Hydrology Report No 416000.PR/6

6.29 Reach 29 – 416204a Weir at Gunn Bridge (Headwater)

6.29.1 Reach Description

This section of the report presents the data and flow calibration results for the from the headwaters to Gunn Bridge (416204A). The reach includes 6 subareas in Table 1.220 in the Upper Weir (see Figure 1.160). This reach has a total catchment area of 4,424 km2.

Table 1.220: Reach 29 (416204a) subareas

Subarea State Description Area (km2) UW1 Qld Subareas upstream of 416204A (Weir River at 826.59 Gunn Bridge) UW2 Qld 1,287.43 UW3 Qld 30.18 UW4 Qld 1,331.08 UW5 Qld 848.57 UW6 Qld 100.03 Area Total 4,423.88

Figure 1.160: Reach 29 (416204a) map

329 Department of Environment and Science

6.29.2 Reach Structure

The model structure for this reach is shown in Figure 1.161.

Figure 1.161: Reach 29 (416204A) model structure

6.29.3 Calibration Methodology

The reach calibration was undertaken as outlined in Section 4. Key points specific to this reach are:  Historical diversions within this reach are unavailable, and were therefore assumed to be zero during the flow calibration.  A detailed model of the Upper Weir was developed as part of the previous IQQM model of the Border . This model included a number of breakouts, floodplain storages, routing, and losses. This model was ported to Source to provide the structure and the initial representation of the hydrology within the reach.  A Sacramento rainfall-runoff model was developed to simulate the runoff from the six subareas in this reach UW1–UW6. It was calibrated to the historical flows at Gunn Bridge (416204A) using an optimisation tool.

330 Hydrology Report No 416000.PR/6

6.29.4 Data

6.29.4.1 Streamflow Data and Calibration Period

Daily streamflow data was obtained from the Hydstra database. The period of available data for each gauge in this reach is outlined in Table 1.221 and shown in Figure 1.162.

The downstream gauge, Gunn Bridge (416204A), provides 15 years of continuous record. The short periods of record provided by the flood warning gauges were insufficient to support subdivision of this reach into multiple independent calibration regions. Instead, data from those gauges was used to derive and/or validate the routing and breakouts modelled within the reach.

Table 1.221: Reach 29 (416204A) flow data

Gauge Gauge Name Start Date End Date Number Flood Warning Gauges 416950 Weir River at O'Connor 08/07/1993 27/05/2000 416952 Weir River at Retreat Bridge 22/09/1990 30/06/2000 416953 Weir River at Ballymena 25/09/1990 19/03/2000 Downstream Gauge 416204A Weir River at Gunn Bridge 01/07/2000 31/12/2015

Figure 1.162: Reach 29 (416204a) gauge data availability

6.29.4.2 Climate Data

Rainfall and potential-evapotranspiration data has been used in the Sacramento rainfall-runoff models. Evaporation data has also been used in the representation of evaporative losses from storage surfaces. Daily rainfall and evaporation was obtained from the SILO database for these purposes. The rainfall data was reviewed to ensure that there were no unexplained trends in the data which may be introduced by deficient infilling/extension.

Rainfall stations were chosen based on their location, their correlation with the target gauge flows (416204A), and their length of record.

331 Department of Environment and Science

The rainfall data for the selected rainfall stations were extended to the full simulation period by infilling with raw rainfall data from other nearby stations and Patched Point data (from the original station if possible). This is shown in Table 1.222 and Figure 1.163.

Potential evapotranspiration data for PO Station (053004) was used for rainfall-runoff modelling. Details about the station are given in Section 5.4.

Table 1.222: Reach 29 (416204a) rainfall data

Station # Station Name Rainfall Infilling 041058 Kindon r041058, r041139, r041348, r041394, p041058 041110 Turallin r041110, r041069, r041127, p041110 041152 Langley TM r041152, r041058, p041127 041349 Mundagai r041349, r041508, r041021, r041468, p041349 041545 Dunmore Exchange TM r041545, r041025, p041374 Note: “r” refers raw data and “p” refers Patched Point data.

332 Hydrology Report No 416000.PR/6

Figure 1.163: Reach 29 (416204a) rainfall infilling

6.29.4.3 Water Infrastructure

The instream infrastructure in this reach consists of Brown Storage (173 ML) and a Floodplain Storage of 150 ML (Breakout 2 Low Storage) associated with a low-flow breakout in the reach. Further information about this infrastructure can be found in Section 5.6.

333 Department of Environment and Science

6.29.4.4 Historical Extraction Data

Queensland users have access to unregulated flows in this reach. However no records of historical extractions are available, and extractions were assumed to be zero during the flow calibration.

6.29.4.5 Knows Effluents

There are breakouts immediately upstream of the Gunn Bridge (416204A) gauge. These have been modelled as per the previous IQQM model. There is a low-effluent and a high-effluent component. These are described below.

A low-effluent component of the breakout commences at 400 ML/day and fills a nearby floodplain which drains back into the stream. The breakout relationship for this component was adopted from the previous IQQM model, and is shown in Table 1.223. Further information about this storage can be found in Section 5.6.

Table 1.223: Reach 29 (416204a) low-effluent breakout

Upstream Flow (ML/d) Branch flow (ML/d) 0 0 400 0 1,820 1,220 3,600 2,300 9,200 2,950 100,000,000 2,950

A high-effluent component of the breakout commences at 24,000 ML/day and diverts a portion of the high-flows around the Gunn Bridge (416204A) gauge into the downstream reach. The breakout relationship for this component was adopted from the previous IQQM model, and is shown in Table 1.224.

Table 1.224: Reach 29 (416204a) high-effluent breakout

Upstream Flow (ML/d) Branch flow (ML/d) 0 0 24,000 0 25,000 1,000 35,000 8,000 100,000,000 40,000,000

334 Hydrology Report No 416000.PR/6

6.29.5 Reach Model Calibration

6.29.5.1 Streamflow Routing

Flow routing in this reach was represented using piecewise-linear routing. Following the structure of the previous IQQM model, there are four identical routing links in this reach. These were configured using parameters adapted from the previous IQQM model, and validated based on timing of events observed in flood warning streamflow stations. Reach evaporation was ignored in this reach. The assumed properties for each of the four identical routing links are shown in Table 1.225 and Table 1.226.

Table 1.225: Reach 29 (416204a) routing parameters for each of four identical links

Parameter Adopted Value x (inflow bias) 1 Number of divisions 1 Area (km2) 0 Reach length (km) - Evaporation -

Table 1.226: Reach 29 (416204a) piecewise-linear routing for each of four identical links

Flow (ML) Lag (days) 0 0.25 1,000,000 0.25

6.29.5.2 Residual Loss

Four in-stream losses were adopted from the previous IQQM model. These are shown in Table 1.227.

Table 1.227: Reach 29 (416204a) residual losses

Source Description Loss ratio Node # (constant) 0059 Residual losses to O’Connor (416950) 1.4% 0063 Residual losses from O’Connor (416950) to Retreat Bridge (416952) 13.9% 0064 Residual losses from Retreat Bridge (416952) to Ballymena (416953) 7.7% 0224 Residual losses from Ballymena (416953) to Gunn Bridge (416204A) 3.8%

335 Department of Environment and Science

6.29.5.3 Observed Inflow Sequence

Although this is a headwater reach, the nonlinear processes modelled in this reach mean deriving inflows directly from the Gunn Bridge (416204A) streamflow data is nontrivial. While inflows in this reach were not derived directly from the Gunn Bridge (416204A) data, they will be adjusted on the basis of observed streamflow data at Talwood (416202A) by a flow adjustment procedure. This is described in Section 7.4.

6.29.6 Sacramento Model Calibration

The Sacramento model for this reach was calibrated to the observed flows at Gunn Bridge (416204A) using an optimisation tool. The calibrated inflow sequence was checked visually to ensure good fit in both high- and low-flow regimes during the period of record.

The effective catchment rainfall was estimated using a weighted combination of the rainfall sequences developed in the previous section. The weighting factors were determined as part of the calibration process and are shown in Table 1.228.Table 6.4 The table shows the mean rainfall in each of the contributing sequences, and in the final sequence.

The mean annual rainfall over the period 1920–1969 is estimated to be 591 mm/y by the 50-Year Isohyet Maps. The adopted catchment rainfall sequence has a mean of 821 mm/y over the same period, which is 39 percent higher.

Table 1.228: Reach 29 (416204a) rainfall sequences details

Infilled Rainfall Sequence Mean Rainfall (mm/y) Contributing Factor 07/1889 to 06/2014 041058 582 0.167 041110 663 0.084 041152 667 0.452 041349 605 0.467 041545 685 0.165 Catchment rainfall Sequence 849 -

The calibrated Sacramento parameter values were manually checked for anomalies. Table 1.229 shows the final Sacramento model parameters for the reach.

Figure 6.1 shows the report card comparing the observed flows and modelled flows for this reach for the calibration period. Table 1.230 shows the summary results. The modelled flows reproduce the gauged flows reasonably well. The model has trouble in the very low flows (below about 1 ML/day) possibly due to the representation of the floodplain return in this reach. The daily flow- exceedance curve displays good agreement between the modelled and recorded flows. The model reproduces a total inflow volume very close to the observe volume over the full period of calibration. The univariate model calibration goodness of fit indicators range from good to excellent, while the bivariate statistics are rated as poor.

336 Hydrology Report No 416000.PR/6

After the Sacramento model calibration, the calibrated model was verified. The purpose was to confirm that the parameters could be used for other periods with an acceptable level of accuracy. The whole flow sequence was used for the calibration, and the available record was split into two subsets (post- and pre-1/1/2008) for verification. Table 1.230 presents the performance statistics for the verification periods. The model has a volume bias of 14.6 percent over the first verification period (which is a dry period) and −5.5 percent over the second verification period (which is a wet period. The model performs acceptably over both verification periods.

Table 1.229: Reach 29 (416204a) Sacramento model parameters

Parameter Adopted Value Adimp 0.00035326 Lzfpm 55.96 Lzfsm 26.00 Lzpk 0.002581 Lzsk 0.07727 Lztwm 290.4 Pctim 0.0003284 Pfree 0.05585 Rexp 2.301 Sarva 1.489E-06 Side 0.0003421 Ssout 0.02022 Uzfwm 15.76 Uzk 0.4083 Uztwm 56.05 Zperc 57.80 uh0 0 uh1 0.1906 uh2 0.8094

337 Department of Environment and Science

Table 1.230: Reach 29 (416204a) Sacramento model results

Parameter Model Model Model Calibration Verification 1 Verification 2 Period 01/07/2000 01/07/2000 to 01/01/2008 to to31/12/2015 31/12/2007 31/12/2015

Total observed flow volume (ML) 1,249,997 298,534 951,463

Modelled flow bias (%) −0.7 14.6 −5.5

Nash-Sutcliffe efficiency 0.85 0.86 0.85

6.29.7 Reach Inflow Sequence

The Sacramento model described above was run for the period 01/07/1889 to 30/06/2014 to generate full-length inflow sequences for the subareas UW1 to UW6. The historical flow records from the Gunn Bridge (416204A) gauge were not directly used in the final inflow sequences. However the inflows to this reach, as well as those to in the downstream reach were adjusted to align the model with the recorded flows at Talwood (416202A). This is dicussed further in Section 7.4 below.

Table 1.231 lists the model inflows, and corresponding subareas.

Table 1.231: Reach 29 (416204a) model inflows

Model Inflow Subareas 0310 UW 416204a Headwater Inflow (UW1) UW1

0311 UW 416204a Headwater Inflow (UW2) UW2

0312 UW 416204a Residual Inflow (UW3) UW3

0314 UW 416204a Residual Inflow (UW4) UW4

0316 UW 416204a Residual Inflow (UW5) UW5

0318 UW 416204a Residual Inflow (UW6) UW6

338 Hydrology Report No 416000.PR/6

Figure 6.1: Reach 29 (416204a) report card comparing Sacramento inflow to observed flow

339 Department of Environment and Science

6.30 Reach 30 – 416204a Weir River at Gunn Bridge to 416202a Weir River at Talwood (Residual)

6.30.1 Reach Description

This section of the report presents the data and flow calibration results for the Weir River from Gunn Bridge (416204A) to Talwood (416202A) residual reach. The reach includes 8 subareas (see Table 7.1) which appear in the Upper Weir and Lower Weir as shown in (see Figure 1.2). This reach has a total catchment area of 7,755 km2.

Table 7.1: Reach 30 (416202a) subareas

Subarea State Description Area (km2) UW7 Qld 90.78 UW8 Qld 1,202.20 UW9 Qld 628.52 LW1 Qld Subareas between 416204A (Weir River at 2,199.54 Gunn Bridge) and 4162021A (Weir River at LW2 Qld Talwood). 785.952 LW3 Qld 1,907.39 LW4 Qld 834.54 LW5 Qld 106.49 Area Total 7,755.41

340 Hydrology Report No 416000.PR/6

Figure 1.2: Reach 30 (416202a) map

6.30.2 Reach Structure

A detailed model of the Weir River to Talwood (416202A) was developed as part of the previous IQQM model of the . It includes representations of the known hydrologic features in this reach including inflows, routing, losses, and floodplain breakouts. This model was ported to Source to provide the structure and the initial representation of the reach. No new information pertaining to the structure of the reach was available. The adopted structure is shown in Figure 1.3.

Inflows from the upstream reach are measured at the Gunn Bridge streamflow gauge (416204A), shown near the top-centre of the figure. However, during some high-flow events, flows are known bypass the Gunn Bridge gauge via a floodplain and return to the Weir River upstream of Talwood (416202A). This is captured in the adopted structure.

The reach ends at Talwood (416202A), shown in the bottom-left.

There are eight residual inflows in the reach corresponding to the eight subareas identified in the previous section. Inflow timeseries will be developed for these as part of the calibration discussed in the following sections.

The reach contains several waterholes and floodplains, which behave as passive storages. It also contains several breakouts and losses. These were identified during the development of the previous IQQM surface water planning model, and have been included as shown. Their features have been adapted from the previous IQQM model and revised (as required) as part of the calibration discussed in the following sections.

341 Department of Environment and Science

Figure 1.3: Reach 30 (416202a) model structure

6.30.3 Calibration Methodology

The reach calibration was undertaken as outlined in Section 4. Key points specific to this reach are:  Historical diversions within this reach are unavailable, and were therefore assumed to be zero during the flow calibration.  Piecewise-linear routing was optimised for the reach with manual adjustments made to the final relationship.  A Sacramento rainfall-runoff model was developed to simulate the runoff from the eight subareas in this reach UW7–UW9 and LW1–LW5. This Sacramento model was calibrated to historical flows at Talwood (416202A) using an optimisation tool. During this the inflows from the upstream reach were estimated using the 416204A reach model developed in Section 6.29.

342 Hydrology Report No 416000.PR/6

6.30.4 Data

6.30.4.1 Streamflow Data and Calibration Period

There are seven streamflow gauges in this reach. Daily streamflow data was obtained from the Hydstra database. The period of available data for each gauge in this reach is outlined in Table 1.2 and shown in Figure 1.4.

The Talwood (416202A) streamflow gauge provides over 50 years of record between 03/05/1949 and 31/12/2015 with a notable break from 1955 to 1968. This record includes a number of floods and dry spells and is therefore well-suited as a target gauge for flow calibration. In comparison the upstream station Gunn Bridge (416204A) provides just 15 years of record.

In order to account for the ungauged inflows that bypass Gunn Bridge via a floodplain breakout, the inflows to this reach were estimated using the upstream reach model rather than the Gunn Bridge streamflow gauge. This ensured consistency in the assumptions regarding the Gunn Bridge breakout during the calibration of this reach, and allowed the calibration to be performed over the full length of record available at Talwood (416202A). That period is 03/05/1949 to 31/12/2015.

Table 1.2: Reach 30 (416202a) flow data

Gauge Gauge Name Start Date End Date Number Upstream Gauges 416204A Weir River at Gunn Bridge 01/07/2000 31/12/2015 Intermediate Gauges 4162051 Murri Murri Creek at Murreba 12/02/2009 19/11/2015 4162052 Yambocully Creek at Broomfield Reserve 12/02/2009 19/11/2015 4162053 at Sunny Reserve 12/02/2009 19/11/2015 Flood Warning Gauges 4162054 Yarrill Creek at Medpark Bridge 28/07/1993 21/06/2000 4162056 Weir River at Surrey 24/12/1991 01/06/2000 Downstream Gauge 416202A Weir River at Talwood 03/05/1949 31/12/2015

343 Department of Environment and Science

Figure 1.4: Reach 30 (416202a) gauge data availability

6.30.4.2 Climate Data

Rainfall and potential-evapotranspiration data have been used in the Sacramento rainfall-runoff models. Evaporation data have also been used in the representation of evaporative losses from storage surfaces. Daily rainfall and evaporation were obtained from the SILO database for these purposes. The rainfall data were reviewed to ensure that there were no unexplained trends in the data which may be introduced by deficient infilling/extension.

Rainfall stations were chosen based on their location, their correlation with the target gauge flows (416202A), and their length of record.

The rainfall data for the selected rainfall stations were extended to the full simulation period by infilling with raw rainfall data with Patched Point data from the same station. This is shown in Table 1.3. The figures below show the proportion of raw and Patched Point data for each of the rainfall stations (see Figure 1.5).

Potential evapotranspiration data for Boggabilla PO Station (053004) was used for rainfall-runoff modelling. Details about the station are given in Section 5.4.

Table 1.3: Reach 30 (416202a) rainfall data

Station # Station Name Rainfall Infilling 041128 Wondalli r041128 & r041129 & r041122 & p041128 041370 Yagaburne r041370 & r041395 & r041139 & r041348 & p041370 041397 Burilda r041397 & r042115 & r042080 & r042102 & p041397 041554 Talinga r041554 & r042088 & r041509 & r041349 & p041554 042030 Bungunya School r042030 & r042104 & r042047 & p042030 Note: “r” refers raw data and “p” refers Patched Point data.

344 Hydrology Report No 416000.PR/6

Figure 1.5: Reach 30 (416202a) rainfall infilling

345 Department of Environment and Science

6.30.4.3 Water Infrastructure

Seven storages were simulated within this reach. One of these (Lalaguli storage) is located on the channel of the Weir River. Two are located on the channels of Billa Billa Creek and Yarrill Creek. And the remaining four are off-stream floodplain storages modelled on breakout branches. The storage and spillway relationships for these storages were taken directly from previous IQQM model. All these storages are passive. They are summarised in Table 1.4 and further information can be found in Section 5.6.

Table 1.4: Reach 30 (416202a) storages

Source Node Number Storage Name Storage Volume (ML) 0348 Lalaguli Storage (On-stream) 238 0358 Billa Billa Storage (On-stream) 867 0361 Yarrill Commoron Storage (On-stream) 12,130 0365 Breakout 2 High Storage (Off-stream) 173 0377 Breakout 3 Low Storage (Off-stream) 130 0381 Eurone Swamp (Off-stream) 12,130 0274 Floodplain Storage (Off-stream) 14,400

6.30.4.4 Historical Extraction Data

Queensland users have access to unregulated flows in this reach. However no records of historical extractions are available, and extractions were assumed to be zero during the flow calibration.

6.30.4.5 Known Effluents

There are a number of known effluents in this reach. The breakout relationships have been adapted from the previous IQQM model, and forced during the calibration of this reach. These are shown in Table 1.5.

346 Hydrology Report No 416000.PR/6

Table 1.5: Reach 30 (416202a) effluents

Upstream Flow (ML/d) Breakout (ML/d) Breakout 2 High Effluent (to off-stream storage) Relationship given in Reach 29 Section Billa Billa Breakout Effluent (re-enters below Billa Billa Storage) 0 0 18,300 0 18,320 20 19,350 1,050 20,000 1,500 100,000 10,000 108 107 Breakout 3 Low Effluent (to off-stream storage) 0 0 13,000 0 26,000 4,000 750,000 7,500 107 105 Breakout 3 High Effluent (to off-stream storage) 0 0 17,300 0 17,400 100 108 2,950 Floodplain Storage Effluent (to off-stream storage) 0 0 1,500 0 3,000 500 9,000 2,000 750,000 7,500 107 105 Yarrilwanna Breakout Effluent (to Weir River downstream of Talwood (416202a)) 0 0 32,500 0 35,000 2,500 40,000 7,500 45,000 12,500 50,000 17,500 106 967,500

347 Department of Environment and Science

6.30.5 Reach Model Calibration

6.30.5.1 Streamflow Routing

Flow routing in this reach was represented using piecewise-linear routing. Following the structure of the previous IQQM model, there are 14 routing links in this reach. The piecewise-linear routing tables were optimised against the recorded downstream flows during the period where historical 416202A data was available. A draft rainfall-runoff model was used to simulated inflows from the upstream and residual reaches during routing optimisation. The optimised routing relationships were checked and extrapolated manually. Reach evaporation was ignored in this reach. The reaches are detailed in Table 1.6, and the calibrated piecewise-linear routing tables for each are shown in Table 1.7.

Table 1.6: Reach 30 (416202a) routing links

Routing link x (inflow Number of Area Reach Evaporation bias) divisions (km2) Length Station (km) Billa Billa 1 4 0 30.0 n/a Yar 2 1 4 0 29.5 n/a Yar 1 1 4 0 20.0 n/a Com 1 4 0 100.0 n/a R11 1 4 0 29.5 n/a R22 1 4 0 11.1 n/a Breakout 3 high 1 4 0 - n/a Ds Commoron 1 4 0 8.2 n/a Yam 1 4 0 55.0 n/a Mobandilla 1 4 0 64.2 n/a Breakout 2 high 1 4 0 100.0 n/a R21 1 4 0 22.6 n/a Lalaguli 1 4 0 12.1 n/a Yarrilwanna b/o 1 4 0 90.0 n/a

348 Hydrology Report No 416000.PR/6

Table 1.7: Reach 30 (416202a) piecewise-linear routing

Flow (ML) Lag (days) Billa Billa 0 1.30 136 1.25 362 1.20 905 1.10 2,263 0.95 4,525 0.90 9,050 1.00 18,100 1.25 106 1.25 Yar 2 0 1.20 536 1.10 1,430 1.00 3,575 0.75 8,938 0.70 17,875 0.80 35,750 1.00 71,500 1.15 106 1.15 Yar 1 0 1.25 244 1.20 650 1.10 1,625 0.95 4,063 0.80 8,125 0.75 16,250 0.85 32,500 1.10 106 1.10 Com 0 1.00 964 1.05 2,570 1.10 6,425 1.18

349 Department of Environment and Science

Flow (ML) Lag (days) 16,063 1.30 32,125 1.35 64,250 1.32 128,500 1.25 106 1.25 R11 0 1.05 184 1.07 490 1.10 1,225 1.15 3,063 1.20 6,125 1.15 12,250 0.95 24,500 0.85 106 0.85 R22 0 0.80 758 0.81 2,020 0.82 5,050 0.85 12,625 0.90 25,250 0.97 50,500 1.10 101,000 1.20 106 1.20 Breakout 3 High 0 2.00 106 2.00 DS Commoron 0 0.75 896 0.80 2,390 0.90 5,975 1.00 14,938 1.00 29,875 0.25

350 Hydrology Report No 416000.PR/6

Flow (ML) Lag (days) 59,750 0.25 119,500 1.00 106 1.00 Yam 0 1.13 349 1.15 930 1.17 2,325 1.22 5,813 1.29 11,625 1.32 23,250 1.34 46,500 1.28 106 1.28 Mobandilla 0 0.35 244 0.42 650 0.50 1,625 0.70 4,063 1.00 8,125 1.50 16,250 1.80 32,500 0.10 106 0.10 Breakout 2 High 0 1.20 51 1.21 136 1.21 340 1.22 850 1.24 1,700 1.28 3,400 1.32 6,800 1.25 106 1.25 R21 0 0.25

351 Department of Environment and Science

Flow (ML) Lag (days) 480 0.25 1,280 0.25 3,200 0.27 8,000 0.32 16,000 0.40 32,000 0.65 64,000 1.40 106 1.40 Lalaguli 0 0.10 281 0.10 748 0.10 1,870 0.10 4,675 0.35 9,350 0.90 18,700 1.40 37,400 0.65 106 0.65 Yarrilwanna 0 1.27 679 1.25 1,810 1.22 4,525 1.20 11,313 1.24 22,625 1.27 45,250 1.25 90,500 1.24 106 1.24

352 Hydrology Report No 416000.PR/6

6.30.5.2 Residual Loss

There are 12 residual losses in this reach. The loss relationships were adopted directly from the previous IQQM model. Eleven of these are modelled using linear loss functions (see Table 1.8) and one is a flow-dependent piecewise-linear loss table (Table 1.9).

