EIS 1483
AA0681 11
Water quality in the Snowy River catchment area : report on
1996/97 data; nutrient loads in the Thredbo river; trend
assessment NSW YEPT PRIMARY IRDUSIRIES
I AA0681 11 I LAND &WATER CONSERVATION I I I I I I
d I I I I I I I I NSW Department of Land and Water Conservation I I
I DEPARTMENT OF LAND & WATER CONSERVATION CENTRE FOR NATURAL RESOURCES
I I I WATER QUALITY IN THE SNOWY I RIVER CATCHMENT AREA I - Report on 1996/97 Data - Nutrient Loads in the Thredbo River I - Trend Assessment I I I I I I I I H I I DEPARTMENT OF LAND & WATER CONSERVATION CENTRE FOR NATURAL RESOURCES
WATER QUALITY IN THE SNOWY RIVER CATCHMENT AREA - Report on 1996/97 Data - Nutrient Loads in the Thredbo River - Trend Assessment
Zenita Acaba, Lee Bowling, Lloyd Flack June 1998 and Hugh Jones CNR 99.005 [s9697co2.Doc]
CENTRE FOR NATURAL RESOURCES
© Department of Land & Water Conservation
ISBN 0 7347 5023 4 Public Document
Water Quality in the Snowy River Gatchinent Area, 1996197 Report 1 I I I I I I Cologne I In KOln, a town of monks and bones,
I and pavements fang'd with murderous stones I and rags, and hags, and hideous wenches; I counted two and seventy stenches, I All well defined, and several stinks! I Ye Nymphs that reign o'er sewers and sinks, The river Rhine, it is well known, I Doth wash your city of Cologne; But tell me, Nymphs, what power divine
I Shall henceforth wash the river Rhine? Li I Samuel Taylor Coleridge, 1828 I I I 1 Water Quality in the Snowy River Catchment Area, 1996/97 Report I I I ACKNOWLEDGEMENTS The majority of the sampling for this project was undertaken by staff of the Snowy Mountains Hydro-Electric Authority's Hydrographic Office at Jindabyne, chiefly by Messrs. Phil Boreham, I David Whitfield and Jason Venables. The staff of the Hydrographic Office were also principally involved in the setting up of the automatic water quality samplers and flow gauging stations at the I sites within the Thredbo River subcatchment, which enabled event sampling and load estimates for the project.
I We acknowledge the input of Messrs. Ken Jones and John Denham, of the Snowy Mountains Hydro-Electric Authority, for the co-ordination of the SMII-IEA's input into the project, and for I review of this report. We also acknowledge the continued interest in the project, and in the protection and improvement of water quality in the Snowy Mountains area through total catchment management, of Messrs. I Barry Dunn and Tony Gan, of SMHEA's Operations Planning Division. They are also thanked for their review comments on the report.
I Mr Tom Freece of SMHEA is thanked for providing river flow data for the project. I Sample analysis was undertaken by the Department of Land and Water Conservation's Water Environment Laboratory at Arncliffe. I Finally, we wish to thank Mr Brett Miners of DLWC Cooma (Snowy-Genoa Water Resource Manager), who reviewed the initial draft of this report. I I I I I I I I I V Water Quality in the Snowy River Catchment Area, 1996/97 Report I I I SUMMARY The water quality of Lake Jindabyne and the streams within the upper Snowy River catchment I area was generally good in 1996/97. The lake continued to have low to moderate concentrations of nutrients, and low conductivity and turbidity, as in previous years. There was some oxygen depletion of the deeper waters during summer and autumn whilst the lake was thermally stratified. I There was little algal growth within the reservoir.
The Thredbo River continued to have the best water quality of the three major inflows to the lake, I with low nutrient concentrations and turbidity. The Mowamba Aqueduct had the poorest water quality of these inflows in 1996/97, with Wollondibby Creek being only slightly better. Both had total phosphorus concentrations in the moderate range, and slight turbidity. However, in terms of I nutrient concentrations, all three inflows still met ANZECC (1992) water quality guidelines for the protection of aquatic ecosystems. I Nutrient concentrations and turbidity at the six sampling sites in the Thredbo river subcatchment upstream of Paddys Corner were low during base flow periods, but increased during storm events. I Nutrient loads increased with distance downstream from Deadhorse Gap. Areal load estimates indicate high input of nutrients from the catchment area between Bundilla and just downstream of the Little Thredbo River confluence. This area, which comprises only 1.5% of the area of the I Thredbo River catchment upstream of the Little Thredbo River confluence, was estimated to contribute approximately 24% of the total phosphorus load entering the Thredbo River upstream I of this point, and 2 1 % of the total nitrogen load. Nutrient input from the Little Thredbo River subcatchment was the lowest of any part of the Thredbo River catchment. Impacts from the Lake Crackenback resort were indicated as being I negligible, and there was very little difference in the nutrient input from the Thredbo Village and ski slope area compared to other undeveloped areas of the catchment. Unfortunately the frequent I failure of autosampling equipment at Paddys Corner prevented a number of high flow events from being sampled there, leading to unreliable load estimates in comparison to the other sites. For this reason, nutrient inputs to the Thredbo River from the section of catchment between the Little I Thredbo River confluence and Paddys Corner could not be adequately calculated.
