Monitoring River Nutrient Loads to the Gippsland Lakes 2006–07

ENVIRONMENT REPORT

MONITORING RIVER NUTRIENT LOADS TO THE GIPPSLAND LAKES 2006–07

Report to the Gippsland Lakes Task Force Task RCIP EG-0607-04.018

Publication 1271 February 2009

1 MONITORING RIVER NUTRIENT LOADS TO THE GIPPSLAND LAKES 2006–07

SUMMARY bushfires in the upper catchments and to several major flood events. This report presents the results of a monitoring • Monitoring of nutrient loads immediately program designed to assess the loads of nutrients downstream of the bushfire-affected areas on the entering the Gippsland Lakes during 2006—07. EPA Thomson, Macalister, Avon and Mitchell Rivers Victoria conducted the monitoring program for the confirmed that these areas were a major source Gippsland Lakes and Catchment Task Force. of nutrients to the Gippsland Lakes. Lake The monitoring program measured the loads of the two Glenmaggie appears to have retained a main nutrients (phosphorus and nitrogen) entering the considerable proportion of the nutrients Gippsland Lakes from the six major rivers that drain the transported by the Macalister River following the catchment of the lakes over the 12-month period from fires, reducing the load that would have otherwise 1 July 2006 to 30 June 2007. This report describes the entered the lakes. nutrient loads contributed by the individual rivers, irrigation drains and major flood events. A number of catchments within the region were affected by major bushfires in December 2006 and a major flood (a one-in-100–year event) that commenced in June 2007 and continued into July. While the flood extended beyond the nominal study period, the magnitude of this event warranted a separate assessment of the nutrient loads over the full period of the event. Nutrient loads up to 30 June and the loads associated with the flood event are reported in section 4.3. Additional nutrient load measurements were taken downstream of the fire-affected catchments for the period February 2007 to February 2008. These measurements assist in understanding the significance of the fire-affected catchments as sources of nutrients. The main findings presented in this report are summarised as follows: • The estimated nutrient loads entering the Gippsland Lakes from the six main rivers and Macalister Irrigation District irrigation drains (phosphorus only) in the period 1 July 2006 to 30 June 2007 were:

{ total nitrogen — 2731 tonnes { total phosphorus — 376 tonnes. • The majority of the nutrients were transported during two flood events in June 2007. • The estimated loads occurring during the floods of June/July 20071 were:

{ total nitrogen — 3096 tonnes

{ total phosphorus — 329 tonnes. • The three western rivers contributed less nutrient load than the three eastern rivers. This may reflect differences in rainfall across catchments and the possible role of Lake Glenmaggie in retaining nutrient loads transported by the Macalister River. • The nutrient loads entering the Gippsland Lakes during 2006—07 were higher than in the past few years and this can be attributed to the severe

1 The estimates for the flood event include loads from July 2007 and there are not directly comparable with the loads presented above for the 2006—07 year.

2 MONITORING RIVER NUTRIENT LOADS TO THE GIPPSLAND LAKES 2006–07

1. BACKGROUND The Gippsland Lakes make up an important ecosystem in terms of environmental, economic, The Gippsland Lakes are a series of large, estuarine cultural and social values. Its wetlands are listed lakes situated in the south-eastern corner of Australia under a number international conventions and with a total area of 364 km2 (Figure 1). The lakes are treaties, including the Ramsar convention, the Japan generally shallow, with Lake Wellington in the west Australia Migratory Bird Agreement (JAMBA) and the having an average depth of 2.6 m, Lake Victoria 4.8 m China Australia Migratory Bird Agreement (CAMBA) and Lake King 5.4 m (Webster et al. 2001). At the (Anon. 2003). The Gippsland Lakes are recognized as eastern end of the Gippsland Lakes is a man-made an important nursery ground for a diverse selection channel (Lakes Entrance). Six major rivers drain a total of aquatic species, some with commercial importance catchment area of 20,600 km2, which represents about (Rigby 1982, Ramm 1983, Coutin et al. 1996). nine per cent of the total land area of Victoria (Webster The Gippsland Lakes are an important tourist et al. 2001). destination and tourism is an important element in economic and employment growth of the region (Anon. 2005).

Lake King

Lake Victoria Lakes Entrance

Lake Wellington

Bass Straight

Figure 1. Location of Gippsland Lakes, Victoria and monitoring stations

Figure 1: Location of the Gippsland Lakes, Victoria, Australia.

3 MONITORING RIVER NUTRIENT LOADS TO THE GIPPSLAND LAKES 2006–07

2. WHY MEASURE NUTRIENT LOADS In response, nutrient reduction programs have been initiated throughout the Gippsland region. In addition, ENTERING THE GIPPSLAND LAKES? a more intensive monitoring program was initiated that measures both low and high-flow events to more A number of reviews have considered the major accurately estimate nutrient loads to the Gippsland environmental pressures and associated problems that Lakes. This report presents the results of this more pose a threat to the integrity of the Gippsland Lakes intensive load monitoring program. ecosystem (Harris et al. 1998, Webster et al. 2001). The riverine nutrient load monitoring program The catchment surrounding the Gippsland Lakes has provides an annual measure of nutrient loads to the significantly altered since the settlement of Europeans Gippsland Lakes. These annual estimates can be in the mid-19th century. For example, forests have been compared to past annual loads to see whether cleared to make way for agriculture, mining and nutrient loads are decreasing or not. Variability urbanisation. As a result more nutrients, toxicants and between years means that any evaluation needs to sediment have been washed from the catchment, consider the long-term trend in nutrient loads rather through the rivers and into the lakes. Of considerable than simply comparing one year with the previous environmental and economic concern is the frequency one. and intensity of cyanobacterial blooms. The annual estimate of nutrient loads provides an Increased nutrient loads to the Gippsland Lakes have indicator for environmental managers and funding led to the degradation of the ecosystem. Impacts bodies in assessing whether management actions are include increased algal blooms and decreases in on track to meet the established nutrient reduction seagrass and fish populations. Dissolved oxygen targets. concentrations in the lakes are also affected by increased nutrient loads (Harris et al. 1998). The program is made possible by the continuing commitment of the Victorian State Government The Gippsland Lakes Future Directions and Action Plan through the Gippsland Lakes Rescue Package (RCIP (GLFD&AP 2002) established a target of a 40 per cent EG–0607–04.018). This program operates in reduction in nutrient loads (total nitrogen and total cooperation with the Gippsland Regional Water phosphorus) entering the Gippsland Lakes by 2022. This Monitoring Partnership (GRWMP). reduction is benchmarked against nutrient loads estimated during the CSIRO Lakes Modelling Project (Webster et al. 2001).

4 MONITORING RIVER NUTRIENT LOADS TO THE GIPPSLAND LAKES 2006–07

3 MONITORING AND ESTIMATING LOADS Table 1: Site description and percentage of catchment captured by the nutrient loads monitoring program 3.1 Where did we measure? Percentage Site The Gippsland Lakes catchment contains six major River Location catchment captured number rivers that drain into the lakes (Figure 1). (from Scanlon 2007) Three of these rivers (Thomson including the Macalister 223209 Tambo Battens Landing 77 River, Latrobe and Avon) are situated at the western end of the lakes and drain into Lake Wellington. These 223210 Nicholson Sarsfield 77 rivers also receive drainage from a large proportion of the Macalister Irrigation District (MID). 224217 Mitchell Rosehill 91 The three eastern rivers (Mitchell, Nicholson and 225232 Thomson Bundalaguah 80 Tambo) drain into Lake King towards the eastern end of (Incl. 100% the lakes. Macalister R) Monitoring sites were established in the lower reaches 225234 Avon Clyde Bank 80 of each river in 2004 and 2005 (Figure 2). These sites were chosen on the following criteria: 226227 Latrobe Kilmany South 95 • availability of continuous flow data • capacity to capture nutrient run-off from a high In February 2007 additional monitoring sites were proportion of the catchment established to assess the impact of fire-affected • logistics (occupational health & safety, site access catchments. The results of this monitoring are and establishment costs etc) described in Section 4.5. This additional monitoring • availability of comparable historical data. was not funded by the Gippsland Task Force, but the Site details and the proportion of each catchment results are presented in this report for completeness. captured by each monitoring site are summarised in table 1.

Nicholson River (Sarsfield)

Tambo River Mitchell River (Rosehill) (Battens Landing)

Lake King

Lake Victoria Avon River (Clydebank)

Thomson River (Bundalaguah) Lake Wellington

Latrobe River (Kilmany south)

Figure 2: Monitoring site locations and site names.

5 MONITORING RIVER NUTRIENT LOADS TO THE GIPPSLAND LAKES 2006–07

3.2 How are the nutrients measured? Data from this monitoring program is provided to the Victorian Water Data Warehouse Automated water samplers (ISCO model 6712) are (www.vicwaterdata.net). located at each monitoring station in a protective housing (Figure 3). All monitoring sites have 3.3 Calculation of nutrient loads continuous-flow gauging and telemetry. Strategies for the efficient sampling and techniques Water samplers start collecting water when the river for estimation of load are discussed in Fox (2007). The level rises. Selection of the trigger water level is based recommended strategy for nutrient load ‘estimates’ on historical records of typical base flows. During a for rivers is that the load be considered in two parts: high-flow event, sampling of the rising and falling river the peak flow and ‘non-peak’ or base flow. The ideal level is at specific increments in river level. This sampling strategy for peak-flow events is to obtain intensive sampling strategy has been recommended accurate empirical load estimates using sampling for accurate load determination (Fox 2007). equipment that is automatically triggered at a preset During a high-flow event, a contractor is notified by flow threshold. telemetry to retrieve water samples for nutrient In addition, regular (weekly to monthly) samples analysis. The samples are typically retrieved as soon (automatic or grab) are being collected throughout the as possible after it is safe to access the area. The year to monitor nutrients during base flow. contractor also visits the site weekly to collect water Daily monitoring data provided for the Tambo River, samples for analysis of normal (base) flow conditions. coupled with the peak-event sampling, has been used Water samples are analysed for nutrient to improve the understanding of uncertainty estimates concentrations by a NATA-accredited laboratory for load monitoring programs, including this program (Ecowise). (Fox 2007, Davies & Martinez 2006). This monitoring was conducted within the framework In this report annual nutrient loads are presented for of the Gippsland Regional Water Monitoring the period from 1 July 2006 to 30 June 2007. A Partnership (GRWMP) and sampling and analysis separate assessment of loads during the June/July methods, including the quality assurance protocols 2007 flood event is also presented. agreed between the GRWMP and contractors. Appendices 1 and 2 contain details of calculations and analysis, including quality assurance procedures used in this program.

b) a) c)

Figure 3: Auto water sampler and housing designs (a: Tambo River; b: Mitchell River; c: Avon River).

