total soil loss (kt y -1 ) Proportion of total catchment erosion (%) Average predicted soil loss rate (t ha -1 y -1 ) Native pastures Improved pastures Sugar cane Other crops Forest and reserves Gully Erosion Hazard The spatial pattern of gully density (Figure 9) shows that the majority of gullies occur in western areas of the Catchment and in smaller pockets elsewhere where the land use is dominated by pasture grazing or sparse woodland. The most extensive areas occur in the region of Kilkivan and southwest of Brooweena in the Munna Creek sub-catchment. Much of the remainder of the Mary catchment is relatively unaffected by gully erosion. Throughout areas with gullies, densities are predicted to be of medium to low severity (<0.5 km km -2 ). About 2% of the catchment has medium gully erosion, while 20% faces low gullly erosion and the remaining 78% has little or no gully erosion. Thus, like hillslope erosion, the problems of gully erosion are highly localised and land management that targets these areas would be more successful than uniform management across the catchment. 22

26 A typical gully is considered to have on average an eroded depth of 2 m and a width of 5 m giving a total sediment yield of 10,000 m 3 per kilometre of channel. Thus, areas with a moderate gully density (0.5 km km -2 ) would have produced a total sediment yield of approximately 75 t ha -1 assuming a sediment density of 1.5 g cm -3. This is the total volume of sediment that is carried by the river network. Riverbank Erosion Rivers also carry sediment generated from erosion of the river and stream banks themselves, and this needs to be considered before examining river sediment loads. The Mary River in particular has undergone major channel widening and bed degradation throughout its channel length in historic times. Like many Australian rivers the channel capacity has greatly increased compared with natural conditions to the extent that the river now confines most of its flow, most of the time. This has led to greater levels of bank erosion in the absence of riparian vegetation as the confined flows now have greater energy to erode banks. One of the two main factors controlling riverbank erosion in the model is the extent of riparian vegetation as shown in Figure 10. The State of Rivers Survey, although not ideal, provides the best complete spatial extent of riparian condition and is a great improvement on the national assessment of riparian vegetation that was used in the NLRWA as it is based on direct observations. Figure 10 shows that large sections of the river network have a relatively low riparian cover of between 0 and 40%, and while this is usually the case in agricultural areas, some stream lengths can have good riparian cover (ie., > 60%). In addition streams towards the mouth of the catchment tend to have better riparian cover. River discharge is the other factor controlling the bank erosion rate in the model. When combined with riparian vegetation the results predict that much of the bank erosion occurs along the length of the Mary River with lower rates in tributary channels. This matches casual observation of bank erosion. Along the Mary River, bank erosion is predicted to range on average between 0.06 and 0.09 m y -1. This equates to between 6 and 9 m of average lateral bank erosion for the length of stream links over a 100 year period. Sediment Sources to the Stream Network Each of the sediment sources described deliver sediment to the network of streams and rivers in the Mary River basin. The predicted mean annual amount of sediment supplied to streams from the three processes is shown in Table 4 in comparison with results of the NLWRA project. Clearly, riverbank erosion is the dominant process contributing 87% of sediment to the river. This is largely a result of the high stream power available at bankfull discharge and the general lack of riparian vegetation. Gully and hillslope erosion are also significant sources of sediment. Much of the catchment is predicted to have only a weak dominance of surface wash erosion, while gully erosion is localised and largely restricted to small areas or the larger western sub-catchments. While all erosion processes need to be managed to reduce sediment loads in the catchment, it is clear that bank erosion dominates, and bank erosion control will have the greatest impact on reducing sediment delivery to channels. It should be noted that sheet and rill erosion is believed to only contribute to suspended sediment loads, while gully and riverbank erosion contribute relatively equally to bedload and suspended load. Even when this is taken into account bank erosion remains the most significant contributor to suspended loads in the basin. 23

