Temporal and spatial patterns of salinity in a catchment of the central wheatbelt of Western Australia.
In Australia, dryland salinity is a major land degradation problem with over 2 Mha of broadacre farm land currently affected and scientists predicting up to 6 Mha to be at risk (PMSEIC 1999; NLWRA 2001; ABS 2002). The clearing of deep-rooted native vegetation and its replacement with shallow-rooted annual agricultural plants is the principal cause of dryland salinity.
Various estimates have been made of the current and future likely extent of salinity within Western Australia (e.g. George 1990; George et al. 2005); however, there has been little temporal analysis on individual catchments in Western Australia, with the notable exception of that conducted in the Upper Kent River catchment (Evans et al. 1996). Monitoring changes to the patterns of salt-affected land at these finer scales inform larger scale estimates of the rates of expansion and where in the landscape expansion is occurring (Summerell et al. 2009). Both spatial and temporal patterns, together with local factors such land-use, groundwater trends, variations in micro-relief, and local hydrological processes, will dictate the choice of the most appropriate salinity management response (Pannell 2001).
In the agricultural zone of Western Australia (WA), the state most affected by dryland salinity, information on the spatial and temporal patterns of salinity and landscape condition has been provided by the Land Monitor project (Allen et al. 1999). Land Monitor provides land managers and administrators with baseline salinity data for monitoring changes over time and estimates of areas at risk from secondary or future salinisation. It uses sequences of calibrated Landsat Thematic Mapper satellite images integrated with landform information derived from height data, ground truthing, and other existing mapped datasets to monitor changes in salinity and woody vegetation. The most recent estimates of LandMonitor were made in 1998. In this study we verify the accuracy of the 1998 estimates in one catchment and update it with estimates made by farmers in 2006-07 to gauge the recent rate of spread and determine associations between landscape position, trends in groundwater rise, and history and the appearance of new outbreaks of salinity.
Groundwater levels and their rates of change over time inform estimates of the future area of land at risk from salinisation (George et al. 2008) and were adopted as the basis for a consistent national analysis and reporting of salinity risk (NLWRA 2001). In addition, the drying trend in rainfall for much of the central and northern agricultural zone of WA during the last 30 years (Smith et al. 2000) has been related to the stable or falling trends in groundwater levels since 2000 in the some parts of the northern wheatbelt (George et al. 2008). Linking climate and salinity development has implications for the forecasts for the spread of salinity under an ongoing trend of a drying climate in the north, and more variable climate in the south-east.
This study is conducted in the Wallatin and O'Brien catchments in the low-medium rainfall zone of the central wheatbelt of WA, where mixed cereal and livestock farms predominate. The study catchments occur in the middle of the WA wheatbelt, and while no catchment could be considered typical of the whole wheatbelt region, these catchments may be regarded in terms of landscape, community, and environment as representative of much of it. The catchments are adjacent to the Woolundra Lakes catchment, a palaeodrainage which exhibits landforms associated with salinity (e.g. playas and lunettes), many of which have been reactivated since clearing.
The catchments have comprehensive soil-landscape mapping, history of vegetation clearing, groundwater trends, satellite salinity maps, and recent farmer records of new salinity (Dawes et al. 2007). This enables the project to address the following aims. (1) What have been the spatial and temporal trends in area and location of salt-affected land for 1998-2006 and how does this relate to 'at risk' estimates, soil-landscape patterns, groundwater trends, and underlying hydrogeology? (2) What is the among-farm variation in such trends and what are the implications for the range of salinity management responses required within a catchment? (3) What is the relationship between groundwater rise, time since clearing, and landscape position? (4) Has reduced rainfall changed the patterns in salinity development between 1998 and 2007?
The Wallatin and O'Brien Creek catchments are in the Shire of Kellerberrin 240km east of Perth (Fig. 1). The location represents the transition from the central to the eastern wheatbelt of Western Australia. The total area is 44457ha (including the Woolundra Lakes, a largely salt-affected area); elevation ranges from 380 to 240 m AHD in both catchments; catchment dimensions are 32 by 8-14 km for Wallatin and 18 by 9-16 km for O'Brien; and catchment gradient is 0.43 and 0.78% in Wallatin and O'Brien, respectively.
The long-term (1900-2006) average rainfall is 330 mm/year (recorded at Kellerberrin Post Office) and annual evaporation is ~2100 mm/year. Evaporation exceeds rainfall in all months of the year except June and July. On average, ~75% of the rain is received between May and September. Compared to other areas in Australia, variability in winter rain is relatively low but summer rainfall is variable and sporadic. Rainfall during the last 30 years (1975-2006, av. 304 mm) was 10% lower than the 1900-2006 long-term average with rainfall in June and July having dropped by >25% (Ludwig and Asseng 2006). During the same period there has been a small and not significant increase in summer rainfall. These changes are consistent with broader-scale changes in rainfall over the medium and high rainfall zones of the south-west of WA (Smith et al. 2000) and have related to changes in the rate of rise of watertables (George et al. 2008).
[FIGURE 1 OMITTED]
The broad landscape is described as the Zone of Ancient Drainage with typical landforms described in Table 1. The soil landscape units have been mapped at a scale of 1 : 50 000 and classified using a soil and topologically based classification (Bettenay and Hingston 1964). The Ulva unit, especially the 'upland sandplain', is highly permeable to rainfall and groundwater. Where this occurs alone or adjacent to the large rock monadnocks (Danberrin units), the potential for groundwater recharge is high. The alluvium and colluvium of the major and minor valley floors are most affected by salinity. The gradational soils adjacent to the lake systems have a complex mosaic of salinity that changes seasonally with fluctuating watertables and flooding.