6 MONITORING RIVER NUTRIENT LOADS TO THE GIPPSLAND LAKES 2006–07

The primary data set used in the load computation Four automated water samplers (ISCO model 6712) comes from the sites described in section 3.1. To were installed, one instrument on each of the increase the accuracy of our load estimates, Macalister, Thomson, Avon and Mitchell Rivers. The phosphorus load from the two MID irrigation drains that samplers were located adjacent to the rivers, enter the system below our monitoring sites has also immediately downstream of the fire-impacted areas. been included. The two MID drains are the Central The sampler locations were as follows (Figure 13): Gippsland Drain Number 4 (CGd4) flowing into the • Mitchell River at Lindenow Thomson River, and the Lake Wellington Main Drain • Macalister River at Stringy Bark Creek (LWMD), which empties directly into Lake Wellington. • Avon River at Valencia Creek Initial installation of automated sampling equipment • Thomson River at Cowarr Weir. occurred in September 2004, so annual loads were initially calculated on a 1 September to 31 August basis No base-flow monitoring was undertaken at these to provide a full year of data. This continued in sites. High-flow samples were collected using the 2005—06. In this report the load estimates for 2004—05 same approach as for the Gippsland Task Force- (10 months) and 2005—06 are recalculated for the funded program. These results are presented in period 1 July to 30 June. The 2004—05 period has not Section 4.5. been assessed as a full year, given that data was only Calculation of loads and associated errors available for 10 months. Tan et al. (2005) undertook a review of load Flood impacts estimation methods currently used and developed a Major floods occurred in Gippsland in late June 2007, simplified typology of load estimation techniques continuing into July 2007. Load estimates targeting the (Table 2) to provide guidance when choosing a load impact of flood have been determined and results are computation method. This work suggests that the presented in Section 4.3. most suitable estimation technique for the sampling strategy used during this project involves the Bushfire impacts ‘averaging or ratio estimation methods that consider Major bushfires occurred in the catchments of the seasonal or flow stratification’ (Table 2). Macalister, Thompson, Avon and Mitchell Rivers from This approach was used to estimate the nutrient December 2006 to early 2007. The effect of these loads for all the rivers considered in this project. bushfires on nutrient export from these catchments was These calculations and associated errors were studied for the period from February 2007 to February undertaken by Environmetrics Australia and are 2008. presented in Section 4.1. Details are in Appendix 1.

Sampling regime Relationship Regular sampling between (e.g. weekly, fortnightly) flow and Sparse sampling Continuous sampling concentration (monthly or less frequent) (e.g. daily, near daily) Limited event data Representative event data

No significant Averaging or ratio Averaging or ratio relationship Averaging or ratio • seasonal-stratified Linear interpolation • seasonal-stratified present • flow regime-stratified Significant Averaging or ratio Regression or averaging or ratio relationship Regression • seasonal-stratified Linear interpolation • seasonal-stratified present • flow regime-stratified

Table 2: Typology of load estimation (adapted from Tan et al. 2005).

7 MONITORING RIVER NUTRIENT LOADS TO THE GIPPSLAND LAKES 2006–07

4. WHAT ARE THE GIPPSLAND LAKES The coefficient of variation (CV)2 has been selected as an appropriate indicator of uncertainty. The highest NUTRIENT LOADS? uncertainty in load estimates was for Tambo River TN loads (114.2 per cent) and the Mitchell TP load (427.3 4.1. Overview per cent, Table 3). The estimated nutrient loads to the lakes from 1 July TN loads ranged from 137 tonnes (Nicholson River) to 2006 to 30 June 2007 are presented in Table 3 with 912 tonnes for the Mitchell River. Such differences can their associated errors. also be observed for the TP loads, with the Nicholson Percentage contribution to the total load from each River being the lowest contributor of TP to the lakes river is presented in figures 4 and 5. The total (12 tonnes) and the Mitchell the largest, with more phosphorus (TP) load includes an estimate for the MID than ten times that input (169 tonnes). drains that enter below the monitoring sites included With around three tonnes of TP for the year, the MID in this program (CGd4 and LWMD). The estimate for input appears relatively small compared to the total these drains is based on data available from other load to the lakes or when its relative contribution is programs. Appropriate monitoring data for total compared with each individual river (Figure 5). nitrogen (TN) in the relevant MID drains was not available and hence the estimated nitrogen load from these drains could not be calculated.

Table 3: Gippsland Lakes nutrient loads for 2006–07

2006–07 nutrient load estimates (tonnes)

TN TP Site code River name Estimate Std error CV (%) Estimate Std error CV (%) 223209 Tambo 598.7 683.7 114.2 47.5 28.5 59.9 223210 Nicholson 136.8 70.6 51.6 11.6 2.1 18.1 224217 Mitchell 912.4 750.0 82.2 168.7 721.0 427.3 225232 Thomson 422.9 14.2 3.4 39.2 1.2 2.9 225234 Avon 324.4 9.0 2.8 38.1 0.6 1.6 226227 Latrobe 335.4 16.4 4.9 68.4 2.9 4.2 CGd4 Central Gippsland Drain 4 * * * 0.95a * * LWMD Lake Wellington Main drain * * * 1.89a * * Total 2730.6 376.5

2 The coefficient of variation is the standard deviation expressed as a percentage of the mean. a From WGCMA, 2008.

8 MONITORING RIVER NUTRIENT LOADS TO THE GIPPSLAND LAKES 2006–07

Rivers contribution to the 2006-07 TN load Rivers contribution to the 2006-07 TP load

Western Western Riv ers Rivers 40% 39%

Eastern Eastern Riv ers Riv ers 60% 61%

Figure 4: Nutrient contributions from western and eastern rivers for 2006—07.

Rivers contribution to the 2006-07 TN load Rivers contribution to the 2006-07 TP load

Latrobe MID (Unknown) MID Tambo 12% 0% Tambo Latrobe 1% 13% 22% 18% Nic hols on 3% Avon 12% Nicholson Avon 5% 10%

Thomson 15% Thomson 10% Mitc hell Mitchell 45% 34%

Figure 5: Nutrient contributions from each river for 2006—07 (CGd4 and LWMD are summed under MID).

4.2 Key features of the nutrient Input to the by fires were Macalister, Avon and Mitchell Gippsland Lakes (Appendix 3). Relative contribution of eastern and western rivers Nutrient loads measured on the Thomson River (which includes inputs from the Macalister River) do In 2006—07, the western rivers (Latrobe, Thomson and not show the same impacts as those on the Avon) contributed fewer nutrients (about 40 per cent of unregulated rivers. While the Macalister River the total loads) than the three eastern rivers (Mitchell, catchment was impacted by the fires, Lake Nicholson and Tambo) (Figure 4). The Mitchell River Glenmaggie appears to have retained a substantial contributed the highest loads of nitrogen (34 per cent) proportion of the nutrient load (sediment-bound), and phosphorus (45 per cent) to the Gippsland Lakes removing particulate nutrients before these rivers during 2006—07 (Figure 5). entered the Gippsland Lakes (SRW 2007a). Nutrient It is likely that large quantities of nutrients would have concentrations found in samples from the Macalister been released in the bushfire-affected areas within the River upstream of the dam are not reflected in the upper catchments and then transported to the lakes by loads detected at the Thomson River site the floods in June 2007. The catchments most affected downstream of the confluence (Figure 14 and 15).

9 MONITORING RIVER NUTRIENT LOADS TO THE GIPPSLAND LAKES 2006–07

The water supply catchment of the Thomson dam Run-off from the 19 June event and the flood at largely escaped the impacts of the bushfires (Miletic & Licola in January 2007 were not reflected in Doherty 2006). Similarly, the Latrobe River is the increased nutrient loads measured downstream of dominant western river and it avoided both fire and the Macalister river confluence with the Thomson major flood impacts (Figure 7; note the difference in River, as the Macalister River flow did not fill Lake scale). This is reflected in figure 5, where the nutrient Glenmaggie and sediment-bound nutrients were able loads in the western rivers represent a lower proportion to settle out. The regulated western rivers (Thomson of total loads than seen in previous studies. For and Latrobe) also showed little effect from the first example, Grayson (2006) found that the western rivers high rainfall, as water from the catchments first filled (Latrobe, Thomson and Avon) contributed 62 per cent water storages to capacity. 3 of TSS , 65 per cent of TP and 56 per cent of TN. In contrast, the hydrology of the unregulated rivers High-rainfall event shows two sharp river flow peaks corresponding to the two rainfall events (Figures 8—11). The Tambo The nutrient loads entering the Gippsland Lakes during River exhibited a small increase in river flow from the 2006—07 largely occurred in two high-flow events: the first rainfall event (Figure 11) — this was most likely first on 19 June 2007 and the second on 27 June 2007. associated with rainfall patterns over this catchment. Of the 377 tonnes of TP the rivers delivered to the Gippsland Lakes in 2006—07, 58 per cent was The second major rainfall event resulted in water transported in an 11-day period (19—30 June 2007). storages overflowing (SRW 2007b, Knowles 2007) and nutrients from this event were carried down to Similarly, of the 2731 tonnes of TN entering the the Gippsland Lakes in the flood, rather than as a Gippsland Lakes, 67 per cent was associated with the gradual release of water with elevated TN (SRW June floods. Figures 6—11 show the hydrographs for 2007a). each river from 1 July 2006 to 30 June 2007, including The nutrient percentages shown in Figures 6 and 7 the percentage contribution of each high-flow event on 4 the total nutrient load. relate only to nutrients added before 30 June 2007 . It is likely that the first large rainfall event flushed For a long period (most of the year) there had been away part of those stored nutrients from the upper little rain and, thus, little transport of nutrients from the catchment and delivered them to water storages in catchment. This led to an increase in nutrients stored in the catchment (in other words, to Lake Glenmaggie) the upper catchment, readily available for transport. and to the lakes. The major bushfires in early 2007 provided an extra source of nutrients for mobilisation following heavy rains (see section 4.5).

4 Refer to Section 4.4 for further discussion of nutrient loads associated specifically with the June/July 2007 flood event. The flood commenced during the reporting period, though the impacts carried into the 2007—08 period; however, the scale of this event warrants consideration 3 Total suspended solids. as a single event.

10 MONITORING RIVER NUTRIENT LOADS TO THE GIPPSLAND LAKES 2006–07

Thomson River Mitchell River

35000 160000 49 % TP 30000 120000 71% TN 25000 61 % TP 20000 51 % TN 80000 15000

Flow [Ml/d] Flow [Ml/d] 40000 10000 5000 0 0 1-Jul-06 29-Sep-06 28-Dec-06 28-Mar-07 26-Jun-07 01-Jul-06 29-Sep-06 28-Dec-06 28-Mar-07 26-Jun-07 Date Date

Figure 6: Thomson River hydrograph. Figure 9: Mitchell River hydrograph. Percentage TP and TN loads from a flood occurring Percentage TP and TN loads from two floods in the year ending 30 June 2007. occurring in the year ending 30 June 2007.

