Environ Monit Assess (2016) 188:308 DOI 10.1007/s10661-016-5310-7
Validation of a model with climatic and flow scenario analysis: case of Lake Burrumbeet in southeastern Australia Yohannes Yihdego & John Webb
Received: 22 June 2015 / Accepted: 15 March 2016 # Springer International Publishing Switzerland 2016
Abstract Forecast evaluation is an important topic that addresses the development of reliable hydrological probabilistic forecasts, mainly through the use of climate uncertainties. Often, validation has no place in hydrology for most of the times, despite the parameters of a model are uncertain. Similarly, the structure of the model can be incorrectly chosen. A calibrated and verified dynamic hydrologic water balance spreadsheet model has been used to assess the effect of climate variability on Lake Burrumbeet, southeastern Australia. The lake level has been verified to lake level, lake volume, lake surface area, surface outflow and lake salinity. The current study aims to increase lake level confidence model prediction through historical validation for the year 2008–2013, under different climatic scenario. Based on the observed climatic condition (2008–2013), it fairly matches with a hybridization of scenarios, being the period interval (2008–2013), corresponds to both dry and wet climatic condition. Besides to the hydrologic stresses uncertainty, uncertainty in the calibrated model is among the major drawbacks involved in making scenario simulations. In line with this, the uncertainty in the calibrated model was tested using
Y. Yihdego (*) : J. Webb Environmental Geoscience, La Trobe University, Melbourne, Victoria 3086, Australia e-mail:
[email protected] Y. Yihdego Snowy Mountains Engineering Corporation (SMEC), Sydney, New South Wales 2060, Australia
sensitivity analysis and showed that errors in the model can largely be attributed to erroneous estimates of evaporation and rainfall, and surface inflow to a lesser. The study demonstrates that several climatic scenarios should be analysed, with a combination of extreme climate, stream flow and climate change instead of one assumed climatic sequence, to improve climate variability prediction in the future. Performing such scenario analysis is a valid exercise to comprehend the uncertainty with the model structure and hydrology, in a meaningful way, without missing those, even considered as less probable, ultimately turned to be crucial for decision making and will definitely increase the confidence of model prediction for management of the water resources. Keywords Hydrology . Verification . Validation . Prediction . Wetland management . Climate . Lake Burrumbeet . Australia
Introduction In natural conditions, lake levels vary on different temporal scales from days to centuries. These changes in lake water levels are due to many natural causes (climate, catchment area, topography, lake size) and anthropogenic pressures such as climate change, groundwater extraction or inflow regulation (Enwright et al. 2011; Juma et al. 2014; Hudak 2014; Haghighi and Kløve 2015). A decrease in water level can influence the physical environment, biota and ecosystem, with
308
Page 2 of 14
impacts on a number of lake ecosystem functions. Severe impacts in lake ecological and socio-economic status have been reported for many large and small lakes worldwide. Different lakes can display different responses to external impacts, with the littoral zone and its habitats typically being most easily affected (Yihdego and Becht 2013). There is a need to better understand the vulnerability of lake water levels to external pressures and to develop methods to relate catchment water use to changes in lake levels. Potential impacts of climate change must also be better understood and predicted. The most obvious method to estimate lake levels is the water balance equation, where water input and output result in lake storage and water level changes. However, all water balance components cannot always be quickly assessed, such as evaporation due to expansion of irrigated areas or lake-groundwater interactions. A method that assesses general changes in lake level can be a useful tool in examining why different lakes have different lake level variation patterns and why the water disappears from some lakes. Assessment methods using climate data can provide important insights into variations in lake levels in different parts of the world. Identifying the trends of climate and hydrological changes is important for developing adaptive strategies for effective water resources management (Dewi et al. 2009; Yihdego and Webb 2013). Many studies focused on the prediction of future climate at a regional/global scale using general circulation models (GCM) or these models’ downscaled outcomes (Dinka et al. 2014; Wu et al. 2014; Yihdego et al. 2015). The lack of methods for the evaluation of hydrological forecasts may hinder acceptance of those forecasts by the public. Previous studies on the basalt plains of western Victoria, Australia (e.g. Coram 1996; Tweed et al. 2009; Yihdego and Webb 2012; Yihdego et al. 2016) have identified several different hydrogeological controls on the lakes there. Lake Burrumbeet was successfully modelled by Yihdego and Webb (2012, 2015) with a record of historical lake level (1998–2007), salinity (1991–2007) and gauged inflow (1975–2007). However, the parameters of a model are uncertain, and the structure of the model can be incorrectly chosen, because validation has little place in hydrology for most of the time. Using the model in a historically predictive model and comparing it with an observed data is a valid exercise, it makes a lot of sense being it increases our confidence in its value (Weijs et al. 2010). The current study aims to increase confidence model prediction by
Environ Monit Assess (2016) 188:308
validating the lake level under different climatic scenario, using the calibrated and verified historical lake level.
