- Email: [email protected]
The relationship between climate and streamflow in the Namoi Basin
Environment International,
Vol. 25, No. 6/7, pp. 827439, 1999 Copyright 01999 Elsevier Science Ltd Printed in the USA. All rights reserved 01604120/99/$-see front matter
Pergamon
PI1SO160-4120(99)00048-3
THE RELATIONSHIP BETWEEN CLIMATE AND STREAMFLOW IN THE NAM01 BASIN P.F. Crapper CSIRO Land and Water, GPO Box 1666, Canberra, ACT 2601, Australia
S.G. Beavis and Li Zhang CRES, Australian National University, Canberra, ACT 0200, Australia
EI 9808-201 M(Received
31 August 1998; accepted I3 March 1999)
The Namoi River Basin (42000 km’) is located in the Murray-Darling Basin, west of the Great Australian Dividing Range in northeast New South Wales (NSW), and includes some of the most fertile agricultural lands in Australia. One of the environmental concerns for this basin is erosion and its effects on downstream water quality. Models that relate climate, land use, and these concerns require measurements of climate (rainfall and temperature) and streamflow. These measurements were examined as a preliminary to the modelling. The residual mass technique was used to examine the temporal variation of annual rainfalls over the approximately 100 years of available data, and significant spatial variations were found in annual rainfall trends over the catchment. Streamflow was examined at key river gauging stations. The impact ofrecent large-scale irrigation operations was clearly observed. The impact of changing land use and land management on runoff ratios was examined for eight subcatchments. Temperature variations were examined for the four major towns in the catchment. The duration of the temperature data is too short to make any comments about long-term trends. Significant variations were observed in an east to west direction. 01999 Elsetier Science Ltd
INTRODUCTION In recent years, the management of the water and land resources of the Namoi Basin have come under increasing public scrutiny. This has come about in part because of a wide belief that the present agricultural practices are not sustainable, and in part because of a concern for the apparent deterioration of water quality. The work described in this paper is part of a larger study on the sources and transport of sediments and nutrients in the Namoi Basin. Additional information on the land and water resources of the Namoi Basin is contained in Schroder and Glennon (1995) and the Namoi Community Catchment Plan (1996), and a location map is shown in Fig. 1.
Rainfall is probably the most widely measured meteorological parameter and is one of the major determinants of erosion. Jakeman et al. (1998) presented a framework to model and predict the effects of rainfall and land use on erosion and downstream water quality. A crucial input is accurate estimation of area1 rainfall. This is not easy, as it is temporal and spatial heterogeneity is high. The locations of rainfall measurement sites, within Australia at least, are generally selected for convenience, not representativeness. The Namoi Basin is acknowledged as one with a highly variable rainfall and as having extremely variable hydrologic characteristics (Findlayson and McMahon 1988). 827
P.F. Crapper et al.
828
w
Lagend .. . .
stream lines
_...
catchment
boundary
__ . ..’ . ...“.
KILOMETERS
Fig. 1. The location of the Namoi Basin.
RAINFALL
DATA
The rainfall data set for Gunnedah (Fig. 1) was examined for initial analysis because Gunnedah is approximately in the middle of the catchment and because it has the longest climate data set in the catchment. It is a composite data set from the New South Wales (NSW) Department of Land and Water Conservation, Gunnedah Research Station and the Australian Bureau of Meteorology at the Gunnedah Post Office. This data set, and data sets for approximately 16000 stations around Australia, are available on Metaccess (1996). The data set records daily rainfall for the period 1 January 1877 to 3 1 March 1997. The annual rainfall for this period is shown in Fig. 2, where the straight line of best fit is also plotted. Gunnedah has, in the last 120 y, experienced 3 y with rainfall in excess of 1000 mm (1879 - 1047 mm, 1890 - 1133 mm, and 1950 - 1150 mm) and 1 y with rainfall less than 250 mm (1946 - 247 mm). The average annual rainfall for the 120-y data set is 6 11.6 mm and the standard deviation is 17 1.1 mm, giving a coefficient of variation of 0.28. The line of best tit reveals, on average, a slow increase of about 0.4 mm/y in average annual rainfall
over this period. This apparent increase in annual rainfall is too small to be statistically significant and could easily be explained by replacement of existing rain gauges, relocation of rain gauges, or growth in surrounding vegetation. The cumulative sum (CUSUM) technique (MC Gilchrist and Woodyer 1975), also known as the residual mass technique, was used to examine the temporal sequence of rainfall. CUSUM techniques have proved to be a valuable tool in detecting intermediate term changes in the mean value of a sequence of regularly spaced observations. The CUSUM si can be defined as: , s, = c (x, - 2) j=l
where 2 =‘& I2 i=l
where xi is the i th observation in_a series of n regularly spaced observations with mean x . The CUSUM distribution is a normalised distribution and it reveals runs of observations greater than the long-term mean with a positive slope and less than the long-term mean with a negative v slone. Such oersistent nositive or negative 1 1 1
Climate and streamflow in the Namoi Basin, Australia
i iz
829
600
z $ 400
200
0 1877
1 a97 1887
1917 1907
1937 1927
1957 1947
1977 1967
1987
YEARS Fig. 2. Annual rainfall distribution at Gunnedah for the period 1877 to 1996. The line of best fit is also plotted.
slopes can be used to detect intermediate term changes in the mean value. The actual ordinate values are not relevant; it is the sign of the slope (i.e., either positive or negative) and the duration for which the sign of the slope remains the same that is important. The CUSUM distribution for the annual rainfall at Gunnedah is shown in Fig. 3. An examination of Fig. 3 reveals that annual rainfalls for the 116 y of record can be treated as 3 distinct average rainfall periods. The first period from 1877 to 1900 consisted of a period of greater than average rainfall, with some uncertainty in the 1880s; the second period from approximately 1900 to 1945 was a period of below average rainfall; and from 1945, there was a period of average rainfall, followed by above average rainfall in the early 1970s and then in the 1980s. These trends are apparent in Fig. 2 but more difficult to discern. Three additional rain gauge stations were examined in detail. These stations were Tamworth (average annual rainfall 661 mm), Narrabri (636 mm), and Walgett (472 mm) (see Fig. 1 for locations). For periods of no record, stations within a 20 km radius were examined and the average rainfall at these stations was used. A moving 5-y average of s, is plotted on Fig. 4. The locations Gunnedah, Tamworth, and Walgett show the
same three major trends in rainfall, i.e., a period of above average rainfall at the end of last century, followed by a period of below average rainfall until 1945 and then a period of above average rainfall up to the present. However, the durations of these periods varied for different locations. Narrabri has a rainfall behaviour significantly different from the other locations. Narrabri showed a period of above average rainfall at the end of the last century, followed by a period of approximately uniform rainfall up to 1935. This pattern was then followed by below average rainfall until 1945, then above average rainfall until 1955. From 1955 until 1980, the rainfall was approximately uniform and then there was a period of below average rainfall in the 1980s. The reasons for the different behaviour at Narrabri are discussed later in the temperature data section. The annual rainfall amounts presented in Figs. 2, 3, and 4 conceal all within-year variations. In Fig. 5, the average monthly rainfalls for each of the four locations for the entire record are presented. The monthly rainfalls for each location were normalised by the highest monthly rainfall for that location. The highest monthly rainfall was chosen as the normalising parameter so that all normalised monthly rainfalls would
P.F. Crapper et al.