Latrobe River Nicholson River

7000 7000

6000 22 % TP 6000 22 % TP

5000 17% TN 5000 17% TN 4000 4000 3000 3000 Flow [Ml/d] Flow [Ml/d] Flow 2000 2000

1000 1000

0 0 1-Jul-06 29-Sep-06 28-Dec-06 28-Mar-07 26-Jun-07 1-Jul-06 29-Sep-06 28-Dec-06 28-Mar-07 26-Jun-07 Date Date

Figure 7: Latrobe River hydrograph. Figure 10: Nicholson River hydrograph. Percentage TP and TN loads from a flood for the Percentage TP and TN loads from two floods year ending 30 June 2007. occurring in the year ending 30 June 2007.

Avon River Tambo River

35000 70000

60000 30000 83 % TP 78 % TP 50000 25000 99 % TN 89 % TN 20000 40000 15000 30000 Flow [Ml/d] Flow [Ml/d] 10000 20000

5000 10000

0 0 1-Jul-2006 29-Sep-2006 28-Dec-2006 28-Mar-2007 26-Jun-2007 1-Jul-06 29-Sep-06 28-Dec-06 28-Mar-07 26-Jun-07 Date Date

Figure 8: Avon River hydrograph. Figure 11: Tambo River hydrograph. Percentage TP and TN loads from two floods Percentage TP and TN loads from two floods occurring in the year ending 30 June 2007. occurring in the year ending 30 June 2007.

11 MONITORING RIVER NUTRIENT LOADS TO THE GIPPSLAND LAKES 2006–07

4.3 How did the major flood in June/July 2007 (GLTF)-funded programs have estimated nutrient affect nutrient loads to the Gippsland Lakes? loads for the 2004—05, 2005—06 and 2006—07 periods (Figure 12). In June/July 2007 a major flood affected the Gippsland region (Appendix 3). A small flood occurred The load estimates for the various programs used a on 19 June and further major flooding began late on number of different monitoring and modelling the 27 June and persisted for several days, with approaches (including SedNet modelling using long- floodwater starting to recede by early July term average loads of sediments based on monitoring, (Thankappan 2007). The period of flooding extended land use and catchment slope; water quality beyond the program-reporting period (30 June 2007). monitoring; and flow monitoring) to estimate loads Due to the significance of large floods in delivering based on assumptions of loads associated with high- nutrient loads to the lakes (Harris et al. 1998; Davies & flow events (Grayson 2006). Martinez 2006), the total input from this specific flood The previous nutrient load estimates (EPA, Grayson’s event has been included in this report as a specific and previous GLTF) vary substantially from year to event crossing over reporting periods. year (Figure 12). This large annual variation is Nutrient loads from the flood have been calculated for dependent on many factors, such as rainfall, storm the Gippsland Lakes and each river (Table 4). In total, events and sampling method (for example, some an estimated 329 tonnes of TP and 3096 tonnes of TN estimates missed high-flow events). If these previous entered the Gippsland Lakes from this flood event. The estimates were averaged (represented as horizontal monitoring data indicates that the majority of the lines), the average TN load would be 1963±1185 tonnes nutrient loads delivered to the Lakes was associated and 226±150 tonnes for TP. The nutrient loads with the fast-rising phase of the flood. estimated for 2006—07 are slightly above these averages (Figure 12). The load estimates include loads from the onset of fast-rising rivers (19 June 2007) to when river height Average loads are included for comparison purposes receded to approximately base flow or half-peak flow. only. They are based on data included in Grayson Some regulated rivers continued to release dam water (2006) and provide a general guide, as the load well into July, meaning the river flows did not return estimates are drawn from a range of different to normal for some time. programs, based on different techniques and different levels of certainty. 4.4 How do nutrient loads compare with previous The estimated loads for 2006—07 differ from previous estimates? load estimates in that the calculations are based on Nutrient loads to the Gippsland Lakes have been weekly base-flow measurements of nutrient previously estimated for the periods 1977—78, concentrations; and peak event samples collected 1978—79, 1980—81, 1984—85, 1988—89 and 1989—90 automatically on the rising and falling sections of the (Grayson et al. 2006). Additionally, Grayson et al. hydrograph. This is expected to provide a more (2001) also estimated nutrient loads based on accurate assessment of river loads (Grayson 2006, catchment model results for the periods 1975—99, Fox 2007). 1995—99 and 1997—99. The Gippsland Lakes Taskforce

Table 4: TP and TN load during the June/July 2007 flood.

Site code River and flow period measured TP (tonnes) TN (tonnes)

225232 Thomson (19/6/07 — 31/7/07) 64 734

226227 Latrobe (19/6/07 — 31/7/07) 35 303

225234 Avon (19/6/07 — 31/7/07) 65 587

224217 Mitchell (19/6/07 — 31/7/07) 115 670

223210 Nicholson (19/6/07 — 31/7/07) 9 125

223209 Tambo (19/6/07 — 31/7/07) 41 677

Total to Gippsland Lakes 329 3096

12 MONITORING RIVER NUTRIENT LOADS TO THE GIPPSLAND LAKES 2006–07

4.5 What was the impact of major wildfires on addition, the flood — which commenced on 26 June nutrient loads? 2007 — was of such a magnitude that the equipment on the Mitchell River at Lindenow was washed away The loads estimate for 2006—07 followed both an and the equipment on the Macalister River at Stringy extended period of drought and severe bushfires that Bark Creek, Avon at Valencia Creek and the Thomson affected large sections of the Gippsland Lakes River at Cowarr Weir were inundated and no samples catchment. Bushfires have previously been identified were collected. Therefore, the data assessed is only as increasing nutrient loads to the Gippsland Lakes for the smaller-scale events in the study period. and changing water quality (Sheridan et al. 2006, Feikema et al. 2005). With the exceptions of the issues noted above, the samplers captured most of the base load from the In light of the fires and floods of 2006—07, it is not bushfire-affected area and show similar increases in surprising that nutrient loads have increased nutrient loads at the downstream sites (with a time lag compared to the previous few years, where loads were — Figures 14 and 15). estimated based on similar data to that for 2006—07. The calculated loads for 2004—05 and 2005—06 The exception was the Macalister River. While there is occurred during a drought period and show historically no sampling site on the Macalister River immediately low nutrient loads. downstream of Lake Glenmaggie, the levels upstream were more than 100 times those measured in the To measure the impact of fire-affected catchments on Thomson River downstream of the confluence with the nutrient loads entering the Gippsland Lakes, auto Macalister River. The Thomson River loads were samplers were installed immediately downstream of accounted for but the Macalister River loads were not. fire-impacted catchments. Locations are marked on It is likely that Lake Glenmaggie was acting as a the map in figure 13. retention basin for nutrients (especially the sediment- The installation of automatic sampling equipment was bound fraction), as the lake’s turbidity levels increased not completed prior to the major flood of the significantly during this period (SRW 2007a). Macalister River at Licola on 26 February 2007. In

5000 Total Nitrogen Total Phosphorus

4000

3000

2000 Tonnes/Year

1000

0 75-99 95-99 97-99 77-78 78-79 80-81 84-85 88-89 89-90 04-05 05-06 06-07 Grayson et al. EPA Monitoring GLTF Year

Figure 12: Historical estimates of total nitrogen and phosphorus entering the Gippsland Lakes Represents average TN loads over all years Represents average TP loads over all years

13 MONITORING RIVER NUTRIENT LOADS TO THE GIPPSLAND LAKES 2006–07

225221

225255

225228 225234 225232

226227

Figure 13: Fire impact sampling sites (red spots) in Thomson Basin. Catchment load sites (green spots).

14 MONITORING RIVER NUTRIENT LOADS TO THE GIPPSLAND LAKES 2006–07

a) Load TN (Tonnes)

140 1080 tonnes 220 tonnes 224219 (Mitchell R) 120 225221 (Macalister R) 225228 (Thomson R) 225255 (Avon R) 100

0 Jan-07 Feb-07 Mar-07 Apr-07 May-07 Jun-07 Jul-07 Aug-07 Sep-07 Oct-07 Nov-07

b) Load TN (Tonnes)

1200 224219 (Mitchell R) 225221 (Macalister R) 1000 225228 (Thomson R) 225255 (Avon R) 800

600

400

200

0 Jan-07 Feb-07 Mar-07 Apr-07 May-07 Jun-07 Jul-07 Aug-07 Sep-07 Oct-07 Nov-07

Figure 14: Total nitrogen loads downstream of fire-impacted catchments (a: modified scale; b: full scale).

4.6 What are the uncertainties in nutrient load for total nitrogen of between two and five per cent for estimates? the Latrobe, Thomson and Avon Rivers, while the coefficient of variation for the Mitchell was 82 per The nutrient load estimates for 2006—07 include a cent, the Nicholson 52 per cent and the Tambo 114 per number of uncertainties that need to be considered. cent. The error estimation in quantifying nutrient loads can Similarly the coefficient of variation for total be reduced with a greater knowledge of the transport phosphorus for the Latrobe, Thomson and Avon varied processes, variability of sample data and errors from 1.6 to 4.2 per cent, while the coefficient of associated with measurement (Etchells et al. 2006). variation for the Mitchell was 427 per cent, the The intensity of monitoring data provided by this Nicholson 18 per cent and the Tambo 59.9 per cent. program is helping to build understanding in sample data variability and the link between flow and The results are reasonably consistent in that the concentrations for the rivers flowing into the western rivers’ loads are considered to be of high Gippsland Lakes. reliability for both total nitrogen and total phosphorus. Estimations undertaken using the full suite of methods However, the eastern rivers, especially the Mitchell undertaken by Envirometrics Australia in and the Tambo, provided a challenge in that the nature 2007(Appendix 3c) indicated a coefficient of variation of the loads results from few events and is difficult to

15 MONITORING RIVER NUTRIENT LOADS TO THE GIPPSLAND LAKES 2006–07

a) Load TP (Tonnes)

20 965 tonnes 30 tonnes 224219 (Mitchell R) 18 225221 (Macalister R) 225228 (Thomson R) 16 225255 (Avon R) 14

0 Jan-07 Feb-07 Mar-07 Apr-07 May-07 Jun-07 Jul-07 Aug-07 Sep-07 Oct-07 Nov-07

b) Load TP (Tonnes)

1200 224219 (Mitchell R) 225221 (Macalister R) 1000 225228 (Thomson R) 225255 (Avon R)

800

600

400

200

0 Jan-07 Feb-07 Mar-07 Apr-07 May-07 Jun-07 Jul-07 Aug-07 Sep-07 Oct-07 Nov-07

Figure 15: Total phosphorus loads downstream of fire-impacted catchments (a: modified scale; b: full scale).

quantify using approaches based on grouping of monitoring data on a daily, weekly or monthly basis with application of mean-based estimators and ratios. The monitoring approach used on the Tambo has included daily sampling in conjunction with triggered, high-flow events and this level of monitoring may assist in modelling the loads for this system to provide more accurate estimations. Scanlon (unpublished data) investigated uncertainty in flow measurements at 80 Gippsland River sites and found the greatest uncertainties in low or high-flow measures, indicating that rivers such as the Tambo, which has the majority of the nutrient loads delivered in a few high-flow events are subject to greater errors.