Regional setting Lake Burrumbeet is located in central western Victoria and is the largest of four shallow lakes in the Ballarat region (Fig. 1), with an area of ∼23 km2. The lake is the major wetland for the region and has been utilized for recreational boating, fishing and camping. Lake Burrumbeet lies within the upper Hopkins River catchment, which in this area is generally gently undulating with an average elevation of only 385 m Australian Height Datum (AHD). The lake catchment, which has an area of 298 km2, is bounded in the north by the Great Dividing Range. The lake is fed by Burrumbeet Creek, the major inflow, from the east and Canico Creek from the south (Fig. 1). The lake outlet is situated on the southwest shore of the lake, and flows into Baille Creek, which is a tributary of Mt Emu Creek. The Lake Burrumbeet outlet structure is a 40-m wide weir consisting of a concrete slab with an outlet height of 378.7 m AHD. The weir can be adjusted with a removable wooden plank to elevate the outflow level to 379.1 m AHD (Yihdego and Webb 2015). Lake Burrumbeet There has been no detailed previous work on the hydrology of Lake Burrumbeet, except the recent study by Yihdego and Webb (Yihdego and Webb 2011a, 2011b; 2012; 2015). Many of the lakes in the region, including Lake Burrumbeet, have been dry in the recent past drought. The low lake levels are attributed to the prolonged period of below average rainfall and human activity such as the extraction of groundwater. An additional factor affecting the water budget of the lake is approximately 2000 × 10 3 m 3 /year of treated waste water discharged from the Ballarat North Water Waste Treatment Plant to Burrumbeet Creek. This volume is estimated to provide approximately 5.6 % of the mean annual volume entering Lake Burrumbeet but does not constitute part of the natural flow of Burrumbeet Creek, as the waste water is not drawn; rather, the water is originally drawn from another
Environ Monit Assess (2016) 188:308
basin and therefore represents an artificial interbasin transfer. Lake Burrumbeet catchment has a complex geology. The basement rocks outcrop over limited areas and are overlain by an extensive cover of Cenozoic basalt, colluvium and alluvium. Most of the catchment is covered by basalt and scoria of the Pleistocene Newer Volcanics, which erupted during three phases, ranging from about six million to a few thousand years ago. Callendar Bay, a small semi-circular bay on the northwestern edge of Lake Burrumbeet (Fig. 1), occupies the site of an explosive eruption point that produced tuff and basaltic agglomerate (Yihdego and Webb 2015).
Page 3 of 14 308
Climate Based on data from Ballarat, the weather station in the area with the longest continuous record, the respective hottest and coldest months are February and July with mean daily maximum and minimum temperatures of 25 and 3.2 °C, respectively. Average rainfall over the catchment is 699 mm (Fig. 2), but is highly variable. Over the last decade, there has been a substantial drop in rainfall, with annual rainfall below the long term average, in response to the recent past drought. Thus, long-term rainfall trends are likely to be changing the hydrological balance across the catchment.
