You are on page 1of 9

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/360783371

The Impacts of Climate Change and Reforestation on Future Water Availability


of the Collie River Catchment, Western Australia

Conference Paper · May 2011

CITATIONS READS

0 23

3 authors:

Mohammed Bari Richard P. Silberstein


Bureau of Meteorology Edith Cowan University
114 PUBLICATIONS 1,368 CITATIONS 140 PUBLICATIONS 3,082 CITATIONS

SEE PROFILE SEE PROFILE

Santosh Aryal
The Commonwealth Scientific and Industrial Research Organisation
47 PUBLICATIONS 1,254 CITATIONS

SEE PROFILE

All content following this page was uploaded by Mohammed Bari on 23 May 2022.

The user has requested enhancement of the downloaded file.


The Impacts of Climate Change and Reforestation on Future Water
Availability of the Collie River Catchment, Western Australia

M. A. Bari1, R. P. Silberstein2, S. K. Aryal2


1
Climate and Water Division, Bureau of Meteorology, PO Box 1370, West Perth, W.A. 6872.
2
CSIRO Land and Water, Private Bag No. 5, Wembley WA 6913.
E-mail: m.bari@bom.gov.au

Abstract: The Collie River catchment (2830 km2) is one of the five Water Resources Recovery
Catchments in Western Australia. About 10,000 ha of land was reforested for salinity reduction during
the 1990s. This together with the reduction in rainfall over the last 30 years has resulted in a
noticeable decrease in the average inflow to the reservoir. General Circulation Models (GCM) predict
further reductions in rainfall in the future. The LUCICAT model was applied to the catchment to assess
the impact of climate change and reforestation on future water availability in the form of reservoir
inflows. Three scenarios were modelled (i) the historical sequence representing the 1975-2007 climate
(Scenario A) (ii) recent climate representing the 1997-2007 climate (Scenario B), and (iii) future
climate representing a 33-year climate series derived from different GCM projections of ~2030 climate
(Scenario C). Results showed that there could be a further 9% reduction in rainfall and
approximately30% reduction in inflows into the reservoir by 2030.

Keywords: Collie River, LUCICAT model, Climate change, Reforestation, Water availability, Rainfall-
Runoff modelling.

1. INTRODUCTION
Collie River catchment (2830 km2) was the main water supply catchment for the Great Southern
Towns Water Supply Scheme (GSTWS) and to the Collie River Irrigation District in the south west of
Western Australia. In 1933 the Wellington Dam with a storage capacity of 186 Giga Litres (GL) was
built on the river. As the salinity of the reservoir increased to an unacceptable level due to clearing of
native forest another reservoir (Harris) was built in 1990 to replace the supply to GSTWS. The
Western Australian Government enacted legislations to control further large-scale agricultural clearing
and a rehabilitation program was initiated to return the catchment as a potable water supply condition
for the future. The rehabilitation program includes (i) planting of deep-rooted trees, (ii) management of
remnant vegetation, and (iii) improved agricultural practices. More than 10,000 hectares of cleared
and salt affected land were reforested with deep-rooted trees (Mauger et al., 2001).

Mean annual rainfall of the catchment has decreased significantly over the last 30 years compared to
the last 100-year average. A combination of forest clearing, reforestation and possibly lower rainfalls
possibly due to the effects of climate change may have significant impact on the streamflow of the
Collie River catchment. The General Circulation Models (GCM) project further decrease in rainfall
(relative to the 1975-2007 period) in the south-west of Western Australia in the next 30-50 years (IOCI,
2002). At the same time, demand in water resources will increase due to the population growth,
agricultural and industrial development in the south-west.

In the south-west of Western Australia, there have been a few studies assessing the impacts of
climate change on water resources (Bari et al., 2005; Charles et al., 2007; Kitsios et al., 2008; Smith et
al., 2009). However, most of these studies were focused on particular catchments or specific climate
scenarios. This paper focuses on the impacts of reforestation and climate change, as derived from the
15 GCMs and recent climates, on streamflow through the application of LUCICAT (Land Use Change
Incorporated CATchment) model (Bari & Smettem 2006a, b).