16 MONITORING RIVER NUTRIENT LOADS TO THE GIPPSLAND LAKES 2006–07

5 CONCLUSION 6 ACKNOWLEDGEMENTS

As noted by Grayson et al. (2006), the EPA monitoring EPA Victoria thanks David Fox and Roger Grayson network for nutrient loads to the Gippsland Lakes from from Melbourne University, who provided valuable each of the major basins provided the data needed to expert advice and guidance. confirm the overall contributions of TSS, TP and TN at Warren Davies and Guillaume Martinez (EPA) were the basin scale; other monitoring programs have not responsible for management, data analysis and been at the required frequency to establish loads at a reporting for this program. similar accuracy. Ecowise Environmental (Victoria) collected sample in To improve the confidence in the estimation of loads base-flow conditions and conducted nutrient analysis entering the lakes there would need to be monitoring for all samples. Analysis was conducted using NATA- locations duplicating the EPA ‘end-of-basin’ network at accredited methods (See Appendix 3 for details). (or close to) the outlets of each catchment. These monitoring tasks would be expensive and would need Thiess Services undertook collection of peak-flow to be in place for five to 10 years before reliable event samples from the automatic sampling estimates of average load could be computed. Both equipment once conditions were safe and conducted Grayson and Fox indicate that the monitoring maintenance and calibration of field equipment. approach used in this program is the appropriate methodology for improving the accuracy of loads estimations. Nutrient loads in 2006—07 were higher than in the previous few years, as a result of major bushfires followed by severe floods. The bushfires and floods make it difficult to assess the effectiveness of nutrient reduction programs. In February 2007, EPA installed water samplers just below the fire-affected catchments in west Gippsland and the results from this program may help separate nutrient loads from the fires and nutrient loads from the MID. The major rainfall event in June/July 2007 added a substantial load of nutrients to the Gippsland Lakes. The scale of the monitoring makes it possible to calculate changes in overall loads entering the lakes. However, as all work prior to this program was modelled and based on limited water quality data (especially in relation to high-flow events), it provides a baseline for comparison that may include an uncertainty greater than 100 per cent (Webster et al. 2001). To determine the impact of specific management actions in reducing nutrient loads, much more focused monitoring would be required before and after each management intervention. This would allow the assessment to remove confounding variables from other actions within a catchment.

17 MONITORING RIVER NUTRIENT LOADS TO THE GIPPSLAND LAKES 2006–07

REFERENCES Letcher RA, Jackeman AJ, Merritt WS, McKee LJ, Eyre BD, Baginska B (1999). Review of techniques to Anon. (2002). Gippsland Lakes Future Direction and estimate catchment exports. EPA NSW. Actions Plan. Department of Natural Resources and Littlewood IG (1992). Estimating contaminant loads in Environment. pp. 1—38. rivers: a review. Report 117. Institute of Hydrology, Anon. (2003). Gippsland Lakes Ramsar site. Strategic Oxfordshire, UK management plan. Department of Sustainability and Miletic D, Doherty B (2007). Crunch time today for Environment. Thomson Dam, The Age Newspaper, 14 December Anon. (2005). East Gippsland Regional Catchment 2006. Strategy 2005. East Gippsland Catchment Ramm D (1983). An ecological survey of postlarval and Management Authority. juvenile fish in the Gippsland Lakes (Victoria). Davies W, Martinez G (2006). Estimation of riverine Gippsland Regional Environmental Study Ministry for nutrient input into the Gippsland Lakes during 2004— Conservation, Victoria Report. pp. 1—25. 05. May 2006, EPA Victoria. Rigby BA (1982). An ecological study of the estuarine Coutin P, Walker S, Morison A (ed.) (1996). Black bream fish assemblage in the Gippsland Lakes. Internal 1996. Assessment Report Number 14. Bay and Inlet Report No. 3. Marine Science Laboratories, Ministry for Fisheries and Stock Assessment Group. Marine and Conservation. pp. 1-55. Freshwater Resources Insititute, Queenscliff, Fisheries Scanlon (2007). Unpublished data, Melbourne Victoria. pp. 1—83. University. Environmetrics Australia (2008). Unpublished Report Sheridan G, Lane P, Grayson R, Noske P, Feikema P, to EPA Victoria. 2006/07 nutrient load estimates. Sherwin C (2006). Estimated changes in stream water Etchells T, Tan KS, Fox D (2006). Quantifying the quality following the 2003 bushfires in eastern uncertainty of nutrient load estimates in the Victoria. University of Melbourne. Research report for Shepparton irrigation region. Proc. MODSIM 2005 the bushfire recovery program. International Congress on Modelling and Simulation. SRW (2007a). Water quality report card — Lake Feikema P, Sheridan G, Lane P, Argent R, Grayson R Glenmaggie April—June 2007. Southern Rural Water. (2005). Modelling the impacts of the 2003 bushfires on SRW (2007b). Water quality report card — Lake water quality in the Gippsland Lakes catchments. Glenmaggie July—September 2007. Southern Rural School of Forest and Ecosystem Science, The Water. University of Melbourne. Research report for the bushfire recovery program. Tan KS, Etchells T, Fox DR (2005). User guide and reference manual for GUMLEAF v0.1 alpha: generator Fox DR (2007). Water quality monitoring in the for uncertainty measures and load estimates using Gippsland Lakes catchments — Efficient sampling alternative formula, Australian Centre for strategies and load estimation techniques. The Environmetric, University of Melbourne, June 2005. Australian Centre for Environmetrics. Occasional Report, May 2007, University of Melbourne, Parkville, Thankappan M (2007). AusGeo News, Issue 87, Australia. September 2007, Geoscience Australia, Australian Government Grayson RB, Tan KS, Western AW (2001). Estimation of sediment and nutrient loads into the Gippsland Lakes. Webster IT, Parslow JS, Grayson RB, Molloy RP, Centre for Environmental Applied Hydrology Report Andrewartha J, Sakov P, Tan KS, Walker SJ, 2/01, University Of Melbourne, September 2001. Wallace BB (2001). Gippsland Lakes environmental study assessing options for improving water quality Grayson R (2006). Prioritising nutrient reduction for and ecological function. CSIRO. Report to the GCB. the Gippsland Lakes and catchments. Part 1 —Loads and sources. Catchment to Seas report to the Gippsland Lakes Task Force. Harris G, Batley G, Webster I, Molloy R, Fox D (1998). Review of water quality and status of the aquatic ecosystems of the Gippsland Lakes. CSIRO. Report to the GCB. Ladson T, Tilleard J (2006). BMPs for reducing phosphorus loads to the Gippsland Lakes. Report on findings from expert panel. Moroka. Report to the Gippsland Lakes Task Force.

18 MONITORING RIVER NUTRIENT LOADS TO THE GIPPSLAND LAKES 2006–07

APPENDIX 1: ESTIMATION AND UNCERTAINTY CALCULATIONS IN NUTRIENT LOADS (ENVIROMETRICS AUSTRALIA PTY LTD)

19 2006/07 Nutrient Load Estimates Site: 223209

Nutrient Load Estimation for selected tributaries of the Gippsland Lakes

Site 223209 July 1 2006 - June 30 2007

Tambo River

Prepared for the Victorian EPA

Environmetrics Australia Pty Ltd

June 2008TT

1. Load Estimation A list of some 24 computational techniques for estimating a load was provided in Letcher et al. (2002). Most of these formulae can be classified as belonging to one of the groupings: mean‐based estimators; ratio estimators; and regression estimators. In this paper we consider a class of load estimators given by equation 1.

n ⎛⎞nc ⎛⎞q ˆ (1) LK= ⎜⎟∑∑ wcii⎜⎟ vq j j ⎝⎠ij==11⎝⎠

th th where ci is a measured concentration on the i occasion; q j is a measured flow on the j

1 occasion and wi and v j are weights . K is a constant that reconciles the sampling time-step with the period of interest (eg. if concentrations and flows represent daily values and an annual load estimate is required, then K=365).

2. Theoretical mean and variance

Before turning our attention to the properties of load estimators, it will be useful to develop some theoretical results for the expected value and variance of a load under certain distributional assumptions. In what follows we assume (not unreasonably), that the distribution of concentration ()C and flow ()Q are well described by the bivariate lognormal distribution given by equation 2 and that load, LCQ= .

fCQ, ()cq, =

2 2 11⎧⎫⎪⎪⎡⎤⎛⎞⎛⎞ln(c)− μ ln(cq )−−μμ⎛⎞⎛⎞ ln( ) ln( q ) − μ exp −−C 2ρ QQ +2 Q ⎨⎬2 ⎢⎥⎜⎟⎜⎟ 2 2(1− ρσ ) ⎜⎟⎜⎟ σ σ σ 21πσcq CQσ − ρ ⎩⎭⎪⎪⎣⎦⎢⎥⎝⎠⎝⎠CCQQ⎝⎠⎝⎠ (2) where μ and σ are the mean and standard deviation of the log-transformed data and ρ is the correlation between log concentration and log flow. Fox (2004) showed that the expected load is given by equation 3.

1 The weights are somewhat arbitrary although values are usually determined by the nature of the sampling scheme. For example, a constant weight of 1/n implies a simple average while flow weighted averaging implies weights are determined on the basis of observed flow (higher flow implying higher weight).