830
BGUNNEDAH I 1880
1
oTAMb%ORTH I
1900 1890
I
I
1920 1910
tN4RRABRl I
I
1940 1930
,
+WALG.EIT ,
1960 1950
,
,
1980 1970
J
2000 1990
YEARS
-31 ’ 1677
’
’
1897 1007
’
’
1917 1907
’
’
1937 1927
’
’
1947
’
’
1977
1957 1967
1997
Fig. 4. A moving 5-y mean of si for Gunnedah, Tamworth, Narrabri, and Walgett.
1987
YEARS Fig. 3. A residual mass plot of Gunnedah annual rainfall.
fall between zero and one and thus allow easier comparison For all locations, it can be seen that January has the highest average rainfall; and, with the exception of Walgett in November and December, the monthly rainfall distributions appear similar. The rainfall for all locations is summer-dominated, with January being the highest rainfall month. This rainfall peak occurs because the summer rainfalls are produced by inflows of tropical air masses associated with tropical cyclones. Long-term annual average rainfall isohyets are shown in Fig. 6. There is an obvious decrease in rainfall from east to west. The highest rainfall occurs in the southeast corner of the catchment on the Peel River catchment. STREAMFLOW
DATA
Streamflow has been measured at 8 1 locations in the Namoi Basin, by the NSW Department of Land and Water Conservation and its predecessors, for varying periods of time using mostly natural controls, although there are a small number of concrete structures. There are also six weir gauges in the catchment measuring the depth of water in the weirs. The daily streamflow records for all these gauging stations are available on Pinneena (1996). Unfortunately, all stations contain some periods of no record and, unless otherwise stated, these periods were set equal to zero. There is no correct
way of handling the problem of missing data. Alternative approaches are to replace the missing values with the mean or median of weekly, monthly, seasonal, or annual values. Also a rainfall-runoff model can be used to construct the streamflow, although in this case, rainfall and streamflow data would be needed for the calibration period and rainfall for the simulation period. In the present case, the simplest approach of replacing the missing values with zeros was chosen, as the accuracy of these streamflows will be subsequently checked using water balance techniques. The minimum main channel streamflow gauging station data set is shown diagrammatically in Fig. 7, and the station names and periods of record are given in Table 1. This data set includes the four major tributaries to the Namoi River (Peel River, Mooki River, Coxs Creek, and Maules Creek) and the river gauging stations along the main river channel, with the exception of the stations at Narrabri. The Narrabri data set is difficult to interpret because the township of Narrabri is on an anabranch of the Namoi River called Narrabri Creek. There are significant periods of no record in either, and sometimes both, of the channels. It is also impossible to infer flow in one channel from flow in the other channel, because there is no relationship between flow in the channels. This absence of a relationship is caused by variations in the channel cross sections at the bifurcation point produced by floods, mining operations, and earth moving equipment.
Climate and streamflow in the Namoi Basin, Australia
831
61 b’
2
E 2
ar
o.6
8
2
0.4
Lz
F
GUNNEIMH
9 0.2
I
12
I
0 TAMdVORTti
I
I
I
3
4
5
tNARR4BRl
I
I
6
I
7
-+WALGEl7 1
a
I
I
9
lo
11
12
rmNlH
Fig. 5. Long-term monthly average rainfalls.
500-600
mm
600-700
mm
700-800
mm
800-900
mm
Fig. 6. Annual rainfall contours for the Namoi Basin.
1 Namoi Rv
Mooki
Rv
006 ICI
Fig. 7. River reach nodes.
1
P.F. Crapper et al.
832
Table 1. River gauging stations and period of record. 4 1900 1 Namoi River at Gunnedah
1 Dee 1891 to 21 Apr 1990
4 19006 Peel River at Cam01 Gap
4 Dee 1923 to 13 Sept 1991
4 19007 Namoi River d/s Keepit Dam
4 Dee 1923 to 21 Nov 1991
4 190 12 Namoi River at Boggabri
1 Jan 1913 to 29 Sept 1989
4 19026 Namoi River at Goangra
1 Aug.1954 to 11 Apr 1990
4 19027 Mooki River at Breeza
1 Sept 1957 to 5 Feb 1992
4 19032 Coxs Creek at Boggabri
6 Dee 1965 to 27 Feb 1990
4 19039 Namoi River at Mollee
1 Ott 1965 to 30 Ott 1989
4 19049 Pian Creek at Waminda
11 Dee 1972 to 13 Aug 1983
4 1905 1 Maules Creek at Avoca East
8 June 1972 to 26 May 1990
3
1973
1975
1977
1979
1981
1983
YEAR +
NODE ONE (6+7+27)/l
o
NODETWO (1+32)/12
-A- NODE THREE ( 12+51)/39
8
NODE FOUR (26+49)/39
Fig. 8. Normalised nodal discharge.
There are four locations (or nodes) along the main channel where water balance techniques can be used to test the accuracy of the streamflow data set. In Fig. 8, the normalised discharge at these four locations is plotted for the 10 y of common record for annual flow (1973 to 1982). All periods of no record were set equal to zero for this plot. At node one, the discharges at 419006 + 419007 + 419027 were normalised by the discharge at 419001 (Fig. 7). Assuming there were no other inputs (e.g., other streams or sewage inputs) or outputs (e.g., irrigation pumping or town water) and no periods of no record, then the normalised discharge should be slightly less than one to account for evaporation from the river surface and deep drainage. Also, the measurement of streamflow, especially with a
natural control, is inexact. There are two major sources of error. The first error occurs in the measurement of discharge for a given stage using current meters. Maidment ( 1993) suggested that, at best, the discharge estimate is accurate to *4.3%. The second source of error relates to the accuracy of the rating table. The magnitude of this error is somewhat harder to predict because, for natural control, the rating table varies with time. Kazmann (1965) believed that, at best, streamflow measurements with natural controls are accurate to *lo%. Thus, a range of normalised discharge from 0.8 to 1.2 or ?? 20% (as terms in both the numerator and denominator are *lo%) is as good as can be reasonably expected. In fact, Fig. 8 shows that the normalised discharge for node one varies from 0.80 ( 1974) to 1.19
Climate and streamflow in the Namoi Basin, Australia
833
3 ; Od y
0.6
d 4
0.4
0.2
I
1973
I
1975
1977
I
1
1979
1981
YEAR
+
NCEE ONE (6+7+27)/l
-0 NOGEM
(1+32)/12
-A- NCDE THREE (12+51)/39
8
NODE FOUR (26+49)/39
Fig. 9. Corrected normalised nodal discharge.
(1980). In 1974, there were no periods of no record, and in 1980, station 419007 had a period of no record from 8 July 1980 until 12 August 1980, inclusive. This period of no record should have reduced the normalised discharge as the term appears in the numerator, although July is a low flow month. A possible explanation for the comparatively high value for the normalised discharge for 1980 is the abstraction of irrigation water between Keepit Dam and Gunnedah. Rainfall for Gunnedah in 1980 (429.0 mm) was the least of the study period but was only marginally less than 1982 (439.4 mm), whereas the nodal discharge for node one for 1982 was 1.10. At nodes two and three, a major problem occurred during 1979, at station 4 19012, and was caused by a period of no record from 1 January 1979 until 21 February 1979. This no record occurs during the high flow period and produced an overestimation of the normalised discharge at node two and an underestimation at node three. If the normalised discharge for both nodes is set equal to one and the two estimates of the annual discharge at station 4190 12 averaged, then the corrected normalised nodal discharge is given in Fig. 9. The normalised nodal discharges for nodes one, two, and three are all within the ranges 0.8 to 1.2. The norrnalised discharge at node four is different. There are several reasons for this result. Whilst stations 4 19026 and 419049 are comparatively close together (approx 25 km, but on different river channels), they
are separated from station 4 19039 by at least 170 km along the river channel, assuming a sinuosity factor of 1.3. Thus, the opportunities for loss of streamflow (as nodal discharge of less than one implies) by the mechanisms mentioned above are substantial. Additionally, large amounts of cotton are grown along this stretch of the Namoi River and cotton is a vast user of water. For the study period, limited data is available on irrigation water extractions and the irrigation licensing authority believes that data to be unreliable. Further, all stations had significant periods of no record. In fact, there was no single year in which all three stations worked for the whole year. The annual flow rates (using corrected flow for stations 419012) for the river gauging stations along the main channel are shown on Fig. 10. There are several interesting features to note. The contributions from the Peel River (station 4 19006, catchment area 4670 km’) and the upper Namoi River (station 4 19007, catchment area 5700 km2) are similar. The Peel catchment is smaller but it does include the highest rainfall area in the south-east corner of the Namoi Basin (Fig. 6). Flow rates at Gunnedah, Boggabri, and Mollee increase slightly in the downstream direction in high flow years and are approximately the same in low flow years. The flow down the Namoi River at Goangra is substantially greater than the flow in the Pian Creek for all years, although the discharge data for 419026 and 4 19049 are dubious. The low values in the normalised
834
P.F. Crapper et al.