Regolith thickness in the upper Wallatin Creek catchment (<15 m) is less than that in comparable wheatbelt catchments (20-30m), as indicated by drilling and geomorphological analysis, due to the abundance of Danberrin and Booraan soil systems (George 1992; McFarlane and George 1992). Much deeper regolith (40-50m) occurs in the lower Wallatin and O'Brien catchments due to the prevalence of deep alluvial sediments. However, the general characteristics of the terrain are typical of the region (George 1992).
Aquifers in the catchment can be classified as local groundwater flow systems (George et al. 1997), where most of the recharge within one sub-catchment becomes discharge within the same sub-catchment. Groundwater catchment size is a function of terrain, geology, and weathering history. In the study area, most local flow systems are of the scale of <100 ha in the uplands and 1000ha in valleys. Catchment salinity mechanisms are dominated by 4 causal processes (after Nulsen and Henschke 1981; George et al. 1997; Coram et al. 2000; Clarke et al. 2002): Type 1, bedrock highs; Type 2, local break of slope; Type 6, perched aquifers; Type 7, geological structures. Salinity is generally expressed in the catchment where there is a reduction in flow that results from shallow bedrock, convergent topography, changes in surface slope, or the presence of faults or dykes (McFarlane and George 1992; George et al. 1997). In the medium rainfall zone of the wheatbelt, recharge to groundwater aquifers is estimated to have increased from 0-1 to 10-30mm/year after clearing (George 1992; George and Coleman 2001).
Farming supports 27 families within the 2 catchments. The dryland fanning systems are based on grain production, predominately wheat and barley with lesser amounts of lupins and canola (Robertson et al. 2009). Livestock systems vary between properties but are not dominant in this area. Over 70% of natural vegetation was cleared for agriculture by the 1940s and now ~11% remains in scattered and generally small patches. There are 2 large nature reserves, both within the Wallatin Creek catchment: Durokoppin Reserve (1030ha) and Kodj Kodjin Reserve (204 ha) (Fig. 1). The catchments are atypical of the wheatbelt in that, to 2006, landholders have planted 1750 ha of woody perennial vegetation adding 4.9% further cover to the 9.4% cover of uncleared remnant native vegetation (Smith 2008). However, the number and area of new tree plantings have reduced during the last 5 years (Smith 2008). The area mapped as saline in the shire in 1998 is 6% (Caccetta and Beetson 2000); however, individual farms especially in the valley floor can have a large proportion of area affected.
Area and spatial pattern of salt-affected land
The LandMonitor dataset for area of consistently low productivity (AOCLP) in the catchments was available for 1989 and 1998 (Caccetta and Beetson 2000). AOCLP is defined as areas of consistently low normalised difference vegetation index (NDVI) across at least 3 consecutive spring images, ground-truthed by field inspection. Estimates of AOCLP were verified in the field and errors of commission and omission registered against mapped areas of dryland salinity. The Wallatin Creek catchment was selected as one of several state-wide to assess the methodology. Caccetta and Beetson (2000) define the methodology and results in detail.
An aerial photo of each farm, flown in November 2004, was superimposed with LandMonitor 1998 AOCLP and presented to each farmer. Farmers were first asked to verify the accuracy of the 1998 LandMonitor estimates against what land on their farm they remember as being saline/not saline in 1998. Farmers were then asked to state whether land that had been classified as saline by LandMonitor in 1998 was in fact not saline (errors of commission). Farmers were also asked to identify land that was saline in 1998 (according to their memory) but was not classified as such by LandMonitor (errors of omission). Next, farmers were asked to add to the current aerial photograph any new salinity that had appeared since the 1998 estimates, with their definition of saline-affected land being guided by the LandMonitor classification in 1998. This ensured that definitions of salinity were consistent between LandMonitor and the farmers. They were also asked to indicate if any saline areas present in 1998 had since disappeared; however, there were no recordings of this kind. Farmer hand-drawn estimates of new salinity were digitised and added to the 1998 estimates.
The updated map of salt-affected land was then analysed as follows. Each discrete polygon of salt-affected land, deemed to have consistently low production, was attributed an area and then classified as being a new outbreak (>100 m from an existing saline patch) or an outgrowth from an existing patch previously recorded by LandMonitor. Polygons were also classified as was whether or not they were located within the valley floor zone (below the break of slope). Polygons that straddled the valley floor boundary were classified according to where the dominant proportion fell. Polygons were classified for location on this basis because, as a generalisation, topographically low areas are the first where watertables come close to the soil surface, with associated water logging and salinity problems. This is the rationale behind using area of the valley floor zone as a measure of potential salinity hazard (George et al. 2005). The valley floor zone is defined using height data and is described in more detail by Caccetta et al. (2010). Using this definition, the valley floor area occupies 25 and 28% of Wallatin and O'Brien catchments, respectively.