EL[]=

⎪⎧ 1 22 ⎪⎫ exp ⎨ μμ++⎡⎤ρσσ ++ ρσσρρσσρσσ +−2 + +⎬ ()CQ 2 ⎣⎦⎢⎥()()()()QC CQ QC CQ ⎪⎩ 21()− ρ ⎭⎪ (3)

Furthermore, it can be established that the second (uncorrected) moment is:

2 EL⎣⎦⎡⎤=

⎧⎪ 2 22 ⎪⎫ exp ⎨22μμ++⎡⎤ ρσσ ++ ρσσρρσσρσσ +− + +⎬ ()CQ 2 ⎣⎦⎢⎥()()()()QC CQ QC CQ ⎪⎩ ()1 − ρ ⎭⎪ (4) and so the variance is given as

2 2 Var[] L=− E⎣⎦⎡⎤ L() E[] L (5)

Data summary

fn= "e:\vicepa\223209.dat" Filename

n= 365 number of rows in data file

K= 361 number of non-missing flow readings

number of non-missing TN concentration readings n1= 73 number of non-missing TP concentration readings n2= 73 number of contemperaneous flow and TN measurements Kn1= 72

Kn2= 72 number of contemperaneous flow and TP measurements

Kn12= 73 number of contemperaneous TN and TP measurements

Time-series plot of flow and TN 5 110× 5 Flow (ML/day) TN (mg/L) 4 110× 4

3 110× 3

100 Flow concen 2 10

1 1

0.1 0 0 100 200 300 400 Time (days from 30/6/2006)

Time-series plot of flow and TP 5 110× 0.5 Flow (ML/day) TP (mg/L) 4 110× 0.4

3 110× 0.3

100 Flow concen 0.2 10

0.1 1

0.1 0 0 100 200 300 400 Time (days from 30/6/2006)

Analysis: TN

μq = 3.7422 σq = 1.9513 μcn −= 0.6690 σcn = 0.7950 ρqtn = 0.6650

Bivariate log-normal distribution for flow and concentration

EL μ , σ , μ , σ , ρ = 558.081 ()cn cn q q qtn Theoretical mean load (tonnes)

8 V()μcn, σcn, μq, σq, ρqtn = 2.073888× 10 Theoretical variance 4 V()μcn, σcn, μq, σq, ρqtn 1.44×= 10 Theoretical standard error

EL()μcn, σcn, μq, σq, ρqtn Theoretical coefficient of variation = 0.039 V()μcn, σcn, μq, σq, ρqtn

Empirical estimates

Total discharge Qtot = 196961.20 number of readings K= 361

365 Estimated annual total discharge Q ⋅:= Q Q = 199143.60 an K tot an

T Q tn − 3 Estimated load ⋅Qan⋅10 = 598.7 tonnes ∑Q

2 ⎛ 365 ⎞ 2 ⎜ ⎟ ⋅ Q ⋅K ⎝ Kn1 ⎠ ∑ V()μcn, σcn, μq, σq, ρqtn Estimated variance of estimate ⋅ = 467427.7572 2 6 ⎛ ⎞ 10 ⎜ Q⎟ ⎝∑ ⎠

2 ⎛ 365 ⎞ 2 ⎜ ⎟ ⋅ Q ⋅K ⎝ Kn1 ⎠ ∑ V()μcn, σcn, μq, σq, ρqtn ⋅ = 683.687 Estimated standard error of estimate 2 6 ⎛ ⎞ 10 ⎜ Q⎟ ⎝∑ ⎠

2 ⎛ 365 ⎞ 2 ⎜ ⎟ ⋅ Q ⋅K ⎝ Kn1 ⎠ ∑ V()μcn, σcn, μq, σq, ρqtn ⋅ 2 6 ⎛ ⎞ 10 ⎜∑Q⎟ coefficient of variation ⎝ ⎠ = 1.142 T ⎛ Q tn − 3⎞ ⎜ ⋅Qan⋅10 ⎟ ⎜ Q ⎟ ⎝∑ ⎠

Copyright (2008) Environmetrics Australia Pty. Ltd. Page 6 of 8 2006/07 Nutrient Load Estimates Site: 223209 Analysis: TP

μq = 3.7422 σq = 1.9513 μcp −= 3.4317 σcp = 0.9956 ρqtp = 0.3314

Bivariate log-normal distribution for flow and concentration

EL μ , σ , μ , σ , ρ = 28.608 ()cp cp q q qtp Theoretical mean load (tonnes)

V()μcp, σcp, μq, σq, ρqtp = 359150.108265 Theoretical variance

V()μcp, σcp, μq, σq, ρqtp = 599.291 Theoretical standard error

EL()μcp, σcp, μq, σq, ρqtp Theoretical coefficient of variation = 0.048 V()μcp, σcp, μq, σq, ρqtp

Empirical estimates

Total discharge Qtot = 196961.20 number of readings K= 361

365 Estimated annual total discharge Q ⋅:= Q Q = 199143.60 an K tot an

T Q tp − 3 Estimated load ⋅Qan⋅10 = 47.5 tonnes ∑Q

2 ⎛ 365 ⎞ 2 ⎜ ⎟ ⋅ Q ⋅K ⎝ Kn2 ⎠ ∑ V()μcp, σcp, μq, σq, ρqtp Estimated variance of estimate ⋅ = 809.4784 2 6 ⎛ ⎞ 10 ⎜ Q⎟ ⎝∑ ⎠

2 ⎛ 365 ⎞ 2 ⎜ ⎟ ⋅ Q ⋅K ⎝ Kn2 ⎠ ∑ V()μcp, σcp, μq, σq, ρqtp ⋅ = 28.451 Estimated standard error of estimate 2 6 ⎛ ⎞ 10 ⎜ Q⎟ ⎝∑ ⎠

2 ⎛ 365 ⎞ 2 ⎜ ⎟ ⋅ Q ⋅K ⎝ Kn2 ⎠ ∑ V()μcp, σcp, μq, σq, ρqtp ⋅ 2 6 ⎛ ⎞ 10 ⎜∑Q⎟ coefficient of variation ⎝ ⎠ = 0.599 T ⎛ Q tp − 3⎞ ⎜ ⋅Qan⋅10 ⎟ ⎜ Q ⎟ ⎝∑ ⎠

Nutrient Load Estimation for selected tributaries of the Gippsland Lakes

Site 223210 July 1 2006 - June 30 2007

Nicholson River

Prepared for the Victorian EPA

Environmetrics Australia Pty Ltd

June 2008

n ⎛⎞nc ⎛⎞q ˆ (1) LK= ⎜⎟∑∑ wcii⎜⎟ vq j j ⎝⎠ij==11⎝⎠

th th where ci is a measured concentration on the i occasion; q j is a measured flow on the j

2. Theoretical mean and variance

fCQ, ()cq, =

EL[]=

⎪⎧ 1 22 ⎪⎫ exp ⎨ μμ++⎡⎤ρσσ ++ ρσσρρσσρσσ +−2 + +⎬ ()CQ 2 ⎣⎦⎢⎥()()()()QC CQ QC CQ ⎪⎩ 21()− ρ ⎭⎪ (3)

Furthermore, it can be established that the second (uncorrected) moment is:

2 EL⎣⎦⎡⎤=

⎧⎪ 2 22 ⎪⎫ exp ⎨22μμ++⎡⎤ ρσσ ++ ρσσρρσσρσσ +− + +⎬ ()CQ 2 ⎣⎦⎢⎥()()()()QC CQ QC CQ ⎪⎩ ()1 − ρ ⎭⎪ (4) and so the variance is given as

2 2 Var[] L=− E⎣⎦⎡⎤ L() E[] L (5)

Data summary

fn= "e:\vicepa\223210.dat" Filename

n= 365 number of rows in data file

K= 310 number of non-missing flow readings

number of non-missing TN concentration readings n1= 88 number of non-missing TP concentration readings n2= 87 number of contemperaneous flow and TN measurements Kn1= 80

Kn2= 79 number of contemperaneous flow and TP measurements

Kn12= 87 number of contemperaneous TN and TP measurements

Time-series plot of flow and TN 5 110× 4 Flow (ML/day) TN (mg/L) 4 110×

3 110×

100 2 Flow concen

0.1 0 0 100 200 300 400 Time (days from 30/6/2006)

Time-series plot of flow and TP 5 110× 0.4 Flow (ML/day) TP (mg/L) 4 110×

0.3

3 110×

100 0.2 Flow concen

0.1

0.1 0 0 100 200 300 400 Time (days from 30/6/2006)

Analysis: TN

μq = 1.6232 σq = 2.0319 μcn −= 0.8101 σcn = 0.6425 ρqtn = 0.8238

Bivariate log-normal distribution for flow and concentration

EL μ , σ , μ , σ , ρ = 64.024 ()cn cn q q qtn Theoretical mean load (tonnes)

6 V()μcn, σcn, μq, σq, ρqtn = 3.301097× 10 Theoretical variance 3 V()μcn, σcn, μq, σq, ρqtn 1.817×= 10 Theoretical standard error

EL()μcn, σcn, μq, σq, ρqtn Theoretical coefficient of variation = 0.035 V()μcn, σcn, μq, σq, ρqtn

Empirical estimates

Total discharge Qtot = 41176.20 number of readings K= 310

365 Estimated annual total discharge Q ⋅:= Q Q = 48481.65 an K tot an

T Q tn − 3 Estimated load ⋅Qan⋅10 = 136.8 tonnes ∑Q

2 ⎛ 365 ⎞ 2 ⎜ ⎟ ⋅ Q ⋅K ⎝ Kn1 ⎠ ∑ V()μcn, σcn, μq, σq, ρqtn Estimated variance of estimate ⋅ = 4981.5421 2 6 ⎛ ⎞ 10 ⎜ Q⎟ ⎝∑ ⎠

2 ⎛ 365 ⎞ 2 ⎜ ⎟ ⋅ Q ⋅K ⎝ Kn1 ⎠ ∑ V()μcn, σcn, μq, σq, ρqtn ⋅ = 70.580 Estimated standard error of estimate 2 6 ⎛ ⎞ 10 ⎜ Q⎟ ⎝∑ ⎠

2 ⎛ 365 ⎞ 2 ⎜ ⎟ ⋅ Q ⋅K ⎝ Kn1 ⎠ ∑ V()μcn, σcn, μq, σq, ρqtn ⋅ 2 6 ⎛ ⎞ 10 ⎜∑Q⎟ coefficient of variation ⎝ ⎠ = 0.516 T ⎛ Q tn − 3⎞ ⎜ ⋅Qan⋅10 ⎟ ⎜ Q ⎟ ⎝∑ ⎠

Copyright (2008) Environmetrics Australia Pty. Ltd. Page 6 of 8 2006/07 Nutrient Load Estimates Site: 223210 Analysis: TP

μq = 1.6232 σq = 2.0319 μcp −= 4.0652 σcp = 0.9328 ρqtp = 0.3742

Bivariate log-normal distribution for flow and concentration

EL μ , σ , μ , σ , ρ = 2.152 ()cp cp q q qtp Theoretical mean load (tonnes)

V()μcp, σcp, μq, σq, ρqtp = 2832.089869 Theoretical variance

V()μcp, σcp, μq, σq, ρqtp = 53.217 Theoretical standard error

EL()μcp, σcp, μq, σq, ρqtp Theoretical coefficient of variation = 0.04 V()μcp, σcp, μq, σq, ρqtp

Empirical estimates

Total discharge Qtot = 41176.20 number of readings K= 310

365 Estimated annual total discharge Q ⋅:= Q Q = 48481.65 an K tot an

T Q tp − 3 Estimated load ⋅Qan⋅10 = 11.6 tonnes ∑Q

2 ⎛ 365 ⎞ 2 ⎜ ⎟ ⋅ Q ⋅K ⎝ Kn2 ⎠ ∑ V()μcp, σcp, μq, σq, ρqtp Estimated variance of estimate ⋅ = 4.3936 2 6 ⎛ ⎞ 10 ⎜ Q⎟ ⎝∑ ⎠