12
1973
1974
1875
1978
1977
IQJJE: 18JY
1980
I!981
FZi 419006
PEEL AT CARROL
Clll 419007
NAtvlOl DF.a I’*EEPIT
=
419001
NAMOI
0
413012
NAtvlGl AT EvWGAHRl
AT GUNNEDAH
L=l413039
NkMOl
AT MOLLEE
F?J 419026
NAtvIOl AT GOANGRA
168 419049
PI&N AT L’VAMINDA
If182
YEAR
Fig. 10. Corrected main channel annual discharge.
U
IQ73
19741975
IQ76
19771978
IQ79
IQ80
MOOKI
AT BREEZA
IDI 419032
419027
COXS
AT BOGGABRI
=
MAULES
419051
AT AVOCA
E
lQSllQ82
YEAR
Fig. 11. Major tributary contributions.
nodal discharge at node four (Fig. 9, 1973, 1975, and 1979 onward) all coincide with low flow years. This supports the hypothesis that irrigation pumping along the Namoi River and the Pian Creek downstream of Mollee is a major and unquantified factor in the water balance. As discussed above, existing irrigation abstraction data for the study period are not deemed reliable enough to be useful. The inputs from the three major tributaries are shown in Fig. 11 using the same scale as Fig. 10. These tributaries contribute insignificant amounts of flow in the
low flow years but in the high flow years (e.g., 1977), their contribution can be significant. For example, in 1977 (the highest flow year), the contribution to flow from the Mooki River and Coxs Creek was approximately equal to the upper Namoi or Peel Rivers. EFFECT OF LAND USE ON RUNOFF
In theNamoi Basin, rainfall amounts generally correlate with longitude, with the highest rainfalls occurring at the catchment boundary in the south-east and the
Climate and streamflow in the Namoi Basin, Australia
835
Table 2. Rainfall and discharge character for eight upland subcatchments in the Namoi Basin.
Station name and no.
Area (km21
Annual rainfall (mm)
Annual discharge (mm)
Annual yield %
Rain gauge coverage (km’/gauge)
Longitude
Period of record
Coxs Creek at Tambar Springs 419033 Mooki River at Caroona 419034 Warrah Creek at Old Warrah 419076 Goonoo Goonoo Creek at Timbumburi 419035 Peel River at Piallamore 419015 Peel River at Bowling Alley Point 419004 Mulla Creek at Goldcliff 419055 MacDonald River at Woolbrook 419010
1450
550
27.15
4.94
725.00
149:53:09
9June 196530 June 1993
2540
652
24.95
3.83
181.43
150:25:47
1 Jan 196630 June 1993
150
930
62.27
6.70
75.00
150:38:34
15June 198231 Dee 1991
503
700
52.16
7.45
100.60
150:54:55
16June 196531 Dee 1990
1140
774
99.41
12.84
142.50
151:03:05
1 July 193614 Feb 1996
310
910
173.97
19.12
155.00
151:08:35
1 Jan 194031 Dee 1969
254
897
142.48
15.88
254.00
151:08:46
21 May 19749 Feb 1989
829
845
156.32
18.50
138.17
151:20:45
1 Jan 197031 Dee 1991
lowest in the western section of the basin (Fig. 6). Rainfall-runoff (or yield) relationships have been characterised for a number of upland catchments in the Namoi Basin (Table 2). Yield is a complex and nonlinear function of many catchment characteristics. In the following, this relationship is explored. The data suggest that both average rainfall and catchment area are good first order predictors of average yield, with correlations of 0.678 (Fig. 12) and -0.741 (Fig. 13), respectively. The outlier in Fig. 12 represents Warrah Creek catchment, which has experienced significant land use and land management changes (Beavis et al. 1997). These data suggest that factors affecting the hydrological response in the eight study catchments include: rainfall amount and temporal distribution; and landscape attributes of the catchment including land cover and catchment area. Agricultural expansion since European settlement (approximately 1865) has been associated with extensive clearing of forests and woodland and subsequent improvement of native grassland with exotic species. The Namoi Basin covers an area of over 40 000 km2 and a wide range of land covers. Maps and descriptions
of these land covers are contained in the Namoi Community Catchment Plan (1996). The application of fertilisers has also impacted vegetation density. These land cover changes have a number of complex impacts on catchment hydrology: 1) removal of the tree canopy reduces interception losses (Langford et al. 1980; Ruprecht and Schofield 1989) and increases surface runoff; 2) replacement of deep rooting native pasture by shallow rooting improved pasture reduces the extraction of soil moisture by plants, particularly during dry periods, leading to aquifer recharge (Boughton 1970); 3) introduced plant species transpire moisture at greater rates than native species, thereby reducing runoff (Ring et al. 1984); 4) improved pasture increases groundcover density and reduces runoff (Ring and Fisher 1985); and 5) water storages, including farm dams, reduce streamflow particularly during low flow conditions (Ockenden and Kotwicki 1982; Cresswell 1991; Good 1992). This complexity is compounded by the inverse relationship between infiltration rates and intensity of grazing as a function of soil compaction. Furthermore,
836
P.F. Crapper et al.
O400
, 500
700
600
800
900
1000
Annual rainfall (mm) Fig. 12. Yield vs. annual rainfall for catchments in Table 2.
-
0
0
4
I
500
700
600
800
900
1000
Annual rainfall (mm) Fig. 13. Catchment area vs. annual rainfall for catchments in Table 2.
the percentage groundcover in cropping conditions, which varies temporally, is significant: bare or fallow ground creates the most runoff, whilst closely grown crops result in least runoff (Ring and Fisher 1985). In response to land degradation problems, often initiated by clearing and inappropriate management practices, farmers have been encouraged to construct contour banks, grassed waterways, and farm dams to
form integrated erosion control networks. These structures modify catchment surfaces, impede the movement of water within a catchment, and ultimately reduce streamflow. In a catchment with conservation treatments including perennial pasture, 3-y crop rotations and extensive contour banking, a 24% reduction in runoff was measured, with the figure almost doubling to 43% for drier periods (Moore and Morgan 1969).
Climate and streamflow in the Namoi Basin, Australia
1955
1960
1965
837
1970
1975
1980
1985
1990
1995
YEAR +=JGUNNEDAH
0 TANNVORTH + NARRABRI
+WALGETf
Fig. 14. Mean annual maximum temperature.