The nutrient loads estimated in this study for the entire Thredbo River catchment area were similar I to those calculated in an earlier study by Bowling (1992). Total phosphorus loads exported from the catchment in terms of kilograms per hectare per year were found to be higher than those I reported in the literature for other forested areas of south-east Australia. The high rainfall of this alpine area, much of which falls as snow which captures airborne particles containing phosphorus, has been proposed as the main reason why the Thredbo River annual areal total phosphorus load I coefficients are higher than those measured elsewhere. Areal total nitrogen load export coefficients were also more similar to those reported from urban catchments than for forested catchments.
I Trend assessments were also carried out using nutrient and turbidity data collected at sites where a minimum of at least five years routine water sampling had been undertaken. These sites included Paddys Corner and others upstream on the Thredbo River; Wollondibby Creek at Gunnadoo; I Mowamba Aqueduct at its Lake Jindabyne outfall; Perisher Creek at Blue Cow; and the Snowy River at Dalgety. No trend was apparent in total phosphorus and total nitrogen concentrations at
Water Quality in the Snowy River C'atchment Area, 1996/97 Report 111
I
I any site, even after the data had been flow adjusted. However, many sites on the Thredbo River indicated significant increasing trends in turbidity over the period 1992 to 1997, although turbidity was still very low. I A range of recommendations have been suggested for catchment management purposes, and for the protection of water quality in the area, which remains amongst the best of anywhere in New South I Wales. Routine monitoring of selected sites should continue, to enable better estimates of long term trends in water quality at these locations. However, further nutrient load studies of the Thredbo I River catchment are not seen as a priority, unless carried out for a specific purpose such as monitoring the impacts of further development in the valley, monitoring the results of remedial action recommended from this study, or determining inputs from the section of catchment between I the Little Thredbo River confluence and Paddys Corner.
Possible nutrient load studies on Wollondibby Creek and the Mowamba River have been mooted, I but a decision needs to be reached on environmental flows in the Snowy River downstream of Jindabyne Dam before it is decided whether these studies should proceed or not. This decision will influence the ultimate destination of Mowamba River water, that is, whether it will continue to be L diverted to Lake Jindabyne, or flow directly to the Snowy River, and will therefore be a factor when applying for future funding of water quality studies of the Mowamba River.
I A closer alignment of the water quality monitoring being undertaken throughout the catchment and the requirements of the Snowy-Genoa Total Catchment Management Committee is urged, with increased input and direction from the TCM process. Better documentation and co-ordination of all water quality studies in the area is also desirable, and the involvement or funding assistance in future studies from all agencies with an interest in the water quality of the area is encouraged. I Only in this way will the good water quality of the area be maintained into the future, and the need for subsequent remedial action avoided. I I I I I I I I iv Water Qualily in the Snowy River Catchment Area, 1996/97 Report I I TABLE OF CONTENTS i. INTRODUCTION...... 1
I 2. BACKGROUND ...... 1
3. OBJECTiVES OF THE STUDY...... 2
I 4. FUNDING ARRANGEMENTS...... 3
5. SAMPLING PROGRAM...... 3
I 6. ANALYSIS OF SAMPLES...... 6 6.1. ANALYSES UNDERTAKEN ...... 6 I 6.2. QUALITYASSURANCE ...... 6 6.3. DATA ANALYSIS AND PRESENTATION...... 7
I 6.3.1. Calculation of loads ...... 7
6.3.2. Estimation of the Error Associated with the Calculation of Loads...... 8
I 6.3.3. TrendAssessment...... 9
7. RESULTS...... 11 I 7.1. HYDROLOGY...... 11 7.2. ROUTINE SAMPLES ...... 14 I 7.2.1. Total phosphorus...... 14
7.2.2. Total nitrogen...... 17 I 7.2.3. Turbidity ...... 20
7.2.4. Additional data for Lake Jindabyne...... 23 7.3. EVENT SAMPLES ...... 26 I 7.4. AmrwL LoADs ...... 30 7.4.1. Total phosphorus Loads...... 30
7.4.2. Total nitrogen Loads ...... 33 7.5. TREND ASSESSMENT ...... 37 I 7.5.1. Total phosphorus...... 37 7.5.2. Total nitrogen...... 40 I 7.5.3. Turbidity...... 43
8. DISCUSSION ...... 46
U 9. CONCLUSIONS...... 50
10. RECOMMENDATIONS...... 53
I Viii Water Quality in the Snowy River Catchment Area, 1996197 Report I List of Tables I
Table 1. Sampling sites for 1995/96 and 1996/97 Water Quality Monitoring Program by subcatchment. (R I = Routine monthly samples, E = High flow event samples) 3 Table 2. Total number of blue-green and other algae at Lake Jindabyne, 1996-97 data. 25 I Table 3. Estimates of total phosphorus loads from 26 July 1995 to 30 June 1997. 31
Table 4. Areal load calculations for total phosphorus. 31
I Table 5. Estimates of total nitrogen loads for 26 July 1995 to 30 June 1997. 33
Table 6. Areal loading calculations for total nitrogen. 34
I Table 7. Percentage of total phosphorus and total nitrogen loads contributed by various sections of the Thredbo River subcatchment passing 1). Site 22210011, the Thredbo River downstream of the I Little Thredbo confluence, and 2). Site 222541, the Thredbo River at Paddys Corner. (Site 222541 is further downstream than Site 22210011). 35 I Table 8. Seasonal Kendall test results for log total phosphorus performed on raw and flow adjusted data. 37 Table 9. Seasonal Kendall results for log total nNitrogen performed on raw and flow adjusted data. 40 I Table 10. Seasonal Kendall results for log turbidity performed on raw and flow adjusted data. 43 I I I I I I I I I IX I Water Quality, in the Snowy River CatchmentArea, 1996/97 Report I d List of Figures i Figure 1. Map of sampling sites in the upper Snowy River catchment area. 5