2 ⎛ 365 ⎞ 2 ⎜ ⎟ ⋅ Q ⋅K ⎝ Kn2 ⎠ ∑ V()μcp, σcp, μq, σq, ρqtp ⋅ = 2.096 Estimated standard error of estimate 2 6 ⎛ ⎞ 10 ⎜ Q⎟ ⎝∑ ⎠

2 ⎛ 365 ⎞ 2 ⎜ ⎟ ⋅ Q ⋅K ⎝ Kn2 ⎠ ∑ V()μcp, σcp, μq, σq, ρqtp ⋅ 2 6 ⎛ ⎞ 10 ⎜∑Q⎟ coefficient of variation ⎝ ⎠ = 0.181 T ⎛ Q tp − 3⎞ ⎜ ⋅Qan⋅10 ⎟ ⎜ Q ⎟ ⎝∑ ⎠

Nutrient Load Estimation for selected tributaries of the Gippsland Lakes

Site 224217 July 1 2006 - June 30 2007

Mitchell River

Prepared for the Victorian EPA

Environmetrics Australia Pty Ltd

June 2008

n ⎛⎞nc ⎛⎞q ˆ (1) LK= ⎜⎟∑∑ wcii⎜⎟ vq j j ⎝⎠ij==11⎝⎠

th th where ci is a measured concentration on the i occasion; q j is a measured flow on the j

2. Theoretical mean and variance

fCQ, ()cq, =

EL[]=

⎪⎧ 1 22 ⎪⎫ exp ⎨ μμ++⎡⎤ρσσ ++ ρσσρρσσρσσ +−2 + +⎬ ()CQ 2 ⎣⎦⎢⎥()()()()QC CQ QC CQ ⎪⎩ 21()− ρ ⎭⎪ (3)

Furthermore, it can be established that the second (uncorrected) moment is:

2 EL⎣⎦⎡⎤=

⎧⎪ 2 22 ⎪⎫ exp ⎨22μμ++⎡⎤ ρσσ ++ ρσσρρσσρσσ +− + +⎬ ()CQ 2 ⎣⎦⎢⎥()()()()QC CQ QC CQ ⎪⎩ ()1 − ρ ⎭⎪ (4) and so the variance is given as

2 2 Var[] L=− E⎣⎦⎡⎤ L() E[] L (5)

Data summary

fn= "e:\vicepa\224217.dat" Filename

n= 365 number of rows in data file

K= 364 number of non-missing flow readings

number of non-missing TN concentration readings n1= 91 number of non-missing TP concentration readings n2= 91 number of contemperaneous flow and TN measurements Kn1= 91

Kn2= 91 number of contemperaneous flow and TP measurements

Kn12= 91 number of contemperaneous TN and TP measurements

Time-series plot of flow and TN 5 110× 10 Flow (ML/day) TN (mg/L)

4 110× 8

3 110× 6 Flow concen 100 4

10 2

1 0 0 100 200 300 400 Time (days from 30/6/2006)

Time-series plot of flow and TP 5 110× 3 Flow (ML/day) TP (mg/L)

4 110×

2 3 110× Flow concen 100 1

1 0 0 100 200 300 400 Time (days from 30/6/2006)

Analysis: TN

μq = 5.1953 σq = 1.7846 μcn −= 0.8134 σcn = 1.1528 ρqtn = 0.3690

Bivariate log-normal distribution for flow and concentration

3 EL μ , σ , μ , σ , ρ 1.633×= 10 ()cn cn q q qtn Theoretical mean load (tonnes)

9 V()μcn, σcn, μq, σq, ρqtn = 1.107891× 10 Theoretical variance 4 V()μcn, σcn, μq, σq, ρqtn 3.329×= 10 Theoretical standard error

EL()μcn, σcn, μq, σq, ρqtn Theoretical coefficient of variation = 0.049 V()μcn, σcn, μq, σq, ρqtn

Empirical estimates

Total discharge Qtot = 275993.50 number of readings K= 364

365 Estimated annual total discharge Q ⋅:= Q Q = 276751.72 an K tot an

T Q tn − 3 Estimated load ⋅Qan⋅10 = 912.4 tonnes ∑Q

2 ⎛ 365 ⎞ 2 ⎜ ⎟ ⋅ Q ⋅K ⎝ Kn1 ⎠ ∑ V()μcn, σcn, μq, σq, ρqtn Estimated variance of estimate ⋅ = 562543.0685 2 6 ⎛ ⎞ 10 ⎜ Q⎟ ⎝∑ ⎠

2 ⎛ 365 ⎞ 2 ⎜ ⎟ ⋅ Q ⋅K ⎝ Kn1 ⎠ ∑ V()μcn, σcn, μq, σq, ρqtn ⋅ = 750.029 Estimated standard error of estimate 2 6 ⎛ ⎞ 10 ⎜ Q⎟ ⎝∑ ⎠

2 ⎛ 365 ⎞ 2 ⎜ ⎟ ⋅ Q ⋅K ⎝ Kn1 ⎠ ∑ V()μcn, σcn, μq, σq, ρqtn ⋅ 2 6 ⎛ ⎞ 10 ⎜∑Q⎟ coefficient of variation ⎝ ⎠ = 0.822 T ⎛ Q tn − 3⎞ ⎜ ⋅Qan⋅10 ⎟ ⎜ Q ⎟ ⎝∑ ⎠

Copyright (2008) Environmetrics Australia Pty. Ltd. Page 6 of 8 2006/07 Nutrient Load Estimates Site: 224217 Analysis: TP

μq = 5.1953 σq = 1.7846 μcp −= 2.9404 σcp = 1.5428 ρqtp = 0.4637

Bivariate log-normal distribution for flow and concentration

EL μ , σ , μ , σ , ρ = 552.349 ()cp cp q q qtp Theoretical mean load (tonnes)

9 V()μcp, σcp, μq, σq, ρqtp = 1.023546× 10 Theoretical variance 4 V()μcp, σcp, μq, σq, ρqtp 3.199×= 10 Theoretical standard error

EL()μcp, σcp, μq, σq, ρqtp Theoretical coefficient of variation = 0.017 V()μcp, σcp, μq, σq, ρqtp

Empirical estimates

Total discharge Qtot = 275993.50 number of readings K= 364

365 Estimated annual total discharge Q ⋅:= Q Q = 276751.72 an K tot an

T Q tp − 3 Estimated load ⋅Qan⋅10 = 168.7 tonnes ∑Q

2 ⎛ 365 ⎞ 2 ⎜ ⎟ ⋅ Q ⋅K ⎝ Kn2 ⎠ ∑ V()μcp, σcp, μq, σq, ρqtp Estimated variance of estimate ⋅ = 519715.8038 2 6 ⎛ ⎞ 10 ⎜ Q⎟ ⎝∑ ⎠

2 ⎛ 365 ⎞ 2 ⎜ ⎟ ⋅ Q ⋅K ⎝ Kn2 ⎠ ∑ V()μcp, σcp, μq, σq, ρqtp ⋅ = 720.913 Estimated standard error of estimate 2 6 ⎛ ⎞ 10 ⎜ Q⎟ ⎝∑ ⎠

2 ⎛ 365 ⎞ 2 ⎜ ⎟ ⋅ Q ⋅K ⎝ Kn2 ⎠ ∑ V()μcp, σcp, μq, σq, ρqtp ⋅ 2 6 ⎛ ⎞ 10 ⎜∑Q⎟ coefficient of variation ⎝ ⎠ = 4.272 T ⎛ Q tp − 3⎞ ⎜ ⋅Qan⋅10 ⎟ ⎜ Q ⎟ ⎝∑ ⎠

Nutrient Load Estimation for selected tributaries of the Gippsland Lakes

Site 225232 July 1 2006 - June 30 2007

Thomson River

Prepared for the Victorian EPA

Environmetrics Australia Pty Ltd

June 2008

n ⎛⎞nc ⎛⎞q ˆ (1) LK= ⎜⎟∑∑ wcii⎜⎟ vq j j ⎝⎠ij==11⎝⎠

th th where ci is a measured concentration on the i occasion; q j is a measured flow on the j

2. Theoretical mean and variance

fCQ, ()cq, =

EL[]=

⎪⎧ 1 22 ⎪⎫ exp ⎨ μμ++⎡⎤ρσσ ++ ρσσρρσσρσσ +−2 + +⎬ ()CQ 2 ⎣⎦⎢⎥()()()()QC CQ QC CQ ⎪⎩ 21()− ρ ⎭⎪ (3)

Furthermore, it can be established that the second (uncorrected) moment is:

2 EL⎣⎦⎡⎤=

⎧⎪ 2 22 ⎪⎫ exp ⎨22μμ++⎡⎤ ρσσ ++ ρσσρρσσρσσ +− + +⎬ ()CQ 2 ⎣⎦⎢⎥()()()()QC CQ QC CQ ⎪⎩ ()1 − ρ ⎭⎪ (4) and so the variance is given as

2 2 Var[] L=− E⎣⎦⎡⎤ L() E[] L (5)

Data summary

fn= "e:\vicepa\225232.dat" Filename

n= 365 number of rows in data file

K= 365 number of non-missing flow readings

number of non-missing TN concentration readings n1= 64 number of non-missing TP concentration readings n2= 64 number of contemperaneous flow and TN measurements Kn1= 64

Kn2= 64 number of contemperaneous flow and TP measurements

Kn12= 64 number of contemperaneous TN and TP measurements

Time-series plot of flow and TN 5 110× 4 Flow (ML/day) TN (mg/L)

4 110× 3

3 110× 2 Flow concen

100 1

10 0 0 100 200 300 400 Time (days from 30/6/2006)

Time-series plot of flow and TP 5 110× 0.4 Flow (ML/day) TP (mg/L)

4 110× 0.3

3 110× 0.2 Flow concen

100 0.1

10 0 0 100 200 300 400 Time (days from 30/6/2006)

Analysis: TN

μq = 5.3818 σq = 0.8122 μcn −= 0.7893 σcn = 0.6058 ρqtn = 0.7417

Bivariate log-normal distribution for flow and concentration

EL μ , σ , μ , σ , ρ = 237.638 ()cn cn q q qtn Theoretical mean load (tonnes)

V()μcn, σcn, μq, σq, ρqtn = 270589.629811 Theoretical variance

V()μcn, σcn, μq, σq, ρqtn = 520.182 Theoretical standard error

EL()μcn, σcn, μq, σq, ρqtn Theoretical coefficient of variation = 0.457 V()μcn, σcn, μq, σq, ρqtn