Fig. 1 Map showing the volcanic crater (Callendar Bay) and basalt outcrop at the lake floor (after Yihdego and Webb 2015)
308
Environ Monit Assess (2016) 188:308
Page 4 of 14 1800
1600
1400
Rainfall (mm)
1200
1000
800
600
400
200
2005
2001
1997
1993
1989
1985
1981
1977
1973
1969
1965
1961
1957
1953
1945
1949
1941
1937
1933
1929
1925
1921
1917
1913
1909
0
-200
Annual rainfall
Cumulative deviation from mean annual rainfall
Fig. 2 Annual rainfall at Ballarat with its cumulative deviation from the mean
Maximum rainfall is received over winter and exceeds or equals evaporation (Fig. 3). Pan evaporation is highest from October to April and totals 1162 mm. There are three recording weather stations in the vicinity of Lake Burrumbeet: Burrumbeet, White Swan Reservoir and Ballarat. However, rainfall data at Burrumbeet station is only
Fig. 3 Mean monthly rainfall and pan evaporation at Ballarat
available until 1995, so the precipitation data since 1995 has had to be extrapolated from Ballarat (25 km east of Lake Burrumbeet) by establishing a correlation between rainfall at Burrumbeet station and Ballarat station. The average rainfall at Burrumbeet (1950–1995) is 624 mm which is slightly less than the 700 mm average for
200 180 Rainfall/Evaporation (mm)
160 140 120 100 80 60 40 20 0 Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Month rainfall
Pan evaporation
Sep
Oct
Nov
Dec
Page 5 of 14 308
Lake level (AHD)m
Fig. 4 Lake level with variation in rainfall
378.6
100
378.4
0
378.2
-100
378.0
-200
377.8
-300
377.6 -400
377.4
-500
377.2 377.0
-600
376.8
-700
Monthly cum.dev. rainfall (mm)
Environ Monit Assess (2016) 188:308
376.6 -800 Mar-97 Jul-98 Dec-99 Apr-01 Sep-02 Jan-04 May-05 Oct-06 Feb-08 Lake level
Monthly cumulative dev. rainfall
level has fallen at 22 cm/year since records began in 1998, with the lake going dry in 2003 and autumn 2004. There was minimal refilling in 2005. Lake Burrumbeet is usually moderately saline (median 3.74 mS/cm) and varies seasonally. A noticeable rise in salinity level has occurred since 2002 (up to 21,950 EC) as the lake level decreased due to abnormally dry conditions and lake levels of less than 0.4 m (Yihdego and Webb 2015).
Ballarat. Monthly evaporation is taken from the weather station at White Swan Reservoir. The variation in level of Lake Burrumbeet is strongly correlated with the monthly cumulative deviation of rainfall (Fig. 4), showing a similar pattern of peaks and troughs, and a long-term fall since the onset of the significant dry period in 1997. Lake Burrumbeet has a median depth of 0.7 m, but typically varies seasonally by 0.2–1 m (Fig. 4). The lake
40
% change on lake level
20 0 -25
-20
-15
-10
-5
0
5
10
15
20
25
-20 -40 -60 -80 -100
% change in water budget variables Surface inflow
Precipitation
Evaporation
Aquifer-lake conductance
Fig. 5 Relative sensitivity of Lake Purrumbete water-balance model to changes in the water budget components
Aquifer area
308
Environ Monit Assess (2016) 188:308
Page 6 of 14
Fig. 6 Monthly flow at Mount Emu Creek (gauge no. 236203; Fig. 1) and simulated surface outflow from the lake
Model calibration and verification A spreadsheet water balance model developed by Yihdego and Webb (2012, 2015) was used to examine the impact of the projected historical climate scenarios on Lake Purrumbete lake level. The spreadsheet model (Yihdego and Webb 2012) used in this study has the capability to simultaneously solve the conservative solute mass balance along with the water mass balance of the lake, on a monthly basis, providing additional constraints on the water balance. Model calibration has been carried out by adjusting model input parameters so that the simulated lake levels and salinities fit the observed lake levels and salinities.