Figure 2. Mean daily flows at sites within the Thredbo River subcatchment, and at the major inflow points I to Lake Jindabyne. 13 Figure 3. Boxplot of total phosphorus data for river and stream sites, routine sampling during the period I July 1996 to June 1997. 15 Figure 4. Boxplot of total phosphorus data for Lake Jindabyne sites, routine sampling during the period 1 July l996toJune 1997. 15 Figure 5. Boxplots of total phosphorus for river and stream sites, routine sampling comparing data collected during the periods 1995/96 and 1996/97. 16 I Figure 6. Boxplot of total nitrogen for river and stream sites, routine sampling during the period July l996toJune 1997. 18 I Figure 7. Boxplot of total nitrogen data for Lake Jindabyne sites, routine sampling during the period July l996toJune 1997. 18 l Figure 8. Boxplots of total nitrogen for river and stream sites, routine sampling comparing data collected during the period 1995/96 and 1996/97. 19 I Figure Boxplot of turbidity data for river and streams sites, routine sampling during the period July 1996 to June 1997. 21
Figure Boxplot of turbidity data for Lake Jindabyne sites, routine sampling during the period July I 1996 to June 1997. 21
Figure 11. Boxplots of turbidity for river and stream sites, routine sampling comparing data collected I during the periods 1995/96 and 1996/97. 22 Figure 12. Thermal and dissolved oxygen profiles for Dam Wall Station, Lake Jindabyne. 24
I Figure 13. Total phosphorus load (in dotted lines) and hourly flow (in solid lines) during event sampling of September 1996 to October 1996. 28 I Figure 14. Total nitrogen load (in dotted lines) and hourly flow (in solid lines) during event sampling of September 1996 to October 1996. 29 I Figure Map showing the annual total phosphorus and total nitrogen loads for each site in the Thredbo River catchment. 36 I Figure LOESS trend curve fitted to raw total phosphorus data, 1992-1997. 38 Figure 17. LOESS trend curve fitted to flow adjusted total phosphorus data, 1992-1997. 39 I Figure 18. LOESS trend curve fitted to raw total nitrogen data, 1992-1997. 41 Figure 19. LOESS trend curve fitted to flow adjusted total nitrogen data, 1992-1997. 42 I Figure 20. LOESS trend curve fitted to raw turbidity data, 1992-1997. Turbidity values above 100 NTU was not included in the graph. 44
Figure 21. LOESS trend curve fitted to flow adjusted turbidity data, 1992-1997. 45 I 1 x Water Quality in the Snowy River Catchnzent Area, 1996197 Report I I I List of Appendices I Appendix 1. Boxplot defined 58 I Appendix 2. Ratio Method for Calculating Loads 59 Appendix 3. Summary statistics of routine samples, 1996 to 1997 60 I Appendix 4. Summary statistics of event samples. 64 I Appendix 5. Total phosphorus and total nitrogen concentration at the start of an event, by site 66 I I I I I I I I I I I I I Water Quality in the Snowy River Catchment Area, 1996/97 Report I I I GLOSSARY Anoxic. The absence of oxygen in the water.
I Campaign sampling. Special sampling when water samples are taken for specific purposes outside of the routine sampling program, if one exists. Campaign samples may be taken as special I samples in locations where there is no routine sampling program. Cyanobacteria. Blue-green algae. Excessive growth of cyanobacteria is usually a symptom of I high nutrient pollution in lakes, reservoirs and slow-flowing streams. They can produce toxins, form scums, and cause taste and odour problems in drinking water supplies. Their decomposition after a bloom can reduce dissolved oxygen concentrations in water. I Epilimnetic. Referring to the surface layers of water in a lake or reservoir during thermal stratification. These waters overlay the thermocline, and are usually warmer, and have higher I dissolved oxygen concentrations than deeper waters below the thermocline, with which they do not mix.
I Eutrophic. A condition where there are high concentrations of plant nutrients in a water body, often accompanied by excessive algal and macrophyte growth, deoxygenation of the bottom waters due to the decomposition of this growth, and other water quality problems. Waters are often I considered to be eutrophic if total phosphorus concentrations exceed 0.035 mg/L, and total nitrogen concentrations exceed 1.00 mg/L.
Hypolimnetic. Referring to the bottom stratum of water in a lake or reservoir which is thermally stratifled—that is the waters below the thermocline. These waters usually remain cold, and can I suffer oxygen depletion and associated water quality problems as they have no contact with the atmosphere for oxygen replenishment, as they are overlain by waters of the thermocline and I Macrophyte. A large aquatic plant growing in a water body. These may be rooted in the sediments I at the bottom, or free-floating. Mesotrophic. The nutrient status of a stream, lake or reservoir where plant nutrients are present at moderate levels, but are neither scarce nor abundant. The nutrients are usually present at sufficient I concentrations to allow some algal and macrophyte growth, but not excessive growth or algal blooms. Mesotrophic waters are often considered to have total phosphorus concentrations ranging I between 0.010 and 0.035 mg/L, and total nitrogen concentrations ranging between 0.60 and 1.00 mg/L.