2. CATCHMENT DESCRIPTION
The Collie River catchment is situated about 200 km south of Perth (Figure 1). Mean annual rainfall
within the catchment ranges from 600 mm to about 1200 mm and annual pan evaporation ranges from
1650 mm to 1400 mm. The native forest of the Collie River catchment is dominated by Jarrah
(Eucalyptus marginata) and Marri (E. calophylla). Most of the cleared areas are under agricultural
crops and pastures. Reforestation by the Government has mainly occurred in the eastern lower rainfall
zones of the catchment (Figure 1). The soil pattern of the Collie River catchment outside the Collie
Coal Basin is controlled by the topographic position. The basement of the catchment is comprised of
crystalline rocks in the Archaean shield. The soil profile consists of a highly conductive 2-4 m thick
surface soil, overlying about 20m of less permeable subsurface materials produced by in situ
weathering of the basement material. Rainfall intensity generally does not exceed the infiltration
capacity of the surface soils. The Collie Coal Basin contains sequences of Permian sediments
including coal seams.

Figure 1 Location of the Collie River catchment (Bari & Smettem, 2003)

3. METHODOLOGY
This study investigated the potential impacts of projected climate change and reforestation on long
term water availability of the Collie River catchment in Western Australia using the LUCICAT rainfall
runoff model. The three main components to this study are: (a) calibration of the LUCICAT model
using observed streamflow data (b) simulation of the impacts of projected climate change scenarios
(A, B and C) on streamflow (c) analysis and assessment of the projected streamflow data.

3.1. Projection Scenarios


The climate scenarios and generation of synthetic climates from GCM forecasts are described in more
detail elsewhere (Charles et al, 2009). Scenario A is based on the historical climate sequence where
rainfall and potential evaporation (PE) for 1975-2007 was repeated from 2008 onward. Under this
scenario, surface water development and allocations remained unchanged at the 2007 level. This is
consistent with the current planning of the Department of Water in Western Australia. The second
scenario (B) based on ‘recent climate’ – rainfall and PE data of recent 11 years (1997-2007) were
repeated three times from 2008. This scenario is more likely to happen and therefore used to assess
the water availability in the future if the climate remains similar to the current decade. The third
scenario (C) is based on the forecast by 15 GCMs for a low (0.70C), medium (1.00C) and high (1.30C)
global warming over the next 30 years (CSIRO, 2009). The projected increases in temperature and
reductions in rainfall were applied to scale the historical climate (Scenario A rainfall and Morton wet
surface potential evaporation) to generate a future climate sequence of 45 runs (Charles et al., 2009).
Land use remained the same for scenarios A, B and C. Results are presented for wet extreme,
median and dry extreme future climates (Cwet, Cmid and Cdry representing 90th, 50th and 10th percentile
of rainfall or runoff respectively).

4. LUCICAT MODEL CALIBRATION

4.1. The model


The LUCICAT model is a distributed lumped conceptual hydrological model. A large catchment is
divided into smaller Response Units (RU) to take into account the spatial distribution of rainfall, pan
evaporation and land use. Each of the RU is represented by the ‘open-book’ configuration and a
fundamental ‘building-block’ model is applied. Catchment attributes such as soil depth, rainfall, pan
evaporation, land use change and groundwater levels are incorporated into the building-block model
(Bari & Smettem 2006a, b).

The building-block model consists of: (i) Dry, Wet and Subsurface stores, (ii) a saturated Groundwater
Store, and (iii) a transient Streamzone Store. The transient Streamzone Store represents the
groundwater induced ‘saturated areas’ along the stream zone. The fluxes between the top layer Dry
and Wet Stores represent the water movement in the unsaturated zone. The dynamically varying
saturated stream zone areas are responsible for surface runoff. The Groundwater Store controls the
groundwater flux to stream zone. Generated flow from each of the RU is routed downstream by the
Muskingum-Cunge routing scheme. Water and salt balances of the lakes and reservoirs in the
catchment are also computed.

4.2. Model setup and Calibration


The model setup involved data preparation, delineating the Response Units, calibration and validation.
Previous studies were reviewed, and streamflow data and rainfall data in the catchments collated
(Mauger et al., 2001, Blake & Owens, 2010). The Collie River catchment was divided into 126
Response Units, based on digital elevation models. Daily rainfall and Morton Wet Surface Potential
Evaporation (Morton, 1983) series on a 5 km grid was obtained from the Bureau of Meteorology and
Queensland Department of Natural Resources’ SILO data set
(http://www.longpaddock.qld.gov.au/silo/) and then calculated for each of the Response Units using
the reciprocal distance weighting method. Response Unit attributes, stream characteristics, surface
topography, land use history and Leaf Area Indices were developed using ArcGIS. The model was
run using LUCICAT Live framework (Bari et al., 2009) and was calibrated against the available
streamflow data obtained from all gauging stations within the catchment including the reservoirs.