The long-term mean monthly volumetric components of the Lake Burrumbeet water budget were calculated for 1976–2007, in order to include a wet period with above average rainfall, so that the long-term mean values adequately represent an average of the wet and dry periods (Yihdego and Webb 2012, 2015). There are several possible sources of error within the model, and the reliability of the parameter estimates given above can be assessed based on their ability to cause change within the model. Therefore, errors in the model can largely be attributed to erroneous estimates of evaporation and rainfall and surface inflow to a lesser extent (Fig. 5). However, inaccuracies in estimating lake water budget components are not likely to alter the order
Table 1 Actual lake area compared with simulated lake area Date
Actual area (km2)
Simulated area (km2)
Error (%)
18 September 1977
22.74
22.68
0.26
26 February 1980
22.56
22.65
0.43
26 December 1984
22.20
22.66
2.06
8 December 1989
22.60
22.65
0.25
12 January 1991
22.36
22.66
1.34
18 February 1993
22.55
22.66
0.46
24 February 1995
22.10
22.66
2.51
Environ Monit Assess (2016) 188:308
Page 7 of 14 308
Fig. 7 Simulated lake level backward from 1976 to 1998
Model verification
of magnitude difference between the individual water balance components (Coram 1996). Running the model on a monthly basis will minimize any errors associated with time lags between a rainfall event and its ultimate capture within the lake.
Model protocol requires that before any predictive simulation is made, the model should be calibrated and verified to assess the potential environmental effects of
1187 987
Rainfall (mm)
787 587
387 187 -13
Fig. 8 Cumulative annual residual rainfall at Burrumbeet station
Jan-06
Jan-03
Jan-00
Jan-97
Jan-94
Jan-91
Jan-88
Jan-85
Jan-82
Jan-79
Jan-76
Jan-73
Jan-70
Jan-67
Jan-64
Jan-61
Jan-58
Jan-55
Jan-52
Jan-49
-213
308
Page 8 of 14
Environ Monit Assess (2016) 188:308
Fig. 9 Scenario 1-continuing dry years
climate and land use change. A calibrated but not verified model can be used to make predictions as long as careful sensitivity analyses of both calibrated model and predicted model are performed and evaluated. Predictions resulting from calibrated but unverified
Fig. 10 Scenario 2–5 years wet climate
models generally will be more uncertain than predictions derived from the verified models. Two major pitfalls involved in making predictions are uncertainty in the calibrated model and uncertainty about the hydrologic stresses.
Environ Monit Assess (2016) 188:308
Page 9 of 14 308
379.5
379.0
Lake height (AHD) m
378.5
378.0
377.5
377.0
376.5 Dec-73
Jan-78
Feb-82
Observed level
Mar-86 May-90
Jun-94
Calculated level
Jul-98
Sep-02 Oct-06
Max outlet height
Nov-10
Min outlet height
Fig. 11 Scenario 3a-simulated lake levels for 5 years dry period with reduction in the waste treatment plant discharged to Burrumbeet Creek
The models can be calibrated to known existing conditions in the flow system and then used to predict with a different set of historical climatic conditions. At Lake Burrumbeet, the model is calibrated to conditions at the lake prior to the recent past drought, and then used to simulate the response of the system to various proposed altered historical climatic conditions, as well as alteration of flow from Burrumbeet Creek, including that due pumping of wells from the Cardigan bore field. Before a historical predictive simulation can be performed, reasonable values for model/system parameters (such as aquifer hydraulics and conductivity, lake bed conductivity and other boundary conditions) are chosen based on the previous model that matches the lake levels and salinity. This process of verification to past conditions is critical to producing meaningful predictions.