I Oligotrophic. The nutrient status of a stream, lake or reservoir where plant nutrients are only present at very low concentrations, and are insufficient to allow much plant or algal growth. Waters are commonly considered to be oligotrophic when total phosphorus concentrations are less I than 0.010 mg/L, and total nitrogen concentrations are less than 0.60 mg/L.
Phytoplankton. Microscopic algae that float within the water column of a lake, reservoir, or I stream. They can be single celled, colonial, or filamentous. I Water Quality in the Snowy River Catchment Area, 1996/97 Report vi I
Thermal stratification. The division of a waterbody into horizontal layers of different densities, I due to the differential uptake of solar radiation by water at different depths. Surface waters heat more than deeper waters, and become less dense. This density difference prevents the complete I mixing of the lake over all depths. Usually three strata are formed: the epilinmion or surface layer; the thermocline where there is greatest change in temperature with depth; and the hypolimnion, towards the bottom. Thermal stratification generally commences in spring and breaks down in 1 autumn.
Thermocline. The stratum of a thermally stratified lake or reservoir over which the maximum I temperature change with depth occurs. It is the layer of water that lies between the epilimnion and the hypolinmion, and is sometimes also called the metalimnion.
I I LI I 1 I I I I I I I I Water Quality in the Snowy River Catchment Area, 1996197 Report Wi I
I 1. INTRODUCTION
Lake Jindabyne and the Snowy River form an important aesthetic, visual and recreational amenity I on the eastern approaches to the Snowy Mountains ski-fields and Kosciusko National Park in south east New South Wales. Increasing tourist development within this area has raised concerns that the water quality of the river and lake, and their tributary streams, may deteriorate.
Investigations to determine the water quality status of Lake Jindabyne, the Snowy River and their U tributary streams, have been under way for a number of years now. Initial discussion of the water quality investigation program took place between what was then the NSW Department of Water Resources (DWR) (now part of the Department of Land and Water Conservation (DLWC)) and I the Snowy Mountains Hydro-Electric Authority (SM}iEA), with some input from the National Parks and Wildlife Service earlier in the program. The major aspects of these discussions were I detailed in a report to the Snowy Mountains Council (Operation Engineers Committee Report No. 145 - Appendix T5, August 1992). This has subsequently been updated in an ongoing water quality management program. Points of agreement between the two Agencies covered the 1 objectives of the program, the sampling program and on joint funding arrangements based on the benefits received by the respective parties.
I This report presents the water quality data collected for the Snowy Mountains Water Quality Program during the period 1995/96 and 1996/97. This report is intended to:
I Provide information on all water quality data collected from the area in 1995/96 and 1996/97 following earlier annual progress reports;
I Report the results of nutrient load studies at sites within the Thredbo River subcatchment during 1995/96 and 1996/97; and
I Assess trends in water quality data for the 3 major inflows to Lake Jindabyne:- Thredbo River at Paddys Corner (Site 222541), Wollondibby Creek at Gunnadoo (Site 222544) and I Lake Jindabyne Mowamba Aqueduct Outflow (Site 22210142). 1 2.. BACKGROUND Concerns of deteriorating water quality in Lake Jindabyne led to a joint SMHEA!DWR study of the lake and its major inflows in 1989 (Bowling 1993). This examined the water quality of Lake I Jindabyne during the period 1989-1992 and calculated a nutrient budget for the lake, including inputs from each of the major inflows. The study found that Lake Jindabyne was in an advanced 1 mesotrophic state in terms of total phosphorus, with concentrations approaching 0.03 mg/L. Total nitrogen concentrations were low, however, and algal presence sparse. While Wollondibby Creek had the highest nutrient concentrations of the major inflows, the Thredbo River contributed the I greatest amount of the total phosphorus load to the lake, in keeping with its high hydraulic loading. Unexpectedly, the Thredbo River sub-catchment also had high annual areal total phosphorus loads. The study concluded that although the water quality of Lake Jindabyne was amongst the best in H New South Wales, there was considerable potential for its sudden deterioration should nutrient I inputs increase and nutrient concentrations within the lake rise. 1 Water Quality in the Snowy River Catchment Area, 1996/97 Report I
Additional water quality studies of the Snowy River catchment area were commenced by the DWR I in 1991. These studies included sampling programs in streams within the Kosciusko National Park, as well as in the Snowy River and tributaries downstream of Jindabyne Dam, in the Dalgety and Berridale areas (Bate 1992). Bate (1992) found generally good water quality during low flow [1 periods in streams of the Kosciusko National Park, but that nutrient concentrations and turbidity in these streams increased during high flow events, especially downstream of resorts. The study also I found nutrient concentrations in the Snowy River at Dalgety to be higher than those in Lake Jindabyne, although turbidity was similar. Both turbidity and nutrient levels were deemed I satisfactory for water quality purposes except during high flows following storni events. The two studies were combined in 1992 as the Snowy River Catchment Water Quality Study, under the OEC Report No. 145 - Appendix T5, August 1992 agreement between SMHEA and I DWR. A number of changes have occurred since then, with greater emphasis on determining nutrient and sediment sources in the Thredbo, Mowamba and Wollondibby subcatchments of Lake El Jindabyne. Since 1994, less emphasis has been given to streams in the Kosciusko National Park and downstream of Jindabyne, with monitoring at only one site in each area. Studies since then have concentrated more on nutrient loads within the Thredbo River. Annual progress reports have I been completed for 1992/93 (Bowling et al. 1993), 1993/94 (Bowling and Acaba 1995), 1994/95 (Kinross and Acaba 1996) and 1995/96 (Maim et al. 1997). Water quality monitoring and nutrient load investigations during the 1995/96 and 1996/97 period are detailed in this report. The program I is still evolving, with water quality concerns downstream of Jindabyne Dam again becoming a higher profile issue.