8
t 1955
1960
1965
1970
1980
1975
1985
1990
1995
2000
YEAR e_ GUNNEIXH
0 TAMdVORTH
t
NARRABRI
+ WALGETT
Fig. 15’.Mean annual minimum temperature.
Furthermore, the diversion of water for irrigation reduces downstream flow, and can contribute to potential conflicts between users. Consequently, whilst the initiation and development of erosional networks is associated with changes in land cover (Beavis et al. 1997), sustainable land management practices also impact the erosional drainage network by reducing runoff.
TEMPERATURE
DATA
The temperature records for the catchment are up to 40 y in length. The annual mean of the daily maximum and minimum temperatures (subsequently called mean annual maximum and minimum temperature) are shown in Figs. 14 and 15, respectively. The duration of
P.F. Crapper et al.
838
p! ZJ
25
5
20
iit E l5 :
10 5 01
1
12
I
1
I
3
4
5
I
I
I
1
,
I
6
7
a
9
10
11
12
fvmtTH e GUNMDAH
Q TAMv’JORTH+NARRABRI
+ WALGEll-
Fig. 16. Mean monthly maximum and minimum temperatures.
the data is too short to make any observations about longer-term (-100 y) trends. However, it is clear that maximum temperatures increase in the westerly direction across the catchment, and 1965, 1973, 1980, and 1982 had higher mean annual maximum temperatures. Again, the duration of the minimum temperature data is too short to make any statements about longerterm trends. The minimum temperatures generally increase in a westerly direction across the catchment, with the exception that Narrabri is cooler than Gunnedah.Narrabri (elevation 2 10 m) is approximately 90 km north-west of Gunnedah (elevation 300 m). However, Narrabri is close to the Nandewar Range, which forms part of the catchment boundary and contains Mount Kaputar (1508 m), whereas Gunnedah is surrounded by plains with the exception of King Jack Mountain (761 m). Hence, it is to be expected that Narrabri would experience considerably more cool air drainage than Gunnedah. Cooler years were 1970, 1976, 1984, and 1994. The mean monthly maximum and minimum temperature are plotted in Fig. 16. These monthly values were obtained by averaging the monthly maximum and minimum values for the period of common record (in this case, 1965 to 1992 inclusive). The previous observations about westward temperature increases and Narrabri being cooler than Gunnedah were confirmed. CONCLUSIONS
Climatic parameters are highly variable and vary with different timescales. The rainfall for the Namoi Basin is highly variable. The rainfall data set contains
only approximately 120 y of records and hence it is not possible to make any statistically significant statements about long-term trends (>lOO y). However, three distinct shorter-term trends (up to 50 y) were evident in the four data sets studied in detail. At the end of the last century, there occurred a period of above average rainfall, which was followed by a period of below average rainfall until 1945 and then a period of above average rainfall up to the present. However, the durations of these periods varied for different locations. Also, Narrabri has a significantly different rainfall behaviour to the other locations. The streamflow data, being a function of rainfall and runoff ratio, were more variable than rainfall as there have been significant variations in land cover over the period ofrecord. A simplified river network containing all the major tributaries was examined in detail. The major tributaries to the network are the upper Namoi River and the Peel River. The other two tributaries of significance are the Mooki River and Coxs Creek. They only contribute significant quantities of water in the high flow years when their combined contribution is approximately equal to either the upper Namoi River or the Peel River. A significant increase in mean annual daily maximum and minimum temperature occurred from east to west across the catchment. The major exception to this rule was that Narrabri was cooler than Gunnedah, and this is believed to be due to the proximity of Narrabri to the edge of the catchment and a high mountain range. Acknow~edgmenr-The authors gratefully acknowledge many useful discussions with P. Mick Fleming of CSIRO Land and Water and thank Andrew Brissett from the Department of Land and Water Conservation in Armidale for supplying stream cross-section information. This work is funded by the Murray-Darling Basin Commission’s Natural Resources Management Strategy.
REFERENCES Beavis, S.G.; Jakeman, A.J.; Zhang, L.; Gray, S.D. Erosional history of selected upland subcatchments in the Liverpool Plains, New South Wales. In: McDonald, A.D. et al., eds. Proc. intl. congress on modelling and simulation, MODSIM 1997. University ofTasmania. 8-l 1 Dee 1997. Available from: Modelling and Simulation Society of Australia, Canberra, Australia. Boughton, WC. Effects of land management on quantity and quality of available water. Water Research Laboratory. Report No. 120. Sidney, Australia: University of New South Wales; 1970. Cresswell, D.J. Integrated management of farm dams in the Barossa Valley. ESW 87/54. Adelaide, Australia: South Australia, Engineering and Water Supply Department; 1991.
Climate and streamflow in the Namoi Basin, Australia
Findlayson, B.L.; McMahon, T.A. Australia v the world: A comparative analysis of streamflow characteristics. In: Warner, R.F., ed. Fluvial geomorphology ofAustralia. Sydney, Australia: Academic Press; 1988: 17-40. Good, R.K. The impact of development on streamflow in the Marne River. ESW 92/23. Adelaide, Australia: South Australia, Engineering and Water Supply Department; 1992. Jakeman,A.J.;Green,T.R.;Beavis,S.G.;Zhang,L.;Dietrich,C.R.; Crapper, P.F. Modelling upland and in-stream erosion, sediment and phosphorus transport in a large catchment. Hydrol. Processes 13: 745-752; 1998. Kazmann, R.G. Modem hydrology. New York, NY: Harper; 1965. Langford, K.J.; Moran, R.J.; O’Shaughnessy, P.J. The North Maroondah Experiment pre-treatment phase comparison of catchment water balances. J. Hydrol. 46: 123-145; 1980. Maidment, D. R. Handbook of hydrology. New York, NY: McGraw-Hill; 1993. McGilchrist, CA.; Woodyer, K.D. Note on a distribution free CUSUM technique. Technometrics 17: 321-325; 1975. Metaccess, vl .Ol, Metaccess for Windows. Produced by Horizon Technology, Australia, http:Nwww.hzn.com.au 1996. Moore, W.L.; Morgan, C. Effects of watershed changes on streamflow. Austin, TX: University of Texas Press; 1969. Namoi Community Catchment Plan. Stage 1. Situation statement. Prepared for the Namoi Catchment Planning Taskforce by Donaldson Planning and Management Services. 1996. Available
839
from: NSW Department of Land and Water Conservation, Tamworth, NSW, Australia: Ockenden, A.P.; Kotwicki, V. The effects of farm dams on flows in the North Para River, South Australia. Adelaide, Australia: Engineering and Water Supply Department 82/59; 1982. Pinneena, Version 5. Produced by Department of Land and Water Conservation. Australia: New South Wales Government; June 1996. Ring, P. J.; Banens, R.J.; Perens, S.J.; Burton, J.R. The influence of changes in agricultural land use on runoff from relatively large catchments: A preliminary appraisal. Conference on Agricultural Engineering, Bundaberg. 27-30 August 1984. Available from: University of New England, Department of Resource Engineering, Armadale, Australia. Ring, P.J.; Fisher, I.H. The effects of changes in land use on runoff from large catchments in the Upper McIntyre Valley, NSW. Hydrology and Water Resources Symposium, Sydney, 14-16 May, 1985. University ofNew England, Department of Resource Engineering, Armadale, Australia. Ruprecht, J.K.; Schofield, N.J. Analysis of streamflow generation following deforestation in southwest Western Australia. J. Hydrol. 105: 1-17; 1989. Schroder, D.; Glennon, C. Water resources issues in the Namoi Basin. In: Proc. international congress on modelling and simulation. MODSIM 1995. Newcastle, November 1995. 1995: 134138. Available from: Modelhng and Simulation Society of Australia, Canberra, Australia.