27 Table 4: Components of the sediment budget for the Mary River Basin. Sediment budget item Predicted mean annual rate (kt y -1 ) This survey NLWRA Hillslope delivery Gully erosion rate Riverbank erosion rate Total sediment supply Total suspended sediment stored Total bed sediment stored Sediment delivery from rivers to estuaries Total losses Table 5: Components of the nutrient budget for the Mary River Basin. Nutrient budget item Predicted mean annual rate (t y -1 ) Total N Total P Hillslope to stream delivery Gully erosion Riverbank erosion Dissolved runoff Total supply Floodplain and reservoir storage Denitrification Dissolved export Particulate export Total losses

28 MARYBOROUGH KILKIVAN GYMPIE Legend Gully Density (km/km2) >0.6 Streams KENILWORTH MALENY Kilometers Figure 9: Density of gully erosion based on the Land Disturbance Survey of the Mary River Catchment (Pointon, 1998). 25

29 MARYBOROUGH KILKIVAN GYMPIE Legend % Riparian Vegetation Mary Catchment KENILWORTH MALENY Kilometres Figure 10: Mapped amount of intact riparian vegetation. 26

30 MARYBOROUGH KILKIVAN GYMPIE Legend Bank Erosion (m/yr) Mary Catchment KENILWORTH MALENY Kilometres Figure 11: Predicted bank erosion. 27

31 MARYBOROUGH KILKIVAN GYMPIE Legend Dissolved N input (ug/l) ,500 Streams KENILWORTH MALENY Kilometers Figure 12: Pattern of dissolved N input to streams. 28

32 MARYBOROUGH KILKIVAN GYMPIE Legend Dissolved P input (ug/l) Streams KENILWORTH MALENY Kilometers Figure 13: Pattern of dissolved P input to streams. 29

33 Nutrient Sources Each of the sediment sources described above together with dissolved contributions from hillslope runoff deliver nutrients to the network of streams and rivers in the Mary River catchment. Table 5 shows the predicted mean annual amount of Total N and Total P from particulate (erosion) and dissolved (runoff) sources. The spatial patterns of dissolved N and P contribution are shown in Figures 12 and 13 respectively. Both show the same broad pattern. Forested land is contributing the least dissolved input of nutrients per unit area while cropping, improved pasture, and urban land form localised hot spots which lie near to the major rivers and streams. Grazing of native pastures, which has a similar areal extent to forested land, is contributing a proportionately higher level of dissolved P than N when compared with cropping land. In contrast improved pastures provide additional hot spots for dissolved N, making a similar level of contribution as cropping and sugar cane. Of the 363 t y -1 of Total P supplied to streams 76% is derived from the particulate sources of hillslope, gully and riverbank erosion. As would be expected, riverbank erosion contributes the majority of particulate N and P. In the case of Total N the relative contribution from particulate sources is lower at 53%. This is because a higher proportion of Total N is supplied by runoff. The ratio of predicted Total P to Total N supply (0.17) is similar to the relative nutrient concentrations ( 0.10) observed in rivers and streams in the Mary catchment. The slightly lower relative nutrient concentrations measured in streams is partly an artefact of the lack of samples being taken at higher flows when particulate sources of phosphorous dominate Total P concentrations and the proportion of Total P to Total N is greater. Sediment Delivery through the River Network On-site erosion hazard is of concern for continued productivity of the land but can only be translated to downstream impacts if the eroded sediment is transported along the river network. The modelled sediment budget for the basin predicts that around 50 % of sediment delivered to streams is exported at the coast. The rest is stored on floodplains or on the bed of streams, with some storage in the basin also occurring in reservoirs. This is a relatively high contribution of sediment to the coast and again reflects the dominance of riverbank erosion and lack of storage for suspended sediment as the present channel confines most flows and there is little opportunity for overbank deposition. River morphology and stream power have a strong bearing on the channels ability to transport sediment through the river network. High stream banks and wide channels have a greater ability to confine flows, producing reduced overbank flooding and higher stream power during floods. This tends to reduce both the storage of fine and coarse sediment along the channel. Surveyed cross-sections exist for each of the river gauging stations. Measurements of bank height and channel width were made from plots of the cross-sections and these are shown in Figure 14 and 15 in relation to catchment area. The dramatic increase in bank height with catchment area, with high banks along the mid to lower Mary River, results in little overbank deposition of suspended sediment along this portion of the catchment. This is incidentally where flood plain extents are greatest and a high proportion of suspended sediment would be expected to be deposited under natural catchment conditions. Overall only 9% of suspended sediment supplied to the channel is deposited on the limited floodplain extent along the Mary River and its tributaries. 30