Table 1.8: Reach 30 (416202a) linear residual losses

Source Node Description Loss Ratio Number (constant) 0366 Breakout 2 high loss 50.63% 0386 Residual losses from Gunn Bridge (416204A) to Commoron 5.494% 0075 Residual losses on Billa Billa 5.585% 0113 Residual losses on Yarrill Creek 1 3.759% 0079 Residual losses on Yarrill Creek 2 7.376% 0372 Residual losses on Commoron Creek 40.73% 0114 Residual losses from Commoron confluence to Yambocully 9.602% confluence 0382 Breakout 3 high loss 9.460% 0117 Resdual losses on Yambocully Creek 25% 0134 Residual losses from Yambocully confluence to Mobandilla 28.52% 0040 Residual losses on Yarrilwanna Creek 85.11%

Table 1.9: Reach 30 (416202a) residual losses from Mobandilla to Talwood (416202A)

Upstream Flow (ML/d) Residual loss (ML/d) Loss ratio 0 0 100% 2 2 100% 1,813 1,028 56.70% 2,571 1,478 57.49% 3,179 1,741 54.77% 3,434 1,759 51.22% 3,518 1,800 51.17% 32,459 9,000 27.73% 1010 9,000 0%

353 Department of Environment and Science

6.30.5.3 Observed Inflow Sequence

Due to relatively short historical record available at Gunn Bridge (416204A), the historical flow records at Talwood (416202A) and Gunn Bridge (416204A) were not directly used to derive an observed inflow sequence by mass balance. However the inflows to this reach, as well as those to the upstream reach, will be adjusted on the basis of recorded flows at Talwood (416202A) by a flow adjustment procedure. This is described in Section 7.4.

6.30.6 Sacramento Model Calibration

The Sacramento model for this reach was calibrated to the observed flows at Talwood (416202A) using an optimisation tool. The calibrated inflow sequence was checked visually to ensure good fit in both high- and low-flow regimes during the period of record.

The effective catchment rainfall was estimated using a weighted combination of the rainfall sequences developed in the previous section. The weighting factors were determined as part of the calibration process and are shown in Table 1.10.Table 6.4 The table shows the mean rainfall in each of the contributing sequences, and in the final sequence.

The mean annual rainfall over the period 1920–1969 is estimated to be 549 mm/y by the 50-Year Isohyet Maps. The adopted catchment rainfall sequence has a mean of 753 mm/y over the same period, which is 37 percent higher.

Table 1.10: Reach 30 (416202a) rainfall sequence details

Infilled Rainfall Sequence Mean Rainfall (mm/y) Contributing Factor 07/1889 - 06/2014 041128 610 0.230 041370 571 0.100 041397 577 0.352 041554 590 0.200 042030 534 0.487 Catchment rainfall Sequence 779 -

The calibrated Sacramento parameter values were manually checked for anomalies. Table 1.11 shows the final Sacramento model parameters for the reach.

Figure 1.6 shows the report card comparing the gauged and modelled flows for this reach over the calibration period. Table 1.12 shows the summary statistics for the calibration and two validation periods. The model fails to reproduce the heights of the two biggest events in the calibration period (1975 and 2013) and this is highlighted by the flow-exceedance curve for high flows. Aside from these, the model does a good job of reproducing the hydrographs of large events up to 25,000 and the flow-exceedance curve reflects this. The flood recession curves are reproduced very well. Model performance around 100 ML/d is important as this flow rate triggers water-harvesting events for many QLD irrigators. The model performs very well at this level. The frequency and timing of cease-to-flows is good, although the model slightly overestimates the number of nonzero flow days, and the low-flow volume is correspondingly overestimated. The model reproduces a total inflow volume equal to the derived residual volume during the full period of calibration.

354 Hydrology Report No 416000.PR/6

Table 1.11: Reach 30 (416202a) Sacramento model parameters

Parameter Adopted Value Adimp 0.00008281 Lzfpm 25.32 Lzfsm 56.36 Lzpk 0.004846 Lzsk 0.03373 Lztwm 102.6 Pctim 0.00006577 Pfree 0.08345 Rexp 5.891 Sarva 7.107x10-6 Side 0.0008442 Ssout 0.04807 Uzfwm 45.80 Uzk 0.1200 Uztwm 53.07 Zperc 115.6 uh0 0.1923 uh1 0.8077 uh2 0

Table 1.12: Reach 30 (416202a) Sacramento model results

Parameter Model Model Model Calibration Verification 1 Verification 2 Period 02/03/1937 to 02/03/1937 to 01/07/1976 to 30/06/2014 30/06/1976 30/06/2014

Total observed flow volume (ML) 8,133,740 2,271,889 5,826,518

Modelled flow bias (%) 0.0 13.3 −4.5

Nash-Sutcliffe efficiency 0.78 0.75 0.80

6.30.7 Reach Inflow Sequence

The Sacramento model described above was run for the period 01/07/1889 to 30/06/2014 to generate full-length inflow sequences for the subareas UW7 to UW9 and LW1 to LW5.

355 Department of Environment and Science

The historical flow records from Talwood (416202A) and Gunn Bridge (416204A) gauge were not directly used to calculate a residual inflow sequence by mass balance. However the inflows to this reach, as well as those in the upstream reach were adjusted on the basis to align the model with the recorded flows at Talwood (416202A). This is dicussed further in Section 7.4 below.

Table 1.231 lists the model inflows, and corresponding subareas.

Table 1.13: Reach 30 (416202a) model inflows

Model Inflow Subareas 0322 UW 416202a Residual Inflow (UW7) UW7

0323 UW 416202a Yarrill Headwater Inflow (UW8) UW8

0357 UW 416202a Billa Billa Headwater Inflow (UW9a) UW9 0325 UW 416202a Yarrill Residual Inflow (UW9b)

0330 LW 416202a Commoron Headwater Inflow (LW1) LW1

0331 LW 416202a Yambocully Headwater Inflow (LW2) LW2

0329 LW 416202a Residual Inflow (LW3) LW3

0343 LW 416202a Residual Inflow (LW4 LW5) LW4, LW5

356 Hydrology Report No 416000.PR/6

Figure 1.6: Reach 30 (416202a) report card comparing Sacramento inflow to observed flow

357 Department of Environment and Science

6.31 Reach 31 – 416034 Croppa Creek at Tulloona Bore (Headwater)

6.31.1 Reach Description

Croppa Creek at Tulloona Bore (416034) is a headwater catchment. There are no streamflow gauges upstream and flows are not affected by diversions or regulation. It has a catchment area of 1,192 km2. A map of the catchment is shown in Figure 1.7. The reach subareas are identified in Table 1.14 and are also shown on the map.

Table 1.14: Reach 31 (416034) subareas

Subarea State Description Area (km2) M24 NSW Croppa Creek to Tulloona Bore (416034) 1,192.3 Area Total 1,192.3

Figure 1.7: Reach 31 (416034) map

358 Hydrology Report No 416000.PR/6

6.31.2 Data

6.31.2.1 Streamflow Data and Calibration Period

Daily streamflow data was obtained from the Hydstra database. The observed inflow sequence for this reach is outlined in Table 1.15. All of the available data was used for the Sacramento calibration.

Table 1.15: Reach 31 (416034) flow data

Gauging Number Gauge Name Date 416034 Croppa Creek at Tulloona Bore 28/06/1972 to 16/02/1989

6.31.2.2 Climate Data

The selected rainfall stations for rainfall-runoff modelling for this catchment are detailed in Table 1.16. The table also shown is the infilling. Figure 1.8 shows the infilling visually.

Potential evapotranspiration data for Research Centre (56018) was used for rainfall-runoff modelling. Details about the station are given in Section 5.4.

Table 1.16: Reach 31 (416034) rainfall data

Station # Station name Rainfall infilling stations 53041 Tulloona (Coolanga) r53041, r53018, 53041p 53018 Croppa Creek (Krui Plains) r53018, r53041, 53018p 54130 Croppa Creek (Belford Street) r54130, r54129, r54124, r53095, 53095p 54129 Croppa Creek (Rawdon) r54129, r54124, r53095, r54032, 53095p 54124 Crooble Station r54124, r54129, r53029, r53033, 53029p 54029 Warialda PO r54029, r54017, 54029p 53047 North Star PO r53047, r53095, 53095p Note: “r” refers raw data and “p” refers Patched Point data.

359 Department of Environment and Science

Figure 1.8: Reach 31 (416034) rainfall infilling

360 Hydrology Report No 416000.PR/6

6.31.3 Sacramento Model Calibration

The Sacramento model for this reach was calibrated to the observed flow at the downstream gauge using an optimisation tool. The calibrated flow sequence was checked visually to ensure good fit in both high- and low-flow regimes during the period of record.

The effective catchment rainfall was estimated using a weighted combination of the rainfall sequences developed in the previous section. The weighting factors were determined as part of the calibration process and are shown in Table 1.17.Table 6.4 The table shows the mean rainfall in each of the contributing sequences, and in the final sequence.

The mean annual rainfall over the period 1920–1969 is estimated to be 667 mm/y by the 50-Year Isohyet Maps. The adopted catchment rainfall sequence has a mean of 654 mm/y over the same period, which is 12 percent lower.

Table 1.18 shows the final Sacramento model parameters for the reach and key calibration statistics are provided in Table 1.19. The report card is presented in Error! Reference source not found..

Table 1.17: Reach 31 (416034) rainfall sequence details

Rainfall Station Mean Rainfall (mm/y) Factor 07/1889 to 06/2014 54124 580 0.390 53047 619 0.190 53018 586 0.119 54029 688 0.130 54129 613 0.153 54130 613 0.071 53041 584 0.078 Catchment rainfall Sequence 686 -

361 Department of Environment and Science

Table 1.18: Reach 31 (416034) Sacramento model parameters

Parameter Adopted Value Adimp 0.000365 Lzfpm 2.614 Lzfsm 14.563 Lzpk 0.00468 Lzsk 0.266 Lztwm 268.095 Pctim 0.000233 Pfree 0.0175 Rexp 2.826 Sarva 1.52E-06 Side 0.000188 Ssout 0.00216 Uzfwm 20.967 Uzk 0.355 Uztwm 55.422 Zperc 33.998 uh0 0 uh1 0.615 uh2 0.385

Table 1.19: Reach 31 (416034) Sacramento model results

Parameter Model Calibration Period 28/06/1972 to 16/02/1989

Total observed flow volume (ML) 330,889

Modelled flow bias (%) 0.0

Nash-Sutcliffe efficiency 0.71

6.31.4 Reach Inflow Sequence

The Sacramento model described above was run for the period 01/07/1889 to 30/06/2014. The simulation period inflow sequence for this reach was created by infilling and extending the observed inflows with the aforementioned Sacramento flow. The composition of the final inflow file is shown in Figure 1.9.

362 Hydrology Report No 416000.PR/6

A single inflow file was used in the final representation of this reach.

Figure 1.9: Reach 31 (416034) inflow composition

363 Department of Environment and Science

Figure 1.10: Reach 31 (416034) report card comparing Sacramento inflow to observed flow

364 Hydrology Report No 416000.PR/6

6.32 Reach 32 – 416202a Weir River at Talwood, 416046 Macintyre River at Boonanga Bridge, 416034 Croppa Creek at Tulloona Bore, 416037 at Boomi Weir Offtake, Yarrilwanna Creek breakout and Whalan Creek breakout to 416028 Boomi River at Neeworra and 416001 Barwon River at (Residual)

6.32.1 Reach Description

This section of the report presents the data and flow calibration results for the following reach in the Border Rivers:  Macintyre River from Boonanga Bridge (416046) to Mungindi (416001), including the Weir River below Talwood (416202A), the Boomi River below the Boomi Weir Offtake (416046), and Croppa Creek below Tulloona Bore (416034).

The residual catchment of this reach includes 12 subareas (see Table 1.20) and is shown in Figure 1.11). The residual catchment is low-lying and has a total area of 5,891.37 km2.

365 Department of Environment and Science

Table 1.20: Reach 32 (416001 & 416028) subareas

Area Subarea State Description (km2) Qld catchment of the Macintyre River between Boonanga Bridge M6 Qld 444.1 (416046) and Kanowna (416048) including Booberanna Creek Weir River catchment between Talwood (416202a) and the Newinga M7 Qld 108.9 breakout NSW catchment of the Macintyre River between Boonanga Bridge M12 NSW 148.8 (416046) and Kanowna (416048) including Gnoura Gnoura Ck Weir River catchment between the Newinga breakout & Jericho M13 Qld 60.0 (416205a) M14 NSW The Boomi River catchment (d/s Boomi Weir offtake) part 1 444.5 M15 NSW The Boomi River catchment (d/s Boomi Weir offtake) part 2 80.8 The Weir River catchment between Jericho (416205a) and Mascot M16 Qld 1,203.7 (416207a) Qld catchment of the Macintyre River between Kanowna (416048) and M17 Qld 94.6 Yarrowee (416051) NSW catchment of the Macintyre River between Kanowna (416048) and M18 NSW 23.8 Yarrowee (416051) Whalan Creek upstream of Neeworra (416028) excluding Croppa Creek M19 NSW 3,115.2 upstream of Tulloona Bore (416034) Qld catchment of the Macintyre River downstream of Yarrowee M20 Qld (416051) and the Weir River downstream of Mascot (416207a) to 88.0 Mungindi (416001) NSW catchment of the Macintyre River downstream of Yarrowee M21 NSW 78.9 (416051) to Mungindi (416001) Area Total 5,891.4

366 Hydrology Report No 416000.PR/6

Figure 1.11: Reach 32 (416001 & 416028) map

6.32.2 Reach Structure

This calibration reach comprises the most downstream regions of the Border Rivers water plan area. It has two gauged endpoints, the Barwon River at Mungindi (416001) and Boomi River at Neeworra (416028). This reach receives flows from:

 the Weir River via Talwood (416202A)  the Weir River via the Yarrilwanna Creek breakout  the Macintyre River via Boonanga Bridge (416046)  the Macintyre River via the Boomi River at Boomi Weir Offtake (416037)  the Macintyre River via the Whalan Creek breakout  Croppa Creek via Tulloona Bore (416034).

It is the second largest reach in the system, and arguably the most complex due to the number of contributing streams and a number of bifurcations and breakouts activated in high-flow events. The reach structure is shown in Figure 1.12.

367 Department of Environment and Science

Figure 1.12: Reach 32 (416001 & 416028) model structure also viewable online

The adopted structure is similar to that of the previous IQQM model, and incorporates the key hydrologic processes occurring in this reach as listed below.

1. Streamflow routing in 21 locations o Macintyre River between Boonanga and Kanowna (3 parts) o Macintyre River between Kanowna and Mungindi (4 parts) o Weir River between Talwood and Jericho (2 parts) o Weir River between Jericho and Mascot o Boomi River to Neeworra (5 parts) o Whalan Creek (4 parts) o Croppa Creek o Boomangera Creek 2. Lumped streamflow losses o Macintyre River between (416201A) and Terrewah (416047) o Macintyre River between Terrewah (416047) and Boomi Weir (416043) o Callandoon Creek between the offtake and Carana Weir (416203A) o Callandoon Creek between Carana Weir (416203A) and Oonavale (416306A) 3. Floodplain breakouts identified by the NSW Healthy Floodplains Project

368 Hydrology Report No 416000.PR/6

o NSW Yarrowee floodplain breakout o NSW Whalan floodplain breakout o NSW Croppa-Whalan floodplain breakout o NSW Boomi-Whalan floodplain breakout o NSW Boomangera floodplain breakout 4. Effluents o Newinga breakout I – Macintyre to Weir River o Newinga breakout II – Weir River to Macintyre o Little Barwon Creek effluent o Unnamed effluent (Whalan Creek to Boomi River) o Boomangera Creek effluent o effluent 5. Lumped diversions o QL9 on-allocation diversions o QL9 off-allocation diversions o QL10 and QL11 off-allocation diversions o QL10 and QL11 off-allocation diversions o NS10 on-allocation diversions o NS10 off-allocation diversions o NS11 and NS12 on-allocation diversions o NS11 and NS12 off-allocation diversions 6. Residual inflows for subareas o M6 o M7 o M12 o M13 o M14 and M15 o M16 o M17 and M18 o M19 o M20 and M21

Amongst those features are five NSW floodplain breakouts that were identified by the NSW Healthy Floodplains Project. The properties of those breakouts have been adopted directly from that work, and have not been revised as part of this flow calibration.

The flow calibration described in the following sections includes the development of residual inflow timeseries, and calibration of the streamflow routing, effluents, and losses (excluding the NSW Healthy Floodplains breakouts) in this reach.

6.32.3 Calibration Period

Daily streamflow data was obtained from the Hydstra database. The period of available data for each gauge in this reach is outlined in Table 1.21Error! Reference source not found. and shown in Figure 1.13.

369 Department of Environment and Science

Table 1.21: Reach 32 (416001 & 416028) flow data

Gauge Gauge Name Start Date End Date Number Upstream Gauges 416202A Weir River at Talwood 03/05/1949 05/04/2016 416046 Macintyre River at Boonanga Bridge 25/08/2011 09/02/2016 416037 Boomi River at Boomi Weir Offtake 06/04/1973 17/11/2015 416034 Croppa Creek at Tulloona Bore 28/06/1972 16/02/1989 Intermediate Gauges 416048 Macintyre River at Kanowna 19/02/1987 30/06/2014 416205A Weir River at Jericho 05/07/2002 05/04/2016 416029 Boomi River at Kanowna 05/04/1968 31/12/1987 416207A Weir River at Mascot 11/01/2007 16/03/2016 Downstream Gauge 416001 Barwon River at Mungindi 02/12/1889 10/12/2015 416028 Boomi River at Neeworra 04/04/1968 23/11/1994

Figure 1.13: Reach 32 (416001 & 416028) gauge data availability Barwon River at Mungindi (416001) is the primary downstream gauge for this reach. It has near- complete records from 02/12/1889 and remains open today. Boomi River at Neeworra (416028) serves as another downstream gauge for this reach. In contrast to Mungindi, it has fragmentary records from 04/04/1968 to 23/11/1994 and is currently inactive.

Only two of the four upstream gauges contribute to the flows at Mungindi, and those are the Weir River at Tallwood (416202A) and Macintyre River at Boonanga Bridge (416046). The Tallwood gauge has significant records starting on 03/05/1949. The Boonanga Bridge gauge has only been operating since 2011 and, to improve the calibration period, the Boonanga Bridge records have been extended back to 03/05/1949 using simulated flows from the upstream reach model (Reach 28; driven by observed flows at Goondiwindi, Terrewah, and Boomi).

370 Hydrology Report No 416000.PR/6

The calibration techniques outlined in Section 0 were applied to constrain the various hydrologic processes sequentially. Starting from the top of the reach, streamflow routing, losses and breakouts were calibrated progressively using the relevant available streamflow records in each step. Once those were determined, Sacramento rainfall-runoff models were calibrated to represent the residual inflows. Key points specific to this reach are:

 Historical diversions were forced using daily time series data.  Floodplain breakout relationships previously identified by the NSW Healthy Floodplains Project were forced using loss or splitter relationships.  Additional lumped streamflow loss and effluent relationships were calibrated by comparing observed upstream and downstream flows.  Piecewise-linear routing was used and calibrated by optimisation. The final relationships were checked and extrapolated manually. Reach evaporation was simulated assuming a constant reach surface area.  Five Sacramento rainfall-runoff models were calibrated to represent the runoff from different regions in this reach.

Five Sacramento rainfall-runoff models were developed for this reach. Sacramento 1 simulates the runoff from subareas upstream of Kanowna (M6, M7, M12, M14, M15). It was calibrated over the period 19/02/1987 to 30/06/2014, limited by the records at Kanowna. Sacramento 2 simulates runoff from subareas between Kanowna and the Weir River confluence (M13, M16, M17, M18). This model was calibrated over the period 19/02/1987 to 30/06/2014, similarly limited by the records at Kanowna. Sacramento 3 simulates runoff from the upper 50 percent of subarea M19, and Sacramento 4 simulates runoff from the lower 50 percent of subarea M19. These were both calibrated over the period 19/02/1987 to 23/11/1994, limited by the records at Neeworra. Sacramento 5 simulates runoff from the subareas immediately above Mungindi (M20 and M21). These were calibrated using the full period record at Mungindi, 02/12/1889 to 30/06/2014.

6.32.4 Data Review

6.32.4.1 Flow Data

The gauges used for this residual reach calibration are shown in Table 1.21 above. Further information about the data availability and quality can be found in Section 5.2.

6.32.4.2 Rainfall and Evaporation Data

Rainfall and potential-evapotranspiration data have been used in the Sacramento rainfall-runoff models. Evaporation data has also been used in the representation of evaporative losses from reach and storage surfaces. Daily rainfall and evaporation were obtained from the SILO database for these purposes. The rainfall data was reviewed to ensure that there were no unexplained trends in the data which may be introduced by deficient infilling/extension.

Rainfall stations were chosen based on their location and their length of record.

The rainfall data for the selected rainfall stations were extended to the full simulation period by infilling with Patched Point data from the same station. This is shown in Table 1.22. The figures below show the proportion of raw and Patched Point data for each of the rainfall stations (see Figure 1.14).

371 Department of Environment and Science

Table 1.22: Reach 32 (416001 & 416028) rainfall data

Rainfall station Rainfall station name Rainfall infilling stations number 041521 Goondiwindi Airport r041521, r041038, r053004, p041521 042027 Talwood State School r042027, r042030, r042104, p042027 042104 Surrey TM r042104, r042030, p042030 042116 Arden Downs TM r042116, r042050, p042051 052004 Boomi (Barwon Street) r052004, r052050, r053042, p052004 052020 Mungindi Post Office r052020, p052020 053041 Tulloona (Coolanga) r053041, r053018, r053040, p053041 053042 Garah (Ulinga) r053042, r053085, p053042 053076 North Star (Bonanza) r053076, r053047, r053041, p053076 Note: “r” refers raw data and “p” refers Patched Point data.

372 Hydrology Report No 416000.PR/6

Figure 1.14: Reach 32 (416001 & 416028) rainfall infilling

373 Department of Environment and Science

Daily evaporation data were extracted from the SILO Patched Point database. Seven-day rolled average Morton’s wet evaporation data from Boggabilla Post Office Station (053004) werer used for Sacramento rainfall-runoff modelling in this reach. Daily Morton’s lake evaporation at the same station was used for the reach evaporation. See Table 1.23.

Table 1.23: Reach 32 (416001 & 416028) evaporation data

Evaporation Site Station Number 053004 Station Name Boggabilla Post Office % dataset which has 9 am cloud okta data 0

6.32.4.3 Water infrastructure and Storages

There is no infrastructure in this reach.

6.32.4.4 Historical extraction data

Queensland and NSW users have access to water in this reach (see Table 1.24). Historical on- allocation and off-allocation water use data in the Border Rivers is available from 1985 to 2014. The historical data is quarterly for Qld and summed over various time steps for NSW. The data for the 1985 to 1996 period were disaggregated to daily extractions for the previous Border Rivers calibration. The daily data was used for this calibration for both NSW and Queensland. The Qld records from 01/07/1997 to 30/06/2014, and NSW records from 01/07/1997 to 30/06/2009 were disaggregated to daily based on the gauged flow at Goondiwindi (416201B). The remaining NSW records (01/07/2009 to 30/06/2014) were disaggregated by NSW DPI Water based on historical order records and gauge data. Refer to Section 5.7 for further information about the disaggregation process.

Table 1.24: Reach 32 (416001 & 416028) historical diversions

Historical Diversion Location NSW and Qld Between Boomi Weir (416043) and Kanowna (416048) (on and off allocation diversion) NSW and Qld Between Kanowna (416048) and the Weir River Junction (on and off allocation diversion) NSW and Qld Between the Weir River Junction and Mungindi (416001) (on and off allocation diversion)

374 Hydrology Report No 416000.PR/6

6.32.4.5 NSW Healthy Floodplain Breakouts

The NSW DPI Healthy Flooplains Project identified four floodplain breakouts in this reach. These breakouts operate in extremely high flows and are significant from the point of view of high-flow losses and floodplain harvesting in NSW. The breakout relationships have been adopted directly from that work, and are as follows.

Table 1.25: Reach 32 (416001 & 416028) NSW Healthy Floodplains breakouts

Upstream Flow (ML/d) Breakout (ML/d) Yarrowee 0 0 1,800 0 100,000 10,000 108 107 Croppa-Whalan 0 0 6,000 0 100,000 10,000 108 107 Boomi-Whalan 0 0 10,000 0 100,000 10,000 108 107 Boomangera 0 0 1,800 0 100,000 10,000 108 107

6.32.5 Reach Model Details

6.32.5.1 Storage Inflow Derivation

No storage inflow derivation was required for this reach.