3. OBJECTIVES OF THE STUDY I The objectives of the study during 1995/96 and 1996/97 were: Upper Snowy Catchment Area (Snowy Catchment above Lake Jindabyne). I Monitor Perisher Creek to detect changes and trends in nutrient concentrations; [1 Middle Snowy Catchment Area (Lake Jindabyne and its catchment). Monitor the water quality in Lake Jindabyne to detect changes in the overall "health" of the reservoir; I Monitor nutrient concentrations in the major inflows of Lake Jindabyne to detect changes and trends in the water quality of these inflows; I Identif' sources of pollution within the Thredbo River sub-catchment. Similar investigations within the Wollondibby, Mowamba and Cobbin subcatcbments are proposed for future years, following completion of the work within the Thredbo I subcatchment. Lower Snowy Catchment Area (Snowy River downstream of Jindabyne Dam). I Monitor the Snowy River at Dalgety to detect changes in water quality I
I 2 Water Quality in the Snowy River Catchment Area, 1996/9 7 Report I I I 4. FUNDING ARRANGEMENTS Funding arrangements for 1996/97 were the same as those for the previous year (Maini et al I 1997). Funding was divided as follows:- DLWC SMHEA
I Upper Snowy Catchment Area 80% 20% Middle Snowy Catchment Area I Lake Jindabyne 10% 90% Major inflow points 40% 60% I Subcatchment streams 40% 60% Lower Snowy Catchment Area 100% 0%
I Of these, studies in the Middle Snowy Catchment Area accounted for approximately 95% of the total budget (of which 10% was for Lake Jindabyne, the remainder for the inflow streams).
I Note: funding for the Lower Snowy River Catchment Area was undertaken entirely by the DLWC, since this part of the study was outside the scope of the OEC agreements.
I 5. SAMPLING PROGRAM I The sampling sites for the 1995/96 and 1996/97 water quality monitoring program are listed in Table 1 and are illustrated in Figure 1.
I Table 1. Sampling sites for 1995/96 and 1996/97 Water Quality Monitoring Program by subcatchment. (R = Routine monthly samples, E = High flow event samples)
I Area Site Site Name Sampling Number. Status Upper Snowy I Perisher Creek 222513 Perisher Creek at Blue Cow R - Middle Snowy Thredbo River 22210019 Thredbo River at Deadhorse Gap R E I 22210016 Thredbo River D/S Thredbo STW R E 22210012 Thredbo River at Bundilla R E 22210017 Little Thredbo River at Alpine Way R E I 22210015 Little Thredbo River at Walking Track Bridge R E 22210011 Thredbo River D/S Little Thredbo River R E 222541 Thredbo River at Paddys Corner R E I Wollondibby 222544 Wollondibby Creek at Gunnadoo R Creek Lake Jindabyne 22210001 Lake Jindabyne at Dam R I 22210002 Lake Jindabyne at Jindabyne East R 22210003 Lake Jindabyne at Hiawatha R Mowamba River 22210142 Lake Jindabyne Mowamba Aqueduct Outflow R - I Lower Snowy Snowy River 222006 Snowy River at Dalgety R - I Water Quality in the Snowy River Catchment Area, 1996197 Report I I
I Water samples are collected at all of the above sites for the analysis of total phosphorus concentration, total nitrogen concentration and turbidity. The samples collected were frozen before being sent to the Department's Water Environmental Laboratory at Arncliffe for analysis. I Nutrients samples were taken from the surface, middle and bottom of Lake Jindabyne. In addition, thermal and dissolved oxygen profiles, electrical conductivity, pH and turbidity were measured in situ in Lake Jindabyne.
The sampling program to monitor water quality at these sites had two major components:
I Routine sampling - Samples were taken at all sites monthly, except for the Snowy River at Dalgety where only eight samples were taken in 1995/96, and 10 in 1996/97. Long term routine I sampling is undertaken to enable statistical analyses to determine trends in the water quality parameters measured at each site over time. The sites of specific interest for trend analysis include the three major iaflows to Lake Jindabyne: Thredbo River at Paddys Corner (Site 222541), Fi Wollondibby Creek at Gunnadoo (Site 222544) and Lake Jindabyne Mowamba Aqueduct Outflow (Site 22210142), as well as Perisher Creek at Blue Cow (Site 222513) and the Snowy River at Dalgety (Site 222006). A minimum of five years of data is needed to undertake trend analyses. To Li date, six years of data have been collected for these sites. This report will incorporate trend analysis for the sites where sufficient data are available.
I Routine monthly samples were taken for two years at the sites within the Thredbo River sub- catchment, to provide sufficient temporal coverage of base flow water quality conditions at these ri sites for use in load estimates. Routine sampling was undertaken at the three sites on Lake Jindabyne during the summer of I 1995/96. Samples were also collected monthly at Dam Station on Lake Jindabyne during 1996/97, and at two-monthly intervals at Jindabyne East and Hiawatha Stations. No sampling was undertaken between July 1996 and December 1996, however, as the instrument used to collect I profile data was not working.
