Monitoring changes in groundwater levels is helpful to detect the effectiveness of the rehabilitation treatments. Many researchers have studied groundwater level changes and attempted to explain them statistically. The common statistical analysis do not explain the changes in groundwater levels very well if the watertable is close to the soil surface. We present an approach for statistically estimating changes in groundwater levels including the impacts of treatments on those levels. This approach explains the changes in groundwater levels very well. The approach not only separates the effect of rainfall events from any underlying time trend but it also estimates the impact of other factors, such as evaporation and saltbush on groundwater levels. Some examples of application of the approach are presented.

Keywords: Saltbush; Atriplex; Groundwater; Monitoring; Treatment; Salinity; Rehabilitation of saline land; Sustainability indicators

By 1994, an estimated 1.8 million hectares of cleared land in Western Australia was affected by secondary dryland salinity to some extent, representing 9.4 percent of agricultural land in the state (Ferdowsian et al., 1996). This area is likely to double in the coming 20 years (to around 3.3 million ha) and may redouble again before a new equilibrium is reached. The salt-affected areas are usually the least productive part of the landscape. Many farmers have planted saltbush (Atriplex species to rehabilitate the saline land and change them to productive areas. Monitoring changes in groundwater levels is helpful to detect the effectiveness of these rehabilitation treatments.

Ferdowsian et al., (2001) presented a new approach called HARTT (Hydrograph Analysis: Rainfall and Time Trends) for statistically estimating groundwater levels. Their method differentiates between the effect of rainfall fluctuations and the underlying trend of groundwater levels over time. Rainfall is represented as an accumulation of deviations from average rainfall and the lag between rainfall and its impact on groundwater is explicitly represented. The HARTT method provides high quality fits to observed data in all but shallow bores (Ferdowsian et al., 2001), and even in these cases, the approach is superior to the time-trend-only approach.

The common statistical approach (and even the HARTT method) does not well explain the changes in groundwater levels under saltbush plantations. This is because groundwater levels under these areas are usually close to the soil surface and are affected by a diversity of factors such as rainfall, evaporation, vegetation, plant density, hydraulic gradient, hydraulic conductivity and depth to groundwater. We need a refined model for explaining changes in watertable under saltbush plantations.

The Bundilla farm (118° 48¢ E, 32° 55¢ S) is located 62 Km north east of Lake Grace in the wheat belt of Western Australia and in a 330 mm annual rainfall zone. The landscape consists of flats with level (<1%) plains and broad (between 1 to >3 Km wide), poorly defined stagnant sedimentary depressions. The broad depressions gradually change to stagnant broad flats with naturally saline areas and saline shallow playas. The upper parts of the broad depressions, which did not have primary salinity, have been cleared for agriculture. These areas have shallow (A horizon <0.25 m) duplex soils (Northcote, 1979). Excessive recharge due to common agricultural practices has caused the levels of very saline groundwater to rise and come close to soil surface (<1.0 m) resulting in expanding soil salinity. The aquifer is stagnant with level (<1%) hydraulic gradient and very little or no flow going out of the area. Consequently the groundwater levels have only one dimensional (up and down) movements.

The soils of the valley floor contain high levels of salt (Figure 1). At the soil surface, the salt storage is less than 5 kg m^-3, but it increases to greater than 15 kg m^-3 below 2 metres depth.

Figure 1: Salt storage increases with depth and large amount of salt is stored in the soil profile.

The farmer has planted large areas (hundreds of hectares) under saltbush to prevent soil salinity and change the area to productive land. The site is an important and a common place for Field Days and has attracted hundreds of farmers and researches to observe productivity of the site. More than 10 shallow bores (< 5 m) were drilled in the area in early 1994, to observe and record changes in groundwater levels and document the effect of saltbushes on watertable. The landholder has records of monthly rainfall and groundwater levels with only a few missing data in groundwater levels. The records are valuable data set for documenting the effectiveness of saltbush in soil salinity prevention. We have used the rainfall data and the records of 5 bores to document the changes in groundwater levels in various growth periods of saltbush.

The five selected bores were all drilled in early 1994 but saltbush planting and their present conditions varied (Table 1).

Table 1: The year and method of planting saltbush and their present conditions.

Watertable depth was hypothesised to depend on rainfall, time and saltbush according to the following functional relationship.

In some cases, the S_t variable was separated into two variables, representing the first 30 months of saltbush establishment, and subsequent months of saltbush. This separation allowed us to test whether the influence of saltbush on watertable depth changed over time.