375 Department of Environment and Science

6.32.5.2 Streamflow routing

Streamflow routing was represented using piecewise-linear routing. Altogether, there are 21 routing links in this reach. These were calibrated by optimising the timing and shapes of simulated flows against observed flows at downstream gauges. The optimised routing relationships were checked and extrapolated manually. Reach evaporation is simulated assuming constant surface area.

Ten routing links are used on the Weir and Macintyre Rivers. These are well constrained by the long period of record at Mungindi (416001) and complementary records at the intermediate stations: Jericho (416205A), Mascot (416207A), and Kanowna (416048). These links are listed in Table 1.26 below.

Table 1.26: Reach 32 (416001 & 416028) routing on the Weir and Macintyre Rivers

x (inflow Number Area Evaporation Length Routing link bias) of (km2) station (km) divisions Boonanga to Kanowna 1of3 1 10 0.6 30 053004 mlake Boonanga to Kanowna 2of3 1 6 0.26 13 053004 mlake Boonanga to Kanowna 3of3 1 4 0.06 3 053004 mlake Jericho to Mascot 1 6 0.774 39 053004 mlake Kanowna to Mungindi 1of4 1 4 0.16 8 053004 mlake Kanowna to Mungindi 2of4 1 6 0.94 47 053004 mlake Kanowna to Mungindi 3of4 1 2 0.06 2 053004 mlake Kanowna to Mungindi 4of4 1 10 0.34 17 053004 mlake Talwood to Jericho 1of2 1 12 0.6 30 053004 mlake Talwood to Jericho 2of2 1 4 0 31 053004 mlake

Eleven routing links are used on the eastern streams, including the Boomi River, Croppa Creek, Whalan Creek, and Boomangera Creek. These streams do not contribute to the flows at Mungindi, and routing is poorly constrained by the fickle records at Neeworra (416028). Consequently, the routing links here have lower quality calibrations. Table 1.27 shows the 11 routing links occurring on these streams.

376 Hydrology Report No 416000.PR/6

Table 1.27: Reach 32 (416001 & 416028) routing on Boomi River, Croppa Creek, Whalan Creek, and Boomangerra Creek

x (inflow Number of Area Length Evaporation Routing link bias) divisions (km2) (km) station Boomangera Creek 1 12 0 49 053004 mlake Boomi River 1of5 1 12 0 40 053004 mlake Boomi River 2of5 1 10 0 35 053004 mlake Boomi River 3of5 1 10 0 17 053004 mlake Boomi River 4of5 1 10 0 16 053004 mlake Boomi River 5of5 1 18 0 24 053004 mlake Croppa Creek 1 10 0 27 053004 mlake Whalan Creek 1of4 1 10 0 52 053004 mlake Whalan Creek 2of4 1 10 0 11 053004 mlake Whalan Creek 3of4 1 10 0 98 053004 mlake Whalan Creek 4of4 1 14 0 30 053004 mlake

The calibrated piecewise-linear routing relationships for all the links listed above (Table 1.26 and Table 1.27) are show in Table 1.28 below.

377 Department of Environment and Science

Table 1.28: Reach 32 (416001 & 416028) piecewise-linear routing

Flow (ML) Lag (days) Boomangera Creek 0 5.20 50 4.20 120 3.50 270 2.60 620 2.00 1,400 1.80 3,300 3.20 999,999 3.20 Boomi River 1 of 5 0 2.50 1,000 3.70 2,000 5.80 4,500 4.60 10,000 4.10 20,000 3.70 44,000 3.00 999,999 3.00 Boomi River 2 of 5 0 2.20 950 0.40 2,200 0.30 5,000 1.80 11,500 3.30 26,500 4.20 60,800 3.80 999,999 3.80 Boomi River 3 of 5 0 1.30 1,500 0.50 3,400 1.40 7,800 3.50 17,900 3.80 41,100 4.50 94,500 3.70

378 Hydrology Report No 416000.PR/6

Flow (ML) Lag (days) 999,999 3.70 Boomi River 4 of 5 0 2.40 900 1.00 2000 0.70 4500 0.70 10,400 3.20 24,000 3.60 55,100 4.30 999,999 4.30 Boomi River 5 of 5 0 1.00 1,900 1.50 4,400 3.30 10,100 5.20 25,000 8.60 55,000 2.30 125,000 2.20 999,999 2.10 Boonanga to Kanowna 1 of 3 0 2.60 500 1.40 2,000 0.50 4,000 0.30 7,500 1.40 15,000 4.20 38,000 3.00 78,000 2.80 999,999 2.80 Boonanga to Kanowna 2 of 3 0 0.30 2000 0.90 5,000 1.10 10,000 0.50 20,000 1.90

379 Department of Environment and Science

Flow (ML) Lag (days) 30,000 2.40 40,000 2.40 60,000 2.00 999,999 1.80 Boonanga to Kanowna 3 of 3 0 0.65 2,000 0.55 4,000 0.85 8,000 1.20 12,000 0.80 20,000 0.40 30,000 0.20 45,000 0.10 999,999 0.10 Croppa Creek 0 3.60 1,800 3.90 4,100 4.20 9,500 4.50 21,700 3.70 50,000 4.20 115,000 3.60 999,999 3.60 Talwood to Jericho 1 of 2 0 0.10 2,500 0.70 5,000 1.30 10,000 2.40 20,000 5.20 30,000 1.80 40,000 0.30 88,000 0.20 999,999 0.20 Talwood to Jericho 2 of 2 0 1.20

380 Hydrology Report No 416000.PR/6

Flow (ML) Lag (days) 2,500 0.90 5,000 0.60 8,000 0.30 12,000 0.70 16,000 2.00 24,000 0.40 50,000 0.30 999,999 0.30 Jericho to Mascot 0 0.30 2,500 0.50 5,000 1.20 10,000 1.40 15,000 1.50 24,000 2.10 38,000 2.70 70,000 1.30 999,999 1.30 Kanowna to Mungindi 1 of 4 0 1.40 500 1.00 1,000 0.70 2,500 0.20 4,000 0.50 6,000 1.20 8,000 1.50 10,000 1.50 18,000 1.20 999,999 1.20 Kanowna to Mungindi 2 of 4 0 1.30 500 1.30 1,000 0.90 2,500 0.30 4,000 0.40

381 Department of Environment and Science

Flow (ML) Lag (days) 6,000 0.70 8,000 1.80 10,000 2.20 18,000 1.80 999,999 1.80 ,Kanowna to Mungindi 3 of 4 0 1.00 1,000 0.60 2,000 0.30 5,000 0.20 10,000 0.20 25,000 0.50 42,000 1.00 68,000 0.40 999,999 0.40 Kanowna to Mungindi 4 of 4 0 0.40 2,000 1.40 5,000 3.50 10,000 4.20 15,000 3.30 20,000 1.40 30,000 1.00 80,000 0.60 999,999 0.60 Whalan Creek 1 of 4 0 3.80 3,200 4.00 7,300 4.30 16,800 4.50 38,800 4.20 89,100 4.30 205,000 4.00 999,999 4.00 Whalan Creek 2 of 4

382 Hydrology Report No 416000.PR/6

Flow (ML) Lag (days) 0 3.80 1,200 4.00 2,800 4.30 6,400 4.80 14,800 4.30 34,000 3.80 78,200 3.70 999,999 3.70 Whalan Creek 3 of 4 0 3.20 900 4.10 2,100 3.90 4,900 3.60 11,200 3.80 26,000 4.80 59,500 3.80 999,999 3.80 Whalan Creek 4 of 4 0 7.00 5,500 6.40 12,700 5.70 29,200 4.20 67,100 3.00 154,000 5.40 355,000 3.50 999,999 3.50

383 Department of Environment and Science

6.32.5.3 Effluents

In addition to the floodplain breakouts identified above, there are several other locations in this reach where breakouts are known to occur.

The two-way breakout at Newinga is particularly important because it shares high-flows between the Weir River and Macintyre River, and affects the accessibility of water on those rivers below Newinga. This two-way system is modelled using a pair of splitter nodes to transfer water in both directions between the two rivers. The associated breakout relationships were calibrated over the period 07/11/1984 to 30/06/2014 using observed streamflow records at the downstream gauges on the Macintyre River at Kanowna (416048) and the Weir River at Jericho (416205A). The initial breakout is well-understood by the scheme operators, and the present calibration did not find any evidence to warrant modification of the initial breakout relationship used in the previous IQQM model. At higher flows, however, the relationship was revised slightly from the previous model, and this was supported by streamflow records from the new gauge at Jericho (416205A). Table 1.29: Reach 32 (416001 & 416028) Newinga two-way breakout

Upstream Flow (ML/d) Breakout (ML/d) Newinga Breakout I (Macintyre River to Weir River) 0 0 500 0 2,500 1,000 3,000 1,200 6,000 2,400 12,000 6,400 25,000 12,000 50,000 25,000 999,999 500,000 Newinga Breakout II (Weir River to Macintyre River) 0 0 450 0 650 20 4,000 1,600 9,000 3,800 18,000 11,200 30,000 15,800 44,000 20,000 80,000 34,000 999,999 400,000

384 Hydrology Report No 416000.PR/6

This reach model also captures one-way breakouts from the Macintyre River above Kanowna into Little Barwon Creek (the “Little Barwon Creek Effluent”), from the Macintyre River above its confluence with the Weir River into Boomangerra Creek (the “Boomangerra Creek Effluent”), from the Barwon River above Mungindi into the Little Weir River (the “Little Weir River Effluent”), and from Whalan Creek near Caloona Road into Boomi River (the “Unnamed Effluent on Whalan Creek”). These were calibrated using observed data as listed in the table below.

Table 1.30: Reach 32 (416001 & 416028) calibration of one-way effluents

Effluent Calibration Gauge Calibration Period Little Barwon Creek Effluent Macintyre River at Kanowna (416048) 07/11/1984 to 30/06/2014 Boomangerra Creek Effluent Barwon River at Mungindi (416001) 19/02/1987 to 30/06/2014 Little Weir River Effluent Barwon River at Mungindi (416001) 19/02/1987 to 30/06/2014 Unnamed Effluent on Whalan Whalan Creek at Neeworra (416028) 04/04/1968 to 23/11/1994 Creek

Calibrated effluent relationships for these breakouts are given in the table below.

385 Department of Environment and Science

Table 1.31: Reach 32 (416001 & 416028) one-way effluents

Upstream Flow (ML/d) Breakout (ML/d) Little Barwon Creek Effluent 0 0 1,620 0 2,950 480 4,000 1,150 6,100 2,000 8,100 3,300 10,900 5,400 12,200 6,000 25,000 18,000 44,000 36,000 999,999 760,000 Boomangera Creek Effluent 0 0 2,100 0 2,500 200 3,000 390 4,000 600 6,000 780 8,200 1,330 14,000 4,000 999,999 300,000 Little Weir River Effluent 0 0 4,800 0 6,000 300 8,000 680 10,000 1,200 20,000 3,800 40,000 4,600 60,000 5,400 999,999 80,000 Unnamed Effluent on Whalan Creek 0 0 500 0

386 Hydrology Report No 416000.PR/6

Upstream Flow (ML/d) Breakout (ML/d) 650 20 1,000 150 2,000 400 26,000 5,200 999,999 200,000

6.32.5.4 Residual losses

Residual losses, notionally representing in-stream losses and non-returning breakout losses, are represented at six locations. These are listed in the table below.

Table 1.32: Reach 32 (416001 & 416028) residual loss calibration

Residual Loss Description Calibration 040 Yarrilwanna Breakout Loss Losses on Yarrilwanna Creek Adopted directly from IQQM 186 Weir River Loss Losses on the Weir River Adopted directly from IQQM between Jericho and Mascot 097 Boomi–Kanowna Loss Losses on the Macintyre River Adopted directly from IQQM between Boomi and Kanowna 104 Kanowna–Weir River Loss Losses on the Macintyre River Adopted directly from IQQM between Kanowna and the Weir River confluence 110 Weir River–Mungindi Loss Losses on the Macintyre River Adopted directly from IQQM between the Weir River confluence and Mungindi 220 Boomi Losses D/S Whalan Losses on the Boomi River and Calibrated using observed flows Creek Whalan Creek upstream of from the Whalan Creek at Neeworra Neeworra (416028)

With the exception of “220 Boomi Losses D/S Whalan Creek”, the residual loss relationships were adopted directly from IQQM. The loss “220 Boomi Losses D/S Whalan Creek” was adjusted on the basis of observed flows at Neeworra (416028) after accounting for upstream routing, historical diversions and effluents. The adopted/calibrated loss relationships are given in the table below.

Table 1.33: Reach 32 (416001 & 416028) residual losses on Yarrilwanna Creek

Upstream Flow (ML/d) Residual loss (ML/d) Loss ratio (%) 0 0 40.0 1,000,000 400,000 40.0

387 Department of Environment and Science

Table 1.34: Reach 32 (416001 & 416028) residual losses between Jericho and Mascot

Upstream Flow (ML/d) Residual loss (ML/d) Loss ratio (%) 0 0 12.5 16 2 12.5 30 3 10.0 50 4 8.0 70 6 8.6 120 10 8.3 170 13 7.6 210 14 6.7 310 19 6.1 650 42 6.5 900 47 5.2 6,400 47 0.7 19,000 100 0.5 1037 5.26 x 1034 0.5

Table 1.35: Reach 32 (416001 & 416028) residual losses between Boomi and Kanowna

Upstream Flow (ML/d) Residual loss (ML/d) Loss ratio (%) 0 0 20.0 40 8 20.0 80 11 13.8 115 14 12.2 190 23 12.1 250 30 12.0 400 31 7.8 520 51 9.8 1,000 65 6.5 1,400 65 4.6 2,000 108 5.4 3,000 112 3.7 80,000 112 0.1 1037 1.40 x 1034 0.1

388 Hydrology Report No 416000.PR/6

Table 1.36: Reach 32 (416001 & 416028) residual losses between Kanowna and the Weir River confluence

Upstream Flow (ML/d) Residual loss (ML/d) Loss ratio (%) 0 0 27.5 40 11 27.5 80 14 17.5 115 17 14.8 190 28 14.7 250 37 14.8 400 38 9.5 520 63 12.1 1,000 82 8.2 1,400 82 5.9 2,000 134 6.7 3,000 139 4.6 80,000 139 0.2 1037 1.74 x 1034 0.2

Table 1.37: Reach 32 (416001 & 416028) residual losses between the Weir River confluence and Mungindi

Upstream Flow (ML/d) Residual loss (ML/d) Loss ratio (%) 0 0 100.0 7 7 100.0 62 7 11.3 100 13 13.0 245 13 5.3 413 43 10.4 1,400 43 3.1 2,350 200 8.5 80,000 200 0.3 1037 2.50 x 1034 0.3

389 Department of Environment and Science

Table 1.38: Reach 32 (416001 & 416028) residual losses upstream of Neeworra

Upstream Flow (ML/d) Residual loss (ML/d) Loss ratio (%) 0 0 100.0 25 25 100.0 65 46 70.8 200 140 70.0 1,000 280 28.0 5,000 1,200 24.0 10,000 3,500 35.0 20,000 12,500 62.5 45,000 28,000 62.2 100,000 56,000 56.0 1037 56,000 0.0

6.32.5.5 Observed inflow sequence

No residual inflow sequence could be calculated from the available streamflow data. One reason for this was the limited overlap between historical streamflow records, and another was the structural complexity of this reach.

Given this, the residual inflow sequences for this reach will be comprised entirely simulated runoff from calibrated Sacramento rainfall-runoff models. The calibration of those models is described in the following section.

6.32.5.6 Sacramento model calibration

Five Sacramento rainfall-runoff models were developed to estimate runoff from the subareas in this reach. That number was warranted by the large size and distributed nature of this reach, and to some degree supported by the available observed streamflow records. Each Sacramento model was fitted with rainfall data chosen for the subareas it represents. Each Sacramento model was allowed independent parameters in calibration, although these were manually validated afterwards.

390 Hydrology Report No 416000.PR/6

Table 1.39: Reach 32 (416001 & 416028) Sacramento models

Represented subareas Calibration Calibration Period

Gauge 1 M6, M7, M12, M14, M15 416001 19/02/1987 to 30/06/2014 2 M13, M16, M17, M18 416001 19/02/1987 to 30/06/2014 3 M19 upper 50% 416028 04/04/1968 to 23/11/1994 4 M19 lower 50% 416028 04/04/1968 to 23/11/1994 5 M20, M21 416001 02/12/1889 to 30/06/2014

6.32.5.6.1 Sacramento model for M6, M7, M12, M14 and M15 (Sac. 1 of 5)

A Sacramento rainfall-runoff model was developed to estimate runoff from subcatchments of the Macintyre River between Boonanga Bridge (416046) and Kanowna (416048), subcatchments of the Weir River between Talwood (416202A) and the Newinga Breakout, and subcatchments of the Boomi River between the Boomi Weir Offtake and Kanowna (416029). Collectively, these are the subareas M6, M7, M12, M14 and M15.

This model was calibrated using an optimisation tool to fit the observed flows at Mungindi (416001) over the period 19/02/1987 to 30/06/2014. The calibration method is detailed in Section 4.1.3.

The effective catchment rainfall was estimated using a weighted combination of the rainfall sequences developed in the previous section. The weighting factors were determined as part of the calibration process and are shown in Table 1.40.Table 6.4 The table shows the mean rainfall in each of the contributing sequences, and in the final sequence.

Table 1.40: Reach 32 (416001 & 416028) rainfall sequence details (Sac. 1 of 5)

Infilled Rainfall Sequence Mean Rainfall (mm/y) Contributing Factor 07/1889 to 06/2014 052004 562.5 0.24 041521 616.9 0.61 042027 564.2 0.14 042104 644.7 0.42 Catchment rainfall Sequence 861.0 -

The calibrated Sacramento parameter values were manually checked for anomalies. Table 1.41 shows the final Sacramento model parameter values.

391 Department of Environment and Science

Table 1.41: Reach 32 (416001 & 416028) Sacramento parameter values (Sac. 1 of 5)

Parameter Adopted value Adimp 0.0015 Lzfpm 2.8 Lzfsm 18 Lzpk 0.003 Lzsk 0.3 Lztwm 201 Pctim 0.0005 Pfree 0.014 Rexp 1.35 Sarva 0.00001 Side 0.0018 Ssout 0.0015 Uzfwm 48 Uzk 0.29 Uztwm 174 Zperc 570 uh0 0 uh1 0.15 uh2 0.85

6.32.5.6.2 Sacramento model for M13, M16, M17 and M18 (Sac. 2 of 5)

A Sacramento rainfall-runoff model was developed to estimate runoff from subcatchments of the Weir River below the Newinga Breakout, and subcatchments of the Macintyre River between Kanowna (416048) and Yarrowee (416051). These are the subareas M13, M16, M17 and M18.

This model was calibrated using an optimisation tool to fit the observed flows at Mungindi (416001) over the period 19/02/1987 to 30/06/2014. The calibration method is detailed in Section 4.1.3.

The effective catchment rainfall was estimated using a weighted combination of the rainfall sequences developed in the previous section. The weighting factors were determined as part of the calibration process and are shown in Table 1.42.Table 6.4 The table shows the mean rainfall in each of the contributing sequences, and in the final sequence.

392 Hydrology Report No 416000.PR/6

Table 1.42: Reach 32 (416001 & 416028) rainfall sequence details (Sac. 2 of 5)

Infilled Rainfall Sequence Mean Rainfall (mm/y) Contributing Factor 07/1889 to 06/2014 052004 562.5 0.57 042027 564.2 0.64 042116 538.2 0.35 Catchment rainfall Sequence 870.0 -

The calibrated Sacramento parameter values were manually checked for anomalies. Table 1.43 shows the final Sacramento model parameter values.

Table 1.43: Reach 32 (416001 & 416028) Sacramento parameter values (Sac. 2 of 5)

Parameter Adopted value Adimp 0.001 Lzfpm 24 Lzfsm 6.4 Lzpk 0.003 Lzsk 0.72 Lztwm 155 Pctim 0.0002 Pfree 0.05 Rexp 1.2 Sarva 0.0001 Side 0.0008 Ssout 0.0005 Uzfwm 101 Uzk 0.13 Uztwm 174 Zperc 570 uh0 0 uh1 0 uh2 0.15 uh3 0.85

393 Department of Environment and Science

6.32.5.6.3 Sacramento model for the upper 50% of M19 (Sac. 3 of 5)

A Sacramento rainfall-runoff model was developed to estimate runoff from the upper 50 per cent of subarea M19. This portion of M19 is intended to approximate the area of Whalan Creek upstream of the point where it breaks out into Boomi River near Caloona Road.

This model was calibrated using an optimisation tool to fit the observed flows at Neeworra (416028) over the period 19/02/1987 to 23/11/1994. The calibration method is detailed in Section 4.1.3.

The effective catchment rainfall was estimated using a weighted combination of the rainfall sequences developed in the previous section. The weighting factors were determined as part of the calibration process and are shown in Table 1.44.Table 6.4 The table shows the mean rainfall in each of the contributing sequences, and in the final sequence.

Table 1.44: Reach 32 (416001 & 416028) rainfall sequence details (Sac. 3 of 5)

Infilled Rainfall Sequence Mean Rainfall (mm/y) Contributing Factor 07/1889 to 06/2014 053076 587.0 1.66 053041 584.2 0.23 Catchment rainfall Sequence 1,108.8 -

The calibrated Sacramento parameter values were manually checked for anomalies. Table 1.45 shows the final Sacramento model parameter values.

394 Hydrology Report No 416000.PR/6

Table 1.45: Reach 32 (416001 & 416028) Sacramento parameter values (Sac. 3 of 5)

Parameter Adopted value Adimp 0.00024 Lzfpm 13.7 Lzfsm 141 Lzpk 0.0025 Lzsk 0.074 Lztwm 370 Pctim 0.00175 Pfree 0.094 Rexp 1.77 Sarva 2.15 x 10-6 Side 0.00046 Ssout 0.0007 Uzfwm 17.6 Uzk 0.187 Uztwm 135 Zperc 379 uh0 0 uh1 0 uh2 0.15 uh3 0.85

6.32.5.6.4 Sacramento model for the lower 50% of M19 (Sac. 4 of 5)

A Sacramento rainfall-runoff model was developed to estimate runoff from the lower 50 percent of subarea M19. This portion of M19 is intended to approximately exclude the areas of Whalan Creek upstream of the point where it breaks out into Boomi River near Caloona Road.

This model was calibrated using an optimisation tool to fit the observed flows at Neeworra (416028) over the period 19/02/1987 to 23/11/1994. The calibration method is detailed in Section 4.1.3.

The effective catchment rainfall was estimated using a weighted combination of the rainfall sequences developed in the previous section. The weighting factors were determined as part of the calibration process and are shown in Table 1.46.Table 6.4 The table shows the mean rainfall in each of the contributing sequences, and in the final sequence.

395 Department of Environment and Science

Table 1.46: Reach 32 (416001 & 416028) rainfall sequence details (Sac. 4 of 5)

Infilled Rainfall Sequence Mean Rainfall (mm/y) Contributing Factor 07/1889 to 06/2014 052004 562.5 0.53 052020 506.9 0.36 053042 536.7 0.87 Catchment rainfall Sequence 947.5 -

The calibrated Sacramento parameter values were manually checked for anomalies. Table 1.47 shows the final Sacramento model parameter values.

Table 1.47: Reach 32 (416001 & 416028) Sacramento parameter values (Sac. 4 of 5)

Parameter Adopted value Adimp 0.00054 Lzfpm 4.7 Lzfsm 246 Lzpk 0.0012 Lzsk 0.208 Lztwm 240 Pctim 0.00035 Pfree 0.0189 Rexp 2.57 Sarva 2.21 x 10-6 Side 0.00033 Ssout 0.0006 Uzfwm 16.4 Uzk 0.254 Uztwm 92 Zperc 106 uh0 0.1 uh1 0.9

6.32.5.6.5 Sacramento model for M20 and M21 (Sac. 5 of 5)

A Sacramento rainfall-runoff model was developed to estimate runoff from subcatchments of the Macintyre River between Boonanga Bridge (416046) and Kanowna (416048), subcatchments of

396 Hydrology Report No 416000.PR/6

the Weir River between Talwood (416202A) and the Newinga Breakout, and subcatchments of the Boomi River between the Boomi Weir Offtake and Kanowna (416029). Collectively, these are the subareas M6, M7, M12, M14 and M15.

This model was calibrated using an optimisation tool to fit the observed flows at Mungindi (416001) over the period 02/12/1889 to 30/06/2014. The calibration method is detailed in Section 4.1.3.