Event sampling - Samples were collected during periods of high flows at all seven sites within the I Thredbo River subcatchment in both 1995/96 and 1996/97 using 'Gamet' automatic water samplers. Sampling was triggered when there was a rise in the water level. Six high flow events I were sampled in 1995/96, while more than 10 events were sampled at each site during 1996/97. No campaign sampling was done in the two-year period from 1995 to 1997.
I I I I Water Quality in the Snowy River Catchment Area, 1996/97 Report ------
/ / I
N
LSO NS xD3 -VALLEY - \ __- -•.s 13ERR?EALE ) h--- 7. __ 1 (\ / .AO - 02 1 SMNS
PERIS HER / - r1 '001 Oil — p142 _..__-_- JINE 0115 012 7 1_7 \ / j( / I
016 i1- f / -fl-1R'OnO -- \ ,2000 0ALGEry 7) oig / .- 9 10 -— KItorr,ofro0 SornIin0 SiIe,
Figure 1. Map of sampling sites in the upper Snowy River catchment area.
Water Quality in the Snowy River Catchment Area, 1996197 Report I
6. ANALYSIS OF SAMPLES 1 6.1. Analyses Undertaken Water samples from the surface, mid-depth and bottom at sites on Lake Jindabyne were analysed for total phosphorus, total nitrogen and turbidity. Additional surface samples from the lake which I had been preserved with Lugols iodine solution were used for the identification and enumeration of algae. i Routine water samples from catchment streams were analysed for turbidity and total phosphorus (TP) following standard methods (APHA 1992), and for total nitrogen (TN) following the method I of D 'Elia et al. (1977). Event samples obtained using the "Gamet" autosamplers were analysed for total phosphorus, total nitrogen and turbidity. I Water samples were analysed at the DLWC's Water Environment Laboratory at Arncliffe.
6.2. Quality Assurance
Documentation of samples using water quality sample log sheets filled out at the time of sampling I allowed easy identification of most samples from sampling, dispatch and laboratory analysis, and assisted subsequent data retrieval. The site, date, time and method of sampling were verified for all data used in this report by referencing to the Sample Log Sheets. Event samples and routine I samples were also identified. If data could not be verified, it was excluded from the data base and not used in the data analyses undertaken for this report (e.g. the absence of a date on the sample I log sheet, and unusually high turbidity due to the sampling probe entering bottom sediments). No data has been rejected due to being unverifiable for the two years reported here.
I The data verification stage also includes checking the sample preservation method and the time taken between preservation and laboratory analysis (see Australian standard "Selection of sample containers and preservation method"). This is done by checking the number of days elapsed using I the fields "Date Sampled" and "Date Analysed" for each record. For frozen total phosphorus (TP) and total nitrogen (TN), standard guidelines allow up to 28 days from the time the water sample is collected to the time the sample is analysed at the laboratory. For the 1996/97 TP and TN data, I almost 40% of samples collected were analysed at least 28 days or more after collection. Instead of rejecting these data, further checking and verification of each record was conducted e.g. these data I were compared to the range of TP and TN values measured in previous years. If the data fell within the range of values for TP and TN for a particular site, these data were utilised in the report. A similar data verification procedure was applied for turbidity. No data was rejected from I the 1996/97 data set.
The laboratory engaged to conduct the chemical and turbidity analyses was the DLWC's Arncliffe I laboratory. It is registered by NATA, the national accreditation agency that ensures high quality analytical results. Analytical errors in total phosphorus, total nitrogen and turbidity analyses are I all less than 5 per cent.
I 6 Water Quality in the Snowy River Catchment Area, 1996197 Report I
The quality assurance procedures undertaken ensured that both the quantity and quality of data I from the Thredbo River and other sites reported below was of the highest standard, and that the adequacy of these data for the statistical analyses performed is beyond doubt. I 6.3. Data Analysis and Presentation I For the purposes of statistical analysis, observations that were below the detection limit for a particular parameter were shown as half the detection limit. For example, the detection limit for total phosphorus is 5 .tg[L, so data observations below this were shown as 2.5 j.tg/L. This method I of treating less than the detection limit (LOD) values by halving the LOD was suggested by Nehis and Aldand (cited by Gilbert 1987, p.lTl). This assumes a uniform distribution of values between zero and the less than detection limit (LOD). This procedure will have a minimal impact on the I estimate of the mean unless the maximum value of the observations above the detection limit is small. This procedure can also result in severe underestimation of the variance if the number of I values less than detection comprise a large proportion of the total data set. Except for Deadhorse Gap (Site 22210019) and Thredbo River DIS STW (Site 22210016), all of the sites had less than 5% of the total number of samples with LOD values for total phosphorus. LOD values for total 1 phosphorus at Deadhorse Gap (Site 2210019) and Thredbo River D/S STW comprised of approximately 20% and 7% of the total samples respectively. This may affect the resulting load estimates at this site. However, as this site had a relatively low total phosphorus load this should I not have a big impact on the overall interpretation of these load estimates.
The use of non-parametric measures of spread like the interquartile range will circumvent 1 underestimation of variances. Trend results should not be affected as the Seasonal Kendall test method is a non-parametric technique for analysing trends.
Previous reports presented values for total phosphorus and total nitrogen in milligrams per litre (mg/L). For this report, these values were converted into micrograms per litre (.tg/L), as this form I of presentation leads to numbers that are easier to comprehend at a glance'; for example, 5 .Lg/L rather than 0.005 mgIL.