The model was verified against the following data: & & & & &
Lake level; Lake volume; Lake surface area; Surface outflow; and Lake salinity (refer to Yihdego and Webb 2012, 2015).
The results of the calibrations in lake level are discussed below and demonstrated a genuine correspondence between the behaviour of the model and the reality. For the surface outflow calibration, the outflow from the lake to Baillie Creek which eventually joins Emu Creek was compared with the recorded flow data, about 30 km at Mount Emu Creek at Skipton further downstream (Fig. 1). The peaks and overall pattern of flow in Mount Emu Creek
308
Environ Monit Assess (2016) 188:308
Page 10 of 14
379.5
379.0
Lake height (AHD) m
378.5
378.0
377.5
377.0
376.5 Dec-73
Jan-78
Feb-82
Observed level
Mar-86 May-90
Calculated level
Jun-94
Jul-98
Sep-02 Oct-06
Max outlet height
Nov-10
Min outlet height
Fig. 12 Scenario 3b-simulated lake levels for 5 years wet period with reduction in the waste treatment plant discharged to Burrumbeet Creek
match closely the simulated outflow from the lake (Fig. 6). Lake area Calibration of lake surface area was carried out by comparing the independent data set of the lake area estimated from Landsat images by ENVI software with the lake area calculated during modelling (Table 1). Thus, during the period when the simulated and observed lake levels differ by 10 cm, the model has been able to reproduce the lake area with less than 3 % error (Table 1). Similarly, the simulated and estimated lake volumes for the corresponding dates overestimate/underestimate the volume less than 2 % higher/lower than observed. The difference could be attributed, among other things, to the uncertainties on the lake bottom morphology from which the lake level area-volume expression was derived.
Lake level/stream discharge The calibrated model was simulated backward from 1998 (the earliest year with lake level data) to 1976 to calculate the lake levels over this period (Fig. 7); the extended modelling period includes additional lake salinity data back to 1991. To verify the model, the outflow from the lake to Baillie Creek, which joins Mount Emu Creek, was compared with the recorded flow data in Emu Creek, about 30 km downstream at Skipton (gauge no. 236203; Fig. 1). The flow in Mount Emu Creek closely matches the modelled outflow from the lake (Fig. 6), confirming the validity of Lake Burrumbeet water budget model.
Model historical validation Lake Burrumbeet rainfall has declined by 15 % over the last 30 years (before the drought broke in 2010),
Environ Monit Assess (2016) 188:308
Page 11 of 14 308
379.5
379.0
Lake height (AHD) m
378.5
378.0
377.5
377.0
376.5 Dec-73
Jan-78
Feb-82
Observed level
Mar-86 May-90
Calculated level
Jun-94
Jul-98
Sep-02 Oct-06
Max outlet height
Nov-10
Min outlet height
Fig. 13 Scenario 4a-simulated lake levels due to the effect of climate change and wet period rainfall
resulting in a >90 % fall in average flows in to Lake Burrumbeet. A non-linear relationship exists between climate change and the water resources system, such that a small change in climatic parameter would cause a great variation in hydrologic regime. A 10 % change in precipitation means 15–25 % reduction in stream flow, and a 10 % reduction in precipitation combined with 2 °C rise in temperature (and hence evaporation) means a 25–35 % reduction in regional runoff. Using the updated model, several simulations were carried out by selecting wet and dry climate periods of below and above average rainfall, from the recent past; 1976– 1981 and 2002–2007 were selected as wet and dry periods, respectively (see Fig. 8). Scenario 1-continuing dry years with low rainfall and stream flow. To run the model, rainfall, evaporation and stream flow, data were used from 2002 to 2007. The results show that there would be a