Vol. 25, No. 6/7, pp. 827439, 1999 Copyright 01999 Elsevier Science Ltd Printed in the USA. All rights reserved 01604120/99/$-see front matter
Pergamon
PI1SO160-4120(99)00048-3
THE RELATIONSHIP BETWEEN CLIMATE AND STREAMFLOW IN THE NAM01 BASIN P.F. Crapper CSIRO Land and Water, GPO Box 1666, Canberra, ACT 2601, Australia
S.G. Beavis and Li Zhang CRES, Australian National University, Canberra, ACT 0200, Australia
EI 9808-201 M(Received
31 August 1998; accepted I3 March 1999)
The Namoi River Basin (42000 km’) is located in the Murray-Darling Basin, west of the Great Australian Dividing Range in northeast New South Wales (NSW), and includes some of the most fertile agricultural lands in Australia. One of the environmental concerns for this basin is erosion and its effects on downstream water quality. Models that relate climate, land use, and these concerns require measurements of climate (rainfall and temperature) and streamflow. These measurements were examined as a preliminary to the modelling. The residual mass technique was used to examine the temporal variation of annual rainfalls over the approximately 100 years of available data, and significant spatial variations were found in annual rainfall trends over the catchment. Streamflow was examined at key river gauging stations. The impact ofrecent large-scale irrigation operations was clearly observed. The impact of changing land use and land management on runoff ratios was examined for eight subcatchments. Temperature variations were examined for the four major towns in the catchment. The duration of the temperature data is too short to make any comments about long-term trends. Significant variations were observed in an east to west direction. 01999 Elsetier Science Ltd
INTRODUCTION In recent years, the management of the water and land resources of the Namoi Basin have come under increasing public scrutiny. This has come about in part because of a wide belief that the present agricultural practices are not sustainable, and in part because of a concern for the apparent deterioration of water quality. The work described in this paper is part of a larger study on the sources and transport of sediments and nutrients in the Namoi Basin. Additional information on the land and water resources of the Namoi Basin is contained in Schroder and Glennon (1995) and the Namoi Community Catchment Plan (1996), and a location map is shown in Fig. 1.
Rainfall is probably the most widely measured meteorological parameter and is one of the major determinants of erosion. Jakeman et al. (1998) presented a framework to model and predict the effects of rainfall and land use on erosion and downstream water quality. A crucial input is accurate estimation of area1 rainfall. This is not easy, as it is temporal and spatial heterogeneity is high. The locations of rainfall measurement sites, within Australia at least, are generally selected for convenience, not representativeness. The Namoi Basin is acknowledged as one with a highly variable rainfall and as having extremely variable hydrologic characteristics (Findlayson and McMahon 1988). 827
P.F. Crapper et al.
828
w
Lagend .. . .
stream lines
_...
catchment
boundary
__ . ..’ . ...“.
KILOMETERS
Fig. 1. The location of the Namoi Basin.
RAINFALL
DATA
The rainfall data set for Gunnedah (Fig. 1) was examined for initial analysis because Gunnedah is approximately in the middle of the catchment and because it has the longest climate data set in the catchment. It is a composite data set from the New South Wales (NSW) Department of Land and Water Conservation, Gunnedah Research Station and the Australian Bureau of Meteorology at the Gunnedah Post Office. This data set, and data sets for approximately 16000 stations around Australia, are available on Metaccess (1996). The data set records daily rainfall for the period 1 January 1877 to 3 1 March 1997. The annual rainfall for this period is shown in Fig. 2, where the straight line of best fit is also plotted. Gunnedah has, in the last 120 y, experienced 3 y with rainfall in excess of 1000 mm (1879 - 1047 mm, 1890 - 1133 mm, and 1950 - 1150 mm) and 1 y with rainfall less than 250 mm (1946 - 247 mm). The average annual rainfall for the 120-y data set is 6 11.6 mm and the standard deviation is 17 1.1 mm, giving a coefficient of variation of 0.28. The line of best tit reveals, on average, a slow increase of about 0.4 mm/y in average annual rainfall
over this period. This apparent increase in annual rainfall is too small to be statistically significant and could easily be explained by replacement of existing rain gauges, relocation of rain gauges, or growth in surrounding vegetation. The cumulative sum (CUSUM) technique (MC Gilchrist and Woodyer 1975), also known as the residual mass technique, was used to examine the temporal sequence of rainfall. CUSUM techniques have proved to be a valuable tool in detecting intermediate term changes in the mean value of a sequence of regularly spaced observations. The CUSUM si can be defined as: , s, = c (x, - 2) j=l
where 2 =‘& I2 i=l
where xi is the i th observation in_a series of n regularly spaced observations with mean x . The CUSUM distribution is a normalised distribution and it reveals runs of observations greater than the long-term mean with a positive slope and less than the long-term mean with a negative v slone. Such oersistent nositive or negative 1 1 1
Climate and streamflow in the Namoi Basin, Australia
i iz
829
600
z $ 400
200
0 1877
1 a97 1887
1917 1907
1937 1927
1957 1947
1977 1967
1987
YEARS Fig. 2. Annual rainfall distribution at Gunnedah for the period 1877 to 1996. The line of best fit is also plotted.
slopes can be used to detect intermediate term changes in the mean value. The actual ordinate values are not relevant; it is the sign of the slope (i.e., either positive or negative) and the duration for which the sign of the slope remains the same that is important. The CUSUM distribution for the annual rainfall at Gunnedah is shown in Fig. 3. An examination of Fig. 3 reveals that annual rainfalls for the 116 y of record can be treated as 3 distinct average rainfall periods. The first period from 1877 to 1900 consisted of a period of greater than average rainfall, with some uncertainty in the 1880s; the second period from approximately 1900 to 1945 was a period of below average rainfall; and from 1945, there was a period of average rainfall, followed by above average rainfall in the early 1970s and then in the 1980s. These trends are apparent in Fig. 2 but more difficult to discern. Three additional rain gauge stations were examined in detail. These stations were Tamworth (average annual rainfall 661 mm), Narrabri (636 mm), and Walgett (472 mm) (see Fig. 1 for locations). For periods of no record, stations within a 20 km radius were examined and the average rainfall at these stations was used. A moving 5-y average of s, is plotted on Fig. 4. The locations Gunnedah, Tamworth, and Walgett show the
same three major trends in rainfall, i.e., a period of above average rainfall at the end of last century, followed by a period of below average rainfall until 1945 and then a period of above average rainfall up to the present. However, the durations of these periods varied for different locations. Narrabri has a rainfall behaviour significantly different from the other locations. Narrabri showed a period of above average rainfall at the end of the last century, followed by a period of approximately uniform rainfall up to 1935. This pattern was then followed by below average rainfall until 1945, then above average rainfall until 1955. From 1955 until 1980, the rainfall was approximately uniform and then there was a period of below average rainfall in the 1980s. The reasons for the different behaviour at Narrabri are discussed later in the temperature data section. The annual rainfall amounts presented in Figs. 2, 3, and 4 conceal all within-year variations. In Fig. 5, the average monthly rainfalls for each of the four locations for the entire record are presented. The monthly rainfalls for each location were normalised by the highest monthly rainfall for that location. The highest monthly rainfall was chosen as the normalising parameter so that all normalised monthly rainfalls would
P.F. Crapper et al.
830
BGUNNEDAH I 1880
1
oTAMb%ORTH I
1900 1890
I
I
1920 1910
tN4RRABRl I
I
1940 1930
,
+WALG.EIT ,
1960 1950
,
,
1980 1970
J
2000 1990
YEARS
-31 ’ 1677
’
’
1897 1007
’
’
1917 1907
’
’
1937 1927
’
’
1947
’
’
1977
1957 1967
1997
Fig. 4. A moving 5-y mean of si for Gunnedah, Tamworth, Narrabri, and Walgett.