I Summary statistics were calculated for each site and for each water quality determinand. Boxplots and time series plots of data were also generated and presented in this report. For a full I explanation of a boxplot, see Appendix 1. 6.3.1. Calculation of loads
I Nutrient loads were calculated for both years to determine the influence of different input sources on the quality of water downstream. Load estimation requires both nutrient concentration data and instantaneous flow data (defined as the flow measured at the date and time the samples were I collected). Reliable load estimates require sampling of both low, or base-flow, and high flow events. Hence, load calculations were only undertaken for those sites that were sampled for both I base flow and storm events. As the instantaneous flow data at the time of sampling were incomplete for some sites, the flow I data for the hour of the day closest to the sampling time was used. Alternatively, if no hourly flow I Water Quality in the Snowy River Catchment Area, 1996/97 Report I
I data were available, the mean daily flow for the day the water quality sample was collected was substituted. I The total annual flow was needed to estimate loads but short gaps were present in the daily discharge record for most sites (Figure 2). The missing daily flows were estimated from the hydrographs of neighbouring stations by regressing the discharge of the neighbouring site with the I discharge of the target site and using this relationship to predict missing stream discharge values. The best models were using time series models based on the logarithms of stream discharge and fitted with autoregressive error terms of up to lag 3. The Autoregressive time series model is used I if the successive values of the error term are not independent that is the value which the error term (pt) assumes in any one period is independent from the value which it assumed in any previous I period. If this assumption of independence is not satisfied, then we say that there is autocorrelation of the error term (Koutsiyannis, 1977). Because we are dealing with time series data such as stream discharge, often the error term is not independent. Durbin-Watson statistics was used to test H for autocorrelation. The plots of partial autocorrelation function (PACF) in SAS software provides information on the number of lags the current time series value are correlated. The number of lags were between 1 and 3 for all sites. More complicated transfer function time series models were not I developed because regression models with autocorrelated errors provided good predictions of stream discharge at neighbouring sites.
I Continuous flow data were available beginning 25 July 1995 for Little Thredbo River at Walking Track Bridge site (Site 2210015). Therefore the starting date for the calculation of loads was at I this date, giving total load estimates for 706 days. Annual loads were calculated by the ratio method (Cochran 1977). Using this method, load is I calculated as the product of the flow-weighted average of the constituent concentration and the total annual discharge for the site. For more information on this method of calculating loads refer to Appendix 2. Some sites were sampled more frequently than others resulting in biased estimates, I particularly for sites with fewer samples during storm events. Also, estimates made by combining base flow data with high flow data tends to result in the overestimation of loads, as very high values measured during flow events can have undue influence on the estimates. To minimise this I bias and to improve the precision of the load estimates, stream discharge was stratified as base flow, the rising limb of the hydrograph, and the falling limb of the hydrograph. The component I loads were then determined for each stratum and summed to provide a total load estimate.
Load values were converted to areal estimates by dividing the load by the area of each local I catchment. The areas of the various sections of the Thredbo River subcatchment between each of the sampling sites were marked on the relevant 1:50,000 scale topographic maps, and their areas I measured using a planimeter. 6.3.2. Estimation of the Error Associated with the Calculation of Loads
I 'Be errors of the load estimates were estimated by a bootstrapping algorithm (Efron and Gong, 1983). Bootstrapping is a non-parametric technique that is appropriate when the sampling distribution is unknown or the calculation of errors is difficult or intractable. A random sample of I size n is taken with replacement from the original sample of n observations. The ratio estimate of I 8 Water Quality in the Snowy River Catchment Area, 1996/97 Report F'j
load is calculated from the bootstrap sample and the resampling procedure is repeated until there I are B bootstrap estimates of the load. I The bootstrap BCa method was used to calculate 95% confidence intervals from B1000 bootstrap replicates (Efron and Tibshirani 1993). The bootstrap mean and standard error of the load estimates were also calculated.
6.3.3. Trend Assessment I The Seasonal Kendall trend test was used to detect trends in total phosphorus, total nitrogen and turbidity. Data collected for the period 1992 to 1997 for the seven sites within the Thredbo River catchment, Wollondibby Creek at Gunnadoo, the Mowamba Aqueduct at the Lake Jindabyne I outfall, Perisher Creek at Blue Cow, and the Snowy River at Dalgety were analysed for trend.
The Seasonal Kendall test is a non-parametric analysis for monotonic trend. Apart from detecting I trend, it calculates a Seasonal Kendall slope estimator which provides an estimate and the probability of the magnitude of the trend. The Seasonal Kendall test is not exact, however, when the data is serially correlated. Statistical significance can be overestimated if a positive correlation I is present, as found by Cunningham and Morton (1983). While the data were not examined for serial correlation, it was considered that this was unlikely to be an issue in small, unregulated, upland streams. Previous experience has shown that monthly sampling intervals are adequate in avoiding serial correlation in samples collected from these types of streams.
The Seasonal Kendall test has been used by several authors to assess trends in water quality, including water quality reports done for the Key Sites Program by Preece et al. (1994) and the Salt I Trends Report by Williamson et al. (1997). To be able to assess trends using the Seasonal Kendall test, the data set for each water quality indicator for each site has to span a period of at least four years. Also, at least 24 data points need I to be present over a 5-year period (the maximum possible is 60 points - that is one data point for each month for 5 years).