The effective catchment rainfall was estimated using a weighted combination of the rainfall sequences developed in the previous section. The weighting factors were determined as part of the calibration process and are shown in Table 1.48.Table 6.4 The table shows the mean rainfall in each of the contributing sequences, and in the final sequence.

Table 1.48: Reach 32 (416001 & 416028) rainfall sequence details (Sac. 5 of 5)

Infilled Rainfall Mean Annual Rainfall (mm/y) Contributing Factor Sequence 07/1889 to 06/2014 052004 562.5 0.88 052020 506.9 0.20 042027 564.2 0.46 Catchment rainfall Sequence 855.9 -

The calibrated Sacramento parameter values were manually checked for anomalies. Table 1.49 shows the final Sacramento model parameter values.

397 Department of Environment and Science

Table 1.49: Reach 32 (416001 & 416028) Sacramento parameter values (Sac. 5 of 5)

Parameter Adopted value Adimp 0.00023 Lzfpm 16 Lzfsm 157 Lzpk 0.067 Lzsk 0.071 Lztwm 10.4 Pctim 0.0011 Pfree 0.02 Rexp 2.5 Sarva 0.0002 Side 0.000017 Ssout 0.002 Uzfwm 21 Uzk 0.28 Uztwm 12.6 Zperc 567 uh0 0 uh1 0 uh2 0.15 uh3 0.85

6.32.6 Reach Calibration Results and Recommendations

6.32.6.1 Sacramento calibration results

Figure 1.15 compares the modelled flows and observed flows at Mungindi (416001). Here the model assumes gauged inflows at Talwood (416202A), and gauged inflows at Boonanga (416046) which have been extended by the upstream reach model. The model shows reasonable performance in all regimes at Mungindi (416001).

Figure 1.16 compares the modelled flows and observed flows at Neeworra (416028). Here the model assumes gauged inflows at Croppa Creek Tulloona Bore (416034), and gauged inflows at Talwood (416202A), as well as modelled inflows generated by upstream reach models for the Macintyre River at Boonanga (416046), the Boomi Creek inflow and Whalan Creek inflow. The model shows acceptable performance in all regimes at Neeworra (416028).

398 Hydrology Report No 416000.PR/6

Figure 1.15: Reach 32 (416001 & 416028) calibration report card for subareas draining to Mungindi (416001)

399 Department of Environment and Science

Figure 1.16: Reach 32 (416001 & 416028) calibration report card for subareas draining to Neeworra (416028)

400 Hydrology Report No 416000.PR/6

6.32.7 Full length inflow sequence

The five Sacramento models described above were run for the period 01/07/1889 to 30/06/2014 to generate full-length inflow sequences for their corresponding subareas. Observed residual inflow sequences were not derived for any areas this reach and therefore the final inflow sequences are comprised completely of Sacramento modelled flow, as shown in Figure 1.17.

Figure 1.17: Reach 32 (416001 & 416028) inflow composition

401 Department of Environment and Science

7 Basin Model Development and Validation

7.1 Model Structure

A model of the combined reaches of the Border Rivers Basin was developed using the calibrated inflow sequences. The individual reaches were combined into one model.

Three locations were identified where the model could be significantly improved by adjusting the flows to match derived storage inflow or long-term streamflow records. These were:

 Pindari (416030) – Simplification of the model upstream of Pindari Dam on the Severn River allowed for improved model performance at the dam.  Wallangra (416010) – Simplification of the model upstream of Wallangra on the Macintyre River to allow the long-term streamflow record at the gauge to be used.  Talwood (416202A) – Making adjustments to Sacramento inflows when there was long- term flow data further downstream that was not used in inflow derivation, so allowing the final calibration model to better align at the downstream gauges. The final validation model is shown in

Figure 7.5. This was used to validate the flow calibration, and its ability to reproduce recorded flow behaviour in the system.

7.2 Model Simplification at Pindari (416030)

The reaches upstream of Pindari Dam (416030) were not required for the intended purposes of the final model, and were therefore removed. This simplification involved removing Reaches 18, 19 and 20, and replacing them with a single inflow node representing the inflow to Pindari Dam. As well as simplifying the model, this change allowed the derived storage inflows (Pindari SID

402 Hydrology Report No 416000.PR/6

inflows) to be used directly where available. That is, the new inflow to Pindari is comprised of derived storage inflows as available (blue in Figure 7.1) and the previous modelled flow at that location (orange in Figure 7.1). Although Reaches 18, 19 and 20 no longer feature in the model, their calibrations (Sections 6.18, 6.19 and 6.20) are still implicit in the new inflows to Pindari.

Figure 7.1: Final inflow at Pindari

7.3 Model Simplification at Wallangra (416010)

The reaches upstream of Wallangra (416010) were not required for the intended purposes of the final model, and were therefore replaced with a single inflow node. This simplification allowed the significant histroic streamflow record at the Wallangra gauge (416010; 1936–now) to be used as available to directly estimate the catchment inflows at that location. The new inflow at Wallangra is comprised of observed streamflow data (blue in Figure 7.2) and the previous modelled flow at that locaiton (organge in Figure 7.2). Although Reach 23 and 24 are no longer featured in the model, their calibrations (Sections 6.23 and 6.24) are still implicit in the new inflows at Wallangra.

Figure 7.2: Final inflow at Wallangra

7.4 Flow Adjustment at Talwood (416202A)

Long-term streamflow records are available at Talwood on the Weir River (416202A). This record was used in the calibration of Reach 30, however, to take further advantage of this substantial record the model inflows have been adjusted to reproduce the 416202A data exactly. This was done by using an iterative method that compares the modelled flows to the observed flows and then applies appropriate corrections to the model inflows. This is described in more detail in Appendix D.

403 Department of Environment and Science

For the purposes of the flow adjustment, a flow model was developed that combined Reaches 29 and 30 (refer to the structure upstream of 416202A in

Figure 7.5). Using the combined reach model, the inflow sequences were adjusted interatively by adding or subtracting daily volumes until the modelled reproduced the observed flows at 416202A. The required adjustments were distributed amongst the upstream inflows in proportionally. The periods of adjustment were based on the available record at Talwood (416202A) and were the same for all upstream inflows (see Table 7.1). The final inflow sequences are shown in Figure 7.3 and Figure 7.4 (Reaches 29 and 30, respectively).

Table 7.1: Reach 29 and 30 inflow adjustment periods

Period Talwood (416202a) gauge 01/05/1949 to 30/06/2014 open Missing record periods 01/10/1954 to 02/11/1954 01/09/1955 to 05/06/1968 Reach 29 inflow adjustment 03/05/1949 to 30/09/1954 periods 03/11/1954 to 31/08/1955 06/06/1968 to 30/06/2014 Reach 30 inflow adjustment 03/05/1949 to 30/09/1954 periods 03/11/1954 to 31/08/1955 06/06/1968 to 30/06/2014

404 Hydrology Report No 416000.PR/6

Figure 7.3: Reach 29 (416204a) inflow composition

Figure 7.4: Reach 30 (416202a) inflow composition

405 Department of Environment and Science

Figure 7.5: Border Rivers Basin model schematic

406 Hydrology Report No 416000.PR/6

7.5 Validation Modelling

7.5.1 Inflows

7.5.1.1 Comparison of Sacramento and final model inflows

The final inflow sequences are summarised in Table 7.2. These are comprised of observed inflows from a headwater gauge or inferred from the mass balance calculation over a reach, infilled and extended with Sacramento modelled inflows. The Weir River inflows are summarised separately in Table 7.3 – these have recieved individual attention to ensure that the flow adjustments at Talwood have not had unwanted impacts on the spatial distribution of the upstream inflows (refer to Section 7.4).

To give confidence that the model will perform acceptably where the Sacramento data is used, these tables compare the final model inflows with the corresponding full-length Sacramento modelled inflow sequences. There should be little or no difference between the general properties of the two sets of inflows, and indeed there are not. The tables show the catchment area, mean annual rainfall and flow statistics for each inflow. The mean inflows are compared graphically in Figure 7.6, Figure 7.7, Figure 7.8 and Figure 7.9.

The runoff coefficient for each catchment (proportion of runoff to rainfall) provides another good test of the model inflows. The runoff coefficients for all subareas (except M20 and M21) range from 2.3 percent to 9.7 percent and this is a suitable range for catchments in the North of the Murray– Darling Basin. The coefficients for subareas M20 and M21 are improbably high at 46 percent. This high runoff may be compensating for overestimated breakouts or underestimated flood returns below Kanowna and Mascot, and upstream of Mungindi. The movement of floodwater through the network of floodrunners and floodplains in the south-west of the catchment is not well-understood, and is an area of the model that could be improved given more data. The median runoff coefficient amongst headwater reaches (6.4 percent) is higher than that of residual reaches (4.9 percent), and this is an anticipated reflection of the relative efficiency of headwater catchments, which tend to be wetter and have more plant cover.

While Table 7.2 and Table 7.3 show the relative contribution of each inflows to others within the same table, the relative contribution of each inflow to the total model inflow is shown in Table 7.4.

Figure 7.10 shows the composition of each of the final inflow sequences.

407 Department of Environment and Science

Table 7.2: Comparison of Sacramento and final inflows (07/1889 – 06/2014)

Sacramento Inflows Final Inflows13 Final Mean/ % Catchment MARF Reach-Inflow Max Mean Rel Max Mean Sac Data 2 % RO Rel % RO Area (km ) (mm/y) Flow Flow Cont. Flow Flow Coeff Cont.14 Coeff Mean Adj. (ML/d) (ML/y) 14 (ML/d) (ML/y) (%) 1 550 978 65,859 51,086 2.85% 9.50 81,550 51,087 2.86% 9.50 100 65.1 2 531 963 30,696 44,062 2.46% 8.62 28,600 44,028 2.47% 8.61 99.9 37.5 3 1,051 965 65,758 78,451 4.38% 7.73 53,679 78,444 4.39% 7.73 100.0 32.4 4 1,317 890 124,480 75,363 4.20% 6.43 87,049 75,550 4.23% 6.44 100.2 39.8 5 1,296 990 107,600 95,582 5.33% 7.45 90,096 95,579 5.35% 7.45 100 28.1 6 890 926 64,922 67,200 3.75% 8.15 64,922 67,153 3.76% 8.15 99.9 49.0 7 855 1,105 96,557 62,732 3.50% 6.64 96,557 62,732 3.51% 6.64 100 38.8 8 760 936 71,514 68,700 3.83% 9.65 66,905 68,700 3.85% 9.65 100 31.2 9 537 718 58,064 26,263 1.47% 6.81 58,064 26,313 1.47% 6.82 100.2 28.5 10 676 776 79,226 26,239 1.46% 5.00 118,319 26,286 1.47% 5.01 100.2 36.0 11 550 838 21,259 15,749 0.88% 3.42 21,259 15,749 0.88% 3.42 100 0 12 1,663 750 207,881 49,318 2.75% 3.95 207,881 49,505 2.77% 3.97 100.4 12.0 13 681 819 96,506 23,289 1.30% 4.17 96,506 23,308 1.31% 4.18 100.1 12.9 14 395 768 34,347 15,791 0.88% 5.21 33,954 15,790 0.88% 5.21 100 29.5 15 323 561 24,371 8,963 0.50% 4.94 20,403 9,692 0.54% 5.34 108.1 17.4 16 1,087 1,100 44,360 57,241 3.19% 4.79 44,360 57,241 3.21% 4.79 100 19.1 17 314 626 10,575 6,589 0.37% 3.35 7,535 6,589 0.37% 3.35 100 7.7

13 Final inflows in this case are gauged flow or observed residual inflow infilled with modelled flows (with the exception of reaches 11, 28 and 32 for which the inflows are comprised completely of Sacramento flow) 14 Relative contribution (%) to total inflow shown in this table – that is, total model inflow excluding Weir River component

408 Hydrology Report No 416000.PR/6

Sacramento Inflows Final Inflows13 Final Mean/ % Catchment MARF Reach-Inflow Max Mean Rel Max Mean Sac Data 2 % RO Rel % RO Area (km ) (mm/y) Flow Flow Cont. Flow Flow Coeff Cont.14 Coeff Mean Adj. (ML/d) (ML/y) 14 (ML/d) (ML/y) (%) 18 19 20 15 2,079 977 121,348 179,493 10.01% 8.83 149,532 177,595 9.95% 8.74 98.9 34.8 21 822 883 51,143 59,497 3.32% 8.20 51,143 58,992 3.31% 8.13 99.2 25.6 22 401 693 15,654 16,790 0.94% 6.04 13,621 16,790 0.94% 6.04 100 14.1 23 24 16 2,139 990 167,662 137,909 7.69% 6.51 120,018 131,802 7.38% 6.22 95.6 58.3 25 1,535 921 202,541 121,268 6.77% 8.58 202,541 121,491 6.81% 8.59 100.2 36.8 26 384 700 34,530 11,338 0.63% 4.22 17,592 10,907 0.61% 4.06 96.2 37.5 27 1,909 969 166,868 94,318 5.26% 5.10 166,868 94,318 5.28% 5.10 100 7.1 28 – D16 M1 95 664 338 2,950 0.16% 4.68 338 2,950 0.17% 4.68 100 0 28 – D15 M2 M3 549 664 1,948 17,023 0.95% 4.67 1,948 17,023 0.95% 4.67 100 0 28 – M4 M5 M8 444 776 74,187 24,805 1.38% 7.20 74,187 24,805 1.39% 7.20 100 0 28 – M9 40 776 6,709 2,243 0.13% 7.23 6,709 2,243 0.13% 7.23 100 0 28 – M11 13 776 2,134 713 0.04% 7.08 2,134 713 0.04% 7.08 100 0 28 – M10 165 776 27,504 9,196 0.51% 7.19 27,504 9,196 0.52% 7.19 100 0 29 30 Inflows for Reach 29 and 30 are shown in Table 7.3 31 1,192 686 114,175 26,209 1.46% 3.21 114,175 26,209 1.47% 3.21 100 4.7 32 – M6 444 861 36,449 8,896 0.50% 2.33 36,449 8,896 0.50% 2.33 100 0 32 – M7 109 861 8,940 2,182 0.12% 2.32 8,940 2,182 0.12% 2.32 100 0 32 – M12 149 861 12,217 2,982 0.17% 2.32 12,217 2,982 0.17% 2.32 100 0

15 Sacramento inflow for this reach is actually end-of-reach modelled flow with implicit routing and losses. See Section 7.2 for description 16 Sacramento inflow for this reach is actually end-of-reach modelled flow with implicit routing and losses. See Section 7.3 for description.

409 Department of Environment and Science

Sacramento Inflows Final Inflows13 Final Mean/ % Catchment MARF Reach-Inflow Max Mean Rel Max Mean Sac Data 2 % RO Rel % RO Area (km ) (mm/y) Flow Flow Cont. Flow Flow Coeff Cont.14 Coeff Mean Adj. (ML/d) (ML/y) 14 (ML/d) (ML/y) (%) 32 – M13 60 870 4,576 2,123 0.12% 4.07 4,576 2,123 0.12% 4.07 100 0 32 – M14 M15 525 861 43,119 10,524 0.59% 2.33 43,119 10,524 0.59% 2.33 100 0 32 – M16 1,204 870 91,857 42,610 2.38% 4.07 91,857 42,610 2.39% 4.07 100 0 32 – M17 M18 118 870 9,041 4,194 0.23% 4.08 9,041 4,194 0.23% 4.08 100 0 32 – M19a 1,558 1,109 174,777 93,176 5.20% 5.39 174,777 93,176 5.22% 5.39 100 0 32 – M19b 1,558 948 298,632 83,475 4.66% 5.65 298,632 83,475 4.68% 5.65 100 0 32 – M20 M21 167 856 32,151 65,972 3.68% 46.16 32,151 65,972 3.70% 46.16 100 0

410 Hydrology Report No 416000.PR/6

Figure 7.6: Comparison of Sacramento and final inflows excluding the Weir River system (page 1)

411 Department of Environment and Science

Figure 7.7: Comparison of Sacramento and final inflows excluding the Weir River system (page 2)

412 Hydrology Report No 416000.PR/6

Table 7.3: Comparison of Sacramento and final inflows in the Weir River (07/1889 - 06/2014)

Sacramento Inflows Final Inflows17 Final Mean/ % Reach- Catchment MARF Max Mean Rel Max Mean Sac Data 2 % RO Rel % RO Inflow Area (km ) (mm/y) Flow Flow Cont. Flow Flow Coeff Cont.18 Coeff Mean Adj. (ML/d) (ML/y) 18 (ML/d) (ML/y) (%) 29 – UW1 827 849 108,925 29,577 5.59% 4.21 131,952 25,492 5.33% 3.63 86.19 28.28 29 – UW2 1,287 849 169,645 46,064 8.70% 4.21 205,508 39,703 8.30% 3.63 86.19 28.36 29 – UW3 30 849 3,988 1,083 0.20% 4.25 4,831 933 0.20% 3.66 86.19 25.42 29 – UW4 1,331 849 175,482 47,649 9.00% 4.22 212,578 41,069 8.59% 3.63 86.19 28.36 29 – UW5 849 849 111,892 30,382 5.74% 4.21 135,546 26,187 5.48% 3.63 86.19 28.29 29 – UW6 100 849 13,199 3,584 0.68% 4.22 15,990 3,089 0.65% 3.64 86.19 27.66 30 – UW7 91 778 7,567 4,478 0.85% 6.33 7,567 4,123 0.86% 5.82 92.08 21.14 30 – UW8 1,202 778 100,226 59,309 11.20% 6.34 100,226 54,610 11.42% 5.84 92.08 22.47 30 – UW9a 550 778 45,854 27,134 5.12% 6.34 45,854 24,984 5.22% 5.84 92.08 22.28 30 – UW9b 79 778 6,564 3,884 0.73% 6.32 6,564 3,577 0.75% 5.82 92.08 20.97 30 – LW1 2,200 778 183,355 108,502 20.49% 6.34 183,355 99,905 20.89% 5.84 92.08 22.56 30 – LW2 786 778 65,532 38,779 7.32% 6.34 65,532 35,706 7.47% 5.84 92.08 22.37 30 – LW3 1,907 778 159,155 94,181 17.79% 6.35 159,155 86,719 18.13% 5.85 92.08 22.56 30 – LW4 LW5 941 778 58,931 34,873 6.59% 4.76 58,931 32,110 6.71% 4.39 92.08 22.35

17 Final inflows in this case are Sacramento flows with periods adjusted to reproduce the historical flows at Talwood (416202A), as described in Section 7.4. 18 Relative contribution (%) to total inflow shown in table – that is, total Weir River inflow

413 Department of Environment and Science

Figure 7.8: Comparison of Sacramento and final inflows in the Weir River system (page 1)

414 Hydrology Report No 416000.PR/6

Figure 7.9: Comparison of Sacramento and final inflows in the Weir River system (page 2)

415 Department of Environment and Science

Table 7.4: Relative contribution of Sacramento and final reach model inflows to total model inflow (07/1889 – 06/2014

Sacramento Inflow – Relative Final Inflow – Relative Reach-Inflow Contribution Contribution

1 2.20% 2.25% 2 1.89% 1.94% 3 3.37% 3.46% 4 3.24% 3.33% 5 4.11% 4.22% 6 2.89% 2.96% 7 2.70% 2.77% 8 2.95% 3.03% 9 1.13% 1.16% 10 1.13% 1.16% 11 0.68% 0.70% 12 2.12% 2.18% 13 1.00% 1.03% 14 0.68% 0.70% 15 0.39% 0.43% 16 2.46% 2.53% 17 0.28% 0.29% 18 19 20 7.72% 7.84% 21 2.56% 2.60% 22 0.72% 0.74% 23 24 5.93% 5.82% 25 5.21% 5.36% 26 0.49% 0.48% 27 4.06% 4.16% 28 – D16 M1 0.13% 0.13% 28 – D15 M2 M3 0.73% 0.75% 28 – M4 M5 M8 1.07% 1.09% 28 – M9 0.10% 0.10% 28 – M11 0.03% 0.03% 28 – M10 0.40% 0.41% 29 – UW1 1.28% 1.13% 29 – UW2 1.99% 1.76%

416 Hydrology Report No 416000.PR/6

Sacramento Inflow – Relative Final Inflow – Relative Reach-Inflow Contribution Contribution

29 – UW3 0.05% 0.04% 29 – UW4 2.06% 1.82% 29 – UW5 1.31% 1.16% 29 – UW6 0.15% 0.14% 30 – UW7 0.19% 0.18% 30 – UW8 2.57% 2.43% 30 – UW9a 1.18% 1.11% 30 – UW9b 0.17% 0.16% 30 – LW1 4.70% 4.44% 30 – LW2 1.68% 1.59% 30 – LW3 4.08% 3.85% 30 – LW4 LW5 1.51% 1.43% 31 1.13% 1.16% 32 – M6 0.38% 0.39% 32 – M7 0.09% 0.10% 32 – M12 0.13% 0.13% 32 – M13 0.09% 0.09% 32 – M14 M15 0.45% 0.46% 32 – M16 1.83% 1.88% 32 – M17 M18 0.18% 0.19% 32 – M19a 4.01% 4.11% 32 – M19b 3.59% 3.68% 32 – M20 M21 2.84% 2.91%

417 Department of Environment and Science

~ The inflows for these reaches were simplified to one inflow (as described in Section 7.2) # The inflows for these reaches were simplified to one inflow (as described in Section 7.3) * These reaches have multiple inflows with the same composition Figure 7.10: Summary of final inflow composition

7.5.1.2 Comparison of model inflows to 50-year Isohyet Rainfall Maps

The final model inflows may be compared to the 50-Year Isohyet Maps by calculating the runoff coefficients over the 01/01/1920 to 31/12/1969 period relative to the mean rainfall as estimated by the 50-Year Isohyet Maps. Refer to Table 7.5. The runoff coefficients calculated here tend to be slightly higher than those calculated in the previous section, as the effective catchment rainfall sequences adopted tended to be wetter than would be inferred from the 50-Year Isohyet Rainfall Maps. Nevertheless the runoff coefficients calculated here are broadly reasonable. The subareas M20 and M21 are alone in having improbably runoff coefficients (77 percent by this estimate). Those were addressed in the previous section. All other runoff coefficients are between 2.4 and 13.4 percent.