Soil survey of salt-affected land
An additional estimate of the area of salt-affected land, independent to that of the farmers and LandMonitor, was conducted via a catchment-wide soil survey. An on-ground survey of surface salinity was conducted, where soil sampling and analysis of soil for electrical conductivity was used to indicate the location of current salinity. This method requires an approach to extrapolate from point samples to an areal estimate, by assigning an average value of salinity to an entire soil landscape unit. A detailed soil survey was commissioned in the catchment (Wells 2004) and soils were mapped at 1 : 50 000 scale. The soil landscape map had 67 soil landscape units mapped, indicating both a landscape position and a probable WA soil group (not differentiated by soil colour) (Schoknecht 2002). Soil profiles, which were sampled at 197 points in the catchment, based on representation of the major landforms (Table 1), were classed into soil landscape units (Schoknecht et al. 2004) and WA soil group (Schoknecht 2002) and measured for soil salinity. Soil salinity was measured in 1 : 5 soil water mixture (EC) then converted into saturated extract values (ECe) using the EC and a soil conversion factor (based on soil type) from DAFWA (2004). This was checked against expected soil properties based on their WA soil group type. The topsoil and subsoil ECe were broken into the following salinity classes (mS/m): fresh <200, slightly saline 200-400, brackish/moderate 400-800, highly saline 800-1600, extremely saline >1600. Land was classed as saline if it was either extremely saline or highly saline.
Each soil-landscape mapping unit was assigned a subsoil and overall salinity based on the most common properties from the point data. Hence, maps of soil salinity are for most probable property for that soil landscape unit.
Time trends and spatial patterns in the area of salt-affected land were analysed by comparing estimates made from LandMonitor, farmer estimates, and hydrograph analysis.
Groundwater catchment areas
Estimates were made of the likely size of the groundwater catchment area underlying saline seeps on each farm in the catchments. A groundwater catchment area is defined as that area where recharge is contributing to discharge at a saline seep, and hence can be thought of as the area requiring management intervention to reduce recharge contributing to the saline area.
A comprehensive survey of groundwater catchments was not conducted; however, a subsample was made, being those groundwater catchments associated with saline seeps that were of current concern to farmers. Of the 494 discrete saline patches identified in 2006, we analysed 17% on 22 farms that were visited over the period November 2006 to January 2007. A discussion was held on the current salinity risk situation being faced by the farmer, and the possible location of options to manage current saline patches or seeps identified by the farmer. The discussion was followed by a field visit to these seeps to evaluate the possible causes of current and future salinity, and hence possible options to manage salinity at the site. An assessment was made of the boundary and area of the groundwater catchment area contributing to the identified seep, using the approach described by Lewis (1991) and informed by information on any nearby bedrock or regolith exposures, the shape and characteristics of the surface of the landscape, and the shapes and locations of any groundwater discharge features. On-site visual inspections were supplemented with air photos flown in 2000 and November 2004, LandMonitor 1998, estimates of salt-affected land, groundwater records, and maps of the main soil types, airborne gamma-K radiometrics, and airborne magnetic intensity. It should be noted that in such landscapes, the groundwater catchment boundaries may not coincide with the boundaries of the surface water catchment (Clarke et al. 2002), although it is the exception that groundwater would flow across surface water catchment boundaries; rather, multiple groundwater catchments could be found within a single surface water catchment. Once the boundary of the contributing groundwater catchment was identified, feasible options to respond to the issues at each site were discussed with the farmer.
Date of clearing
Remnant native vegetation had previously been mapped for the area in detail through farmer interviews, original survey maps, and aerial photography (Arnold and Weeldenburg 1991; G. W. Arnold and J. R. Weeldenburg, unpubl, data), which enabled a date of vegetation clearing to be assigned to individual parcels of land, with an accuracy of [+ or -] 5 years. Clearing dates were aggregated into the following classes for mapping and analysis: 1890-1920, 1921-40, 1941-60, 1961-72, 1973-74, 1975-85, and 2 further classes where the land was uncleared or the date of clearing was unknown. Where possible these dates were checked against the dates of alienation of these parcels from State Lands Department records.
Analysis of groundwater levels
Although Wallatin and O'Brien catchments have been the subject of much hydrological investigation over many years, the hydrograph record was patchy. Data are available from 1985 to 1989, when Department of Agriculture and Food installed the piezometers. However, there is a gap for most years to 2005. Unfortunately, as many of the original piezometers were drilled to investigate the causes of the saline seeps (McFarlane and George 1992), many (68%) of the piezometers are located in lower landscape positions and valley floors, which have been cleared of native vegetation for longest. As a result, this biases the total set of hydrographs towards those that are likely to have stabilised, many of which have watertables near the ground surface.
Hydrographs were inspected and minimum rates of change calculated for the period between the late 1980s, when a large number of observations were made, and 2005, when piezometer readings were resumed. The minimum rate of change was estimated from the average measured groundwater level before 1990, measurements in 2005, and the length of record; there were 12-38 measurements before 1990. To determine whether the level had increased, stabilised, or decreased over time, the average 2005 depth was compared to the average pre-1990 depth in a standard 2-tailed significance test at the 95% confidence level.
The year of clearing at the location of the piezometers, inferred from land alienation records and farmer interviews, was subtracted from 2000 to get an approximate time since clearing. There were 48 bores, 34 of which were from cleared agricultural land, 13 from areas near the border between uncleared and cleared land, and 1 from the interior of a nature reserve and hence classified as 'uncleared'. Correlations were conducted between the minimum rate of change and time since clearing. Bores were also defined by landscape position ranging from the upper slope to the valley floor.
To assist in interpretation of temporal patterns of salinity during the period when there were no groundwater readings (1989-2005), HARTT (Ferdowsian et al. 2001) was used to interpolate the groundwater levels for several bores on the valley floor at Wallatin Creek for the period between September 1989 and February 2005, using data for 1985-89 and 2005-2009. The algorithm separates the influence of rainfall variability from any underlying trend in groundwater levels by using either the accumulated monthly or accumulated annual rainfall residuals as a surrogate for the influence of rainfall variability on groundwater levels.