Empirical estimates

Total discharge Qtot = 300446.70 number of readings K= 365

365 Estimated annual total discharge Q ⋅:= Q Q = 300446.70 an K tot an

T Q tn − 3 Estimated load ⋅Qan⋅10 = 422.9 tonnes ∑Q

2 ⎛ 365 ⎞ 2 ⎜ ⎟ ⋅ Q ⋅K ⎝ Kn1 ⎠ ∑ V()μcn, σcn, μq, σq, ρqtn Estimated variance of estimate ⋅ = 201.1953 2 6 ⎛ ⎞ 10 ⎜ Q⎟ ⎝∑ ⎠

2 ⎛ 365 ⎞ 2 ⎜ ⎟ ⋅ Q ⋅K ⎝ Kn1 ⎠ ∑ V()μcn, σcn, μq, σq, ρqtn ⋅ = 14.184 Estimated standard error of estimate 2 6 ⎛ ⎞ 10 ⎜ Q⎟ ⎝∑ ⎠

2 ⎛ 365 ⎞ 2 ⎜ ⎟ ⋅ Q ⋅K ⎝ Kn1 ⎠ ∑ V()μcn, σcn, μq, σq, ρqtn ⋅ 2 6 ⎛ ⎞ 10 ⎜∑Q⎟ coefficient of variation ⎝ ⎠ = 0.034 T ⎛ Q tn − 3⎞ ⎜ ⋅Qan⋅10 ⎟ ⎜ Q ⎟ ⎝∑ ⎠

Copyright (2008) Environmetrics Australia Pty. Ltd. Page 6 of 8 2006/07 Nutrient Load Estimates Site: 225232 Analysis: TP

μq = 5.3818 σq = 0.8122 μcp −= 2.9138 σcp = 0.6539 ρqtp = 0.3101

Bivariate log-normal distribution for flow and concentration

EL μ , σ , μ , σ , ρ = 23.959 ()cp cp q q qtp Theoretical mean load (tonnes)

V()μcp, σcp, μq, σq, ρqtp = 1792.719856 Theoretical variance

V()μcp, σcp, μq, σq, ρqtp = 42.341 Theoretical standard error

EL()μcp, σcp, μq, σq, ρqtp Theoretical coefficient of variation = 0.566 V()μcp, σcp, μq, σq, ρqtp

Empirical estimates

Total discharge Qtot = 300446.70 number of readings K= 365

365 Estimated annual total discharge Q ⋅:= Q Q = 300446.70 an K tot an

T Q tp − 3 Estimated load ⋅Qan⋅10 = 39.2 tonnes ∑Q

2 ⎛ 365 ⎞ 2 ⎜ ⎟ ⋅ Q ⋅K ⎝ Kn2 ⎠ ∑ V()μcp, σcp, μq, σq, ρqtp Estimated variance of estimate ⋅ = 1.3330 2 6 ⎛ ⎞ 10 ⎜ Q⎟ ⎝∑ ⎠

2 ⎛ 365 ⎞ 2 ⎜ ⎟ ⋅ Q ⋅K ⎝ Kn2 ⎠ ∑ V()μcp, σcp, μq, σq, ρqtp ⋅ = 1.155 Estimated standard error of estimate 2 6 ⎛ ⎞ 10 ⎜ Q⎟ ⎝∑ ⎠

2 ⎛ 365 ⎞ 2 ⎜ ⎟ ⋅ Q ⋅K ⎝ Kn2 ⎠ ∑ V()μcp, σcp, μq, σq, ρqtp ⋅ 2 6 ⎛ ⎞ 10 ⎜∑Q⎟ coefficient of variation ⎝ ⎠ = 0.029 T ⎛ Q tp − 3⎞ ⎜ ⋅Qan⋅10 ⎟ ⎜ Q ⎟ ⎝∑ ⎠

Nutrient Load Estimation for selected tributaries of the Gippsland Lakes

Site 225234 July 1 2006 - June 30 2007

Avon River

Prepared for the Victorian EPA

Environmetrics Australia Pty Ltd

June 2008

n ⎛⎞nc ⎛⎞q ˆ (1) LK= ⎜⎟∑∑ wcii⎜⎟ vq j j ⎝⎠ij==11⎝⎠

th th where ci is a measured concentration on the i occasion; q j is a measured flow on the j

2. Theoretical mean and variance

fCQ, ()cq, =

EL[]=

⎪⎧ 1 22 ⎪⎫ exp ⎨ μμ++⎡⎤ρσσ ++ ρσσρρσσρσσ +−2 + +⎬ ()CQ 2 ⎣⎦⎢⎥()()()()QC CQ QC CQ ⎪⎩ 21()− ρ ⎭⎪ (3)

Furthermore, it can be established that the second (uncorrected) moment is:

2 EL⎣⎦⎡⎤=

⎧⎪ 2 22 ⎪⎫ exp ⎨22μμ++⎡⎤ ρσσ ++ ρσσρρσσρσσ +− + +⎬ ()CQ 2 ⎣⎦⎢⎥()()()()QC CQ QC CQ ⎪⎩ ()1 − ρ ⎭⎪ (4) and so the variance is given as

2 2 Var[] L=− E⎣⎦⎡⎤ L() E[] L (5)

Data summary

fn= "e:\vicepa\225234.dat" Filename

n= 365 number of rows in data file

K= 346 number of non-missing flow readings

number of non-missing TN concentration readings n1= 102 number of non-missing TP concentration readings n2= 102 number of contemperaneous flow and TN measurements Kn1= 97

Kn2= 97 number of contemperaneous flow and TP measurements

Kn12= 102 number of contemperaneous TN and TP measurements

Time-series plot of flow and TN 5 110× 4 Flow (ML/day) TN (mg/L)

4 110× 3

3 110× 2 Flow concen

100 1

10 0 0 100 200 300 400 Time (days from 30/6/2006)

Time-series plot of flow and TP 5 110× 0.5 Flow (ML/day) TP (mg/L)

0.4 4 110×

0.3

3 110× Flow concen 0.2

100 0.1

10 0 0 100 200 300 400 Time (days from 30/6/2006)

Analysis: TN

μq = 3.9758 σq = 1.2975 μcn −= 0.7661 σcn = 0.7052 ρqtn = 0.0832

Bivariate log-normal distribution for flow and concentration

EL μ , σ , μ , σ , ρ = 79.529 ()cn cn q q qtn Theoretical mean load (tonnes)

V()μcn, σcn, μq, σq, ρqtn = 58868.843600 Theoretical variance

V()μcn, σcn, μq, σq, ρqtn = 242.629 Theoretical standard error

EL()μcn, σcn, μq, σq, ρqtn Theoretical coefficient of variation = 0.328 V()μcn, σcn, μq, σq, ρqtn

Empirical estimates

Total discharge Qtot = 133630.20 number of readings K= 346

365 Estimated annual total discharge Q ⋅:= Q Q = 140968.27 an K tot an

T Q tn − 3 Estimated load ⋅Qan⋅10 = 324.4 tonnes ∑Q

2 ⎛ 365 ⎞ 2 ⎜ ⎟ ⋅ Q ⋅K ⎝ Kn1 ⎠ ∑ V()μcn, σcn, μq, σq, ρqtn Estimated variance of estimate ⋅ = 80.2805 2 6 ⎛ ⎞ 10 ⎜ Q⎟ ⎝∑ ⎠

2 ⎛ 365 ⎞ 2 ⎜ ⎟ ⋅ Q ⋅K ⎝ Kn1 ⎠ ∑ V()μcn, σcn, μq, σq, ρqtn ⋅ = 8.96 Estimated standard error of estimate 2 6 ⎛ ⎞ 10 ⎜ Q⎟ ⎝∑ ⎠

2 ⎛ 365 ⎞ 2 ⎜ ⎟ ⋅ Q ⋅K ⎝ Kn1 ⎠ ∑ V()μcn, σcn, μq, σq, ρqtn ⋅ 2 6 ⎛ ⎞ 10 ⎜∑Q⎟ coefficient of variation ⎝ ⎠ = 0.028 T ⎛ Q tn − 3⎞ ⎜ ⋅Qan⋅10 ⎟ ⎜ Q ⎟ ⎝∑ ⎠

Copyright (2008) Environmetrics Australia Pty. Ltd. Page 6 of 8 2006/07 Nutrient Load Estimates Site: 225234 Analysis: TP

μq = 3.9758 σq = 1.2975 μcp −= 3.2814 σcp = 0.7879 ρqtp −= 0.0582

Bivariate log-normal distribution for flow and concentration

EL μ , σ , μ , σ , ρ = 5.971 ()cp cp q q qtp Theoretical mean load (tonnes)

V()μcp, σcp, μq, σq, ρqtp = 281.390044 Theoretical variance

V()μcp, σcp, μq, σq, ρqtp = 16.775 Theoretical standard error

EL()μcp, σcp, μq, σq, ρqtp Theoretical coefficient of variation = 0.356 V()μcp, σcp, μq, σq, ρqtp

Empirical estimates

Total discharge Qtot = 133630.20 number of readings K= 346

365 Estimated annual total discharge Q ⋅:= Q Q = 140968.27 an K tot an

T Q tp − 3 Estimated load ⋅Qan⋅10 = 38.1 tonnes ∑Q

2 ⎛ 365 ⎞ 2 ⎜ ⎟ ⋅ Q ⋅K ⎝ Kn2 ⎠ ∑ V()μcp, σcp, μq, σq, ρqtp Estimated variance of estimate ⋅ = 0.3837 2 6 ⎛ ⎞ 10 ⎜ Q⎟ ⎝∑ ⎠

2 ⎛ 365 ⎞ 2 ⎜ ⎟ ⋅ Q ⋅K ⎝ Kn2 ⎠ ∑ V()μcp, σcp, μq, σq, ρqtp ⋅ = 0.619 Estimated standard error of estimate 2 6 ⎛ ⎞ 10 ⎜ Q⎟ ⎝∑ ⎠

2 ⎛ 365 ⎞ 2 ⎜ ⎟ ⋅ Q ⋅K ⎝ Kn2 ⎠ ∑ V()μcp, σcp, μq, σq, ρqtp ⋅ 2 6 ⎛ ⎞ 10 ⎜∑Q⎟ coefficient of variation ⎝ ⎠ = 0.016 T ⎛ Q tp − 3⎞ ⎜ ⋅Qan⋅10 ⎟ ⎜ Q ⎟ ⎝∑ ⎠

Nutrient Load Estimation for selected tributaries of the Gippsland Lakes

Site 226227 July 1 2006 - June 30 2007

Latrobe River

Prepared for the Victorian EPA

Environmetrics Australia Pty Ltd

June 2008

Copright (2008) Environmetrics Australia Pty Ltd Page 1 of 8 2006/07 Nutrient Load Estimates Site: 226227

n ⎛⎞nc ⎛⎞q ˆ (1) LK= ⎜⎟∑∑ wcii⎜⎟ vq j j ⎝⎠ij==11⎝⎠

th th where ci is a measured concentration on the i occasion; q j is a measured flow on the j