4.2.1. Model Calibration


The LUCICAT Live framework provides a visual interface for LUCICAT calibration. The framework
displays a series of graphs and shows tables of statistical criteria to facilitate performance evaluation
(Bari et al., 2009). The model was calibrated for all catchments by trial and error method based on
statistical criteria including Nash-Sutcliffe Efficiency (NSE), Correlation Coefficient (CC), Explained
Variance (EV), Flow-period Error Index (EI) and Water Balance Error (E). In some instances, NSEs
for daily streamflow were lower than 0.5 but most of the time ranged between 0.6 to 0.85. However,
for monthly and annual streamflow, NSE increased above 0.8. For daily streamflow CC, EI and E were
in the range of 0.5-1.0, 0.7-1.2 and ±0.10 respectively for all seven gauging stations within the
catchment. For monthly streamflow, CC and EV improved significantly were above 0.9 for most of the
gauging stations. The comparison of the observed and predicted streamflow at the James Well
gauging station is shown in Figure 2 as an example. The modeled mean annual runoffs from all 7
gauging stations were within ±0.10 of the observed flows. The annual average inflows to Wellington
and Harris reservoirs calculated from the Water Balance method were 125 GL, 20 GL respectively; the
same from the LUCICAT calibration were and127 GL, 20 GL respectively. The model predicted the
low (10th percentile) and high (90th percentile) of the reservoir inflows very well.
612025 612025 612025
200 50 100
NSE = 0.89 NSE = 0.90 Observed

Modelled monthly runoff (mm)


40 Modelled
Modelled annual runoff (mm)

10
150
NSE = 0.77

Daily runoff (mm)


30 1
100
20 0.1

50
10 0.01

0 0 0.001
0 50 100 150 200 0 10 20 30 40 50 0 25 50 75 100
Observed annual runoff (mm) Observed monthly runoff (mm) Percent time daily runoff is exceeded

Figure 2 Relationship between observed and modelled runoff (a) annual, (b) monthly, and (c)
daily flow-duration at the James Well catchment.

612004
612025
100
80 20
60 15
40 10
Error (mm)..

Error (mm)..

20 5
0 0
-20 -5
-40 -10
-60 -15
-80
-20
-100
-25
1975 1985 1995 2005
1975 1985 1995 2005

Figure 3 Distribution of annual runoff error (a) Hamilton River, (b) James Well catchments

Table 1 Model performance representing climate and land use changes

Catchment Land use Trend Mean annual runoff (mm)

Scenario Cmid
Scenario A

Scenario B
Calibration
Observed

Hamilton River Clearing of about 5% of the catchment area during None 186 176 138 120 105
1950-77. Operational logging and regeneration. Some
pine plantations.
Bingham River Clearing of about 2% of the catchment area during None 9.1 9.3 8.4 7.9 4.3
Stenwood 1950-77. Operational logging and regeneration.
James Well Clearing of about 22% of the catchment area during None 35 38 23 19 15
1950-77. Most of the cleared area replanted by 2000.
Some tree harvesting, perennial pastures.
James Crossing Clearing of about 50% of the catchment area during None 45 48 41 39 29
1950-77. Some of the cleared area replanted by 2000.
Some tree harvesting, perennial pastures.

4.3. Model validation


One of the possible ways of checking how a model represents non-stationary hydrological processes,
particularly due to climate and land use changes, is to analyse prediction errors (the difference
between predicted and observed annual runoff) and find any temporal trends. Errors in annual runoffs
were analysed in some selected catchments based on: (1) no significant land use change to see how
the model represents recent low rainfall and runoff generation, and (2) large amount of forest clearing
and tree planting to see impacts of rainfall and land use changes on runoff generation (Table 1). In
most of the cases there were no time trends in error distributions. Hamilton River is a forested
catchment where some operational logging took place but no significant forest clearing or
reforestation. While there were significant areas of cleared land which were planted in the James Well
catchment. Error distribution in these two catchments was well-balanced and no trends were found
(Figure 3) indicating that the model represented non-stationary land use and rainfall changes very
well.