34 Channel width (m) y = Ln(x) R 2 = Catchment area (km2) Figure 14: Measured bank height at gauging stations in relation to catchment area. River width increases downstream as the greater discharge carves a wider channel. The rate of increase influences the sediment transport capacity of each reach, as wider channels have slightly lower sediment transport capacity. The sediment budget results, however, are relatively insensitive to river width as it is overwhelmed by that of discharge and river slope. Figure 16 shows how the discharge term increases by 3 orders of magnitude while channel width only increases 5 fold. Slope can vary by an order of magnitude from one reach to another. The discharge and river width predictions, together with measured river slope, are combined to determine the sediment transport capacity of each river link. This is the prime control on whether bedload is likely to accumulate as a result of increased supply of sand and gravel. Bank height (m) y = x R 2 = Catchment area (km2) Figure 15: Measured channel width at gauging stations in relation to catchment area. Bedload Deposition The bedload sediment budget predicts the accumulation of sand and gravel on the bed of rivers as a result of increased rates of gully and bank erosion. We consider that where historical bed deposition is in excess of 30 cm, there is likely to be some impact on bed habitats. This might be through filling of pools, smothering of cobble beds with finer grained sediment or reduced diversity of bed forms. 31

35 Our results suggest that only a small percentage of the river network length in the basin has bed deposition in excess of 30 cm (Figure 17). The most significant area impacted is that along the Mary River from below Gympie towards the coast. Localised deposition of bed material of between 0.3 and 2 m is expected in this area where local channel bed slopes are low. Predicted deposition to depths greater than 2 m over short stream links more than likely represents errors in the calculation of bed slope from the 25 m DEM sum Qd Catchment area (km2) Figure 16: Regionalisation of the discharge term of streampower (ΣQ 1.4 ) with catchment area for the Mary catchment. River Suspended Loads The river budget for suspended load predicts mean annual suspended loads through the river network allowing for deposition on floodplains and in reservoirs. The resultant mean annual specific sediment loads for each river link are shown in Figure 18. The predicted mean annual export to the sea is 445 kt y -1 which equates to a specific sediment yield of 0.47 t ha -1 yr -1. While sediment yield increases with catchment area, Figure 16 shows that specific sediment yields of t ha -1 yr -1 are maintained throughout the Mary River, only decreasing once tributary channels are reached. Nutrient Budget The nutrient budget predicts mean annual Total P and Total N loads through the river network allowing for losses on floodplains, in reservoirs and as a result of instream processes of denitrification and p- equilibrium. The predicted mean annual export to the sea of Total P amounts to 344 t y -1 which represents 95% of that produced by erosion processes and hillslope runoff. This is consistent with the lack of floodplain deposition for suspended sediment and relatively small reservoir storages in the catchment. In contrast, 74% or 1541 t y -1 of Total N produced by erosion processes and hillslope runoff reach the coast. The lower proportion of Total N to reach the coast is a reflection of the higher proportion of soluble N and losses due to instream denitrification (490 t y -1 ). Overall however, there is a strong connection between nutrient sources and nutrient export to the coast in the Mary River catchment. Figure 19 shows a clear spatial pattern of the proportion of soluble N to total N supply to streams exists in the catchment, with sub-catchments to the west of the Mary River predicted to have a lower soluble/total N ratio than those to the east. Soluble/Total P shows a similar pattern, although somewhat less accentuated. This pattern is a reflection of the distribution of land uses where western sub-catchments are dominated by grazing of native pastures and those to the east by a mix of land uses including cropping, dairy grazing, horticulture and forestry. It is therefore also a reflection of the dominance of particulate sources for nutrients in western catchments and an increase in soluble nutrients in eastern catchments, partly due to increased yields from cropping and dairying land, but also a reduction in sediment from forested land. 32