418 Hydrology Report No 416000.PR/6

Table 7.5: Validation model runoff coefficients over the 50-Year Isohyet period (01/1920 to 12/1969)

Reach # Reach Subregions CA Mean Mean MARF19 % RO Name Represented (km2) Flow Flow (mm/y) Coeff by Inflow (ML/y) (mm/y) 1 416003 Tenterfield Ck. T1, T2 550 61,028 110.9 893 12.4 at Clifton 2 416023 Deepwater R. T4 531 44,223 83.4 909 9.2 at Bolivia 3 416032 T5, T6, T7, T8, Mole R. at 1,051 75,646 72.0 862 8.3 T9, T10, T11 Donaldson 4 416315A Pike Ck. at G1, G2, G3, G4, 1,317 77,243 58.7 710 8.3 G5, G6 Headwater 5 416310A G7, G8, G9, Dumaresq R. at G10, G11, G12, 1,296 102,843 79.4 775 10.2 Farnbro G13 6 416008 Beardy R. at B1, B2 890 66,033 74.2 774 9.6 Haystack 7 416007 Dumaresq R. at B3, B7 855 52,929 61.9 708 8.7 Bonshaw 8 416011 T3, T12, T13, Dumaresq R. at G14, G15, G16, 760 75,744 99.7 744 13.4 Roseneath B4, B5, B6 9 416410A Macintyre R. at MB1, MB2 537 25,017 46.6 673 6.9 Barongarook 10 416404BC Bracker Ck. at MB5, MB6 676 30,000 44.4 663 6.7 Terraine 11 416409AB Coolmunda MB3, MB4, MB7 550 16,143 29.4 630 4.7 Dam Headwater 12 416402BC MB8, MB9, Macintyre MB10, MB11, 1,663 42,870 25.8 607 4.2 Brook at MB12, MB13 Inglewood

19 Isohyet rainfall – 1920 to 1969

419 Department of Environment and Science

Reach # Reach Subregions CA Mean Mean MARF19 % RO Name Represented (km2) Flow Flow (mm/y) Coeff by Inflow (ML/y) (mm/y) 13 416415A MB14, MB15, Macintyre Brk. MB16, MB17, 681 24,602 36.1 614 5.9 at Booba MB18 Sands 14 416312A Oaky Ck. at D2 395 15,658 39.6 653 6.1 Texas 15 416305B Brush Ck. at D4 323 11,221 34.8 622 5.6 Beebo 16 416040 Dumaresq D1, D3, D5, D6, 1,087 53,314 49.1 640 7.7 River at D7, D9, D10 Glenarbon 17 416036 Campbells D8 314 6,117 19.5 658 3.0 Creek at Near Beebo 18, 19, 20 416022 Severn R. at Fladbury 416039 Severn River at P1, P2, P3, P4 2,079 172,054 82.8 838 9.9 Strathbogie 416030 Pindari Dam Headwater Gauge 21 416021 Frazers Ck. at P7, P8 822 54,142 65.9 780 8.4 Westholme 22 416006 Severn R. at P5, P6, P9 401 15,682 39.1 739 5.3 Ashford 23, 24 416016 P10, P11, P12, Macintyre R. at P13, P14, P15, Inverell P16, P17, P18, 2,139 115,023 53.8 791 6.8 416010 P19, P20, P21, Macintyre R. at P22, P23, P24 Wallangra 25 416012 Macintyre R. at S1, S2, S3, S4 1,535 109,741 71.5 701 10.2 Holdfast

420 Hydrology Report No 416000.PR/6

Reach # Reach Subregions CA Mean Mean MARF19 % RO Name Represented (km2) Flow Flow (mm/y) Coeff by Inflow (ML/y) (mm/y) 26 416020 Ottleys Ck. at S5 384 8,741 22.8 738 3.1 Coolatai 27 416201A D11, D12, D13, Macintyre R. at D14, S6, MB19, 1,909 82,211 43.1 643 6.7 Goondiwindi MB20 28 416046 D15, M2, M3 549 13,590 24.7 604 4.1 Macintyre River D16, M1 95 2,355 24.7 534 4.6 or Brook? at Boonanga M4, M5, M8 444 18,598 41.9 528 7.9 Bridge M9 40 1,682 41.9 578 7.2 M10 165 6,895 41.9 525 8.0 M11 13 535 41.9 567 7.4 29 416204A UW1 827 21,834 26.4 632 4.2 Upper Weir at UW2 1,287 34,005 26.4 596 4.4 Gunn Bridge UW3 30 799 26.5 575 4.6 UW4 1,331 35,175 26.4 575 4.6 UW5 849 22,429 26.4 575 4.6 UW6 100 2,646 26.5 560 4.7 30 416202A UW7 91 3,519 38.8 525 7.4 Weir River at UW8 1,202 46,602 38.8 550 7.0 Talwood UW9a 550 21,321 38.8 534 7.3 UW9b 79 3,052 38.6 534 7.2 LW1 2,200 85,256 38.8 573 6.8 LW2 786 30,471 38.8 556 7.0 LW3 1,907 74,003 38.8 530 7.3 LW4, LW5 941 27,401 29.1 535 5.4 31 416034 Croppa Ck. at M24 1,192 19,372 16.2 667 2.4 Tulloona Bore 32 416001 & M6 444 6,392 14.4 529 2.7 416028 M7 109 1,568 14.4 540 2.7 Barwon R. at Mungindi and M12 149 2,143 14.4 559 2.6 Boomi R. at M13 60 1,578 26.3 533 4.9 Neeworra M14, M15 525 7,562 14.4 573 2.5 M16 1,204 31,664 26.3 504 5.2 M17, M18 118 3,116 26.3 501 5.2

421 Department of Environment and Science

Reach # Reach Subregions CA Mean Mean MARF19 % RO Name Represented (km2) Flow Flow (mm/y) Coeff by Inflow (ML/y) (mm/y) M19a 1,558 68,749 44.1 593 7.4 M19b 1,558 54,167 34.8 593 5.9 M20, M21 167 64,124 384.3 500 76.9

7.5.2 Comparisons to Gauge Records and Storage Behaviours

Table 7.2 compares the flows from the final validation model, the Sacramento validation model and the streamflow gauges in the catchment. The comparisons are made on a mean annual basis at key gauge locations in the catchment.

Very good performance is seen at most stations throughout the system: the final validation model flows are within 2 percent of the observed flows at 20 of the 28 streamflow gauging sites listed here, and the Sacramento validation model flows are within 5 percent of the observed flows at 20 of the 28 streamflow gauging sites listed here. There is perfect agreement between the recorded flows and final validation models at headwater catchments. This is trivial since headwater inflows were defined by gauge data, nevertheless confiming the result here adds confidence that the final validation model was constructed properly from the reach models.

The primary end-of-system is the Barwon River at Mungindi (416001) and at this site the validation model underpredicts the total observed flows overall by around 8 percent. This error is principally associated with misestimation of over-bank flows, and one can confirm the within-bank performance is very good: both the Sacramento validation model and final validation model are within 0.4 percent of observed flows < 8,000 ML/d over the full period of record (1889–2014).

The model underpredicts the total flow at Goondiwindi (416201A) by around 7 percent. This error is associated with both over-bank and within-bank flows.

The largest discreapancy in the fnal validation model is an 18 percent overestimation occuring at Macintyre River at Boonanga Bridge (416046). Although this proved to be a very valuable site for model calibration, with only two years of observed record a large variance in the measured performance may be expected. This site may be of interest as its record grows over the coming years. Error! Reference source not found. to Figure 7.38 show report cards for the streamflow gauge locations listed in the table. They compare of the validation model flows (with Sacramento inflows and final inflows) to the observed streamflow records over the full period of streamflow record. The behaviors

422 Hydrology Report No 416000.PR/6

of Glenlyon Dam, Coolmunda Dam and Pindari Dam are compared to historic recrods (Figure 7.39 to

Figure 7.41).

These results show the model and associated data is able to reproduce the historic behaviour of the system reasonably well. The fact that the Sacramento validation model performs well gives confidence in the accuracy of the Sacramento inflows which were used when observed inflows were unavaialble.

423 Department of Environment and Science

Table 7.2: Comparison between validation model flows and historical observed streamflows over years with complete records using Sacramento inflows and the final inflows

Streamflow Gauge Observed Flows Validation Model Validation Model Bias Modelled Flow (ML/y)

Number Name Period Length in Complete Observed Flow Using Final Using Using Using Water Years Water Years (ML/y) Inlows Sacramento Final Sacramento inlows Inlows inlows 416003 Tenterfield Creek at Clifton 07/1922 to 06/2014 92 82 51,376.660 51,376.660 51,385.050 1 1.000163 416023 Deepwater at Bolivia 07/1967 to 06/2014 47 46 42,807.090 42,807.090 42,534.460 1 0.993631 416032 Mole River at Donaldson 07/1970 to 06/2014 44 40 109,728.800 109,816.900 112,468.400 1.000803 1.024967 416309ab at Glenlyon Dam Tailwater 07/1961 to 06/2013 52 49 64,042.160 64,068.770 63,059.610 1.000416 0.984658 416310a Duamresq River at Farnbro 07/1963 to 06/2013 50 49 81,119.600 81,119.600 80,502.890 1 0.992398 416008 at Haystack 07/1935 to 06/2014 79 54 47,435.360 47,435.360 49,222.160 1 1.037668 416007 at Bonshaw 07/1935 to 06/2014 79 55 388,458.800 389,103.400 392,703.700 1.001659 1.010927 416011 Dumaresq River at Roseneath 07/1937 to 06/2014 77 70 349,473.300 349,545.600 353,472.700 1.000207 1.011444 416410ab Macintyre Brook at Barongarook 07/1967 to 06/2014 47 44 28,833.370 28,833.370 28,109.650 1 0.9749 416404bcd Bracker Creek at Terraine 07/1953 to 06/2014 61 52 32,003.570 32,003.570 32,249.140 1 1.007673 416402bc Macintyre Brook at Inglewood 07/1954 to 06/2014 60 56 105,153.800 122,068.100 123,884.300 1.160852 1.178125 416415a Macintyre Brook at Booba Sands 07/1987 to 06/2014 27 27 101,801.600 114,106.800 116,383.900 1.120875 1.143243 416312a Oaky Creek at Texas 07/1969 to 06/2013 44 44 15,892.120 15,890.870 15,985.590 0.999921 1.005882 416305b Brush Creek at Beebo 07/1969 to 06/2013 44 27 8,170.791 8,170.791 8,305.733 1 1.016515 416040 Dumaresq River at Glenarbon 07/1986 to 06/2014 28 27 349,944.200 348,356.300 347,951.000 0.995462 0.994304 416036 Campbells Creek near Beebo 07/1973 to 06/1995 22 19 6,993.935 6,993.935 7,041.469 1 1.006796 416021 Frazers Creek at Westholme 07/1967 to 06/2014 47 30 58,461.870 58,461.870 60,879.560 1 1.041355 416006 Severn River at Ashford 07/1934 to 06/2014 80 72 257,912.900 232,671.700 232,259.100 0.902133 0.900533 416010 Macintyre River at Wallangra 07/1937 to 06/2014 77 73 116,952.500 116,952.500 125,479.900 1 1.072914 416012 Macintyre River at Holdfast 07/1951 to 06/2014 63 51 397,544.300 394,174.200 407,718.400 0.991523 1.025592 416020 Ottleys Creek at Coolatai 07/1967 to 06/2014 47 40 13,763.140 13,763.140 14,300.680 1 1.039056 416201a Macintyre River at Goondiwindi 07/1925 to 06/2014 89 65 1,052,201.000 972,715.400 978,484.800 0.924458 0.929941 416046 Macintyre River at Boonanga Bridge 07/2012 to 06/2014 2 2 340,147.700 401,812.800 388,196.900 1.181289 1.14126 416204a Weir River at Gunn Bridge 07/2000 to 06/2014 14 14 86,746.700 97,117.580 86,759.600 1.119554 1.000149 416202a Weir River at Talwood 07/1949 to 06/2014 65 51 156,295.500 140,815.300 156,762.300 0.900956 1.002986 416034 Croppa Creek at Tulloona Bore 07/1972 to 06/1988 16 12 19,721.650 19,721.650 16,437.840 1 0.833492 416028 Boomi River at Neeworra 07/1968 to 06/1994 26 10 211,572.900 209,143.800 215,333.200 0.988519 1.017773 416001 Barwon River at Mungindi 07/1890 - 06/2014 124 120 597,497.700 546,514.600 551,241.800 0.914672 0.922584

424 Hydrology Report No 416000.PR/6

Figure 7.11: Validation model report card for Tenterfield Creek at Clifton (416003)

425 Department of Environment and Science

Figure 7.12: Validation model report card for Deepwater at Bolivia (416023)

426 Hydrology Report No 416000.PR/6

Figure 7.13: Validation model report card for Mole River at Donaldson (416032)

427 Department of Environment and Science

Figure 7.14: Validation model report card for Pike Creek at Glenlyon Dam Tailwater (416309ab)

428 Hydrology Report No 416000.PR/6

Figure 7.15: Validation model report card for Duamresq River at Farnbro (416310a)

429 Department of Environment and Science

Figure 7.16: Validation model report card for Beardy River at Haystack (416008)

430 Hydrology Report No 416000.PR/6

Figure 7.17: Validation model report card for Dumaresq River at Bonshaw (416007)

431 Department of Environment and Science

Figure 7.18: Validation model report card for Dumaresq River at Roseneath (416011)

432 Hydrology Report No 416000.PR/6

Figure 7.19: Validation model report card for Macintyre Brook at Barongarook (416410a)

433 Department of Environment and Science

Figure 7.20: Validation model report card for Bracker Creek at Terraine (416404bc)

434 Hydrology Report No 416000.PR/6

Figure 7.21: Validation model report card for Macintyre Brook at Inglewood (416402bc)

435 Department of Environment and Science

Figure 7.22: Validation model report card for Macintyre Brook at Booba Sands (416415a)

436 Hydrology Report No 416000.PR/6

Figure 7.23: Validation model report card for Oaky Creek at Texas (416312a)

437 Department of Environment and Science

Figure 7.24: Validation model report card for Brush Creek at Beebo (416305ab)

438 Hydrology Report No 416000.PR/6

Figure 7.25: Validation model report card for Dumaresq River at Glenarbon (416040) and Mauro (416049)

439 Department of Environment and Science

Figure 7.26: Validation model report card for Campbells Creek near Beebo (416036)

440 Hydrology Report No 416000.PR/6

Figure 7.27: Validation model report card for Frazers Creek at Westholme (416021)

441 Department of Environment and Science

Figure 7.28: Validation model report card for Severn River at Ashford (416006)

442 Hydrology Report No 416000.PR/6

Figure 7.29: Validation model report card for Macintyre River at Wallangra (416010)

443 Department of Environment and Science

Figure 7.30: Validation model report card for Macintyre River at Holdfast (416012)

444 Hydrology Report No 416000.PR/6

Figure 7.31: Validation model report card for Ottleys Creek at Coolatai (416020)

445 Department of Environment and Science

Figure 7.32: Validation model report card for Macintyre River at Goondiwindi (416201a)

446 Hydrology Report No 416000.PR/6

Figure 7.33: Validation model report card for Macintyre River at Boonanga Bridge (416046)

447 Department of Environment and Science

Figure 7.34: Validation model report card for Weir River at Gunn Bridge (416204a)

448 Hydrology Report No 416000.PR/6

Figure 7.35: Validation model report card for Weir River at Talwood (416202a)

449 Department of Environment and Science

Figure 7.36: Validation model report card for Croppa Creek at Tulloona Bore (416034)

450 Hydrology Report No 416000.PR/6

Figure 7.37: Validation model report card for Boomi River at Neeworra (416028)

451 Department of Environment and Science

Figure 7.38: Validation model report card for Barwon River at Mungindi (416001)

452 Hydrology Report No 416000.PR/6

Figure 7.39: Validation model storage behavior at Glenlyon Dam (416315a)

453 Department of Environment and Science

Figure 7.40: Validation model storage behavior at Coolmunda Dam (416409ab)

454 Hydrology Report No 416000.PR/6

Figure 7.41: Validation model storage behavior at Pindari Dam (416030)

455 Department of Environment and Science

7.6 Discussion - Recomendations

It should be noted that while in-bank flows are reasonably well defined, over-bank and high flows are not. This is partly due to the fact that extreme events are able to redefine the path of flow over flat landscapes. While breakouts have been built into the model, it is likely that the behaviour in extreme events will differ to that in the model.

456 Hydrology Report No 416000.PR/6

457 Department of Environment and Science

8 Quality Assurance

Quality assurance procedures were followed out for this calibration and are reported on in the QA report. This was divided into five sections: 1. Model Setup – this ensures that procedures are in place to document decisions made regarding the set-up of the model. This also includes the planning stage of the model work. 2. Data Review – this includes the collation and checking of basic data (streamflow, rainfall, evaporation, etc.), to identify data gaps and data quality issues. 3. IQQM Reach Model Calibration Review - this documents the calibrated reach models ability to reproduce the recorded downstream flows. 4. Rainfall-runoff Model Calibration Review – this documents the Sacramento model parameters and the performance of the Sacramento model in reproducing the recorded or residual inflows. 5. IQQM Validation Model Review – this considers the whole-of-model checks that are performed on the models developed for the full system at completion of the calibration. It considers the match at the calibration gauges. A star system (more stars are better) was used on report cards to indicate the quality of calibrations. The report cards for Sacramento calibrations and Validation model results along with their star ratings are shown in this report.

Ratings are shown for volume ratios for the whole flow range, as well as the low-, mid- and high- flow ranges. The low-, mid- and high-flow statistics indicate how well the models reproduce the observed flows in each range. The low-, mid- and high-flow ranges are defined by the flexion points on the daily flow-exceedance curves.

The performance of the Sacramento model calibrations varies as shown in the report cards. The mid- and high-flow ranges were better reproduced than the low-flow ranges.

The performance of the Validation models against the full period of record at each gauge returned higher ratings for the final flows model than the Sacramento validation as is be expected due to the use of recorded data in the final flows. Once again, the mid- to high-flow ranges were better reproduced.

There were no significant changes recommended as a result of the internal quality assessment review.

458 Hydrology Report No 416000.PR/6

459 Department of Environment and Science

9 Methodology and Data Differences 2002 to 2016

The new model and the 2002 model differ both in their account of the historic behaviour (as descriptive models) and in their response under modified scenarios (as predictive models).

Both models simulate the areas upstream of Mungindi (416001) and Neeworra (416028) using a network of nodes and links to represent hydrological processes, but the representation of most of those processes has been revised through the current recalibration. The new model was calibrated over a longer period, using contemporary methods, and has benefitted from the availability of additional streamflow/rainfall gauges and longer streamflow/rainfall records.

A result of the large number of changes is that it is not always easy to meaningfully disentangle the roots of particular variations between the new and old models. Given the dissimilarities in methods it is more appropriate to work through the methodology used in this model. If the methodology is acceptable and has been applied correctly then the resultant model should be acceptable.

Key differences in the underlying data and methodologies are outlined below.

9.1 Modelling Platform

The preivous model was built using the IQQM modelling platform (NSW DLWC, 1996) and the new model was developed using the Source modelling platform (eWater Ltd., 2016a & 2016b). Although this is a conspicuous difference, both platforms offer similar flow-modelling functionality and therefore the choice of platform has only minor consequences on the flow calibration.

Three minor differences are worth mentioning:  IQQM simulates flows on a sub-daily timestep (a 6-hour timestep) whereas Source simulates flows on a daily timestep. The consequences of this are minimal because the underlying data for both models is daily, and both models use a daily timestep to simulate operational phases including ordering.  IQQM simulates storages by applying various fluxes (inflows, releases, spills, pond diversions, rainfall, evaporation, seepage) sequentially, whereas Source integrates all fluxes simultaneously across the timestep. This impacts storage behaviour when fluxes are large, in particular during flood events. It is unlikely this has significant implications for the intended purposes of the models, however it may prevent affected parts of the calibration from being directly compared or ported between platforms.  Piecewise-linear routing has been adopted by many Source users and was adopted for the current work as it provides good performance and flexibility. This routing method is not available in most versions of IQQM, and the 2002 Border Rivers model used Muskingum routing instead.

While the choice of platform has only minor consequences on the flow calibration, it does have significant practical and technical implications for subsequent modelling tasks including development and calibration of irrigation demand models, implementation of management rules, and the development and use of scenario models. Those topics are beyond the scope of this document.

460 Hydrology Report No 416000.PR/6

9.2 Calibration Period

The 2002 model was calibrated over the period 1985–1996 and the present model was calibrated over the period 1889–2016 (both subject to the data availability within those periods). The longer calibration period is a potential advanatage this model has over the preivous model.

9.3 Rainfall and Evaporation

Different rainfall and evaporation data have been used.

In the current work, rainfall data was sourced from Bureau of Meteorology rainfall stations, and supplemented using SILO Patched Point data. Catchment rainfall was then estimated using a weighted combination of infilled station records, with weighting factors informed through the Sacramento rainfall-runoff calibration process. The method is discussed in this report.

For the previous work, long-term Bureau of Meteorology rainfall stations were selected for each of five climatic zones: Glenlyon Storage, Roseneath to Macintyre Brook, Macintyre Brook to Terrewah, Terrewah to Kanowna and Kanowna to Mungindi. The records from each of these was manually infilled using data from other stations as available. Details of the approach is documented in the Border Rivers System: IQQM Implementation. Vol 2 of 7: Calibration Report: Dumaresq– Macintyre Rivers Subsystem (1986–1991) (NSW DLWC and QLD DNR 1998). Catchment rainfall was estimated using a weighted combination of infilled station records where the weighting factors were determined from Thiessen polygons. The method is documented in Streamflow Derivation for the Border Rivers System (QLD DPI 1995).

Potential evaporatranspiration data for the current work was sourced from the SILO Patched Point dataset. Two Morton evapotranspiration estimates were used – MWET for catchment evaporation, and MLAKE for the evaporation from water surfaces.

In the previous IQQM model development, the required evapotranspiration data was derived from the evaporation stations with long term records that were held by Bureau of Meteorology. Evaporation was applied for two purposes: (1) In the generation of long-term streamflow evaporation stations were chosen to represent the evaporation for the various catchments. For this study the Wet and Dry evaporation method was used to derive the the long-term evaporation. The method is documented in Streamflow Derivation for the Border Rivers System (QLD DPI 1995); (2) In the IQQM model development evaporation records were developed using records from the best available evaporation stations for each of five climatic zones: Glenlyon Storage, Roseneath to Macintyre Brook, Macintyre Brook to Terrewah, Terrewah to Kanowna and Kanowna to Mungindi. Details of the approach are documented in the Border Rivers System: IQQM Implementation. Vol 2 of 7: Calibration Report: Dumaresq–Macintyre Rivers Subsystem (1986– 1991) (NSW DLWC and QLD DNR 1998).

There is little variation between the various evaporation datasets, and most model results are relatively insensitive to variations in evaporation, therefore this difference is unlikely to have a significant effect on the model.

9.4 Flow Data

Recorded flow data has varied in a number of ways:  Different and additional gauges used.  Longer records with flows associated with more extreme weather conditions.

461 Department of Environment and Science

 Rating changes. This will change earlier flow records if the rating curves change.  Data may have been extracted differently. Variations include use of different time offsets and different conversion calculations used to generate flow data from levels. Use of additional flow data and longer records allowed for additional catchment responses to be captured in calibration using the longer datasets.

9.5 Observed Residual Inflows

Observed residual inflows were derived by mass-balance for both the 2002 model and the current model. The periods for which observed residual inflows were calculated is different due to changes in the combinations of gauging stations used in each reach. There are also differences in how the residual inflows were used:

 In the current work the observed residual inflow records were used to calibrate Sacramento rainfall runoff models for the residual catchments, and they were also used directly to develop the inflow sequences.  In the previous IQQM model, the observed residual inflows were used to identify representative inflow(s) amongst the calibrated headwater catchment. The observed residual inflow data was not used to calibrate bespoke Sacramento models for the residual areas, nor to develop the residual inflow sequences. There is a description of the method in the Border Rivers System: IQQM Implementation. Vol 2 of 7: Calibration Report: Dumaresq–Macintyre Rivers Subsystem (1986–1991) (NSW DLWC and QLD DNR 1998).

9.6 Sacramento Calibrations

The new Sacramento models are different to the old ones. They use different catchment areas, rainfall, evaporation, and have been calibrated to different target flow data using a different calibration approach. Whereas the old Sacramento models were manually calibrated, the new ones have been calibrated using numerical global optimisation methods.

9.7 Historical Diversions

Historical water extraction records are available aggregated over intervals of three months or more. Trying to dissagregate these records to a daily sequence is difficult and likely to cause errors in the low flows. The current model uses the daily diversions records that were previously dissagregated for the 2002 model calibration – these end after 31/12/1996. Subsequent periods required dissagreation for the present work, and these may contain errors which may lead to inconsistencies between the new and previous model.

There is some unmetered use in the catchment however it was considered better to not include them in the estimation of inflows. This is unlikely to have had a large effect on the calibration.

9.8 Inflow Adjustment Process

In the current model calibration, the reproduction of historic flows on the Weir River has been improved by applying the flow adjustment process at Talwood (416202A). No flow adjustments were made in the calibration of the previous model.

462 Hydrology Report No 416000.PR/6

463 Department of Environment and Science

10 Conclusions

This report describes the calibration of a Source model for the Border Rivers river system.

Individual models of river reaches between stream gauging stations on the main river and its tributaries were set up. For each reach the following occurred. A reach model was set up and a flow routing model was calibrated using the available flow record. The reach model was then used to estimate inflows and losses. A Sacramento rainfall-runoff model was calibrated for the reach, which in turn was used to infill and extend the observed inflow sequences to cover the full model period 01/07/1889 to 30/06/2014.

This information was used to develop a validation model of the Border Rivers River Basin which was checked for quality of calibration over different periods for each reach. The quality of the data was judged to be satisfactory.

The models developed constitute a whole river system model and are considered adequate for use in Water Resource Planning studies and other water resource investigations.

464 Hydrology Report No 416000.PR/6

465 Department of Environment and Science

11 Recommendations

During the development of the Border Rivers Basin model there were a number of areas where further attention needs to be paid.

The Queensland procedure for estimating the inflows in the catchment models uses gauged streamflow data, so the quality of the model is very dependent on the quality of this data. Therefore, it is recommended that the quality of the current stream gauging network in the Border Rivers be maintained.

The daily rainfall stations used in the Sacramento models also play an important role in maintaining the accuracy of the model. Therefore, it is recommended that the current daily rainfall station network be maintained. Work has been continuing on improving methods for infilling missing rainfall records. This work has demonstrated the difficulty of estimating missing data even when there are nearby rainfall stations. Therefore, it is recommended that the closure of existing stations used in this study be avoided. Work should continue on trying to improve methods of estimating missing rainfall.