1987
YEARS Fig. 3. A residual mass plot of Gunnedah annual rainfall.
fall between zero and one and thus allow easier comparison For all locations, it can be seen that January has the highest average rainfall; and, with the exception of Walgett in November and December, the monthly rainfall distributions appear similar. The rainfall for all locations is summer-dominated, with January being the highest rainfall month. This rainfall peak occurs because the summer rainfalls are produced by inflows of tropical air masses associated with tropical cyclones. Long-term annual average rainfall isohyets are shown in Fig. 6. There is an obvious decrease in rainfall from east to west. The highest rainfall occurs in the southeast corner of the catchment on the Peel River catchment. STREAMFLOW
DATA
Streamflow has been measured at 8 1 locations in the Namoi Basin, by the NSW Department of Land and Water Conservation and its predecessors, for varying periods of time using mostly natural controls, although there are a small number of concrete structures. There are also six weir gauges in the catchment measuring the depth of water in the weirs. The daily streamflow records for all these gauging stations are available on Pinneena (1996). Unfortunately, all stations contain some periods of no record and, unless otherwise stated, these periods were set equal to zero. There is no correct
way of handling the problem of missing data. Alternative approaches are to replace the missing values with the mean or median of weekly, monthly, seasonal, or annual values. Also a rainfall-runoff model can be used to construct the streamflow, although in this case, rainfall and streamflow data would be needed for the calibration period and rainfall for the simulation period. In the present case, the simplest approach of replacing the missing values with zeros was chosen, as the accuracy of these streamflows will be subsequently checked using water balance techniques. The minimum main channel streamflow gauging station data set is shown diagrammatically in Fig. 7, and the station names and periods of record are given in Table 1. This data set includes the four major tributaries to the Namoi River (Peel River, Mooki River, Coxs Creek, and Maules Creek) and the river gauging stations along the main river channel, with the exception of the stations at Narrabri. The Narrabri data set is difficult to interpret because the township of Narrabri is on an anabranch of the Namoi River called Narrabri Creek. There are significant periods of no record in either, and sometimes both, of the channels. It is also impossible to infer flow in one channel from flow in the other channel, because there is no relationship between flow in the channels. This absence of a relationship is caused by variations in the channel cross sections at the bifurcation point produced by floods, mining operations, and earth moving equipment.
Climate and streamflow in the Namoi Basin, Australia
831
61 b’
2
E 2
ar
o.6
8
2
0.4
Lz
F
GUNNEIMH
9 0.2
I
12
I
0 TAMdVORTti
I
I
I
3
4
5
tNARR4BRl
I
I
6
I
7
-+WALGEl7 1
a
I
I
9
lo
11
12
rmNlH
Fig. 5. Long-term monthly average rainfalls.
500-600
mm
600-700
mm
700-800
mm
800-900
mm
Fig. 6. Annual rainfall contours for the Namoi Basin.
1 Namoi Rv
Mooki
Rv
006 ICI
Fig. 7. River reach nodes.
1
P.F. Crapper et al.
832
Table 1. River gauging stations and period of record. 4 1900 1 Namoi River at Gunnedah
1 Dee 1891 to 21 Apr 1990
4 19006 Peel River at Cam01 Gap
4 Dee 1923 to 13 Sept 1991
4 19007 Namoi River d/s Keepit Dam
4 Dee 1923 to 21 Nov 1991
4 190 12 Namoi River at Boggabri
1 Jan 1913 to 29 Sept 1989
4 19026 Namoi River at Goangra
1 Aug.1954 to 11 Apr 1990
4 19027 Mooki River at Breeza
1 Sept 1957 to 5 Feb 1992
4 19032 Coxs Creek at Boggabri
6 Dee 1965 to 27 Feb 1990
4 19039 Namoi River at Mollee
1 Ott 1965 to 30 Ott 1989
4 19049 Pian Creek at Waminda
11 Dee 1972 to 13 Aug 1983
4 1905 1 Maules Creek at Avoca East
8 June 1972 to 26 May 1990
3
1973
1975
1977
1979
1981
1983
YEAR +
NODE ONE (6+7+27)/l
o
NODETWO (1+32)/12
-A- NODE THREE ( 12+51)/39
8
NODE FOUR (26+49)/39
Fig. 8. Normalised nodal discharge.
There are four locations (or nodes) along the main channel where water balance techniques can be used to test the accuracy of the streamflow data set. In Fig. 8, the normalised discharge at these four locations is plotted for the 10 y of common record for annual flow (1973 to 1982). All periods of no record were set equal to zero for this plot. At node one, the discharges at 419006 + 419007 + 419027 were normalised by the discharge at 419001 (Fig. 7). Assuming there were no other inputs (e.g., other streams or sewage inputs) or outputs (e.g., irrigation pumping or town water) and no periods of no record, then the normalised discharge should be slightly less than one to account for evaporation from the river surface and deep drainage. Also, the measurement of streamflow, especially with a
natural control, is inexact. There are two major sources of error. The first error occurs in the measurement of discharge for a given stage using current meters. Maidment ( 1993) suggested that, at best, the discharge estimate is accurate to *4.3%. The second source of error relates to the accuracy of the rating table. The magnitude of this error is somewhat harder to predict because, for natural control, the rating table varies with time. Kazmann (1965) believed that, at best, streamflow measurements with natural controls are accurate to *lo%. Thus, a range of normalised discharge from 0.8 to 1.2 or ?? 20% (as terms in both the numerator and denominator are *lo%) is as good as can be reasonably expected. In fact, Fig. 8 shows that the normalised discharge for node one varies from 0.80 ( 1974) to 1.19
Climate and streamflow in the Namoi Basin, Australia
833
3 ; Od y
0.6
d 4
0.4
0.2
I
1973
I
1975
1977
I
1
1979
1981
YEAR
+
NCEE ONE (6+7+27)/l
-0 NOGEM
(1+32)/12
-A- NCDE THREE (12+51)/39
8
NODE FOUR (26+49)/39
Fig. 9. Corrected normalised nodal discharge.
(1980). In 1974, there were no periods of no record, and in 1980, station 419007 had a period of no record from 8 July 1980 until 12 August 1980, inclusive. This period of no record should have reduced the normalised discharge as the term appears in the numerator, although July is a low flow month. A possible explanation for the comparatively high value for the normalised discharge for 1980 is the abstraction of irrigation water between Keepit Dam and Gunnedah. Rainfall for Gunnedah in 1980 (429.0 mm) was the least of the study period but was only marginally less than 1982 (439.4 mm), whereas the nodal discharge for node one for 1982 was 1.10. At nodes two and three, a major problem occurred during 1979, at station 4 19012, and was caused by a period of no record from 1 January 1979 until 21 February 1979. This no record occurs during the high flow period and produced an overestimation of the normalised discharge at node two and an underestimation at node three. If the normalised discharge for both nodes is set equal to one and the two estimates of the annual discharge at station 4190 12 averaged, then the corrected normalised nodal discharge is given in Fig. 9. The normalised nodal discharges for nodes one, two, and three are all within the ranges 0.8 to 1.2. The norrnalised discharge at node four is different. There are several reasons for this result. Whilst stations 4 19026 and 419049 are comparatively close together (approx 25 km, but on different river channels), they
are separated from station 4 19039 by at least 170 km along the river channel, assuming a sinuosity factor of 1.3. Thus, the opportunities for loss of streamflow (as nodal discharge of less than one implies) by the mechanisms mentioned above are substantial. Additionally, large amounts of cotton are grown along this stretch of the Namoi River and cotton is a vast user of water. For the study period, limited data is available on irrigation water extractions and the irrigation licensing authority believes that data to be unreliable. Further, all stations had significant periods of no record. In fact, there was no single year in which all three stations worked for the whole year. The annual flow rates (using corrected flow for stations 419012) for the river gauging stations along the main channel are shown on Fig. 10. There are several interesting features to note. The contributions from the Peel River (station 4 19006, catchment area 4670 km’) and the upper Namoi River (station 4 19007, catchment area 5700 km2) are similar. The Peel catchment is smaller but it does include the highest rainfall area in the south-east corner of the Namoi Basin (Fig. 6). Flow rates at Gunnedah, Boggabri, and Mollee increase slightly in the downstream direction in high flow years and are approximately the same in low flow years. The flow down the Namoi River at Goangra is substantially greater than the flow in the Pian Creek for all years, although the discharge data for 419026 and 4 19049 are dubious. The low values in the normalised
834
P.F. Crapper et al.