Time trends and spatial patterns in the area of salt-affected land
The accuracy of the 1998 LandMonitor estimates was assessed by farmers for their own farms in 2006. The area of land that had been classified as saline by LandMonitor in 1998 and was in fact not saline (i.e. errors of commission) amounted to a total of 10ha over 28 patches for the combined area of Wallatin and O'Brien, which is a small error (~1%) compared to the total area of 900 ha classified as saline. Farmers were also asked to identify land that was saline in 1998 (according to their memory) but was not classified as such by LandMonitor (errors of omission). This area was 142 ha covering 31 patches, the majority (90ha) being in the valley floor (data not shown). When these corrections are applied to the 1998 LandMonitor AOCLP, then 87% of the salinity in Wallatin and O'Brien occurs within the valley floor area.
The combined area of existing salinity identified by LandMonitor in 1998 for Wallatin and O'Brien Creek catchments is 1045 ha (Table 2). The total for Wallatin of 550ha is close to the 496ha reported by McFarlane and George (1992) in their ground survey of salinity, even though in their survey they only mapped the larger patches of salt-affected land. Net increase in salinity between 1989 and 1998 has been close to zero in the O'Brien catchment and within the error of estimation from LandMonitor (Allen et al. 1999) and an increase of 130 ha in the Wallatin catchment.
The estimates of surface soil salinity derived from soil survey conducted in 2003 were close to those estimated from LandMonitor in 1998 (Table 2). Most of the surface soil salinity was located on the Baandee and CoUgar soil-landscape units (Table 1).
Table 2 includes an update of the area of saline land from 1998 to the present as provided by farmer estimates made in 2006. According to these farmer estimates, between 1998 and 2006, the area of salt-affected land doubled (an increase of 537ha, 49%) in the Wallatin catchment and increased by ~119ha or 20% in O'Brien. An analysis of farm-by-farm area of salt-affected land shows that between 1998 and 2006 the mean increase in the area of salt-affected land per farm was 28 ha and ranged from <10ha on 14 farms up to 133ha on the most affected farm (Fig. 2b). All but 3 farms conformed to the trend that the higher the percent of the farm affected by salinity in 1998, the greater the increase in saline area between 1998 and 2006. There were 3 farms in the valley floor that were notable exceptions to this trend. On 2 of these farms, <60 ha of the farm was salt-affected in 1998, but the increase in saline land between 1998 and 2006 was >120ha. On the third farm, 24% was salt-affected in 1998 but the increase in saline land between 1998 and 2006 has been <5 ha. In 2006, the percentage of each farm with salt-affected land ranged from 1 to 30% with a mean of 5.4%, with two-thirds of farms having <5% salt-affected land (Fig. 2a).
In 1998 there were 131 identifiable discrete patches of salt-affected land on the hill slopes of the Wallatin and O'Brien catchments, with a mean size of 0.8 ha and a median of 0.3 ha. In the valley floor there were 274 patches with a mean area of 3.2 ha and a median of 0.4 ha, indicating that a few large patches were dominating the mean patch size. Since 1998, farmers indicated that there have been 52 new outbreaks of salt-affected land in the valley floor of Wallatin and O'Brien. The mean size of these outbreaks was 10.7 ha, but the median was 3.6ha indicating a skewed distribution (Fig. 3). Outside the valley floor, on the adjacent hill slopes, there were 37 new outbreaks, with a mean and median of 2.8 ha and 2.3 ha, respectively, indicating less skewness in the distribution of salt patches compared to the valley floor. Approximately 80% (552 ha) of the new salinity was expansion of already existing patches, while 20% (107 ha) was in new isolated patches.
Groundwater flow compartments
Analysis of saline seeps on 22 farms in the catchment involved identifying 85 associated groundwater catchment areas. The frequency distribution of the area of these contributing groundwater catchments shows that most are <150ha, many <80ha, and none >500ha (Fig. 4). In addition we did not find any cases where the groundwater catchment on a farm, associated with a saline outbreak, straddled property boundaries.
A plot of the 62 groundwater bores (excluding 2 that had static water levels greater than 15 m before 1989 and also in 2005) shows that there were 36 bores with water levels closer than 2 m of the surface in 1989 and these showed little change in 2005 (Fig. 5a). Bores that had deeper water levels in 1989 had the greatest minimum rate of rise between 1989 and 2005 (Fig. 5b). Of the 62 groundwater bores examined for trends between 1989 and 2005: (i) 16 (26%) had declining trends from -0.005 to -0.15 m/year, (ii) 17 (27%) were stable with trends from -0.005 to +0.005 m/year, and (iii) 29 (47%) were rising at rates from +0.005 to +0.4 m/year (Table 3). Of the 62 bores, 49 (80%) were located in the mid/lower or lower positions of the landscape and around half of these were either stable or declining in trend and the other half were rising.