2. Theoretical mean and variance

fCQ, ()cq, =

Copright (2008) Environmetrics Australia Pty Ltd Page 2 of 8 2006/07 Nutrient Load Estimates Site: 226227

EL[]=

⎪⎧ 1 22 ⎪⎫ exp ⎨ μμ++⎡⎤ρσσ ++ ρσσρρσσρσσ +−2 + +⎬ ()CQ 2 ⎣⎦⎢⎥()()()()QC CQ QC CQ ⎪⎩ 21()− ρ ⎭⎪ (3)

Furthermore, it can be established that the second (uncorrected) moment is:

2 EL⎣⎦⎡⎤=

⎧⎪ 2 22 ⎪⎫ exp ⎨22μμ++⎡⎤ ρσσ ++ ρσσρρσσρσσ +− + +⎬ ()CQ 2 ⎣⎦⎢⎥()()()()QC CQ QC CQ ⎪⎩ ()1 − ρ ⎭⎪ (4) and so the variance is given as

2 2 Var[] L=− E⎣⎦⎡⎤ L() E[] L (5)

Data summary

fn= "e:\vicepa\226227.dat" Filename

n= 365 number of rows in data file

K= 365 number of non-missing flow readings

number of non-missing TN concentration readings n1= 56 number of non-missing TP concentration readings n2= 56 number of contemperaneous flow and TN measurements Kn1= 56

Kn2= 56 number of contemperaneous flow and TP measurements

Kn12= 56 number of contemperaneous TN and TP measurements

Copright (2008) Environmetrics Australia Pty Ltd Page 3 of 8 2006/07 Nutrient Load Estimates Site: 226227

Time-series plot of flow and TN 4 110× 8 Flow (ML/day) TN (mg/L)

3 110× 4 Flow concen

100 0 0 100 200 300 400 Time (days from 30/6/2006)

Time-series plot of flow and TP 4 110× 2 Flow (ML/day) TP (mg/L)

1.5

3 110× 1 Flow concen

0.5

100 0 0 100 200 300 400 Time (days from 30/6/2006)

Copright (2008) Environmetrics Australia Pty Ltd Page 4 of 8 2006/07 Nutrient Load Estimates Site: 226227

Analysis: TN

μq = 6.0056 σq = 0.4392 μcn −= 0.2932 σcn = 0.5783 ρqtn = 0.8644

Bivariate log-normal distribution for flow and concentration

EL μ , σ , μ , σ , ρ = 490.601 ()cn cn q q qtn Theoretical mean load (tonnes)

V()μcn, σcn, μq, σq, ρqtn = 391961.527396 Theoretical variance

V()μcn, σcn, μq, σq, ρqtn = 626.068 Theoretical standard error

EL()μcn, σcn, μq, σq, ρqtn Theoretical coefficient of variation = 0.784 V()μcn, σcn, μq, σq, ρqtn

Copright (2008) Environmetrics Australia Pty Ltd Page 5 of 8 2006/07 Nutrient Load Estimates Site: 226227

Empirical estimates

Total discharge Qtot = 169140.66 number of readings K= 365

365 Estimated annual total discharge Q ⋅:= Q Q = 169140.66 an K tot an

T Q tn − 3 Estimated load ⋅Qan⋅10 = 335.4 tonnes ∑Q

2 ⎛ 365 ⎞ 2 ⎜ ⎟ ⋅ Q ⋅K ⎝ Kn1 ⎠ ∑ V()μcn, σcn, μq, σq, ρqtn Estimated variance of estimate ⋅ = 268.0541 2 6 ⎛ ⎞ 10 ⎜ Q⎟ ⎝∑ ⎠

2 ⎛ 365 ⎞ 2 ⎜ ⎟ ⋅ Q ⋅K ⎝ Kn1 ⎠ ∑ V()μcn, σcn, μq, σq, ρqtn ⋅ = 16.372 Estimated standard error of estimate 2 6 ⎛ ⎞ 10 ⎜ Q⎟ ⎝∑ ⎠

2 ⎛ 365 ⎞ 2 ⎜ ⎟ ⋅ Q ⋅K ⎝ Kn1 ⎠ ∑ V()μcn, σcn, μq, σq, ρqtn ⋅ 2 6 ⎛ ⎞ 10 ⎜∑Q⎟ coefficient of variation ⎝ ⎠ = 0.049 T ⎛ Q tn − 3⎞ ⎜ ⋅Qan⋅10 ⎟ ⎜ Q ⎟ ⎝∑ ⎠

Copright (2008) Environmetrics Australia Pty Ltd Page 6 of 8 2006/07 Nutrient Load Estimates Site: 226227 Analysis: TP

μq = 6.0056 σq = 0.4392 μcp −= 2.6882 σcp = 0.9121 ρqtp = 0.5954

Bivariate log-normal distribution for flow and concentration

EL μ , σ , μ , σ , ρ = 58.458 ()cp cp q q qtp Theoretical mean load (tonnes)

V()μcp, σcp, μq, σq, ρqtp = 11926.385592 Theoretical variance

V()μcp, σcp, μq, σq, ρqtp = 109.208 Theoretical standard error

EL()μcp, σcp, μq, σq, ρqtp Theoretical coefficient of variation = 0.535 V()μcp, σcp, μq, σq, ρqtp

Copright (2008) Environmetrics Australia Pty Ltd Page 7 of 8 2006/07 Nutrient Load Estimates Site: 226227

Empirical estimates

Total discharge Qtot = 169140.66 number of readings K= 365

365 Estimated annual total discharge Q ⋅:= Q Q = 169140.66 an K tot an

T Q tp − 3 Estimated load ⋅Qan⋅10 = 68.4 tonnes ∑Q

2 ⎛ 365 ⎞ 2 ⎜ ⎟ ⋅ Q ⋅K ⎝ Kn2 ⎠ ∑ V()μcp, σcp, μq, σq, ρqtp Estimated variance of estimate ⋅ = 8.1562 2 6 ⎛ ⎞ 10 ⎜ Q⎟ ⎝∑ ⎠

2 ⎛ 365 ⎞ 2 ⎜ ⎟ ⋅ Q ⋅K ⎝ Kn2 ⎠ ∑ V()μcp, σcp, μq, σq, ρqtp ⋅ = 2.856 Estimated standard error of estimate 2 6 ⎛ ⎞ 10 ⎜ Q⎟ ⎝∑ ⎠

2 ⎛ 365 ⎞ 2 ⎜ ⎟ ⋅ Q ⋅K ⎝ Kn2 ⎠ ∑ V()μcp, σcp, μq, σq, ρqtp ⋅ 2 6 ⎛ ⎞ 10 ⎜∑Q⎟ coefficient of variation ⎝ ⎠ = 0.042 T ⎛ Q tp − 3⎞ ⎜ ⋅Qan⋅10 ⎟ ⎜ Q ⎟ ⎝∑ ⎠

Copright (2008) Environmetrics Australia Pty Ltd Page 8 of 8 MONITORING RIVER NUTRIENT LOADS TO THE GIPPSLAND LAKES 2006–07

APPENDIX 2: ANALYSIS METHODS AND QUALITY ASSURANCE

The monitoring program for this work is a component Contract 3 – Module 3: of the Gippsland Regional Water Monitoring Analysis contractors must use analytical laboratory Partnership. procedures and methodologies currently specified as The quality assurance for all phases of the Gippsland part of the Victorian Water Quality Monitoring Regional Water Quality Monitoring Partnership Network protocols. Laboratories must be NATA- program (of which the river nutrient loads sample accredited for each of the parameters assessed. collection and analysis is a component) is managed by The contractor must: DSE under contracts with third-party suppliers, as listed here:. • provide appropriately cleaned and labelled sampling containers • Module 1 (Stream flow, lake level and rainfall monitoring) — Bureau of Meteorology • provide field recording sheets for Module 2 contractors • Module 2 (Water quality sampling and field measurements) — Ecowise and Theiss Systems • analyse core water quality parameters (suspended solids, colour, filterable reactive phosphorus, total • Module 3 (Laboratory water quality analysis and phosphorus, oxidised nitrogen, total Kjeldahl field equipment calibration) — Ecowise. nitrogen, ammonia) (range of other parameters not Contract 1 — Module 1 used in this program - metals and major ions) Establishes contractor requirements for Stream flow, • report possible anomalies in water quality results lake level and rainfall monitoring. • undertake quarterly calibration and labelling of all field equipment used in Module 2 Sampling is in accordance with established quality assurance and quality control protocols established by • prepare chain-of-custody records to record the the Bureau of Meteorology. Details are available from delivery and receipt of items between Modules 2 the Department of Sustainability and Environment in and 3 Contract 1, Gippsland Regional Water Monitoring • upload field measurements and laboratory results Partnership. to the State Data Warehouse and provide results to nominated stakeholders Contract 2 — Module 2 • conduct specified field sampling and laboratory Establishes contractor requirements for water quality analysis quality control provisions sampling and field measurements. • provide advice in relation to data accuracy and The contract requires that 100 per cent of sampling analytical requirements. and records will be delivered to Module 3 contractors, Specified protocols to the prescribed standards and intervals. Analytical procedures must maintain the integrity of All stations will be maintained to a standard that the data set, ensure consistency in approach to all reflects Module 2 contractor usage and landowner laboratory procedures and apply methodologies as requirements. stated in the Victorian Water Quality Monitoring Prior to each field measurement and sampling run, the Network. All procedures are outlined in the contract. Module 2 contractor is to generate summary statistics Quality control must include proficiency trials, for each field measurement parameter at each including: monitoring station. Any new field water quality measurements are to be checked against these • blind trials using a range of samples from natural historic field water quality measurements. If the new waters (including high and low-concentration water quality measurement is more than two standard ranges). deviations from the long-term historic median, then a • standard reference sample recovery over a range further measurement must be immediately undertaken of concentrations. at the monitoring station. Any such additional • blanks and control samples measurement will be considered to be part of the • replicate samples. same sampling run. Occupational health, safety and rehabilitation plans The utmost care is required in the collection and are included as part of the overall quality plan. subsequent handling of the sample to ensure that the analysis correctly identifies the concentrations of the Details of the contracts, including the quality components being tested. All sampling and field assurance and quality control plans, are available from measurements must be undertaken in accordance with DSE. guidelines and standards outlined in the agreement.

68 MONITORING RIVER NUTRIENT LOADS TO THE GIPPSLAND LAKES 2006–07

APPENDIX 3: BUSHFIRE-IMPACTED AREAS 2006–07

Figure 17: Fire boundaries as at February 2007, representing impacted regions for 2006—07 fire season (Source: DSE).

69 MONITORING RIVER NUTRIENT LOADS TO THE GIPPSLAND LAKES 2006–07

APPENDIX 4: FLOOD MAP GIPPSLAND JULY 2007

Figure 18: Gippsland Region flood inundation 29 June 2007. (Source: Thankappan 2007).