5. CLIMATE CHANGE SCENARIO PROJECTIONS


The calibrated LUCICAT model was then run to assess the change in yield that could occur following
a change in climate and land use. The rainfall and Morton PE data was extended to 2040 and all the
management scenarios were introduced in 2008. The Leaf Area Indices and land use remained
unchanged from 2008 onward.

5.1. Rainfall
In Scenario A rainfall during 2008-2040 remains the same as 1975-2007. However, if the ‘recent
climate’ continues (Scenario B) the mean annual rainfall (2008-2040) is predicted to decline (relative
to 1975-2007) in all gauged sub-catchments within the Collie River, with the largest (7%) being in the
Bingham River catchment. Under three different warming scenarios (Scenario C) average annual
rainfall reductions were 6, 9 and 12% respectively. The largest projected reduction of 142 mm would
be at the Hamilton River Worsley Catchment, near the Wellington Reservoir.

5.2. Climate change and reforestation


Reforestation is an important feature of the land use in some of the catchments in the south-west.
More than 10,000 hectares of the cleared areas of the Collie River catchment were planted with trees,
particularly for stream salinity control and management. In some of the forested catchments,
particularly James Well and James Crossing, Scenario A runoff was significantly lower than that of the
calibration period, which could be attributed to the higher LAI in 2007 and greater proportion of cleared
area planted (Table 1), which were kept unchanged for the 2008-2040 projection period. Most of the
cleared areas in the James Well catchment were reforested in the late 1990s (Mauger et al., 2001).
Daily flow-duration was about 50% most of the time and the model projected a decrease to
approximately 40% in future under all climate change scenarios and reforestation. Results are similar
to other applications of the model (Dixon and Bari, 2008; Smith et al., 2006; Bari et al., 2005).

5.3. Streamflow and catchment storage


Under Scenario B spatial average annual rainfall at the Mungalup Tower, upstream gauging station of
the Wellington reservoir (Figure 1), is projected to be 3% lower which translates to a 11% reduction in
streamflow. For other gauging stations, projected reductions in streamflow ranged from 5 to 23%.
Under Scenario C a large range of variations in streamflow reductions were observed, mainly due to
variations in 15 GCMs’ rainfall projections. However, all GCMs projected lower rainfall in all future
scenarios meaning lower streamflow as well. Streamflow response to different climate scenarios at the
Mungalup Tower is shown as an example (Figure 4). Monthly runoff is projected to decline, the largest
being in July-August when most of the runoff being generated (Figure 4). Under scenario Cdry,
projected runoff reductions range from 50-72% between different gauging stations. In some instances
Cwet runoff is projected to be greater than Scenario B (Figure 5a). However, if the median climate
prevails in the future (Cmid), streamflow at Mungalup Tower is expected to be 32.5% lower and only a
few years projected annual streamflow would exceed that of Scenario A (Figure 4).

The impacts of lower rainfall in the future climate scenarios are also evident in conceptual
groundwater levels and catchment storages. Both the groundwater level and catchment-average soil
water storage declined in scenarios B and C. The conceptual groundwater level declines under the
native forest and planted areas in the range of 0.4 to 0.8 m. Similar trends are also evident in the
reduction of catchment average soil water storage. However, the soil water storage of most of the
catchments appears to have stabilized by 2030.
In general, flow durations are projected to decrease in all gauging stations under all future climate
scenarios (e.g. see Figure 4). That means streams would flow less frequently during the year than the
1975-2007 period. Reduction in flow-duration has implications on stream zone ecology, flora and
fauna and environmental water allocation and planning.
250 12
Mean monthly rainfall/APET (mm)

10
APET C range

Mean monthly runoff (mm)


200 Rainfall 10 Cmid C range
1

Daily runoff (mm)


B Cmid
8
150 A B
6 0.1 A
100
4
0.01
50 2
0.001
0 0
0 25 50 75 100
J F M A M J J A S O N D J F M A M J J A S O N D Percent time daily runoff is exceeded

Rainfall 140
Runoff 100 Deviation from Scenario A mean
1000

Annual runoff divergence (mm)


900 120
Annual rainfall (mm)

50
Scenario A

Annual runoff (mm)


100
800
80
700 0
60
600
40 -50
500 20

400 0 -100
1 5 10 15 20 25 30 33 1 5 10 15 20 25 30 33 1 5 10 15 20 25 30 33

1000 140 100

Annual runoff divergence (mm)