[FIGURE 2 OMITTED]
[FIGURE 3 OMITTED]
[FIGURE 4 OMITTED]
[FIGURE 5 OMITTED]
For all hydrographs analysed using HARTT (Ferdowsian et al. 2001), the p statistic for the accumulated annual residual rainfall was highly significant as a determinant for groundwater level, giving confidence that the HARTT analysis could be used to interpolate groundwater levels for the period when there were no readings. Two examples of the interpolated groundwater levels are shown in Fig. 6. In both cases the strong relationship between accumulated annual residual rainfall and the predicted groundwater level is obvious; the p statistics for the residual rainfall were 2.2 x 1045 and 4.1 x 10-5, respectively. The figures also show that when the accumulated residual rainfall was consistently positive between mid-1999 and the end of 2001, groundwater was predicted to be at its shallowest under the valley floor. Furthermore, accumulated annual residual rainfall peaked in January 2000 and April 2000 in response to 95 mm of rain in January 2000 and 55 mm in March 2000.
Clearing and salinity
In the Wallatin and O'Brien Creek catchments, land clearing started in the lower catchment adjacent to the Great Eastern Highway, before the First World War, and was largely completed by the start of the Second World War (Fig. 7). By contrast, large areas of the upper catchment remained under native woodlands and heath until the 1970s. McFarlane and George (1992) linked the delayed response of salinity in the upper Wallatin catchment to the existence of remnant nature reserves (Durokoppin and Kodj Kodjin Reserves, Fig. 1) and this later clearing.
[FIGURE 6 OMITTED]
Land clearing information was used to explicitly examine the link between clearing and trends in groundwater level. Figure 8 shows the simple regression of rate of change in groundwater levels against years since clearing and shows an expected negative trend--the longer the time since clearing the lower the minimum rate of rise as groundwater levels come to equilibrium. For land cleared in the last 70 years, all water levels are rising, whereas for land cleared earlier than this, all water levels are stable or falling. Bores that were located near the border between cleared and uncleared land have a rate of groundwater rise that is above the trend line for bores from cleared land, indicating that such bores are behaving as if they had been cleared more recently than the date of clearing would indicate. This suggests that the adjacent native vegetation is affording some protection, thus slowing rates of watertable rise.
Assessment of salt-affected area
This study verified the overall accuracy of the LandMonitor estimates of salt-affected land (areas of consistently low productivity) through checking against farmer assessments and the soil survey. It gives confidence that the method is reliable, at least in this landscape of the WA wheatbelt, and can be expected to give reasonable updated estimates of saline-affected land if a new survey were conducted. Errors of commission were small and within the expected error of the methodology of 3-13% (Allen et al. 1999). Errors of omission (e.g. the 90 ha in the valley floor of the Wallatin catchment that was saline in 1998 according to farmer memory, but was not classified as such by LandMonitor) were more significant.
The area of salt-affected land as at 2006, with updates from farmers, was assessed to be 4.4% and 5.3% of the Wallatin and O'Brien catchments (Table 2), far less than the 25% and 30% of these catchments that occurs within the valley and indicates strongly that valley area is not equivalent to the area at risk. This also substantiates the use of the term 'Valley Hazard' by George et al. (2005) to denote where shallow watertables (and high subsoil salinity) are likely; they note that the actual area at risk depends on local factors such as land-use, groundwater trends, variations in micro-relief, and local hydrological processes that are not well represented by current mapping and modelling technology.
The increase in saline land since 1998, both in the valley floor and on adjacent slopes, indicates that salinity is an ongoing land degradation problem even in a catchment that had been largely cleared for agriculture by the 1940s and where many of the bores in the valley floor show that groundwater levels have stabilised. Most (80%) of the new salinity is expansion of existing areas in the valley floor, although a significant number of outbreaks have also occurred on the adjacent slopes. As new outbreaks are larger in the valley floor, this confirms that the on-going threat is still greatest in this landscape position. The disproportionate increase in the number of saline patches (associated with later clearing) on the slopes suggests that salinity will continue to spread in this part of the landscape.
[FIGURE 7 OMITTED]
Clearing and salinity
In other areas of the South-West of WA, the dominant driver behind the observed trends in salt-affected area has been shown to be strongly related to time since clearing and groundwater depth (Peck and Williamson 1987). Similarly George et al. (1997) indicates this is a process that underpins the salinisation of the wheatbelt, and as outlined in Fig. 8, is implicated as a major driver in continued salinisation in the study catchments. However, as the catchments approach equilibrium, a process that may take 30-200 years (George et al. 1997), climate begins to dominate changes in water level.
Climate and salinity
A significant finding from this study has been the documented increase in salt-affected land between 1998 and 2006 in both the Wallatin and O'Brien catchments. In particular, during this period, 2 valley floor farms experienced an increase in salt-affected land of >120 ha (to >25% of farm area) where only 5% (<60 ha) of their farms were saline-affected in 1998. This also took place in a period when the valley groundwater bores displayed either a stable or falling trend (Table 3). It also occurred within an area of the landscape that had been largely cleared for >60 years (Fig. 7).
We attribute this response to several factors, in particular the 12 months between March 1999 and April 2000 when 500 mm of rain fell, and to a lesser degree, a similarly wet period from March 2005 to February 2006 when 450 mm fell. Both periods included January rainfalls of 100 mm. Figure 5 shows this trend as a plot of accumulated deviations of monthly rainfall from the long-term mean and resulting interpolated groundwater levels for 2 valley floor bores. The figure shows the relatively low rainfall period (negative slope) from 1975 to 1988 when McFarlane and George (1992) undertook baseline analysis of the area of salinity, and the period of positive slope (rainfall accumulation) in the early 1990s and that identified above. The climate data identify that salinity development is underpinned by groundwater trends, and depth to the watertable, but subject to expansion after wet phases, such as between 1999 and 2000 when the interpolated groundwater levels approached the soil surface. This pattern of wet winters and peak summer rainfalls spikes watertables, as also observed by Speed and Kendle (2008) in WA, and thus exposes the capillary fringe to high evaporative conditions (Nulsen 1981) and results in active salinisation.