36 MARYBOROUGH KILKIVAN GYMPIE KENILWORTH Legend Bedload Accumulation (m/yr) Mary Catchment MALENY Kilometres Figure 17: Predicted bedload deposition. 33

37 MARYBOROUGH KILKIVAN GYMPIE Legend Specific Suspended Sediment Load (t/ha/yr) Mary Catchment KENILWORTH MALENY Kilometres Figure 18: Predicted specific suspended sediment load. 34

38 MARYBOROUGH KILKIVAN GYMPIE Legend Dissolved N : Total N KENILWORTH MALENY Mary Catchment Kilometres Figure 19: Predicted proportion of dissolved N to Total N for the Mary catchment. 35

39 Contribution to Suspended Sediment Export at the Coast Estuaries and coastal waters are of particular concern for suspended sediment as they are the ultimate areas of sediment deposition. Along the tropical coast of Queensland estuaries have strong tidal currents and open mouths. They flush most sediment out to the coastal waters. There, increased rates of deposition can smother significant habitats such as sea grass beds and inshore reef communities. The sediment can also be a supply of nutrient, and change the food web structure. Resuspension of sediment by waves and currents increases the turbidity of water and decreases light penetration, which can inhibit growth of organisms. The sediment budget predicts that 91% of suspended sediment and about 1% of bedload delivered to the river network in any year is exported from the river mouth. The high suspended sediment contribution is due to limited opportunity for storage on floodplains or in reservoirs. While gully and bank erosion contribute evenly to the coarse and fine budgets, by the catchment mouth, bedload makes up only 0.1% of the total load, meeting common observations of suspended load dominance on large rivers (Richards, 1982). Given that around 50% of sediment delivered to streams within the Mary River basin is exported from the mouth, it can be concluded that increased erosion upstream in a sub-catchment results in a significant increase in export to the coast. In other words, within the Mary River there is a moderate degree of connectivity between upstream erosion and contribution to the coast. To further examine sediment export at the coast we have taken the results of our link by link sediment budget and worked back up the river network to trace where the suspended sediment comes from. Each river link carries sediment contributed from its internal sub-catchment and from tributaries to the link. Each river link also deposits a proportion of the mean annual load that it carries. We have worked back up each tributary to find the contribution from each internal sub-catchment to export from the Mary River basin. The result is expressed as a mean annual sediment contribution in t ha -1 y -1 from each sub catchment that reaches the coast. Sub-catchments which make a substantial contribution to the export at the coast are those with high erosion and limited floodplain extent between the source and sea. Sub-catchments close to the coast are more likely to contribute to the coastal export because of limited possibilities for that sediment to be deposited along the way. This is the overall pattern shown by the results (Figure 20). Inland subcatchments will contribute significant amounts of sediment to the coast if the erosion rate is high and the river delivers sediment efficiently. Few areas in the Mary River meet these conditions with the majority of subcatchment areas contributing between 0.1 and 1 t ha -1 y -1. Overall, however, a high proportion of sediment contributed to the coast comes from watershed areas adjacent to the main river channel. These include smaller riverside sub-catchments below Gympie where bank erosion rates are high. The contribution of these areas to the coast is typically > 2 t ha -1 y -1 (Figure 20). A few areas contribute between 5 and 10 t ha -1 y -1. If the goal is to reduce sediment loads to the coast 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 those adjacent to the Mary River. Targeting these areas will obviously have the greatest effect on reducing sediment export to the coast. However, because the majority of the catchment is still making a significant contribution, without more widespread rehabilitation of problem areas, it is unlikely that major reductions in suspended sediment loads to the coast will be achieved. 36

40 MARYBOROUGH KILKIVAN GYMPIE Legend Contribution of suspended sediment to the coast (t/ha/yr) KENILWORTH MALENY Kilometres Figure 20: Predicted contribution of suspended sediment to the coast for sub-catchments within the Mary River catchment. 37