900 120
Annual rainfall (mm)

50
Annual runoff (mm)

100
800
Scenario B

80
700 0
60
600 40 -50
500 20

400 0 -100
1 5 10 15 20 25 30 33 1 5 10 15 20 25 30 33 1 5 10 15 20 25 30 33

1000 140 100


Annual runoff divergence (mm)

900 120
Annual rainfall (mm)

50
Scenario Cmid

Annual runoff (mm)

100
800
80
700 0
60
600 40 -50
500 20

400 0 -100
1 5 10 15 20 25 30 33 1 5 10 15 20 25 30 33 1 5 10 15 20 25 30 33

Year since 2007


Figure 4 Climate change scenarios and its impacts on streamflow at Mangalup Tower

80 80
(a)
(b)
y = 3.84x
Runoff reduction (%)

60 60 R2 = 0.85
Runoff reduction(%)

Wellingto n Dam
Hamilto n River
James Well
B ingham River
40 40

20 20

0 0
A B Cw et Cm id Cdry 0 5 10 15
Rainfall reduction (%)

Figure 5 Runoff reductions (a) under different climate change scenarios, and (b) relationship
with rainfall reduction
Runoff reduction has a strong relationship with the corresponding rainfall reduction. Larger
proportional reductions were reported for the low-runoff producing catchments. In the Cdry climate
change scenario, runoff from Bingham River catchment, located in the low rainfall zone was projected
to decrease by approximately 70% compared to 40% of the Hamilton River, located in the high rainfall
zone of the catchment (Figure 5a). For all the gauged catchments in the Collie River, on average
runoff reduction is approximately 4 times rainfall reduction (Figure 5b). Similar results were also found
in other studies in Western Australia (Charles, et al., 2007; Kitsios et al, 2008; Smith, et al., 2009).

5.4. Reservoir inflow


Inflows to the Wellington and Harris reservoirs were computed for all scenarios. Projected mean
annual inflows (2008-2040) into the two reservoirs under Scenario A were within ±2% of 1975-2007
mean. The variation was mainly due to static LAI and different initial conditions in 2008, when all the
scenario simulations started. The impact of climate change is also evident through the changes in
inflow statistics for the future. Under Scenario B inflow to Wellington reservoir is projected to decrease
from 125 GL to 110 GL. Under Cmid climate scenario the annual inflow to the reservoir is projected to
be 32% lower ranging from 11% (Cwet) to 50% (Cdry). Projected reduction in inflow to Harris reservoir
was similar to the Wellington reservoir. Average inflow is projected to decrease from current 21 GL
(Scenario A) to 15 GL (Cdry). The magnitude of this reduction in streamflow highlights the serious
implications of a change in climate for the abundance and availability of surface water resources.

6. SUMMARY AND CONCLUSIONS


A simple distributed conceptual water balance model, LUCICAT, was calibrated and validated against
observed data at major gauging stations within the Collie River catchment. The predicted mean annual
inflow to Wellington and Harris reservoirs was within ±2% of the reverse Water Balance method. The
predicted streamflow was within ±10% of the observed data for all the gauged sub-catchments.

Three climate change scenarios were used to project future catchment yield and reservoir inflow. The
historical rainfall during 1975-2007 was extended and was considered as the reference base case (A).
The second scenario (B) represented the more recent 1997-2007 climate and was repeated 3 times to
represent climate to 2040. For all gauged sub-catchments, scenario B rainfall was less than 1975-
2007 mean with the largest reduction of 69 mm being in the Hamilton River Worsley catchment, near
to the Wellington Reservoir. The third scenario (C) represents climates of 15 different GCM projections
including low (0.70 C), medium (1.00 C) and high (1.30 C) global warming by 2030s. Rainfalls in the
catchment are projected to decrease by 6, 9 and 12% for the three warming levels respectively.

The LUCICAT model was then used to predict the catchment yield for all three scenarios. In
Scenario A streamflow was slightly different from the calibration period, as the land use and LAI were
kept static at 2007 level rather than varied as through the calibration period. Under the recent climate
(Scenario B) total inflows to the Wellington and Harris reservoirs were projected to decrease to 110 GL
and 20 GL compared 1975-2007 simulated mean of 115 GL and 21 GL respectively. On average the
annual inflow to these two reservoirs under climate change (Cmid) is predicted to be approximately
32% and 30% lower than the inflow for scenario A. However, the reduction in runoff is not uniform
across the catchment. The largest proportional reduction is projected to be at the low rainfall-runoff
generating Bingham River catchment. Catchments with significant area of reforestation were projected
to have larger streamflow reductions.