[FIGURE 8 OMITTED]
In the case of the 2 farms that experienced a rapid expansion in salinity, both noted a difficulty in crop germination coming out of pastures that they attributed to high soil salinity levels. This suggests that salt was not leached from the profile by winter rains as the watertable dropped. With pressure on gross margins due to higher input costs, farmers are reconsidering cropping risky areas. If valley soils with higher watertables are prone to such episodic events, their place as the most reliable and productive parts of the farm business becomes questionable.
The mapping of the size and location of salinity outbreaks and the inferred groundwater catchment areas revealed the highly localised nature of the geo-hydrology of the region. Groundwater catchments contributing recharge to saline outbreaks of concern to farmers were mostly <150 ha and without exception contained within the farm boundaries. Associated with this is the spatial pattern of saline outbreaks. After the 2006 update of salt-affected land we documented 494 discrete patches of saline land in the 35000 ha of Wallatin--O'Brien, and 89 of these had emerged in the 1998-06 period. Although 85% of the new saline area is found in the valley floor (similar to the distribution of preexisting salinity), only 58% of the new patches occur there. Larger patches of new salinity are occurring in the valley floor than on the adjacent slopes, but there are ongoing smaller outbreaks on the adjacent slopes.
Implications for farm management
The spatial pattern of salinity in a farming landscape affects the choice of hydrologically effective and feasible management options that are compatible with farm layout and operations, and the degree to which the benefits of intervention will be captured by the farmer concerned (Pannell et al. 2001). Small and isolated patches, while potentially easy to treat or contain within farm boundaries, may not be compatible with farm layout and operations, particularly in crop-dominant farming systems. On the other hand, spread of salinity adjacent to large outbreaks in the valley floor may require little modification to farm plans because of the ability to manage large contiguous patches of salt-affected land.
The localised nature of groundwater catchments and salinity outbreaks found in these catchments has several implications for managing salinity. Small areas requiring treatment means that options can be deployed at manageable scales and do not require large capital outlay or disruptive changes to farm layout. Benefits will accrue largely within the farm boundaries of where they are treated and hence will not be diluted by benefits accruing to neighbours. This acts as an incentive for farmers to invest their own funds combating problems on their own farms (Pannell et al. 2001). A further implication is that there will be limits to which coordinated action by groups of farmers will have an aggregate impact at the catchment scale that is greater than the sum of its parts. Farmers in these catchments now perceive salinity as a manageable problem and recognise the options that they can adopt can be implemented within the boundaries of their farms.
The negativities associated with the widespread occurrence of small treatment areas are that such areas may be difficult to deal with within the constraints of farm operations, layout, and management. For instance, small areas of lucerne will be difficult to manage with large cropping paddocks and small mobs of sheep. Engineering options such as deep open drains may not be effective along their full length if the characteristics of groundwater flow systems change over short distances. Options that deal with discrete seeps on slopes, such as siphons, may be a better choice.
One apparent precaution to protect against the effects of the wetter years such as 2000 and 2006 would be to both maximise cropping and/or increase the area of perennial pastures (e.g. lucerne or salt-tolerant systems) on the valley floor to maintain depleted soil water conditions to act as a buffer against spiked watertables that are associated with above-average winter and summer rains. This is particularly important in catchments that are likely to have of the order of 25% of the valley areas with shallow watertables and for which projected climate change may deliver more unusual rainfall events (Hatton et al. 2003).
Our results have confirmed that satellite-derived salinity maps provide a reliable base-line for saline mapping, and that groundwater monitoring and time since cleating provide the reliable predictors of salinity risk. This study has highlighted the ongoing threat of salinity in the agricultural areas of Western Australia against a background where much of the landscape has been cleared for 60 years or more, groundwater levels have equilibrated in the valley floor, and rainfall has been below average for the past 30 years. Increases in the area of salinity are changing from that dominated by increases in the valley floor to now being associated with the development of small, isolated outbreaks on the adjacent slopes. Episodic rainfall in areas of shallow watertables is a significant cause of the expansion in observed salinisation. The widespread occurrence of small treatment areas means that such areas may be difficult to deal with within the constraints of farm operations, layout, and management.
This research was supported by the Grains Research and Development Corporation, CSIRO's Water for Healthy Country Flagship, and the Catchment Demonstration Initiative. Thanks to Fay Lewis for data from hydro-geological analysis and many stimulating discussions that influenced this paper. Martin Wells of Land Assessment Pty Ltd provided the soil map for Wallatin--O'Brien. Input from the landholders of Wallatin--O'Brien in mapping salinity is gratefully acknowledged. Hamish Cresswell provided helpful comments on an early draft of the paper.
Manuscript received 14 July 2009, accepted 15 January 2010
ABS (2002) 'Salinity on Australian farms.' Bulletin 4615.0, (Australian Bureau of Statistics: Canberra, ACT)
Allen A, Beetson B, Behn G, Caccetta P, Campbell NEP, Evans F, Furby S, Kiiveri H, Mauger G, McFarlane D, Napier A, Pearce C, Smith R, Wallace J, Wallis R, Watson I (1999) The LandMonitor Project: A Multi-agency project of the Westem Australian Salinity Action Plan supported by the Natural Heritage Trust. In 'Proceedings of WALIS Forum 1999'. Perth, W. Aust., March 1999, pp. 74-77.