Evaporation is the other important parameter used in the model. The evaporation has been estimated using Morton’s method from meteorological data. However, there are only a couple of stations within the catchment that have all the meteorological information needed for the application of this method. Many of the meteorological stations do not measure solar radiation. It is recommended that consideration be given to increasing the number of stations recording all the variables needed to accurate estimate evaporation within the catchment.

The other important ingredient in developing an accurate model is information on water usage. Metering of installations is being undertaken. However, the metering information is not readily available to the modelling group. Work needs to be done in making this information readily available to the modelling group. Ideally, the diversion data should be include the start and end date of the diversion event as well as the volume diverted.

The quality of the diversion data provided by the Resource Operation Licence holder is not adequate for the purposes of the model development. The conditions of the Generation 1 plan only require the provision of the annual diversion from each zone. This information is not suitable for the estimation of low flows in the supplemented schemes and it arrives too late to be of use in the extension of the model for Section 71 reporting under the Basin Plan. It is recommended that the Generation 2 plan insist that the modellers have access to the daily diversion data for each reach in a timely manner.

These recommendations would assist in the continuing improvement of the Border River models for the next plan review.

466 Hydrology Report No 416000.PR/6

467 Department of Environment and Science

12 References

Beven, K. and Biney, A. M. (1992). The future of distributed models: model calibration and uncertainty prediction, Hydrological Processes, 6, 279–98, 1992.

Boughton, W. (2005). Catchment water balance modelling in 1960-2004. Agricultural Water Management, 71: 91–116.

Burnash, R,J.C., Ferral, R.L., and McGuire, R.A. (1973). A Generalised Streamflow Simulation System: Conceptual Modelling for Digital Computers, Joint Federal-State River Forecast Centre, U.S. National Weather Service and California Department of Water Resources, Sacramento, California.

Coron, L., Andrassian, V., Perrin, P., Lerat, J., Vaze, J., Bourqui, M., and Hendrickx, F. (2012). Crash testing hydrological models in contrasted climate conditions: an experiment on 216 Australian catchments. Water Resources Research, 48, W05552.

Duan, Q. (1991). A global optimization strategy for effcient and effective calibration of hydrologic models, Ph.D. thesis, Dept. of Hydrol. and Water Resour., Univ. of Ariz., Tuscon.

Duan, Q., Gupta, V. K., and Sorooshian, S. (1992). A shuffled complex evolution approach for effective and efficient global minimization, J. Optim. Theory Appl.

Duan, Q., Sorooshian, S., and Gupta, V. K. (1994). Optimal use of the SCE-UA global optimization method for calibrating watershed models, Journal of Hydrology, 158, 256–84.

eWater Ltd. (2016a). Source Scientific Reference Guide (v4.1) [Online]. Available: https://wiki.ewater.org.au/display/SD41/Storage+Routing

eWater Ltd. (2016b). Source for Managing Rivers [Online]. Available: http://ewater.org.au/products/ewater-source/for-rivers/

Ellis, R. J., Doherty, J., Searle, R. D., and Moodie, K. (2009). Applying PEST (Parameter ESTimation) to improve parameter estimation and uncertainty analysis in WaterCAST models, MODSIM 2009, Cairns, Australia.

Gupta, V. K., and Sorooshian, S. (1985). The automatic calibration of conceptual catchment models using derivative-based optimization algorithms, Water Resour. Res., 21(4), 473–86.

Gupta, V. K., and Sorooshian, S. (1983). Uniqueness and observability of conceptual rainfall-runoff model parameters: the percolation process examined, Water Resour. Res., 19(1), 269–76.

Hansen, N. and Ostermeier, A. (2001). Completely derandomized self-adaptation in evolution strategies. Evolutionary Computation, 9(2) p159–195.

Hansen, N. (2006). The CMA evolution strategy: a comparing review, Advances on estimation of distribution algorithms, Springer, p1769–1776.

Kavetski, D., Kuczera, G., and Franks, S. W. (2006). Bayesian analysis of input uncertainty in hydrological modeling: 2. Application, Water Resour. Res., 42.

Kisters Pty. Ltd (2010) Hydstra Version 10.

468 Hydrology Report No 416000.PR/6

Kuzmin, V., Dong-Jun, S., and Koren, V., (2008), Fast and efficient optimization of hydrologic model parameters using a priori estimates and stepwise line search, Journal of Hydrology, 353, 109–28.

Lerat, J., Egan, C. A., Kim, S., Gooda, M., Loy, A., Shao, Q., and Petheram, C. (2013). Calibration of river models for the Flinders and Gilbert catchments, CSIRO, Canberra, Australia.

Morton, F.I. (1983). Operational estimates of area evapotranspiration and their significance to the science and practice of hydrology, Journal of Hydrology 66, p 1-76

NSW DLWC. (1996). Integrated Quantity-Quality Model (IQQM) User Manual, Department of Land and Water Conservation, NSW.

NSW DLWC & QLD DNR. (1998). Border Rivers System: IQQM Implementation. Vol 2 of 7: Calibration Report: Dumaresq-Macintyre Rivers Subsystem (1986-1991), Department of Land and Water Conservation, NSW and Department of Natural Resources, QLD.

NSW DLWC & QLD DNR. (1999a). Border Rivers System: IQQM Implementation. Vol 1 of 7: Calibration Report: MacIntyre Brook Subsystem (1987-1996), Department of Land and Water Conservation, NSW and Department of Natural Resources, QLD.

NSW DLWC & QLD DNR. (1999b). Border Rivers System: IQQM Implementation. Vol 3 of 7: Calibration Report: Severn-Macintyre Rivers Subsystem (1986-1990), Department of Land and Water Conservation, NSW and Department of Natural Resources, QLD.

NSW DLWC & QLD DNR. (2000). Border Rivers System: IQQM Implementation. Vol 4 of 7: Model Validation and Testing, Department of Land and Water Conservation, NSW and Department of Natural Resources, QLD.

NSW Government. (2009). Water Sharing Plan for the NSW Border Rivers Regulated River Water Source 2009. . (2003). Water Plan (Border Rivers) 2003.

Queensland Hydrology (2013). Assessment of Sacramento Rainfall Runoff Models Used in Water Planning Models in Queensland, Queensland Hydrology, Brisbane, Australia.

Renard, B., Kavetski, D., Kuczera, G., Thyer, M., and Franks, S. W. (2010). Understanding predictive uncertainty in hydrologic modeling: Le challenge of identifying input and structural errors, Water Resour. Res. Sorooshian, S., Gupta, V. K., and Fulton, J. L. (1983). Evaluation of maximum likelihood parameter estimation techniques for conceptual rainfall-runoff models: influence of calibration data variability and length on model credibility, Water Resour. Res., 19(1), 251–59. Sorooshian, S., and Gupta, V. K. (1983). Automatic calibration of conceptual rainfall-runoff models: the question of parameter observability and uniqueness, Water Resour. Res., 19(1), 260–68. Vrugt, J. A., ter Braak, C. J. F., Gupta H. V., and Robinson B. A. (2009). Equifiniality of formal (DREAM) and informal (GLUE) Bayesian approaches in hydrologic modelling?, Stoch. Environ. Res. Assess. 23, 1011–1026. Zhang, X., Waters, D., and Ellis, R. (2013). Evaluation of Simhyd, Sacramento and GR4J rainfall runoff models in two contrasting Great Barrier Reef catchments, MODSIM 2013, Adelaide, Australia.

469 Department of Environment and Science

13 Abbreviations

AHD Australian Height Datum

AMTD Adopted Middle Thread Distance

APFD Annual Proportional Flow Deviation

BoM Bureau of Meteorology

CA catchment area

CINRS Climate Impacts and Natural Resource Systems (a group within DERM)

Ck. Creek

cumecs cubic metres per second

DERM Department of Environment and Resource Management (Qld)

DLWC Department of Land and Water Conservation (NSW)

DMM Data Modification Module

DPI Department of Primary Industries

D/S downstream

DS dead storage

EFO Environmental Flow Objective

FSA full supply area

FSL full supply level

FSV full supply volume

GL gigalitres

GS Gauging Station

ha hectare

HNFY historical no-failure yield

HW headwater

IQQM Integrated Quantity-Quality Model

IROL Interim Resource Operations Licence

IRM Integrated Resource Management

IWA Interim Water Allocation

km kilometres

km2 square kilometres

470 Hydrology Report No 416000.PR/6

Lat latitude

Long longitude m metres

MAD Mean Annual Diversion

MAF Mean Annual Flow

MAR Mean Annual Rainfall

MARO Mean Annual Runoff

Max maximum

Min minimum

ML megalitres mm millimetres mth month m3/s cubic metres per second n/a not applicable

PET potential evapotranspiration

R River

ROL Resource Operations Licence

ROP Resource Operations Plan (Post-WROL the ROP is split into a water management protocol (rules for unsupp WAs + dealings for supp WAs), ROLs & Operation Manuals (water sharing rules for supp WAs)).

QLD Queensland

Sac. Sacramento model

SID Storage Inflow Derivation

SILO BoM’s Internet website that provides meteorological and agricultural data

TWS town water supply

U/S upstream

WASO Water Allocation Security Objectives

WERD Water Entitlements Registration Database

WP Water Plan

WROL Water Resource Operation Licence

WRP Water Resource Plan

471 Department of Environment and Science

WSI Water Sharing Index

WSS Water Supply Scheme y year

472 Hydrology Report No 416000.PR/6

473 Department of Environment and Science

14 Glossary

Alluvial: Alluvial refers to deposits of clay, silt, sand, gravel, or other particulate material that has been deposited by a stream or other body of running water in a streambed, on a flood plain, on a delta, or at the base of a mountain.

Adopted Middle Thread Distance (AMTD): AMTD is the distance in kilometres, measured along the middle of a watercourse, from the mouth or junction.

Allocation: A water allocation is an authority granted under section 146 or 147 of the Water Act 2000 to take water.

Announced allocation: Announced allocation is a ratio (expressed as a percentage), which is announced from time to time by the Resource Operation Licence holder which sets a limit to the amount of supplemented water which a water allocation holder can divert during the water year as a proportion of the water allocation holder’s nominal volume. The announced allocation may increase but cannot decrease during a water year.

Aquifer: An aquifer is a body of permeable material or rock, capable of transmitting significant amounts of water underlain by impermeable material and through which underground water flows.

Artesian (water): Artesian water is water that occurs naturally in, or is introduced artificially into, an aquifer, which if tapped by a bore, would flow naturally to the surface.

A-depletion: A-depletion is the depletion (expressed in millimetres) in soil moisture from the maximum soil moisture capacity that a crop can withstand before it requires watering to sustain it. Once the A-depletion value falls below the nominated value, the allocation holder starts placing irrigation water orders to restore the soil moisture capacity to the nominated A-depletion value.

Authorisation: An authorisation refers to a licence, permit, interim water allocation or other authority to take water given under the Water Act 2000 or a repealed Water Act, other than a permit for stock or domestic purposes.

Annual Proportional Flow Deviation (APFD): APFD refers to the statistical measure of changes to flow seasonality and volume in the simulation period.

Baseflow: Baseflow is the natural streamflow derived from underground water seepage from aquifers and/or through the lateral movement of water through soils and into the stream. At times of peak flow, baseflow represents only a small proportion of total flow, while in periods of drought, it may represent all of the flow.

Basin: A basin is the total area from which water drains to a river system, or a grouping of adjacent river systems. In geological terms, a basin is defined as either a broad tract of land in which the rock strata are tilted toward a common centre, or a large, bowl-shaped depression in the surface of the land or ocean floor.

Benefited/Supplemented groundwater area: A benefited/supplemented groundwater area contains aquifers that are recharged from augmented surface water supplies from water storage structures.

Bore: A bore is a hole drilled to extract, recharge or investigate groundwater resources. In the Water Act 2000, it means a shaft, well, gallery, spear or excavation and any works constructed in connection with the shaft, well, gallery, spear or excavation, which taps the aquifer.

474 Hydrology Report No 416000.PR/6

Calibration model: A calibration model involves the modelling of flows, extractions, operational rules and infrastructure that occurred historically.

Catchment: A catchment is an area, bounded by natural topographic features such as hills or mountains, from which a drainage system derives its water.

Confluence node: A confluence node is defined as a node representing the confluence of two watercourses. These watercourses may be supplemented or unsupplemented streams.

Current development: The current development case is modelling the existing entitlements within the system, to the degree to which they are presently operating. Authorisations are set to take only the water they are currently accessing, as indicated by data investigation reports and knowledge of the system operation.

Dam: A dam is works that include a barrier, whether permanent or temporary, that does, or could, or would, impound, divert or control water; and the storage area created by the works.

Discharge (water): Discharge is the rate at which a volume of water passes through a cross- section per unit of time; measured in cubic metres per second (m3/s) or in megalitres per day (ML/d).

Distribution efficiency: Distribution efficiency is the efficiency of the system in delivering water from the dams to the users. This is determined by dividing deliveries by releases. (Note: this often excludes hydropower releases and deliveries).

Data Modification Module (DMM): This is a program used to adjust inflows using recorded flows downstream.

Drawdown: Drawdown is the lowering of the water table resulting from the extraction of water.

Entitlement: A water entitlement is a water allocation, interim water allocation or water licence.

Environmental flow: Environmental flow is the flow required to sustain a healthy environment. The release of water from a storage to a stream to maintain the healthy state of the stream.

Environmental Flow Objective (EFO): An EFO is a flow objective associated with a water plan (WP), for the protection of the health of natural ecosystems for the achievement of ecological outcomes.

Event duration: The event duration for a flow at a point in a watercourse, means the period of time when the discharge is greater than or less than the level necessary for a particular riverine process to happen.

Full development case: The full development case is modelling the full use of existing entitlements within the system. Authorisations are set to take all the water they are allowed to, regardless of climate or other factors not specifically mentioned in the licence. Generally, the full development case represents a higher level of use than the current development case, as it can include defunct licences and sleepers.

Headwater: A headwater reach is the source and upper reaches of a stream.

Hydrograph: A hydrograph is a graph showing the change in streamflow discharge at some location over time.

475 Department of Environment and Science

Hydrologic model: A hydrologic model is a computer program that simulates streamflows, water losses, storages, releases, in-stream infrastructure, water diversion and water management rules within a river system.

Infiltration: Infiltration is the downward entry of water into soil through the soil surface.

Integrated Quantity-Quality Model (IQQM): IQQM is a computer program, with associated statistical analysis and reporting programs, which simulates daily streamflows, flow management, storages, releases, in-stream infrastructure, water diversions, water demands and other hydrologic events within a modelled area.

Licence: A water licence is licence granted under chapter 2, part 3, division 2 of the Water Act 2000 for the taking and using of water or for interfering with the flow of water. A water licence does not have a specified performance.

Licence volume: Licence volume is the nominal volume of water that may be taken under a water licence in one water year. The amount drawn may be subject to other licence conditions or allocation rules.

Link: A link in a Source model is a reach of river between two nodes.

Low flow regime: The low-flow regime for a watercourse refers to magnitude, frequency, duration, timing and rate of change of low flow through the watercourse.

Mean Annual Diversion (MAD): The mean annual diversion is the average volume of water taken by an allocation or group of allocations in a year. It is calculated by adding the total volume of water taken over a period of years and dividing by the number of years in that period. The calculation is performed on a water year basis.

Mean Annual Flow (MAF): The mean annual flow is the average volume of water in a year that would flow past a point and is calculated by adding the total volume of flow over a period of years and dividing by the number of years in that period. The calculation is performed on a water year basis.

Node: A node in a Source model is used to represent a point on a river system where certain processes occur. The node type identifies the rules and parameters that are used by the model to simulate the relevant processes at a given location.

Nominal operating volume: A nominal operating volume of a storage is the level that is to be maintained during the specified period by releasing extra water (if available) from the upstream storage.

Nominal volume: A nominal volume is the volume of water, in megalitres, that represents the proportional annual volumetric share of water available to be taken by holders of water allocations in a priority group or a water allocation group.

On-Farm storage: An on-farm storage is a private storage constructed on a property to store water.

Order time: Order time is the number of days in advance that an order has to be made to ensure that the ordered water arrives on time.

Over order factor: An over order factor in a Source model is the factor by which water orders need to be increased to account for operational inefficiencies in a water supply scheme. This factor does not account for transmission losses.

476 Hydrology Report No 416000.PR/6

Overland flow water: Overland flow water is water, including floodwater, flowing over land, other than in a watercourse or lake after having fallen as rain, or after rising to the surface naturally from underground, or in any other way.

Pre-development case: The pre-development case is created by removing all infrastructure, diversions and operation rules from the full development case. No adjustment is made for the effect of land clearing, natural changes in river course, or climate change.

Performance indicators: Performance indicators are measures that are calculated and stated in the WP with the purpose of assessing the effect of allocation and management decisions or proposals on water entitlements and natural ecosystems.

Plan Area: The Plan Area is the total area to be managed under the WP.

Pseudo crop method: The pseudo crop method involves the arrangement of evaporation, crop factors and planted area in a Source model to ensure that the full amount of water allowed to be diverted each year is diverted if available.

Reach: A reach in a Source model is a series of nodes connected by links. A river reach refers to a defined stretch of river.

Recharge (of underground water/aquifer): The replenishment of underground water by the gradual downward movement of water from the soil to the water table, by actions such as rainfall, overland flow or infiltration from streams percolating through the unsaturated zone; the volume of water added to the amount of water stored in the aquifer over a particular period; by artificial means, such as direct injection.

Resource Operations Licence (ROL): An ROL is granted under section 180 or 181 of the Water Act 2000. It authorises the holder of the licence to interfere with the flow of water to the extent necessary to operate the water infrastructure to which the licence applies.

Resource Operations Plan (ROP): An ROP was used to implement a WRP in specified areas. It detailed the operating rules for water infrastructure and other management rules that was applied in the day-to-day management of the flow of water in a reach or subcatchment. ROP specifies water access rules, environmental flow rules, trading rules, the conversion of licences to water allocations and monitoring requirements.

Post-WROL the ROP is split into a water management protocol (rules for unsupp WAs + dealings for supp WAs), ROLs & Operation Manuals (water sharing rules for supp WAs).

Return flow: Return flow is the water that flows out of the end of a channel system and back into a natural river system without being diverted by any user.

Riparian: Riparian refers to the area adjacent to a watercourse. Riparian access refers to an authority for an owner of land abutting a watercourse to take water for stock watering or domestic purposes.

River section: A river section in a Source model comprises a chain of links and nodes commencing with a headwater inflow node or a confluence node and finishing with a confluence or end-of system node.

Riverine: Riverine refers to rivers and their flood plains.

Routing: Routing occurs as water flows from one point to another in a system. Routing is the attenuation (flattening out) of the flow hydrograph as water moves down the system.

477 Department of Environment and Science

Scenario/Simulation model: A scenario/simulation model involves a fixed set of parameters for infrastructure, rules and licences. Scenario/simulation models are used to produce a representation of what may occur in the system, if the selected set of parameters were in place.

Simulation period: The simulation period is defined by the start and end dates of the model.

Sleepers: A sleeper is a licence which is current, but not in use.

Source Rivers (Source): Source is a computer program, with associated statistical analysis and reporting programs, which simulates daily streamflows, flow management, storages, releases, in- stream infrastructure, water diversions, water demands and other hydrologic events within a modelled area.

Subartesian water: Subartesian water is water that occurs naturally in, or is introduced artificially into an aquifer, which, if tapped by a bore, would not flow naturally to the surface.

Subcatchment area (subarea): A subarea is a portion of a catchment within the Plan Area. A subarea may be physically defined or simply a result of breaking the catchment into smaller sections for the purposes of modelling.

Supplemented: Supplemented refers to a water supply where the natural flow is reduced or increased by a dam or some other water storage facility.

Surface water: Surface water is water that is on the earth’s surface, such as in a watercourse, spring, lake or reservoir.

Sustainable management: Sustainable management allows for the allocation and use of water for the physical, economic and social wellbeing of people within limits that can be sustained indefinitely while protecting the biological diversity and health of natural ecosystems.

Transmission losses: Transmission losses are losses from surface water (other than into defined groundwater systems) as it flows from one location in a system to another. This can include evaporation, seepage, uptake by plants and unauthorised usage.

Tributary: A tributary is a stream that joins another stream or body of water.

Tributary recession factor: The tributary recession factor in a Source model specifies the percentage of each tributary inflow which can be used by downstream water users as part of the supplemented water supply.

Underground water: Underground water or groundwater is water found in the cracks, voids or pore spaces or other spaces between particles of clay, silt, sand, gravel or rock within the saturated zone of a geologic formation. In the saturated zone, all cracks, voids or pore spaces are completely filled with water – not to be confused with soil water in the unsaturated zone where voids are filled with both air and water. The upper surface of the saturated zone is called the water table.

Underground water levels: The physical measurement of the distance from the natural surface or reference point to the water surface in a subartesian bore when it is in a fully recovered state. A negative value indicates that the water level is below the reference point. Underground water level measurements provide an estimate of the ‘depth to the water table‘-or upper surface to the saturated zone-where the reference point is the natural surface.

Unsupplemented: Refers to water in a watercourse that is not supplemented from storage or diversion facilities.

478 Hydrology Report No 416000.PR/6

Water year: A water year is a continuous 12 month period starting from a specified month, used for the accounting of entitlements.

Water Allocation (WA): A Water Allocation is an authority granted under the Water Act 2000 to take water, interfere with water or a single combined water licence to authorise both the interference and take of water where these two activities are inextricably linked.

Water Allocation Security Objectives (WASO): WASOs are objectives that may be expressed as performance indicators and are stated in a WRP to ensure protection of a water entitlement to obtain water in accordance with a water allocation.

Water Supply Scheme (WSS): A WSS is a water infrastructure development designed and constructed for storage, supply and distribution of water from and to a watercourse.

Water Resource Operation Licence (WROL): This is a licence granted in relation to existing infrastructure or proposed infrastructure in an area where a resource operations plan had not been approved.

Water harvesting: Water harvesting is an entitlement to take unsupplemented water from a watercourse during specified high-flow events and generally involves diverting water into an on- farm storage for later use. Water harvesting is licensed.

Weir: A weir is a barrier constructed across a watercourse below the banks of the watercourse that hinders or obstructs the flow of water in the watercourse.

479 Department of Environment and Science

Appendix A – Source Modelling Platform

For managing rivers

Source provides Australia’s first nationally applicable integrated modelling software combining river and catchment modelling to support water planning and river operations across the country. Its use in river management is to simulate the physical and management aspects of river systems at a range of spatial and temporal scales. It can be run in one of two interchangeable operations and planning ‘modes’. The first mode is used to inform day-to-day operational decisions. The second mode is used to inform policy decisions relating to the long-term impacts on water and environment resources.

Overview

Source has been developed to address water sharing and savings for river and connected groundwater systems. It offers important new features and capabilities dealing with water reform, climate change and environmental water.

It allows users to:  share water between environmental and irrigation demands  consider what impact climate change will have on water security  manage multiple water owners in storage and in transit in the river system  link existing models to build on current approaches.

Source provides a consistent modelling environment to support transparent river management decisions. Fundamental to this design is the flexibility which makes it readily customisable and easy to update as new science becomes available. New capabilities can be incorporated via plug- ins developed to suit particular needs.

Extensive trials with eWater partners have proved the capabilities of Source in river basins across the country.

Modes for planning and operations

The Source modelling package can be run in one of two inter-changeable modes: 'Operations' to inform day-to-day operational decisions; and 'Planning' to inform policy decisions relating to the

480 Hydrology Report No 416000.PR/6

long-term impacts on water and environment resources. This means that managers dealing with daily operations, water accounting and long-term planning will be able to efficiently and accurately compare analyses using a common platform and river system model.

Planning

For assessing the longer term impacts of water resource policy on river behaviour and all users

The Planning mode is designed to explore how changes in policy and management will affect the behaviour of the river system in the longer term. It can be used by river managers to assess the impact of actual or potential changes to the behaviour of the river system and the probable effect on system storages, flows and water shares. By exploring the options, river planners can decide on which management actions will optimise river performance to meet planning objectives.

Source (Planning) is designed to:  determine which management rules will best meet planning objectives  explore the impact of changes in management, land-use and climate on river behaviour and water availability  model the supply, demand and use of water at a range of time scales  simulate complex management rules, such as continuous sharing  accommodate the needs and conditions of different river catchments across Australia  track and account for water shares and ownership  assess current and future water availability across entire river systems  interact efficiently with river operations.

Operations

For day-to-day operational decisions around the release and storage of water in regulated systems

The Operational mode of Source is designed to inform day-to-day decisions around managing storage releases and delivery of water to meet demands in regulated river systems. It enables the user to build an operational model of the river. The operator can then assess the impact of different operational scenarios on the way water moves through a river system in response to storage releases, tributary inflows, losses, demands and constraints.