The modelling set up could be used as an operational tool for water supply source planning, assessing
both land use management options for additional yield and the effects of climate variability on runoff.

7. ACKNOWLEDGMENTS
We are thankful to Steve Charles for rainfall series (from CSIRO) and to Geoff Mauger (from GIA Ltd)
for model preparation and data analyses. The Department of Water developed an earlier version of
the LUCICAT model for this catchment for salinity recovery work. We also acknowledge reviewers
whose feedback improved this paper - Renee Dixon of Department of Water and Katrina Annan of
Bureau of Meteorology.
8. REFERENCES

Bari, M.A., Shakya, D.M. and Owens, M. (2009), LUCICAT Live – A modelling framework for predicting catchment
management options, In MODSIM 2009, Cains, July 2009, Australia.

Bari, M.A. and Smettem, K.R.J. (2003), Development of a Salt and Water Balance Model for a large Partially
Cleared Catchment, Australian Journal of Water Resources, The Institution of Engineers, Australia, pp.93 -
99.

Bari, M.A., Berti, M.A., Charles, S. Hauck, E. and Pearcy, M. (2005), Modelling of streamflow reduction due to
climate change in Western Australia – A case study, In MODSIM 2005, Melbourne, December 2005,
Australia.

Bari, M.A. and Smettem, K.R.J. (2006a), A conceptual model for daily water balance following partial clearing
from forest to pasture, Hydrol. Earth Syst. Sci., 10, 321-337.

Bari, M.A. and Smettem, K.R.J., (2006b). A daily salt balance model for stream salinity generation processes
following clearing from forest to pasture, Hydrol. Earth Syst. Sci, 10, 519-534.

Blake, G.M. and Owens, M., (2010), Collie Modelling Report: technical documentation for the Collie Salinity
Recovery Plan, Department Of Water, Government of Western Australia.

Charles, S.P., Bari, M.A., Kitsios, A. and Bates, B.C. (2007), Effects of GCM bias on downscaled precipitation and
runoff projections for the Serpentine catchment, Western Australia, Int. J. Climatol., 27, 1673-90.

Charles, S.P., Silberstein, R.P., Chiew, F.H.S., Viney, N.R., Teng, J., Yang, A., Van Niel, T., and Smith, I., 2009
Future Climate Scenarios for the South-West Western Australia, Sustainable Yields Project. CSIRO.

CSIRO(2009), Surface water yields in south-west Western Australia, A report to the Australian Government from
the CSIRO South-West Western Australia Sustainable Yields Project, CSIRO, Australia.

Dixon, N.R.M and Bari, M.A. (2008), Modelling of Stream Salinity Management Options for the Warren River, In:
Hydrology and Water Resources Symposium, The Institution of Engineers, Adelaide, Australia, 2008.

IOCI (2002), Climate variability and change in south-west Western Australia, Indian Ocean Climate Initiative, 34p.

Kitsios, A., Bari, M.A. and Charles, S.P. (2008), Projecting impacts of climate change on the Serpentine
catchment – Donscaling from multiple General Circulation Models, Department of Water, Western Australia,
Water Resource Technical Series, No. WRT36, 87p.

Mauger, G.W., Bari, M.A., Boniecka, L, Dixon, N.R.M. Dogramaci, S.S. and Plat, J. (2001). Salinity situation
statement – Collie River. Water and Rivers Commission, Water resource technical report series No. WRT29,
108pp.

Morton, F.I. (1983), Operational estimates of areal evapotranspiration and their significance to the science and
practice, J. of Hydrology, J. Hydrol., 66, 1-76.

Smith, R., Bari, M.A., Rowlands, D. and Dixon, N.R.M. (2006), Salinity Situation Statement: Helena River,
Department of Water, Division of Water Resources Management, Water Resources Technical Series, 177pp.

Smith, K., Boniecka, L., Bari, M.A. and Charles, S.P. (2009). The impact of climate change on rainfall and
streamflow in the Denmark River catchment, Western Australia, Department of Water, Water Resource
Assessment Branch, Rep. HY30, 58p.

View publication stats

You might also like