Arnold GW, Weeldenburg JR (1991 ) 'The distributions and characteristics of remnant native vegetation in parts of the Kellerberrin-Tammin-Trayning and Wyalkatchem Shires of Western Australia.' (CSIRO: Western Australia)
Bettenay E, Hingston FJ (1964) Development and distribution of soils in the Merredin area, Western Australia. Australian Journal of Soil Research 2, 173-186. doi:10.1071/SR9640173
Caccetta P, Dunne R, George RJ, McFarlane D (2010) A methodology to estimate the future extent of dryland salinity in the south-west of Western Australia. Journal of Environmental Quality 39, 26-34. doi: 10.2134/ jeq2009.0037
Caccetta PA, Beetson B (2000) Mapping and monitoring salinity: the Kellerberrin Landsat TM scene (111-082). A report from the NHT-funded LandMonitor Project. CSIRO Mathematical and Information Sciences Report No. CMIS 2000/202.
Clarke C J, George RJ, Bell RW, Hatton TJ (2002) Dryland salinity in southwestern Australia: its origins, remedies, and future research directions. Australian Journal of Soil Research 40, 93-113. doi: 10.1071/SR01028
Coram JE, Dyson PR, Houlder PA, Evans WR (2000) 'Australian groundwater flow systems contributing to dryland salinity.' (Bureau of Rural Sciences: Canberra, ACT)
Dawes W, Robertson MJ, George R, Lewis F, Hodgson G, O'Connor MH, Wilkes P, Oliver Y (2007) Salinity risk assessment: Methods and case studies from the Wallatin and O'Brien catchments in the Central Wheatbelt of Western Australia. Unpublished report.
Department of Agriculture and Food Western Australia (2004) Salinity measures, units and classes. Available at: www.agric.wa.gov.au/ content/lwe/salin/smeas/salinity_units.htm (accessed 30 June 2008).
Evans FH, Caccetta PA, Ferdowsian R (1996) Integrating remotely sensed data with other spatial data sets to predict areas at risk from salinity. In 'Proceedings of the 8th Australasian Remote Sensing Conference'. Canberra, 1994.
Ferdowsian R, Pannell DJ, McCarron C, Ryder AT, Crossing L (2001) Explaining groundwater hydrographs: Separating atypical rainfall events from time trends. Australian Journal of Soil Research 39, 861-876. doi: 10.1071/SR00037
George R, Coleman M (2001) Hidden menace or opportunity--groundwater hydrology, playas and commercial options for salinity in wheatbelt valleys. In 'Wheatbelt Valleys Conference 2001'. Merredin, W. Aust.
George R, Kingwell R, Hill-Tonkin J, Nulsen B (2005) 'Salinity Investment Framework: Agricultural Land and Infrastructure.' Resource Management Technical Report 270 (Western Australian Department of Agriculture: Perth)
George RJ (1990) The 1989 saltland survey. The Journal of Agriculture, Western Australia (4th series) 31, 159-166.
George RJ (1992) Hydraulic properties of groundwater systems in the saprolite and sediments of the wheatbelt, Western Australia. Journal of Hydrology 130, 251-278. doi:10.1016/0022-1694(92)90113-A
George RJ, McFarlane DJ, Nulsen RA (1997) Salinity threatens the viability of agriculture and ecosystems in Western Australia. Hydrogeology Journal 5, 6-21. doi:10.1007/s100400050103
George RJ, Speed RJ, Simons JA, Smith RH, Ferdowsian R, Raper GP, Bennett DL (2008) Long-term groundwater trends and their impact on the future extent of dryland salinity in Western Australia in a variable climate. In 'Proceedings of the 2nd International Salinity Forum'. 30 March--3 April 2008, Adelaide, Australia. Available at: www. internationalsalinityforum.org/14_final.html
Hatton TJ, Ruprecht J, George RJ (2003) Preclearing hydrology of the Western Australian wheatbelt: Targets for the future? Plant and Soil 257, 341-356. doi:10.1023/A:1027310511299
Lewis MF (1991) Lineaments and salinity in Western Australia carriers or barriers? In 'Institution of Engineers Australia, International Hydrology and Water Resources Symposium'. Perth, W. Aust. 1991, pp. 202-209.
Ludwig F, Asseng S (2006) Climate change impacts on wheat production in a Mediterranean environment in Western Australia. Agricultural Systems 90, 159-179. doi:10.1016/j.agsy.2005.12.002
McFarlane DJ, George RJ (1992) Factors affecting dryland salinity in two wheatbelt catchments in Western Australia. Australian Journal of Soil Research 30, 85-100. doi:10.1071/SR9920085
NLWRA (2001) 'Australian Dryland Salinity Assessment 2000.' (National Land and Water Resources Audit: Canberra, ACT)
Nulsen RA (1981) Critical depth to saline groundwater in non-irrigated situations. Australian Journal of Soil Research 19, 83-86. doi: 10.1071/ SR9810083
Nulsen RA, Henschke CJ (1981) Groundwater systems associated with secondary salinity in Western Australia. Agricultural Water Management 4, 173-186. doi:10.1016/0378-3774(81)90049-4
Pannell DJ (2001) Dryland salinity: economic, scientific, social and policy dimensions. Australian Journal of Agricultural and Resource Economics 45, 517-546. doi:10.1111/1467-8489.00156
Pannell DJ, MacFarlane DJ, Ferdowsian R (2001) Rethinking the externality issue for dryland salinity in Western Australia. Australian Journal of Agricultural and Resource Economics 45, 459-475. doi: 10.1111/14678489.00152
Peck A J, Williamson DR (1987) Hydrology and salinity in the Collie River Basin, Western Australia. Journal of Hydrology 94, 1-198.