Source (Operations) is designed to:  inform decisions on how the system should be operated to deliver water in the short and medium term to consumptive and environmental users  inform decisions on water transfers between catchments, rivers and reservoirs as specified in operation and management plans  inform changes in water delivery requirements as a consequence of external drivers, such as water trading  decide on the optimum storage and weir operations to meet target watering regimes for consumptive and environmental demands  interact efficiently with long-term river system planning.

Information sourced from eWater (2016b)

481 Department of Environment and Science

Appendix B – The Sacramento Model

The Sacramento rainfall-runoff model was developed by Burnash, Ferral and McGuire (1973). It can be implemented through the computer programs WINSAC and/or Source. It is an explicit soil moisture accounting type model developed by the United States National Weather Service and the California Department of Water Resources, originally for flood forecasting applications.

The Sacramento model consists of a number of storages connected by catchment processes. The model components and the relationships between them are shown in Figure B.1.

Rainfall on the catchment is considered as falling on one of two types of surface, permeable areas or impervious areas that are linked to the channel system. Runoff is produced from impervious areas in any rainfall event.

The permeable area, in contrast, produces runoff only when the rainfall is sufficiently heavy. In this portion, initial soil moisture storage (the upper zone tension storage) must be filled before water is available to enter other storages. This represents the depth of precipitation required to meet interception requirements and is water bound closely to soil particles. When this tension storage is filled, water is accumulated in the upper zone free water storage, from where it is free to drain to deeper storages or to move laterally to appear in the stream channel as interflow.

The vertically draining water, or percolation, can enter one of three lower zone storages, the lower zone tension storage (the depth of water held closely by the soil particles) or one of the two lower zone free water storages, primary and supplemental (that are available for drainage as baseflow or subsurface outflow). The two free water storages fill simultaneously but drain independently at different rates to produce the variable baseflow recession.

Evaporation occurs from surface water areas at the potential rate, but in other areas, varies with both evapotranspiration demand and the volume and distribution of tension water storage.

The surface runoff and interflow are routed to the catchment outlet by a non-dimensional unit hydrograph. In catchments where significant nonlinearities may be present, such as extensive flood plains that may alter the mean travel times, a layered Muskingum routing technique, effectively introducing a number of linear storage-discharge relationships, can be used.

To implement the model in a given catchment, a set of 18 parameters must be defined. These parameters define the generalised model for a particular catchment. The parameters are usually derived for a gauged catchment by a process of calibration where the recorded streamflows are compared with calculated streamflows and the parameters are adjusted to produce the best match between the means and standard deviations of the daily streamflows, and reducing the difference in peak flow discharge.

For ungauged catchments, parameter sets from adjacent or nearby gauged catchments may be used. A parameter set may be called a regional parameter set especially if the ungauged catchment is located in the same local region where the catchment with the calibrated parameter set is located.

482 Hydrology Report No 416000.PR/6

Figure B.1: Sacramento Model Schematic

483 Department of Environment and Science

Appendix C – Fors

Introduction

A major challenge in the development of hydrologic models is the derivation of model inflows. Typically, model inflows are derived (in full or part) from historical rainfall and potential evapotranspiration records using conceptual rainfall-runoff models calibrated to simulate the catchment behaviour.

Many conceptual models have been used to simulate Australian catchments over the decades (Boughton 2005). The Sacramento Soil Moisture Accounting model (Burnash et al. 1973; Figure C.1) is consistently amongst the best performing in comparison studies (for example Zhang et al. 2013) and for that reason is the model of choice for rainfall-runoff modelling at Queensland Hydrology.

Figure C.1: The Sacramento soil moisture accounting model simulates catchment runoff using a conceptual network of interconnected soil moisture stores. The major processes included in the model are illustrated. In order for a Sacramento model to aptly represent the behaviour of a particular catchment, it is necessary to estimate suitable values for the model’s many free parameters. There are 17 numerical parameters describing the sizes of various soil moisture stores and the movement of water between them (Adimp, Lzfpm, Lzfsm, Lzpk, Lzsk, Lztwm, Pctim, Pfree, Rexp, Rserv, Sarva, Side, Ssout, Uzfwm, Uzk, Uztwm, Zperc), 1 unit hydrograph kernel and, in some implementations, weighting factors for the input rainfall datasets. These parameters cannot be measured directly and are usually estimated using expert knowledge of the region, and/or inferred such that they best reproduce observed streamflows over a historical “calibration period”. There is no clear consensus on exactly how this should be done, partly because it depends on the model’s purpose, and partly because new methods are continually being researched as enabled by increasing computational capacity.

484 Hydrology Report No 416000.PR/6

This appendix outlines an automated method for Sacramento model calibration developed at Queensland Hydrology and used in this work. In this method the Sacramento model parameters are iteratively adjusted using a global optimisation search algorithm to find the best-fitting parameter values. The specifics are explained in the following 3 sections:  The Calibration Objective – what the calibration aims to achieve  The Model Parameterization – which parameters are adjusted during the calibration, and what are the constraints  The Optimisation Algorithm – by what procedure are the model parameters iteratively adjusted to progressively achieve the calibration objective. Afterwards, the calibration method is demonstrated holistically using two case studies – one with synthetic data and one with real data. Finally some conclusions are drawn and recommendations made for future work.

The Calibration Objective

When calibrating a rainfall-runoff model to observed streamflow records, there are several characteristics of the historical flow that may be targeted including but not limited to:  The reproduction of daily flow records, with a balance of emphasis on both high and low flows  The reproduction of hydrograph characteristics such as the shapes of flood peaks and recession curves, and the occurrence of cease-to-flow periods  The long-term average runoff volume  The statistical distribution of flows of different magnitudes, which includes the median flow and other percentiles. In the automatic calibration paradigm (Gupta and Sorooshian 1983, 1985; Sorooshian et al. 1983a, Sorooshian and Gupta 1983b) these qualitative objectives are amalgamated into a single quantifiable statistic, the so-called “objective function”, which measures the overall goodness-of-fit between modelled results and observed data. An iterative optimisation search algorithm is then used to find the model parameter values that minimize the objective function (assuming the usual convention whereby a lower value = better fit).

Common choices for the objective function include Pearson’s correlation coefficient (R2), and the Nash-Sutcliffe Efficiency statistic calculated in linear space (NSE) or log space (NSELOG). These simplistic choices serve to target good daily flow performance, but tend to give disproportionate weight to either the high flows (in the case of R2 or NSE) or low flows (in the case of NSELOG) and do little to achieve the observed long-term average runoff volume or the statistical distribution of observed flows. More elaborate objective functions have been devised and shown to overcome some of these downfalls (e.g. Ellis et al. 2009).

Recent work by Coron et al. (2012) and Lerat et al. (2013) has explored a family of objective functions of the form

휇 |∑ 푄 − ∑ 푄̂| 2 2 푓 = [1 + ] × [훼 ∑(푄휆 − 푄̂휆) + (1 − 훼) ∑(푅휆 − 푅̂휆) ] ∑ 푄 where 푄 is the set of observed daily flows, 푄̂ is the set of corresponding modelled flows, 푅 is the observed daily flow ranked by amplitude, 푅̂ is the corresponding modelled flow ranked by amplitude, and 휆, 휇 and 훼 are constants governing the relative importance of various parts of the expression. The first term in the expression, 1 + |∑ 푄 − ∑ 푄̂|/ ∑ 푄, measures how well the model reflects the total runoff volume (i.e. bias) during the calibration period. The exponent 휇 regulates

485 Department of Environment and Science

the relative importance of this term, and a value of 휇 = 1 was found to give calibrations with very good bias without significantly impacting the daily or statistical performance. The second term has 2 two parts. ∑(푄휆 − 푄̂휆) measures the difference between the modelled and observed data on a 2 daily basis. ∑(푅휆 − 푅̂휆) measures the difference between the modelled and observed data on a ranked-daily basis. The constant 휆 tunes the relative importance of high flow to low flows in both parts. A larger value of 휆 gives more weight to high flows, and a smaller value gives more weight to low flows. A value of 휆 = 0.5 was found to give a good balance between high and low flow performance for a variety of catchments. The constant 훼 regulates the relative balance between daily performance and statistical performance. A value of 훼 = 0.1 was found to work well across a variety of scenarios.

The work of Coron et al. (2012) and Lerat et al. (2013) as well as work conducted at Queensland Hydrology has demonstrated that this objective function excels in reproducing the target flow characteristics discussed earlier, across a broad range of catchments, and is reasonably robust to common data issues such as misrepresentative rainfall, and transient streamflow gauge issues. This objective function has been adopted for the present work, with the values 휇 = 1, 휆 = 0.5, and 훼 = 0.1. Hereafter it is referred to as SDEB in reference to the performance aspects that it targets: Square root Daily flow, Exceedeance, and Bias.

The figure below illustrates the typical results from a calibration using the SDEB objective function.

Figure C.2: Observed streamflow data at GS925002A (red) and Sacramento modelled flows that maximize the NSE statistic (blue) and SDEB statistic (orange). While both modelled series have similar behaviour at high flows, the recession behaviour is much better represented when optimising SDEB than NSE (left). Moreover the statistical distribution of flows shown by the exceedance curve (right) is also better for SDEB than NSE for both high flows and low flows.

The Model Parameterization

The parameterization of the model – what parameters are adjusted during model calibration and what values they may take – is critical to ensure that unphysical parameter combinations are not explored. The parameterization also affects the efficiency of the optimisation, with the easiest parameters to optimise being the ones that influence the objective function linearly and independently. For these reasons the original Sacramento model parameterization was revised.

486 Hydrology Report No 416000.PR/6

The Sacramento rainfall-runoff model has 17 basic free parameters. These describe the sizes of conceptual stores in the model, and the movement of water between them. In addition there are an arbitrary number of elements jointly defining a unit hydrograph kernel which acts to retard the modelled surface runoff. The eWater Source Sacramento model supports a 5-element unit hydrograph. In addition to these there is a weighting factor for each rainfall sequence contributing to the catchment rainfall. All these parameters are listed in Table C.1 below, along with allowed ranges conservatively devised following a thorough review of previous calibrations undertaken at Queensland Hydrology (QH 2013). The final column shows additional constraints and interdependencies that should also be considered (Burnash, Ferral and McGuire 1973).

To account for the constraints and interdependencies presented in the original parameter set above, the model software was edited and the model was reparameterized.

The requirement that LZPK ≤ LZSK was handled by introducing a new independent parameter LZPKONSK (valid on the range 0.001–1) and subsequently treating LZPK as a derived parameter (LZPK = LZPKONSK * LZSK). Similarly, the requirement that SARVA ≤ PCTIM was handled by introducing a new independent parameter SARVAONPCTIM (valid over the range 0.0001–1) and subsequently treating SARVA as a derived parameter (SARVA = SARVAONPCTIM * PCTIM). The unit hydrograph kernel was replaced with a simple parametric function called the “lag unit hydrograph” (Lerat et al. 2013). This function has just 1 parameter LAGUH which represents the mean hydrograph lag from which the values of all underlying unit hydrograph elements UH1, UH2, UH3, UH4 and UH5 are derived. RSERV was fixed in the model (RSERV = 0.3) and subsequently omitted from the calibration.

During this reparameterisation, logarithmic transformations were used to introduce a logarithmic prior to most parameters, and all parameter ranges were normalized so that the final calibration parameters are all independently valid over the range 0–1. The final set of calibration parameters and their relationships to the original parameters are shown in Table C.2.

The final calibration parameter set consists of 17 independent numerical parameters valid over the range 0–1 plus one additional parameter for each rainfall sequence. The final parameters are related to the original Sacramento model parameters by the transformations described above. The raw values of the final calibration parameters are not directly meaningful, and therefore all calibrations have been reported in terms of the original Sacramento parameters. Nevertheless the transformations above will be of interest to anyone trying to imitate (or improve) this method.

The optimisation algorithm used to determine optimal values for these parameters is discussed next.

487 Department of Environment and Science

Table C.1: Sacramento Parameter Ranges and Dependencies

Original Description Valid Constraints and parameter ranges interdependencies Basic ADIMP Additional impervious fraction 0.00001–0.15 LZFPM Lower zone free primary capacity [mm] 1–300 LZFSM Lower zone free supplementary capacity 1–350 [mm] LZPK Lower zone primary drainage rate 0.00001–0.9 LZPK ≤ LZSK LZSK Lower zone supplementary drainage rate 0.001–0.9 LZPK ≤ LZSK LZTWM Lower zone tension water capacity [mm] 10–600 PCTIM Permanent impervious fraction 0.00001–0.11 SARVA ≤ PCTIM PFREE Proportion of direct percolation to lower 0.01–0.5 zone REXP Exponent in percolation equation 1–6 RSERV Fraction of lower zone with no transpiration 0.3 Fixed value SARVA Fraction covered by water and riparian veg. 1e-8–0.11 SARVA ≤ PCTIM SIDE Channel loss ratio 0.00001–0.1 SSOUT Channel loss rate [mm/day] 0.00001–0.1 UZFWM Upper zone free water capacity [mm] 5–155 UZK Upper zone lateral drainage rate 0.1–1 UZTWM Upper zone tension water capacity [mm] 12–180 ZPERC Change in percolation rate from wet to dry 1–600 Unit hydrograph kernel

UH1 Unit hydrograph element 1 0–1 ∑ 푈퐻푖 = 1 UH2 Unit hydrograph element 2 0–1 푖

UH3 Unit hydrograph element 3 0–1 Credible kernels are UH4 Unit hydrograph element 4 0–1 single-peaked. UH5 Unit hydrograph element 5 0–1 Rainfall weighting factors (one per rainfall dataset) RF1 Rainfall factor 1 03 One rainfall weighting factor per rainfall … … … dataset

488 Hydrology Report No 416000.PR/6

Table C.2: Relationship between Original Sacramento Parameters and Calibration Sacramento Parameters

Final calibration Relationship to original parameters parameters Basic G_ADIMP ADIMP = 10^(G_ADIMP * (Log(0.15) − Log(0.00001)) + Log(0.00001))) G_LZFPM LZFPM = 10^(G_ LZFPM * (Log(300) − Log(1)) + Log(1))) G_LZFSM LZFSM = 10^(G_ LZFSM * (Log(350) − Log(1)) + Log(1))) G_LZPKONSK LZPKONSK = 10^(G_LZPKONSK * (Log(1) − Log(0.001)) + Log(0.001))) LZSK = 10^(G_LZSK * (Log(0.9) − Log(0.001)) + Log(0.001))) G_LZSK LZPK = LZPKONSK * LZSK G_LZTWM LZFSM = 10^(G_ LZTWM * (Log(600) − Log(10)) + Log(10))) G_SARVAONPCTIM SARVAONPCTIM = 10^(G_SARVAONPCTIM * (Log(1) − Log(0.0001)) + Log(0.0001))) G_PCTIM PCTIM = 10^(G_PCTIM * (Log(0.11) − Log(0.00001)) + Log(0.00001))) SARVA = SARVAONPCTIM * PCTIM G_PFREE PFREE = 10^(G_PFREE * (Log(0.5) − Log(0.01)) + Log(0.01))) G_REXP REXP = 10^(G_REXP * (Log(6) − Log(1)) + Log(1))) G_SIDE SIDE = 10^(G_SIDE * (Log(0.1) − Log(0.00001)) + Log(0.00001))) G_SSOUT SSOUT = 10^(G_SSOUT * (Log(0.1) − Log(0.00001)) + Log(0.00001))) G_UZFWM UZFWM = 10^(G_UZFWM * (Log(155) − Log(5)) + Log(5))) G_UZK UZK = 10^(G_UZK * (Log(1) − Log(0.1)) + Log(0.1))) G_UZTWM UZTWM = 10^(G_UZTWM * (Log(180) − Log(12)) + Log(12))) G_ZPERC UZTWM = 10^(G_UZTWM * (Log(600) − Log(1)) + Log(1))) Unit hydrograph kernel G_LAGUH LAGUH = G_LAGUH * 3 UH1 = Max(0, 1 − LAGUH) UH2 = IF(LAGUH < 1, Max(0, LAGUH − 0), Max(0, 2 − LAGUH)) UH3 = IF(LAGUH < 2, Max(0, LAGUH − 1), Max(0, 3 − LAGUH)) UH4 = IF(LAGUH < 3, Max(0, LAGUH − 2), Max(0, 4 − LAGUH)) UH5 = IF(LAGUH < 4, Max(0, LAGUH − 3), Max(0, 5 − LAGUH)) Rainfall weighting factors (one per rainfall dataset) G_RF1 RF1 = G_RF1 * 3 … …

The Optimisation Algorithm

Despite the efforts in the previous section to create an optimisation-friendly set of model parameters, the Sacramento rainfall-runoff model is inherently nonlinear. Conditional operations that take place within the model mean that the objective function result does not change smoothly with respect to the model’s parameters (i.e. its derivative is discontinuous). These nonlinearities cause difficulty for most optimisation algorithms, and especially for gradient-based methods which

489 Department of Environment and Science

depend on the ability to estimate derivatives of the objective function. In addition to being highly nonlinear, the objective function is typically pitted with local optima which can prevent optimisation algorithms from discovering the globally optimum solution, and again these impact gradient-based methods most.

The Shuffled Complex Evolution Algorithm

The Shuffled Complex Evolution (SCE; Duan 1991, Duan et al. 1992) global optimisation algorithm stands out as one of the most capable algorithms for solving these difficult optimisation problems. It is a stochastic evolutionary (non-gradient based) algorithm developed at the University of Arizona specifically for the calibration of watershed models, and Sacramento models in particular (Duan et al. 1994, Soroosihan et al. 1993).

The SCE algorithm starts with the selection of a large number of models with randomly chosen parameter sets. The Latin hypercube sampling method is used for this to ensure that the initial sample has good coverage of parameter space. These models are run and the fitness of each assessed using the objective function. The population is then ranked by fitness and shuffled into q smaller groups, referred to as ‘complexes’. Each complex is evolved using the competitive complex evolution algorithm (CCE) for i ‘breeding iterations’. During these iterations better fitting models are discovered and introduced into the complex, and poorly fitting models are discarded. After i breeding iterations the termination criterion is checked: does the total number of model runs exceed some very large number d. If this criterion is satisfied, the SCE algorithm terminates, otherwise all complexes are recombined, sorted, shuffled into new complexes, and the evolutionary process begins again. When the algorithm terminates the best-fitting model is returned as the ‘optimal model’. Figure C.3 shows the SCE algorithm and the embedded CCE algorithm.

Figure C.3: The Shuffled Complex Evolution algorithm (left) and the Competitive Complex Evolution algorithm that is embedded within it (right).

490 Hydrology Report No 416000.PR/6

Parameters of the Shuffled Complex Evolution Algorithm

The SCE algorithm itself has several parameters which can be used to tune the algorithm for particular optimisation problems. These were introduced in the previous section and are listed in Table C.3 below with their recommended values. The dimensionality of the problem n refers to the number of parameters being optimised, i.e. nominally n = 17 + 1 per rainfall series. The next four parameters, m, s, i, and p, can be calculated from n using relationships recommended by Duan et al. (1994).

Table C.3: Recommended values for SCE algorithm parameters

Parameter Description Value n Dimensionality of problem ? m Number of points in each complex m = 2n + 1 s Total population size over all complexes s = m * n i Number of breeding iterations per shuffle i = m p Number of parents in each simplex p = n + 1 q Number of complexes q = 12 d Minimum number of model runs d = 80,000

Suitable values for the number of complexes q and the minimum number of model runs d (the termination criterion) were determined empirically using hundreds of trial calibration runs with 3, 6, 12, and 24 complexes and d=200,000. When using 3 or 6 complexes, the SCE converged too quickly and regularly became trapped in local optima. Using 12 complexes gave a good balance between reliability and speed and q=12 was adopted as the default. In the tests with 12 complexes, very little improvement in the calibration was observed beyond about 40,000 model runs. A very conservative default value of d=80,000 was adopted to ensure sufficient time for convergence under conditions that differed from our test conditions.

These findings are consistent with Duan etal (1994) who tested calibrations of up to 13 Sacramento parameters using just one year of data and determined that at least four complexes and an average of over 13,000 model runs were required for successful optimisation.

Demonstration of Calibration Method

The calibration method is demonstrated below using two case studies. In the first case study, the calibration method is applied to synthetic streamflow that was generated using a Sacramento model with known parameter values. The ability of the calibration method to reproduce the synthetic streamflow data and retrieve the known parameter values is assessed. In the second case study, the calibration method is applied to real streamflow data at measured at GS925002A on the .

Case 1 – Synthetic data

A synthetic sequence of streamflow data was generated using a Sacramento model previously calibrated for the Barron River at Picnic Crossing GS110003A headwater catchment. The sequence was generated for the period 1/1/1889 to 28/08/2013 and set as the target flow

491 Department of Environment and Science

sequence for 100 individual Sacramento model calibration runs. Each calibration run followed the method outlined in this appendix, using an SCE optimiser to calibrate 18 parameters (since there was just 1 input rainfall sequence) to optimise the SDEB objective function. The same rainfall and potential evapotranspiration data was used in the calibration as was used to generate the target streamflow sequence, and therefore the calibrations were theoretically capable of reproducing the target sequence with indefinite precision.

Figure C.4 shows the synthetic target streamflow data (black line) and the 95th percentile range of calibrated flows (green bands, hidden behind the black line) during the years of the largest two floods. For all practical purposes, the calibrations reproduced the synthetic target streamflow data perfectly with all calibrations yielding NSE=1.000, R^2=1.000 and bias=1.000.

Figure C.4: Synthetic streamflow data (black line) and 95th percentile range of calibrated flows (green band hidden behind the black line).

Since the underlying parameter values were already known, it was possible to compare them to the calibration results. Table C.4 below shows the ideal values, and the mean and standard deviation of the optimised results. The optimised values compare very well to the known underlying values. The only exceptions were ADIMP which had a large standard deviation compared to the ideal value (indicating insensitivity in the parameter in this particular case), and SIDE and SSOUT which both have a similar effect in the model and took variable anticorrelated values.

492 Hydrology Report No 416000.PR/6

Table C.4: Sacramento parameters

Parameter Ideal value (known) Mean optimised value Standard deviation of optimised values ADIMP 0.00002 0.00055 0.00059 LZFPM 250 250.4931 0.5137 LZFSM 100 99.96232 0.17650 LZPK 0.015 0.015005 0.000050 LZSK 0.15 0.15009 0.00038 LZTWM 125 124.819 0.184 PCTIM 0.0006 0.000486 0.000223 PFREE 0.28 0.28024 0.00126 REXP 1.9 1.9027 0.0059 SARVA 0.0015 0.00125 0.00065 SIDE 0.007 0.003747 0.003168 SSOUT 0.0018 0.003125 0.003826 UZFWM 35 34.96864 0.05109 UZK 0.6 0.59961 0.00075 UZTWM 12 12.01651 0.10822 ZPERC 32 32.10454 0.16630 UH1 0 0 0 UH2 0.12 0.1200 0.0001 UH3 0.88 0.8800 0.0001 UH4 0 0 0 UH5 0 0 0 RF1 1.07 1.06849 0.00147

This case study demonstrated the ability of the calibration method to consistently and accurately fit a Sacranento model to (ideal) streamflow data.

Case 2 – Real data

Real historical streamflow data recorded at Wenlock River at Wenlock GS925002A was used to perform 100 Sacramento calibrations of the upstream catchment. Each calibration run followed the method outlined in this appendix, using an SCE optimiser to calibrate 18 parameters (since there was just one input rainfall sequence) to optimise the SDEB objective function. Due to observational errors in the data and limitations inherent in the conceptual model, the calibrations were not expected to reproduce the observed flows exactly.

Figure C.5 shows the synthetic target streamflow data (black line) and the 95th percentile range of calibrated flows (green bands) during the years of the largest two floods.

493 Department of Environment and Science

For reasons which remain unclear, the use of real (not synthetic) streamflow data led to increased variability in the calibration results compared to the previous case study. The reasons may be related to an increased occurrence of local optima in the objective function when using real data, but further study is required to fully understand this effect. Despite the slightly increased variability the modelled flow sequences, the results are consistently of a very high quality, giving confidence that any single calibration is likely to be acceptably close to the optimal solution for practical purposes. In this test all calibrations yielded NSE=0.60±0.01, R^2=0.795±0.005 and bias=1.000±0.001.

Figure C.5: Historical streamflow data from GS925002A (black) and 100 calibrated modelled flows (green band). The variability in the calibration results reflects that the optimisation did not consistently achieve unique optimal Sacramento parameters. Nevertheless the results are consistently of high quality, giving confidence that any single calibration is likely to be acceptably close to the optimal solution.