I Trend is considered to exist if the Seasonal Kendall test result has a probability (p) of occurring by chance of less than 0.05 (p<0.05). The magnitude of trend is given by the Seasonal Kendall slope, I where negative slope represents an improvement in water quality and positive slope represents deterioration. I In order to make the trend slopes more interpretable, concentrations were log-transformed before applying the Seasonal Kendall test. The resulting log concentration slope can be expressed in percent per year (I-{irsch et al., 1991). The estimated slope expressed as the percentage change I from the beginning of any year to the end of that year will be (eB]) x 100 where B is the estimated slope. The percentage change was converted into original units by multiplying (eB1) with the median concentration. The annual trends are point estimates. The 95% confidence intervals I provide an indication of the precision of the trend slope estimates or of the power of the analysis. Thus, a broad confidence interval shows that the point estimate is poor while a narrow confidence I interval shows that the point estimate is good.
Water Quality in the Snowy River Catchment Area, 1996/97 Report H
I The Seasonal Kendall test was calculated for both raw data and for data from which the effect of flow was removed. The benefits in calculating trend from flow adjusted data are: I 1. Flow may be one of the main factors contributing to variation in water quality through time; 2. Removal of variability associated with flow improves the power and efficiency of the trend I test; and 3. It helps to determine whether a trend observed in the raw water quality data is an artefact of I the prevailing flow pattern or is due to other factors. The process of removing the flow effects from the data involves modelling concentration and flow I in an equation. The LOESS (Locally Weighted Scatterplot Smoothing) technique (for an explanation of LOESS; see below) was used to fit log concentrations with log flow. The residuals (the difference between the concentration value and the predicted value) were then used for the I seasonal Kendall test. Another approach to examining trend is the use of a non-parametric smoothing technique called I LOESS (Cleveland, 1979). This graphical technique was also applied to the water quality data. This technique has been recommended as it complements or provides an alternative to the Seasonal Kendall test, and is more sensitive where the trend is not merely a simple change in one overall I direction. (Esterby, 1992). \Vhile the Seasonal Kendall test is limited to testing a monotonic increase or decrease in trend, LOESS is an invaluable complement by not treating trend as constant, and demonstrates trend variability over time (Williamson, 1997). Another advantage of I using LOESS is that it doesn't have an underlying assumption of linearity and normal distributions of data. I LOESS plots have been produced for all sites with sufficient data. They have been used to illustrate graphically the pattern of both raw data and flow adjusted data over time. I I I I I I I I Water Quality in the Snowy River Catchment Area, 1996/97 Report 10 I
I 7. RESULTS 7.1. Hydrology I Hydrographs for sampling sites within the Thredbo River subcatchment, and for the major inflow sites to Lake Jindabyne, from July 1995 to June 1997, are shown in Figure 2. Flow in the Thredbo I River increased considerably with distance downstream from Deadhorse Gap (Site 22210019) to Bundilla (Site 22210012). There appears to be little difference in flow between the two sites on the Little Thredbo River. Likewise there appears to be little difference in flow between sites on the I Thredbo River from Bundilla downstream. Inflow to the Thredbo River from sources between Bundilla (Site 2210012) and Paddys Corner (Site 222541), including the Little Thredbo River, I Sawpit Creek and other streams must be small in comparison to the flow that is already present in the Thredbo River at Bundilla.
Most of the inflow to Lake Jindabyne comes from the Thredbo River, with inflows from Wollondibby Creek (Site 222544) and the Mowainba Aqueduct (Site 22210142) being small in comparison.
Total inflows to Lake Jindabyne were slightly greater in 1995/96 compared to 1996/97. Total flow in the Thredbo River at Paddys Corner was 205,894 ML in 1995/96, compared to 199,560 ML in 1996/97 (Tom Freece, SMI-IEA, pers. comm). The Mowamba Aqueduct contributed 37,536 ML and 32,135 ML in 1995/96 and 1996/97 respectively, while inflows from Wollondibby Creek totalled 7,368 ML and 4,415 ML in these years. There tended to be a longer period of high flow in the Thredbo River in 1995/96 compared to 1996/97, although the actual high flows were of a smaller volume than in the following year.
I
1 I I I
I 11 Wafer Quality in the Snowy River Catchment Area, 1996/97 Report 22210019 - Thredbo River at Deadhorse Gap
5000
4000
3000
2000
1000
0
01JUL95 01JAN96 01JUL96 01JAN97 01JUL97
22210016 - Thredbo River D/S Thredbo STW
5000
4000
3000
2000
1000
0
01JUL95 01JAN96 01JUL96 01JAN97 01JUL97
22210012 - Thredbo River at Bundilla
5000 > 4000
3000
2000
1000
0•
01JUL95 01JAN96 01JUL96 01JAN97 01JUL97
22210017 - Little Threbro River at Alpine Way
5000
4000
3000
2000
1000
0
01JUL95 01JAN96 01JUL96 01JAN97 01JUL97
22210015 - Little Thredbo River © Walking Track Bridge
5000
4000
3000
2000
1000
0 II
01JUL95 01JAN96 01JUL96 01JAN97 01JUL97
Figure 2 continued next page
12 Water Qualn'y in the Snowy River Catchment Area, 1996/97 Report I
I 2221 001 1 - Thredbo River D/S Little Thredbo confluence
5000 > I 4000 3000
2000 I 1000 0 II I 01JUL95 01JAN96 01JUL96 01JAN97 01JUL97 222541 - Thredbo River at Raddys Corner
5000 I > 4000
3000 ZI I 2000 1000