12
1973
1974
1875
1978
1977
IQJJE: 18JY
1980
I!981
FZi 419006
PEEL AT CARROL
Clll 419007
NAtvlOl DF.a I’*EEPIT
=
419001
NAMOI
0
413012
NAtvlGl AT EvWGAHRl
AT GUNNEDAH
L=l413039
NkMOl
AT MOLLEE
F?J 419026
NAtvIOl AT GOANGRA
168 419049
PI&N AT L’VAMINDA
If182
YEAR
Fig. 10. Corrected main channel annual discharge.
U
IQ73
19741975
IQ76
19771978
IQ79
IQ80
MOOKI
AT BREEZA
IDI 419032
419027
COXS
AT BOGGABRI
=
MAULES
419051
AT AVOCA
E
lQSllQ82
YEAR
Fig. 11. Major tributary contributions.
nodal discharge at node four (Fig. 9, 1973, 1975, and 1979 onward) all coincide with low flow years. This supports the hypothesis that irrigation pumping along the Namoi River and the Pian Creek downstream of Mollee is a major and unquantified factor in the water balance. As discussed above, existing irrigation abstraction data for the study period are not deemed reliable enough to be useful. The inputs from the three major tributaries are shown in Fig. 11 using the same scale as Fig. 10. These tributaries contribute insignificant amounts of flow in the
low flow years but in the high flow years (e.g., 1977), their contribution can be significant. For example, in 1977 (the highest flow year), the contribution to flow from the Mooki River and Coxs Creek was approximately equal to the upper Namoi or Peel Rivers. EFFECT OF LAND USE ON RUNOFF
In theNamoi Basin, rainfall amounts generally correlate with longitude, with the highest rainfalls occurring at the catchment boundary in the south-east and the
Climate and streamflow in the Namoi Basin, Australia
835
Table 2. Rainfall and discharge character for eight upland subcatchments in the Namoi Basin.
Station name and no.
Area (km21
Annual rainfall (mm)
Annual discharge (mm)
Annual yield %
Rain gauge coverage (km’/gauge)
Longitude
Period of record
Coxs Creek at Tambar Springs 419033 Mooki River at Caroona 419034 Warrah Creek at Old Warrah 419076 Goonoo Goonoo Creek at Timbumburi 419035 Peel River at Piallamore 419015 Peel River at Bowling Alley Point 419004 Mulla Creek at Goldcliff 419055 MacDonald River at Woolbrook 419010
1450
550
27.15
4.94
725.00
149:53:09
9June 196530 June 1993
2540
652
24.95
3.83
181.43
150:25:47
1 Jan 196630 June 1993
150
930
62.27
6.70
75.00
150:38:34
15June 198231 Dee 1991
503
700
52.16
7.45
100.60
150:54:55
16June 196531 Dee 1990
1140
774
99.41
12.84
142.50
151:03:05
1 July 193614 Feb 1996
310
910
173.97
19.12
155.00
151:08:35
1 Jan 194031 Dee 1969
254
897
142.48
15.88
254.00
151:08:46
21 May 19749 Feb 1989
829
845
156.32
18.50
138.17
151:20:45
1 Jan 197031 Dee 1991
lowest in the western section of the basin (Fig. 6). Rainfall-runoff (or yield) relationships have been characterised for a number of upland catchments in the Namoi Basin (Table 2). Yield is a complex and nonlinear function of many catchment characteristics. In the following, this relationship is explored. The data suggest that both average rainfall and catchment area are good first order predictors of average yield, with correlations of 0.678 (Fig. 12) and -0.741 (Fig. 13), respectively. The outlier in Fig. 12 represents Warrah Creek catchment, which has experienced significant land use and land management changes (Beavis et al. 1997). These data suggest that factors affecting the hydrological response in the eight study catchments include: rainfall amount and temporal distribution; and landscape attributes of the catchment including land cover and catchment area. Agricultural expansion since European settlement (approximately 1865) has been associated with extensive clearing of forests and woodland and subsequent improvement of native grassland with exotic species. The Namoi Basin covers an area of over 40 000 km2 and a wide range of land covers. Maps and descriptions
of these land covers are contained in the Namoi Community Catchment Plan (1996). The application of fertilisers has also impacted vegetation density. These land cover changes have a number of complex impacts on catchment hydrology: 1) removal of the tree canopy reduces interception losses (Langford et al. 1980; Ruprecht and Schofield 1989) and increases surface runoff; 2) replacement of deep rooting native pasture by shallow rooting improved pasture reduces the extraction of soil moisture by plants, particularly during dry periods, leading to aquifer recharge (Boughton 1970); 3) introduced plant species transpire moisture at greater rates than native species, thereby reducing runoff (Ring et al. 1984); 4) improved pasture increases groundcover density and reduces runoff (Ring and Fisher 1985); and 5) water storages, including farm dams, reduce streamflow particularly during low flow conditions (Ockenden and Kotwicki 1982; Cresswell 1991; Good 1992). This complexity is compounded by the inverse relationship between infiltration rates and intensity of grazing as a function of soil compaction. Furthermore,
836
P.F. Crapper et al.
O400
, 500
700
600
800
900
1000
Annual rainfall (mm) Fig. 12. Yield vs. annual rainfall for catchments in Table 2.
-
0
0
4
I
500
700
600
800
900
1000
Annual rainfall (mm) Fig. 13. Catchment area vs. annual rainfall for catchments in Table 2.
the percentage groundcover in cropping conditions, which varies temporally, is significant: bare or fallow ground creates the most runoff, whilst closely grown crops result in least runoff (Ring and Fisher 1985). In response to land degradation problems, often initiated by clearing and inappropriate management practices, farmers have been encouraged to construct contour banks, grassed waterways, and farm dams to
form integrated erosion control networks. These structures modify catchment surfaces, impede the movement of water within a catchment, and ultimately reduce streamflow. In a catchment with conservation treatments including perennial pasture, 3-y crop rotations and extensive contour banking, a 24% reduction in runoff was measured, with the figure almost doubling to 43% for drier periods (Moore and Morgan 1969).
Climate and streamflow in the Namoi Basin, Australia
1955
1960
1965
837
1970
1975
1980
1985
1990
1995
YEAR +=JGUNNEDAH
0 TANNVORTH + NARRABRI
+WALGETf
Fig. 14. Mean annual maximum temperature.
8
t 1955
1960
1965
1970
1980
1975
1985
1990
1995
2000
YEAR e_ GUNNEIXH
0 TAMdVORTH
t
NARRABRI
+ WALGETT
Fig. 15’.Mean annual minimum temperature.