PMSEIC (1999) Dryland salinity and its impact on rural industries and their landscape. Occasional Paper No. 1, Prime Minister's Science, Engineering and Innovation Council, Department of Industry Science and Resources, Canberra.
Robertson MJ, Kingwell R, Measham T, O'Connor M, Batchelor G (2009) Constraints to landholders managing dryland salinity in the Central wheatbelt of Western Australia. Land Degradation and Development 20, 235-251. doi:10.1002/ldr.887
Schoknecht N (2002) 'Soils Groups of Western Australia A simple guide to the main soils of Western Australia.' 3rd edn. Resource Management Technical Report 246. (Agriculture Western Australia: Perth)
Schoknecht N, Tille P, Purdie B (2004) Soil--landscape mapping in South-Western Australia Overview of methodology and outputs. Resource Management Technical Report 280, Department of Agriculture, Western Australia, November 2004.
Smith FP (2008) Who's planting what, where and why and who's paying? An analysis of farmland revegetation in the central wheatbelt of Western Australia. Landscape and Urban Planning 86, 66-78. doi:10.1016/ j.landurbplan.2007.12.008
Smith IN, McIntosh P, Ansell TJ, Reason CJC, McInnes K (2000) Southwest Western Australian winter rainfall and its association with Indian Ocean climate variability. International Journal of Climatology 20, 1913-1930. doi: 10.1002/1097-0088(200012)20:15<1913::AID-JOC 594>3.0.CO;2-J
Speed RJ, Kendle AL (2008) Groundwater declines in response to a drier climate in the Northern Agricultural Region of Western Australia. In 'Proceedings of the 2nd International Salinity Forum'. 30 March-3 April 2008, Adelaide, S. Aust. Available at: www.international salinityforum.org/Final%20Papers/speed_A1.pdf
Summerell GK, Miller M, Beale G, Emery K, Lucas S, Scown J, Spiers P (2009) Current and predicted minimum and maximum extents of land salinisation in the upland NSW portion of the Murray Darling Basin. 1SBN 978-1-74122-872-4, DECC 2009/107, Available at: http:// goulbwb01/salinity/science/assessments.htm
Wells M (2004) Soil landscape units of the Wallatin and O'Brien catchments near Kellerberrin, Western Australia. Report 0412 prepared by Land Assessment Pty Ltd for CSIRO Sustainable Ecosystems.
M. J. Robertson (A,D), R. J. George (B), M. H. O'Connor (A), W. Dawes (C), Y. M. Oliver (A), and G. P. Raper (B)
(A) CSIRO Sustainable Ecosystems, Private Bag 5, PO Wembley, WA 6913, Australia.
(B) Department of Agriculture and Food WA, PO Box 1231, Bunbury, WA 6231, Australia.
(C) CSIRO Land and Water, Floreat, WA, Australia.
(D) Corresponding author. Email: Michael.Robertson@csiro.au
Table 1. Soil landforms of the WallatinO'Brien catchment (based on Bettenay and Hingston 1964) Soil landform Soil landform description Baandee Ancient drainage zone Belka Broad flat alluvial (upper) valley floor (mainly 'sandier' soils) Booraan Upper dissected valley slopes Collgar Lower valley slopes Danberrin Irregular low hills and gentle slopes with granite outcrops Merredin Broad flat alluvial (lower) valley floor (mainly 'heavier' soils) Rocky hills Steeper hills with large areas of granite outcrop Ulva Remnants of lateritic sandplain % of soil Soil Dominant native landform landform vegetation in catchment Baandee Halophytes 3.5 Belka Salmon gum 1.9 Booraan White gum 21.8 Collgar Mallee 13.0 Danberrin York gum 22.1 Merredin Salmon gum 16.4 Rocky hills Tamma 4.1 Ulva Wodjil, grevillea 17.2 Table 2. Estimate of saline land (ha) for the Wallatin and O'Brien catchments Saline Catchment Valley area (A) floor area 1989 1998 Change 1989-98 Wallatin Creek 6237 (25%) 421 (1.7%) 550(2.2%) 129 O'Brien Creek 3407 (30%) 503 (4.4%) 495 (4.3%) -8 Farmer estimates of Saline area (B) Catchment saline area increase Surface Subsurface (1998-06) soil soil Wallatin Creek 537 550(2.2%) 4648 (19%) O'Brien Creek 119 736(6.4%) 2815 (25%) (A) As assessed by LandMonitor area of consistently low productivity. (B) As assessed by catchment-wide soil survey. Table 3. Number of bores with declining, stable, or rising trends between 1989 and 2005 by landscape position Landscape position Total Groundwater Upper/ Mid/ trend Upper mid Middle lower Lower Declining 3 0 0 2 11 16 Stable 2 0 1 3 11 17 Rising 0 2 5 2 20 29
|Printer friendly Cite/link Email Feedback|
|Author:||Robertson, M.J.; George, R.J.; O'Connor, M.H.; Dawes, W.; Oliver, Y.M.; Raper, G.P.|
|Publication:||Australian Journal of Soil Research|
|Date:||Jul 1, 2010|
|Previous Article:||Spectral and wavelet analysis of gilgai patterns from air photography.|
|Next Article:||Surface charge characteristics of variable charge soils in Thailand.|