limited recovery in the lake level and stays dry for most of the time (Fig. 9). Scenario 2-return to wet years like the 1976–1981 data set. This scenario shows that it would actually only need about two consecutive years of normal winter rainfall and stream flow for the lake to refill and overflow (Fig. 10). Scenario 3-under wet and dry climates, the impact of altering flow from Burrumbeet Creek by diverting the outflow from Ballarat waste treatment plant (6ML/day) to Lake Wendouree (Figs. 11 and 12). The inflow is estimated to provide approximately 10 % of the mean annual stream flow volume entering Lake Burrumbeet at present. It is only a minor component of the Lake Burrumbeet water budget (∼35 % of average annual inflow, during wet years). The simulations Figs. 11 and 12 support this in comparison with Figs. 9 and 10, respectively; it
308
Environ Monit Assess (2016) 188:308
Page 12 of 14
379.5
379.0
Lake height (AHD) m
378.5
378.0
377.5
377.0
376.5 Dec-73
Jan-78
Feb-82
Observed level
Mar-86 May-90
Calculated level
Jun-94
Jul-98
Sep-02 Oct-06
Max outlet height
Nov-10
Min outlet height
Fig. 14 Scenario 4b-simulated lake levels due to the effect of climate change and dry period rainfall
has little effect on the lake level (1.5 m, averaging 220 mm annually. Mean monthly local rainfall (Burrumbeet), lake inflow and lake level rises show an approximate monthly time lag (Fig. 15), indicating the natural catchment response for the time taken to restore soil moisture deficit after the dry season. Although the lakes in these regions were shrank and dried in the recent past drought, they will enlarge in years of high flow. Since 1998, the weather has become much drier, and flows entering the lake have decreased. The lake dried up at least six times (on monthly basis) during the 2004 and almost every month since 2005 before the drought broke in 2010. However, Lake Burrumbeet was also refilled for some time during
Page 13 of 14 308
60
0.15
50
0.1
40
0.05
30
0
20
-0.05
10
-0.1
0
Change in lake level (m)
Flow (ML/d) & Precipitation (mm)
Environ Monit Assess (2016) 188:308
-0.15 0
2
4
6
8
10
12
14
Month
Precipitation
Inflow of the Burrumbeet Creek
Variation in lake level
Fig. 15 Mean monthly Burrumbeet Creek flow, rainfall at Burrumbeet and lake level change
heavy rains in 2004 (Fig. 4), indicating for the possibility to restore the lake level if the hydrological condition prevails. Currently in year 2016, the lake has refilled again and is 1.1 m below full, and the water will continue to reduce with the warmer weather (Lake Burrumbeet City of Ballarat 2016).
extreme climate variables (wet, dry, average, climate change), to fluctuations in surface water resources, and improving understanding of the lake-climate/ flow relationship. Acknowledgments Support was provided by the Glenelg–Hopkins Catchment Management Authority.
Conclusion References The base model has been calibrated and verified against lake level, surface outflow lake area/storage and lake salinity (Yihdego and Webb 2012, 2015) and demonstrated a genuine correspondence between the behaviour of the model and the reality, and further validated in this study, using historical data over 5 years (2008–2013), under different climatic and flow conditions. Such approach adds confidence on using the calibrated model for prediction as an instrumental tool and for decision being constrained and matched against historical calibration, verification and validation data set. This model will contribute towards better wetland management such as increase or decrease withdrawal, diversion, drainage control, salinity mitigations, which require accurate knowledge of the relative contributions of diverted/discharge flows and