41 Calibration of Suspended Sediment Load The preceding sections have concentrated on predictions of the river sediment budgets and their implications with little discussion on tests of their accuracy. For the Mary River it is possible to directly compare the results from SedNet with loads determined from sediment rating curves. Figure 21 shows the rating curve for the middle to lower section of the Mary River covered by Miva (138001A) and Fishermans Pocket (138007A) gauging stations. At these two stations suspended sediment sampling has been conducted since the early 1970 s and covers the range of river flows to about a 1 in 10 year recurrence interval event. C 0.62 = Qd (11) The rating curve (Equation 11) when applied to the 90 years of daily flow recorded at the Miva site from January 1910 to March 2000 results in a total mean annual load of 260 kt y -1 estimated to flow past this gauging station Suspended sediment concentration (mg/l) Daily flow (ML/day) Figure 21: Suspended sediment rating curve for the Mary River for combined suspended sediment samples taken at Miva (138001A) and Fishermans Pocket (138007A) gauging stations. The same general trend as shown in Figure 21 was observed for all gauging stations analysed in the Mary River catchment. Principal differences were the mean suspended sediment concentration at low flows and the exponent b in the power function (Equation 7). In general, mean low flow concentrations increase from 6 to 10 mg L -1 when moving from tributaries to the main river channel. The exponent b, which determines the rate of change in LogC against LogQ d, was found to vary from 0.5 to 0.9 for gauging records whose suspended sediment samples cover a reasonable range in river flows. These were, however, not considered to be significantly different due to a standard error for the coefficient of about 0.2 in most cases. At some gauging sites there was poor resolution of the relationship between suspended sediment and daily flow due to few samples being taken at high flows. In these cases, the exponent in the rating curve was fixed to an average value of 0.7, based on results of better rating curves. While this approach avoided underestimation of total load at these sites, it is important to realise that the average annual loads estimated from suspended sediment samples have significant errors associated with them. 38

42 A comparison of predicted average annual suspended sediment loads and estimated loads derived from sediment rating curves are shown for 6 gauging sites within the Mary River catchment in Figure 22. Given the errors associated with estimating sediment loads from rating curves, the loads predicted by SedNet show reasonable agreement with measured loads. The predicted load for the Mary River, in particular, is within ± 20% of the measured load. 300 Annual Suspended Sediment Load (kt/yr) Munna Cr (138004B) Wide Bay Cr (138002C) SedNet Rated Amamoor Cr (138102C) Six Mile Cr (138107B) Gauging Station Mary at Miva (138001A) Mary at Bellbird (138110A) Munna Wide Bay Amamoor Bellbird Miva Six Mile Figure 22: Comparison of average annual suspended sediment loads predicted from SedNet with loads estimated from gauging station (location shown on map) sediment rating curves (eg. Figure 21). The remaining sites are mostly within 1-order of magnitude of the measured loads. The differences to some degree can be explained by errors in estimation of measured loads as described above. It is important to realise also, that SedNet predictions are 100-year mean annual averages as compared with sediment concentration data which have only been collected over the past 30 years. Some of the difference, therefore, can be attributed to temporal variation in the activity of sediment sources. For example, Munna and Wide Bay Creeks, both of which have similar land use patterns, both tend to show a lower measured load when compared with the predicted load. These are also the main source catchments for gully erosion and it would appear that the activity of gully erosion may have declined over the past 30 years relative to the 100-year average. Conversely, the higher measured load at Amamoor Creek could reflect short term acceleration of sediment supply from erosion sources, that is not necessarily representative of the long term average. Overall, the predictions from SedNet appear to slightly underestimate river suspended sediment loads. There are a number of parameters in the SedNet model which influence the predicted value, for which only approximate values are known. The main one of these is the hillslope sediment delivery ratio which was 5% for the present model. Using a higher value for the sediment delivery ratio would have improved agreement between predicted and measured loads along the Mary River. 39