Furthermore, the diversion of water for irrigation reduces downstream flow, and can contribute to potential conflicts between users. Consequently, whilst the initiation and development of erosional networks is associated with changes in land cover (Beavis et al. 1997), sustainable land management practices also impact the erosional drainage network by reducing runoff.
TEMPERATURE
DATA
The temperature records for the catchment are up to 40 y in length. The annual mean of the daily maximum and minimum temperatures (subsequently called mean annual maximum and minimum temperature) are shown in Figs. 14 and 15, respectively. The duration of
P.F. Crapper et al.
838
p! ZJ
25
5
20
iit E l5 :
10 5 01
1
12
I
1
I
3
4
5
I
I
I
1
,
I
6
7
a
9
10
11
12
fvmtTH e GUNMDAH
Q TAMv’JORTH+NARRABRI
+ WALGEll-
Fig. 16. Mean monthly maximum and minimum temperatures.
the data is too short to make any observations about longer-term (-100 y) trends. However, it is clear that maximum temperatures increase in the westerly direction across the catchment, and 1965, 1973, 1980, and 1982 had higher mean annual maximum temperatures. Again, the duration of the minimum temperature data is too short to make any statements about longerterm trends. The minimum temperatures generally increase in a westerly direction across the catchment, with the exception that Narrabri is cooler than Gunnedah.Narrabri (elevation 2 10 m) is approximately 90 km north-west of Gunnedah (elevation 300 m). However, Narrabri is close to the Nandewar Range, which forms part of the catchment boundary and contains Mount Kaputar (1508 m), whereas Gunnedah is surrounded by plains with the exception of King Jack Mountain (761 m). Hence, it is to be expected that Narrabri would experience considerably more cool air drainage than Gunnedah. Cooler years were 1970, 1976, 1984, and 1994. The mean monthly maximum and minimum temperature are plotted in Fig. 16. These monthly values were obtained by averaging the monthly maximum and minimum values for the period of common record (in this case, 1965 to 1992 inclusive). The previous observations about westward temperature increases and Narrabri being cooler than Gunnedah were confirmed. CONCLUSIONS
Climatic parameters are highly variable and vary with different timescales. The rainfall for the Namoi Basin is highly variable. The rainfall data set contains
only approximately 120 y of records and hence it is not possible to make any statistically significant statements about long-term trends (>lOO y). However, three distinct shorter-term trends (up to 50 y) were evident in the four data sets studied in detail. At the end of the last century, there occurred a period of above average rainfall, which was followed by a period of below average rainfall until 1945 and then a period of above average rainfall up to the present. However, the durations of these periods varied for different locations. Also, Narrabri has a significantly different rainfall behaviour to the other locations. The streamflow data, being a function of rainfall and runoff ratio, were more variable than rainfall as there have been significant variations in land cover over the period ofrecord. A simplified river network containing all the major tributaries was examined in detail. The major tributaries to the network are the upper Namoi River and the Peel River. The other two tributaries of significance are the Mooki River and Coxs Creek. They only contribute significant quantities of water in the high flow years when their combined contribution is approximately equal to either the upper Namoi River or the Peel River. A significant increase in mean annual daily maximum and minimum temperature occurred from east to west across the catchment. The major exception to this rule was that Narrabri was cooler than Gunnedah, and this is believed to be due to the proximity of Narrabri to the edge of the catchment and a high mountain range. Acknow~edgmenr-The authors gratefully acknowledge many useful discussions with P. Mick Fleming of CSIRO Land and Water and thank Andrew Brissett from the Department of Land and Water Conservation in Armidale for supplying stream cross-section information. This work is funded by the Murray-Darling Basin Commission’s Natural Resources Management Strategy.
REFERENCES Beavis, S.G.; Jakeman, A.J.; Zhang, L.; Gray, S.D. Erosional history of selected upland subcatchments in the Liverpool Plains, New South Wales. In: McDonald, A.D. et al., eds. Proc. intl. congress on modelling and simulation, MODSIM 1997. University ofTasmania. 8-l 1 Dee 1997. Available from: Modelling and Simulation Society of Australia, Canberra, Australia. Boughton, WC. Effects of land management on quantity and quality of available water. Water Research Laboratory. Report No. 120. Sidney, Australia: University of New South Wales; 1970. Cresswell, D.J. Integrated management of farm dams in the Barossa Valley. ESW 87/54. Adelaide, Australia: South Australia, Engineering and Water Supply Department; 1991.
Climate and streamflow in the Namoi Basin, Australia
Findlayson, B.L.; McMahon, T.A. Australia v the world: A comparative analysis of streamflow characteristics. In: Warner, R.F., ed. Fluvial geomorphology ofAustralia. Sydney, Australia: Academic Press; 1988: 17-40. Good, R.K. The impact of development on streamflow in the Marne River. ESW 92/23. Adelaide, Australia: South Australia, Engineering and Water Supply Department; 1992. Jakeman,A.J.;Green,T.R.;Beavis,S.G.;Zhang,L.;Dietrich,C.R.; Crapper, P.F. Modelling upland and in-stream erosion, sediment and phosphorus transport in a large catchment. Hydrol. Processes 13: 745-752; 1998. Kazmann, R.G. Modem hydrology. New York, NY: Harper; 1965. Langford, K.J.; Moran, R.J.; O’Shaughnessy, P.J. The North Maroondah Experiment pre-treatment phase comparison of catchment water balances. J. Hydrol. 46: 123-145; 1980. Maidment, D. R. Handbook of hydrology. New York, NY: McGraw-Hill; 1993. McGilchrist, CA.; Woodyer, K.D. Note on a distribution free CUSUM technique. Technometrics 17: 321-325; 1975. Metaccess, vl .Ol, Metaccess for Windows. Produced by Horizon Technology, Australia, http:Nwww.hzn.com.au 1996. Moore, W.L.; Morgan, C. Effects of watershed changes on streamflow. Austin, TX: University of Texas Press; 1969. Namoi Community Catchment Plan. Stage 1. Situation statement. Prepared for the Namoi Catchment Planning Taskforce by Donaldson Planning and Management Services. 1996. Available
839
from: NSW Department of Land and Water Conservation, Tamworth, NSW, Australia: Ockenden, A.P.; Kotwicki, V. The effects of farm dams on flows in the North Para River, South Australia. Adelaide, Australia: Engineering and Water Supply Department 82/59; 1982. Pinneena, Version 5. Produced by Department of Land and Water Conservation. Australia: New South Wales Government; June 1996. Ring, P. J.; Banens, R.J.; Perens, S.J.; Burton, J.R. The influence of changes in agricultural land use on runoff from relatively large catchments: A preliminary appraisal. Conference on Agricultural Engineering, Bundaberg. 27-30 August 1984. Available from: University of New England, Department of Resource Engineering, Armadale, Australia. Ring, P.J.; Fisher, I.H. The effects of changes in land use on runoff from large catchments in the Upper McIntyre Valley, NSW. Hydrology and Water Resources Symposium, Sydney, 14-16 May, 1985. University ofNew England, Department of Resource Engineering, Armadale, Australia. Ruprecht, J.K.; Schofield, N.J. Analysis of streamflow generation following deforestation in southwest Western Australia. J. Hydrol. 105: 1-17; 1989. Schroder, D.; Glennon, C. Water resources issues in the Namoi Basin. In: Proc. international congress on modelling and simulation. MODSIM 1995. Newcastle, November 1995. 1995: 134138. Available from: Modelhng and Simulation Society of Australia, Canberra, Australia.