Coram, J. E. (1996). Groundwater-surface water interactions around shallow lakes of the western district plains. Victoria: MSc Thesis, Department of Earth Sciences, Melbourne University. 179 pp. CSIRO (2001). Climate change: projections for Australia CSIRO Climate Impact Group Dewi, G. C. K., Jones, R. N, Kent, D., Leahy, P. (2009). Modelling lake levels under climate change conditions: three closed lakes in Western Victoria. Proceedings 18th World IMACS / MODSIM Congress, Cairns, Australia 13–17 July 2009, 6 pp. accessed from http://mssanz.org. au/modsim09 on the 8/1/2010. Dinka, M. O., Loiskandl, W., & Ndambuki, J. M. (2014). Hydrologic modelling for Lake Basaka: development and application of a conceptual water budget model. Environmental and Monitoring Assessment, 186(9), 5363–79. Enwright, N., Hudak, P. F., & Hatley, D. (2011). Estimating inundation patterns in detention wetlands: methodology and
308
Page 14 of 14
application to north-central Texas. International Journal of Environmental Studies, 68(1), 61–71. Haghighi, A. T., & Kløve, B. (2015). A sensitivity analysis of lake water level response to changes in climate and river regimes. Limnologica, 5, 118–130. Hudak, P. F. (2014). Inundation patterns and plant growth in constructed wetland characterized by dynamic water budget model. Environmental Earth Sciences Journal, 72(6), 1821– 1826. Hughes, L. (2003). Climate change and Australia: trends, projections and impacts. Australian Ecology, 28(3), 423–443. Juma, D. W., Wang, H., & Li, F. (2014). Impacts of population growth and economic development on water quality of a lake: case study of Lake Victoria Kenya water. Environmental Science and Pollution Research International, 21(8), 5737–46. Lake Burrumbeet City of Ballarat (2016). http://www.ballarat.vic. gov.au/lae/lakes/lake-burrumbeet.aspx. Accessed 12 March 2016. Tweed, S., Leblanc, M., & Cartwright, I. (2009). Groundwatersurface water interaction and the impact of a multi-year drought on lakes conditions in South-East Australia. Journal of Hydrology, 379, 41–53. Weijs, S. V., Schoups, G., & van de Giesen, N. (2010). Why hydrological predictions should be evaluated using information theory. Hydrology and Earth System Sciences, 14, 2545– 2558. Wu, Y., Cheng, D., Yan, W., Liu, S., Xiang, W., Chen, J., Hu, Y., & Wu, Q. (2014). Diagnosing climate change and hydrological responses in the past decades for a minimally-disturbed headwater basin in South China. Water Resources Management, 28(12), 4385–4400. Yihdego, Y., & Becht, R. (2013). Simulation of lake–aquifer interaction at Lake Naivasha, Kenya using a three-
Environ Monit Assess (2016) 188:308 dimensional flow model with the high conductivity technique and a DEM with bathymetry. Journal of Hydrology, 503, 111–122. Yihdego, Y., & Webb, J. A. (2011a). Modelling of lake level and salinity for Lake Burrumbeet in western Victoria. Civil engineering transactions. Institute of Engineers Australia, CE8(1), 83–97. Yihdego, Y., & Webb, J. A. (2011b). Modeling of bore hydrograph to determine the impact of climate and land use change in a temperate subhumid region of south-eastern Australia. Hydrogeology Journal, 19(4), 877–887. Yihdego, Y., & Webb, J. A. (2012). Modelling of seasonal and long-term trends in lake salinity in South-western Victoria, Australia. Journal of Environmental Management, 112(2012), 149–159. Yihdego, Y., & Webb, J. A. (2013). An empirical water budget model as a tool to identify the impact of land-use change in stream flow in southeastern Australia. Water Resources Management Journal, 27(14), 4941–4958. Yihdego, Y., & Webb, J. A. (2015). Use of a conceptual hydrogeological model and a time variant water budget analysis to determine controls on salinity in Lake Burrumbeet in southeast Australia^. Environmental Earth Sciences Journal, 73(4), 1587–1600. Yihdego, Y., Webb, J. A., & Leahy, P. (2015). Modelling of lake level under climate change conditions: Lake Purrumbete in south eastern Australia. Environmental Earth Sciences Journal, 73, 3855–3872. Yihdego, Y., Webb, J.A., & Leahy, P. (2016). Modelling of water and salt balances in a deep, groundwater-throughflow lake— Lake Purrumbete southeastern Australia. Hydrological Sciences Journal, 6(1), 186–199. http://dx.doi.org/10.1080/ 02626667.2014.975132.