43 Disaggregation of Annual to Daily Loads Figure 23 illustrates how mean annual loads can be disaggregated into mean daily loads for river links. In this case, the example used is for the Mary River at Miva, but we could have equally used any river link in the catchment. The figure shows how suspended sediment and nutrients are likely to vary through time for a typical year. It demonstrates how storm-flows (ie., peaks on the graphs) transport the majority of suspended sediment and nutrients. The curve for suspended sediment shows the greatest difference in load between peak storm flow and low flow conditions owing to the observed increase in suspended sediment concentration with increasing flow (Figure 21). Relative to the suspended sediment curve, those for Total P and Total N are more attenuated because a higher proportion of nutrients is transported by low flows. FLow (ML/day) Predicted mean load (t/day) Predicted mean load (kg/day) Predicted mean load (kg/day) /21/1999 8/28/1999 9/4/1999 9/11/1999 9/18/1999 9/25/ /2/ /9/1999 Daily Flow < 1 yr recurrence flows Suspended Sediment Load L d = Q 1.7 d /1000 Total-P Load L d = 0.14.Q d Total-N Load L d = 0.65.Q d 10/16/ /23/ /30/ /6/ /13/ /20/ /27/1999 Date 12/4/ /11/ /18/ /25/1999 1/1/2000 1/8/2000 1/15/2000 1/22/2000 1/29/2000 2/5/2000 2/12/2000 2/19/2000 2/26/2000 Figure 23: Variation in predicted mean daily loads from August 1999 to February 2000 for the Mary River at Miva. Daily loads are disaggregated from the mean annual loads predicted by SedNet using rating curves for sediment and nutrient concentrations. Due to a lack of rating data describing the relationship between nutrient concentration and daily flow, the estimates of mean daily nutrient loads shown in Figure 23 assume a constant nutrient concentration with flow, in this case 0.70 mg L -1 Total N and 0.13 mg L -1 Total P. These values were based on mean concentrations determined from loads estimated by SedNet and mean annual discharge at the Miva gauging station. These are approximately 2 and 3 times, respectively, the concentrations measured at low flow and therefore strongly suggest an increase in concentration with increasing daily flow. Compare this to the suspended sediment concentrations which are almost 2 orders of magnitude higher at high compared with low flows. It is likely, therefore, that the actual curve for Total P is more accentuated than that shown in Figure 21 since the majority of phosphate is transported attached to sediment (Table 5). In contrast, the curve for Total N will not show quite the same degree of accentuation. 40

44 It is important to realise that Figure 23 shows predicted mean daily loads only. The load curves simply serve to illustrate how average conditions fluctuate through time. Actual daily loads for this period will have differed from mean values depending on the influence of environmental variables which determine daily fluctuations in sediment and nutrient concentrations (eg., rainfall frequency-duration, spatial pattern of rainfall and source area for flow, antecedent conditions etc). Testing of Land Use Scenarios The results of SedNet so far presented represent those for the current catchment condition and pattern of existing land uses. It is possible to test the effect that different land management scenarios would have on the long-term export of sediment and nutrients from the catchment by reducing or increasing the levels of erosion inputs to channels to reflect changes in land use. By rerunning the sediment and nutrient calculations in SedNet, new estimates of sediment and nutrients loads can be compared with current conditions, bearing in mind that the changes represent long-term annual averages. Changes to flow hydrology or channel morphology are presently not considered for new scenarios. Figure 24: Scenario testing in ArcMap. Proportion of land use, riparian vegetation, gully density and hillslope delivery ratio can be interactively changed for an individual or group of selected basins (as shown in pink) within the Mary River catchment to reflect changed land use conditions and how they might impact on the supply of sediment to the channel. Sediment loads and contribution to the coast are recalculated once changes are made. To this extent a catchment scenario testing tool has been developed which enables the land manager to interactively change selected attributes in the model which effect contribution of sediment to the channel network. At present ArcMap software is used as the visual interface with the SedNet suite of programmes. Users will therefore need an ArcMap licence to run this scenario tool. When first opening 41