On the calibration of rainfall weighting factors

For rainfall-runoff modelling purposes, catchment rainfall is typically estimated using a weighted combination of rainfall data collected at stations in or around the catchment. The weighting factors used are normally calculated prior to calibration on the basis of the proportions of the catchment each rainfall station is thought to represent. This simple idea harbours some subtle problems. For instance it is not straightforward, when attributing fractional areas to each rainfall station, to account for topographical effects such as the influences of mountain ranges, plains, and coastlines – the very popular Thiessen weighting method blatantly ignores these land features. And in cases where the catchment rainfall is suspected to be ubiquitously under- or overestimated by the available rainfall data the modeller may be challenged with contriving a suitable bias adjustment for the measured rainfall data.

494 Hydrology Report No 416000.PR/6

In the present work the rainfall weighting factors have been treated as parameters that can be calibrated and the observed streamflow itself is used to determine how much of the catchment rainfall is represented by each input rainfall sequence.

The catchment average rainfall 푃푖 on each day i is calculated as a weighted average of the data 퐷푖푗 from each station j using calibrated rainfall weighting factors 푅퐹푗.

푃푖 = ∑ 퐷푖푗 푅퐹푗 푗

For quality assurance of the weighting factors, all final weighted catchment rainfall sequences are compared to the corresponding 50-year mean annual isohyets. In most cases they agree well with the estimated isohyet values (within 10–20 percent). Larger discrepancies may call for a review of the assumed catchment area, rainfall station selection, and/or reverting to fixed rainfall weighting factors that are consistent with the estimated isohyets. The runoff coefficients (the ratio of modelled runoff to catchment rainfall) are also calculated to verify that the Sacramento models have physically reasonable levels of runoff efficiency – these coefficients should normally be in the range 0.20–0.35.

Discussion

The method outlined in this appendix uses a Shuffled Complex Evolution optimisation algorithm and the SDEB objective function to calibrate Sacramento rainfall-runoff models to historical streamflow data. The success of this method does depend substantially on the details of the implementation, which have been described.

The primary benefits compared to manual calibration are that the automated method is very fast, repeatable (increasing the consistency between modellers of different levels of experience) and can produce very high-quality results (limited by modelling assumptions and data quality).

The method may be improved in future work. Recommended work includes:  Exploration of alternative optimisation algorithms with the aim to improve convergence performance (speed and reliability) and to better understand how convergence performance decays with increasing numbers of model parameters. Two optimisation algorithms of particular potential are CMA-ES (Hansen and Ostermeier 2001, Hansen 2006) and DREAM (Vrugt et al. 2009).  Investigation of alternative unit hydrograph parameterizations.  Assessment of the benefits and drawbacks of decreasing the number of Sacramento parameters as suggested by Kuzmin et al. (2008). There is ongoing academic work on fully Bayesian calibration frameworks (Beven and Biney 1992, Kavetski et al. 2006, Vrugt et al. 2009, Renard et al. 2010). These frameworks combine model calibration and simultaneous estimation of the uncertainties in the calibrated parameter values, catchment rainfall data and observed streamflow data. Unfortunately this work has not matured and the methods do not yet adequately account for limitations in modelling structure, the complex nature uncertainty in catchment rainfall estimation, and they have not been trialled with complicated models (including the Sacramento model) nor with calibration periods longer than a couple of years. Nevertheless progress is being made, and should be watched carefully over the coming years.

495 Department of Environment and Science

Appendix D – The Flow Adjustment Process (DMM)

RUNDMM2S

RUNDMM2S is a data modification module (DMM) that consists of a number of programs that can be used to adjust model inflows on a daily basis to give good agreement between the Source modelled flow and the flow recorded at a streamflow gauge.

The inflows estimated by the calibrated Sacramento model for each subarea are used in the Source to simulate the flows at the stream gauge for the period of record. The DMM compares the recorded and simulated flow to determine daily factors that are used to adjust the inflow sequences.

When the modelled flow is greater than zero, the daily inflow from each subarea is multiplied by the following factor:

푀푒푎푠푢푟푒푑 푓푙표푤 + 푅푒푠푖푑푢푎푙 푑푖푓푓푒푟푒푛푐푒 퐹푎푐푡표푟 = 푀표푑푒푙푙푒푑 푓푙표푤 + 푅푒푠푖푑푢푎푙 푑푖푓푓푒푟푒푛푐푒 Where:

푅푒푠푖푑푢푎푙 푑푖푓푓푒푟푒푛푐푒 is residual difference in the Source model specified by the user

When there is no modelled flow, a daily flow is added to the appropriate daily flow in each inflow sequence. The amount of flow added to a particular subarea inflow is determined by the difference between the measured flow and the modelled flow scaled by a factor. The scaling factor is usually estimated by dividing the subarea area by the total catchment area upstream of the gauge.

The DMM process is undertaken in two steps. In the first step, the factors are estimated from the measured and modelled flow. In the second step, the factors are applied to the inflow sequences allowing for any lag caused by routing in Source. In the second step, the user can define the periods of time that the DMM factors are to be applied.

It should be noted that the Source is nonlinear because of routing, impacts of weirs and losses that depend on the flow. The DMM process is essentially a linear process. Therefore in most situations it may be necessary to iterate the process a number of times. In some situations, smoothing may have to be used to smooth out oscillations in the low flows.

Residual Catchments

In adjusting the subarea inflows for residual catchments, which are catchments between two stream gauges, the process needs to take into account the flows recorded at the upstream gauge (or gauges). Because these flows have been recorded, they cannot be adjusted. All adjustments have to be carried out on the subarea inflows downstream of the upstream gauge.

The formula used to calculate the adjustment factors in this situation are as follows.

When the modelled flow is greater than the upstream flow, the daily flow from each subarea is multiplied by the following factor:

푀푒푎푠푢푟푒푑 푓푙표푤 − 푈푝푠푡푟푒푎푚 푓푙표푤 + 푅푒푠푖푑푢푎푙 푑푖푓푓푒푟푒푛푐푒 퐹푎푐푡표푟 = 푀표푑푒푙푙푒푑 푓푙표푤 − 푈푝푠푡푟푒푎푚 푓푙표푤 + 푅푒푠푖푑푢푎푙 푑푖푓푓푒푟푒푛푐푒

Where: 푅푒푠푖푑푢푎푙 푑푖푓푓푒푟푒푛푐푒 is residual difference in the Source model specified by the user

496 Hydrology Report No 416000.PR/6

When the modelled flow is less than the upstream flow, a value is added to each subarea inflow as described above.

If there is routing and lag between the upstream gauge and the downstream gauge, the upstream flow sequence should be routed through Source before being used in the program.

When there are inconsistencies between the rating curves of the two gauges, the DMM process will try to compensate. For example, if the rating curve of the upstream gauge underestimates the flow, then the DMM process will increase the flow in the downstream catchments to ensure that the predicted flow at the downstream gauge matches the upstream flow. A small discrepancy can be almost impossible to detect. If the rating curve of the upstream gauge overestimates the flow, the DMM process will reduce the flow in the downstream catchments. If the problem is severe, there will be no flow in the downstream catchments. This situation is easier to detect. Any suspicions about the stream gauge ratings are referred to the hydrographers.

However, it cannot change the poorly-routed flow from the upstream gauge. This usually leads to an overestimation of the flows. This can be dealt with using an overall adjustment process built into the software.

Multiple Reaches

The DMM process is carried out in each reach upstream of a gauge. When this process has been completed for each reach, a daily inflow sequence is created for each subarea upstream of the stream gauge consisting of flows originally estimated using the Sacramento model. In some periods, the flow has been adjusted using the DMM process to give good agreement to the flows recorded at the downstream gauge. For the periods of time when there areis no recorded data at the gauge, the flows are purely Sacramento model estimates.

In the final Source model, the flow at a downstream gauge is an accumulation of all the subarea inflows from all the reaches upstream. Sometimes there is a long-term gauge at the end of system and a comparison between the predicted flow and the recorded flow shows considerable losses in the period where the upstream subarea flows are based purely on the Sacramento model. In this situation, the DMM process can be applied to all the subarea inflows upstream. This is done only for the periods when there is no local stream gauge data to undertake a local DMM process.

497 Department of Environment and Science

Appendix E – Flow Difference (NEGFLO6S)

NEGFLO6S can be used to calculate the difference between two Source format files. It is especially useful for calculating the residual catchment inflow by subtracting the upstream gauged flow routed through a Source model from the flow recorded at the downstream gauge.

The subtraction can generate negative flows due to differences in routing. If the flows are set to zero, the resultant modelled flows can significantly exceed the measured downstream flow. This can cause a particular problem if the reach flows into a dam, so the final model tends to predict higher dam levels than the recorded levels. Therefore, NEGFLO6S includes a number of methods for correcting the positive flows.

498 Hydrology Report No 416000.PR/6

499 Department of Environment and Science

Appendix F - Storage Volume-Area-Discharge Relationships

Table F.1: Glenlyon Dam storage curve

Level (m, SD) Glenlyon Volume (ML) Glenlyon Area (ha)

365.25 0 0.0

365.64 0 0.0

369.00 185 0.2

370.00 345 0.2

371.00 555 0.2

372.00 827 0.3

373.00 1,170 0.4

374.00 1,600 0.5

375.00 2,110 0.6

376.00 2,720 0.7

377.00 3,430 0.8

378.00 4,260 0.9

379.00 5,220 1.0

379.80 6,090 1.2

380.60 7,070 1.3

381.30 8,030 1.5

382.00 9,120 1.7

382.60 10,160 1.9

383.10 11,100 2.0

383.60 12,110 2.1

384.10 13,190 2.2

384.50 14,100 2.3

384.90 15,040 2.4

385.30 16,020 2.5

385.70 17,050 2.6

386.10 18,120 2.7

386.50 19,230 2.9

386.80 20,110 3.0

500 Hydrology Report No 416000.PR/6

Level (m, SD) Glenlyon Volume (ML) Glenlyon Area (ha)

387.50 22,260 3.3

388.10 24,270 3.5

388.60 26,050 3.7

389.20 28,330 3.9

389.70 30,340 4.1

390.10 32,010 4.3

390.60 34,180 4.4

391.10 36,430 4.6

391.50 38,280 4.7

391.90 40,200 4.9

392.30 42,170 5.1

392.70 44,210 5.2

393.10 46,330 5.4

393.50 48,520 5.7

393.80 50,240 5.8

394.60 55,080 6.3

395.40 60,310 6.8

396.10 65,180 7.2

396.80 70,290 7.5

397.50 75,610 7.8

398.10 80,350 8.0

398.70 85,250 8.4

399.30 90,340 8.6

399.90 95,600 9.0

400.40 100,140 9.3

401.00 105,770 9.6

401.50 110,620 9.9

402.00 115,620 10.2

402.50 120,780 10.5

402.90 125,020 10.8

403.40 130,480 11.1

501 Department of Environment and Science

Level (m, SD) Glenlyon Volume (ML) Glenlyon Area (ha)

404.30 140,730 11.7

405.10 150,300 12.3

405.90 160,290 12.8

406.70 170,660 13.3

407.40 180,040 13.7

408.20 191,280 14.3

408.80 200,040 14.9

409.50 210,710 15.7

410.10 220,260 16.2

410.70 230,150 16.8

411.30 240,390 17.3

411.90 250,930 17.8

412.09 – FULL SUPPLY 254,320 18.0

412.50 261,730 18.3

413.00 270,950 18.7

413.50 280,370 19.0

415.00 309,900 20.4

415.50 320,210 20.9

416.50 341,560 31.8

417.00 362,600 31.8

502 Hydrology Report No 416000.PR/6

Table F.2: Glenlyon Dam spillway curve

Level (m, SD) Glenlyon Spillway Discharge (ML/d)

365.25 0

412.09 – FULL SUPPLY 0

412.10 400

412.24 1,361

412.38 2,322

412.55 3,603

412.77 5,524

413.00 9,047

413.19 13,211

413.47 19,936

413.77 27,302

414.11 36,269

414.44 45,556

414.84 57,406

415.25 69,896

415.64 82,386

416.08 97,438

416.95 129,143

417.46 149,960

417.91 170,777

418.65 208,887

419.86 271,017

421.44 356,845

423.14 458,687

423.78 499,680

503 Department of Environment and Science

Table F.3: Glenlyon Dam valve capacity

Level (m, SD) Glenlyon Valve Capacity (ML/d)

365.25 0

368.55 0

381.41 1,998

384.45 2,224

387.50 2,441

390.55 2,646

393.60 2,840

396.65 3,026

399.69 3,202

402.74 3,368

405.79 3,527

408.84 3,681

411.73 3,824

411.89 3,831

414.93 3,979

417.98 4,125

421.03 4,269

424.08 4,410

427.13 4,547

430.17 4,679

433.22 4,807

434.75 4,807

504 Hydrology Report No 416000.PR/6

Table F.4: Pindari Dam storage curve

Level (m, AHD) Pindari Volume (ML) Pindari Area (ha)

443 0 0

444 15 0.95

446 68 4.23

449 400 18.9

451 925 33.4

454 2,298 58.1

458 5,387 98.4

464 13,913 189

473 36,581 315

481 65,978 421

487 93,638 506

495 138,967 630

502 187,142 751

516 – FULL SUPPLY 312,321 1,048

529 469,075 1,379

540 638,537 1,714

505 Department of Environment and Science

Table F.5: Pindari Dam spillway curve

Level (m, AHD) Pindari Spillway Discharge (ML/d)

443 0

516 – FULL SUPPLY 0

517 48,383

518 137,351

519 252,759

520 389,507

521 544,585

522 716,123

523 902,657

524 1,103,072

525 1,316,455

526 1,542,062

527 1,779,250

528 2,027,516

529 2,286,336

530 2,555,323

531 2,834,115

532 3,122,354

533 3,419,724

534 3,725,965

535 4,040,881

536 4,364,207

537 4,695,723

538 5,035,238

506 Hydrology Report No 416000.PR/6

Table F.6: Pindari Dam valve capacity

Level (m, AHD) Pindari Valve Capacity (ML/d)

443 0

446.2 0

455.6 1,590

458.6 1,737

461.7 1,945

464.8 2,092

467.8 2,250

470.9 2,397

473.9 3,048

477.0 4,235

480 4,970

490 6,551

500 7,824

510 8,783

520 9,801

530 10,676

540 11,499

541 11,499

507 Department of Environment and Science

Table F.7: Coolmunda Dam storage curve

Level (m, AHD) Coolmunda Volume (ML) Coolmunda Area (ha)

297 0 0

300.4 110 10.7

301.19 212 15.2

302 355 20.6

302.4 445 24.9

303 618 33.2

303.4 765 40.4

303.8 946 50.6

304 1,053 56.4

304.4 1,305 70.1

304.8 1,617 86.9

305 1,802 98.6

305.2 2,012 112

305.4 2,254 131

305.6 2,537 151

305.8 2,859 172

306 3,232 202

306.2 3,671 236

306.4 4,179 271

306.6 4,754 303

306.8 5,385 330

307 6,076 362

307.2 6,834 395

307.4 7,652 424

307.8 9,461 482

308 10,454 511

308.4 12,612 566

309.2 17,473 626

309.6 20,103 676

310 22,902 722

310.6 27,482 807

508 Hydrology Report No 416000.PR/6

Level (m, AHD) Coolmunda Volume (ML) Coolmunda Area (ha)

311 30,842 877

311.2 32,636 917

311.6 36,470 1,004

312 40,694 1,106

312.2 42,956 1,156

312.4 45,316 1,205

312.6 47,778 1,256

312.8 50,340 1,307

313 53,005 1,358

313.2 55,771 1,408

313.4 58,635 1,458

313.6 61,610 1,518

313.8 64,702 1,574

314 67,903 1,628

314.07 – FULL SUPPLY 69,061 1,645

314.2 71,212 1,681

314.4 74,624 1,732

314.8 81,744 1,780

315 85,455 1,883

509 Department of Environment and Science

Table F.8: Coolmunda Dam spillway curve

Level (m, AHD) Coolmunda Spillway Discharge (ML/d)

297 0

314.07 – FULL SUPPLY 0

314.18 393

314.22 2,376

314.27 12,182

314.3 24,149

314.4 72,403

314.6027 118,713

314.8 171,417

314.85 214,963

314.9 258,940

314.95 306,288

315 339,379

315.1 355,449

315.11 465,004

510 Hydrology Report No 416000.PR/6

Table F.9: Coolmunda Dam valve capacity

Level (m, AHD) Coolmunda Valve Capacity (ML/d)

297 0

301.19 0

304 220

305 230

306 255

307 270

308 290

309.2 315

310 330

311 345

312 360

313 380

314.07 – FULL SUPPLY 390

511 Department of Environment and Science

Table F.10: Volume-area-discharge curve for Whetstone Weir

Level (m) Volume (ML) Area (ha) Spillway Valve Discharge Discharge (ML/d) (ML/d)

257.4 0 0 0 0

257.66 0

257.67 1,000,000

259.25 36 0.8

259.5 48 1.7

259.7 62 2.4

259.9 78 5

260.1 97 9.1

260.5 140 13

261.1 220 14.2

261.5 280 18

262.45 506 25 0 1,000,000

262.46 1,000,000,000

1,000,000,000 1,000,000,000

512 Hydrology Report No 416000.PR/6

Table F.11: Volume-area-discharge curve for Ben Dor Weir

Level (m) Volume (ML) Area (ha) Spillway Valve Discharge Discharge (ML/d) (ML/d)

244.4 0 0 0 0

245.74 0

245.75 1,000,000

246.05 24.6 3.2

246.65 51.7 8.1

247.5 135 13

248.1 226 17.8

248.9 394 24.3

249.3 492 28.3

249.65 615 32.4

249.9 700 36.4 0 1,000,000

249.95 1,000,000,000

1,000,000,000 1,000,000,000

513 Department of Environment and Science

Table F.12: Volume-area-discharge curve for Boggabilla Weir

Level (m) Volume (ML) Area (ha) Spillway Valve Discharge Discharge (ML/d) (ML/d)

207.1 0 0 0 0

208.84 87 10

209.77 250 25

210.2 0

210.21 1,000,000

210.8 645 52

211.81 1,280 73

212.8 2,100 94

213.82 3,170 115

214.81 4,410 135

215.8 5,850 155

216 0 1,000,000

216.8 7,510 180

217.78 9,500 225 86,000

262.47 300,000

263.1 100,000 225 100,000,000

514 Hydrology Report No 416000.PR/6

Table F.13: Volume-area-discharge curve for Brown Storage

Level (m) Volume (ML) Area (ha) Spillway Discharge (ML/d)

0 0 0 0

0.23 0.28 0.12

0.63 1.8 0.49

1.2 5.4 1.1

2.0 12.2 1.9

3.0 23.1 3.0

4.2 39.1 4.4

5.6 61.1 6.0

7.2 90.2 7.8

9.0 127 9.8

11.0 173 12.1

12.7 1

13.2 200 13.4 6

18.4 10,000 200 9,806

29.7 100,000 1,000 99,806

52.2 1,000,000 5,000 999,806

80.0 3,000,000 12,000 10,000,000

515 Department of Environment and Science

Table F.14: Volume-area-discharge curve for Breakout 2 Low Storage

Level (m) Volume (ML) Area (ha) Spillway Discharge (ML/d)

0 0 0 0

0.03 0.21 0.75

0.08 1.3 3.0

0.15 4.1 6.8

0.25 9.1 12.0

0.37 17.3 18.8

0.51 29.3 27.1

0.68 45.8 36.9

0.87 67.6 48.1

1.1 95.4 60.9

1.3 130 75.2

1.5 0

1.6 150 82.9 5

1.9 200 100 55

2.4 1,000 250 855

3.1 10,000 1,500 9,855

7 10,000,000 10,000 10,000,000

516 Hydrology Report No 416000.PR/6

Table F.15: Volume-area-discharge curve for Breakout 2 High Storage

Level (m) Volume (ML) Area (ha) Spillway Discharge (ML/d)

0 0 0 0

0.23 0.28 0.12

0.63 1.8 0.49

1.2 5.4 1.1

2.0 12.2 1.9

3.0 23.1 3.0

4.2 39.1 4.4

5.6 61.1 6.0

7.2 90.2 7.8

9.0 127.3 9.8

11.0 173.3 12.1

12.7 1

13.2 200 13.4 6

18.4 10,000 200 9,806

29.7 100,000 1,000 99,806

52.2 1,000,000 5,000 999,806

517 Department of Environment and Science

Table F.16: Volume-area-discharge curve for Billa Billa Storage

Level (m) Volume (ML) Area (ha) Spillway Discharge (ML/d)

0 0 0 0

0.39 1.4 0.36

1.1 8.7 1.4

2.1 27.0 3.2

3.5 60.9 5.7

5.2 115 8.9

7.2 195 12.8

9.6 306 17.4

12.3 451 22.7

15.4 636 28.8

18.8 867 35.5

22.3 1

22.5 1,000 39.2 6

33.6 10,000 120 9,006

46.9 100,000 800 99,006

75.0 1,000,000 4,000 999,006

518 Hydrology Report No 416000.PR/6

Table F.17: Volume-area-discharge curve for Yarrill Commoron Storage

Level (m) Volume (ML) Area (ha) Spillway Discharge (ML/d)

0 0 0 0

0.06 19.3 31.0

0.17 122 124

0.34 378 279

0.56 853 495

0.83 1,616 774

1.2 2,735 1,114

1.5 4,278 1,517

2.0 6,313 1,981

2.5 8,908 2,507

3.0 12,130 3,095

3.58 1

3.61 14,000 3,412 101

4.3 15,000 3,550 1,101

5.4 50,000 7,000 36,101

7.4 100,000 9,500 86,101

9.9 500,000 25,000 486,101

13.1 1,000,000 41,000 986,101

519 Department of Environment and Science

Table F.18: Volume-area-discharge curve for Breakout 3 Low Storage

Level (m) Volume (ML) Area (ha) Spillway Discharge (ML/d)

0 0 0 0

0.2 0.21 0.1

0.6 1.3 0.4

1.1 4.1 0.9

1.9 9.1 1.6

2.8 17 2.51

3.8 29 3.61

5.1 46 4.91

6.6 67.6 6.42

8.2 95.4 8.12

10.0 130.0 10.0

11.5 1

11.9 150 11.1 6

17.2 10,000 200 9,856

28.4 100,000 1,000 99,856

50.9 1,000,000 5,000 999,856

520 Hydrology Report No 416000.PR/6

Table F.19: Volume-area-discharge curve for Eurone Swamp Storage

Level (m) Volume (ML) Area (ha) Spillway Discharge (ML/d)

0 0 0 0

0.16 19.3 12.4

0.43 122 49.5

0.85 378 111

1.4 853 198

2.1 1,616 309

2.9 2,735 446

3.9 4,278 607

5.0 6,313 792

6.2 8,908 1,003

7.6 12,130 1,238

9.0 1

9.0 14,000 1,365 101

10.6 15,000 1,430 1,101

14.0 50,000 2,450 36,101

20.7 100,000 3,200 86,101

36.0 500,000 5,800 486,101

65.5 1,000,000 7,500 986,101

521 Department of Environment and Science

Table F.20: Volume-area-discharge curve for the Lower Weir Floodplain Storage

Level (m) Volume (ML) Area (ha) Spillway Discharge (ML/d)

0 0 0 0

0.05 20.7 43.2

0.13 131 173

0.26 405 389

0.43 914 691

0.64 1,731 1,080

0.89 2,930 1,555

1.2 4,583 2,117

1.5 6,764 2,765

1.9 9,544 3,500

2.3 12,998 4,321

2.6 14,400 4,764 1

2.7 100

3.0 14,500 4,790 5,000

3.4 25,000 7,500 10,000

3.7 40,000 12,000 15,000

4.7 60,000 14,000 18,000

5.7 80,000 16,000 20,000

6.7 100,000 18,000

15.0 160,000 25,000 10,000,000

522 Hydrology Report No 416000.PR/6

Table F.21: Volume-area-discharge curve for Lalaguli Storage

Level (m) Volume (ML) Area (ha) Spillway Discharge (ML/d)

0 0 0 0

0.909 0.8 0.09

2.0 3.7 0.35

3.3 9.3 0.79

4.8 18.4 1.4

6.5 31.7 2.2

8.3 49.8 3.2

10.4 73.5 4.3

12.7 103 5.6

15.2 140 7.1

17.9 185 8.8

20.8 238 10.6

23.6 1

23.8 300 12.6 6

29.1 1,000 26 706

39.8 10,000 110 9,706

62.9 100,000 500 99,706

108 1,000,000 2,500 999,706

230 10,000,000 8,000 10,000,000

523