The statistical analysis showed that despite having shallow groundwater levels, the explanatory power of the model was very good (Table 2). All parameters listed in Table 2 had significant (P<0.02) effect on groundwater levels. Between 68% and 84% of changes in groundwater levels in the 5 selected bores could be explained (Table 2). We consider these to be very good results in terms of statistical fit. In all cases monthly rainfall increased groundwater levels significantly and its effect lasted for 3 to 4 month after the event. To describe the effect of saltbush during its life cycle each bore will be reviewed separately.

Table 2: Statistical analysis results of the 5 selected bores in saltbush.

Notes: Monitoring period has been since 1994.

n.s. is for parameters which, were not significant (P>0.04) and were removed from the model. All parameters included in the table were significant with P values <0.02.

Bore ML9 was under saltbush for only 32 months and was in a location where saltbush plants were relatively sparse. The model explained 68% of the variation in watertable depth, despite having a few "dry" records where the watertable was below the depth of the bore (Figure 2).

There was no underlying time trend for groundwater level prior to planting saltbush. The watertable fluctuated above a –1.66 m baseline, with depth at any point in time depending on rainfall over the previous three months. Table 2 shows that for three of the five bores, the influence of rainfall on watertable depth was greatest for the most recent rainfall and progressively smaller for rainfall in the more distant past. Bore ML 9 was an exception to this; its parameters for monthly rainfall were similar for the three most recent months.

Saltbush caused a significant reduction (-0.016 m/month; 0.19 m/year) in groundwater level during its 32-month period of effect. The thin line to the right of September 1999 shows the result of using the estimated statistical model to predict what the groundwater depth would have been during this period if saltbush had not been established. By the end of the 32-month period, the total effect of the saltbush on groundwater level was estimated to be 0.5 m. In view of the relative sparseness of saltbush at this location, such a fall in level is encouraging. Given that the data set includes a large number of "dry" readings, the true impact on groundwater is likely to have been greater than the 0.5 m estimated.

Like bore ML9, bore ML10 was under saltbush for only 32 months and often during this period the watertable dropped below the drilling depth (Figure 3). The model explained 72% of the variation in watertable depth, as a function of rainfall and the presence of saltbush. Like bore ML9, the estimated influence of rainfall did not decline greatly with time (Table 2), extending to four months in this case. There was again no underlying time trend for groundwater level prior to planting saltbush. Watertable dropped by -0.031 m/month (0.37 m/year) during the period of saltbush, around twice the estimated rate for bore ML9.

Figure 2: Hydrograph for bore ML 9 showing the net effect of saltbush on groundwater levels. (Final effect was a drop of 0.5m). During a number of months groundwater dropped below the bore’s maximum depth of 1.8m.

Figure 3: Hydrograph for bore ML 10 showing the net effect of saltbush on groundwater levels. (Final effect was a drop of 0.9m). During a number of months groundwater dropped below the bore’s maximum depth of 2.34m.

Prior to planting saltbush there was an underlying time trend of 0.105 m rise per year in the groundwater level. For around 30 months after the establishment of saltbush, the estimated trend reversed and fell at 0.13 m per year. After that, it appears that the groundwater trend stabilised, with rainfall causing rises above a constant base depth of 1.4 m. Thus, in this case, the saltbush was able to prevent groundwater from rising to near the soil surface, but it was not able to lower the watertable below a depth of 1.4 m.

The thin line to the right of September 1997 shows the predicted watertable depth in the absence of saltbush. According to the model, groundwater would have been discharging at the surface for two periods. Note that the predicted depths for the without-saltbush case do not take account of the loss of water from these periods of discharge.

Bore ML 8

As expected, in the environment of the study area (shallow watertable, very low or no hydraulic gradient), groundwater levels have strong seasonal fluctuations due to local rainfall. The seasonality in groundwater levels could be significantly explained by monthly rainfall over the preceding four months. In two of the five case studies there was a slight rising trend in groundwater level prior to the establishment of saltbush, while in the other three, the long-term trend was flat.

The analysis show that saltbush can lower groundwater levels to below the capillary fringe and thus prevent ongoing worsening of soil salinity at the surface. Judging from the two bores that had been established for 56 months, this effect of saltbush on watertable effect was mostly completed within 30 months after planting. After that period, saltbush managed to keep groundwater levels at bay and prevent them from rising again.

There are impediments preventing saltbush from further lowering the watertable. Groundwater salinity (27,000 mg/L) and high salt storage (i.e. >15 Kg/m³) are likely to be prominent amongst these limiting factors. The same factors may have affected the natural vegetation prior to clearing. In these cases the saltbush plantations may have reverted the hydrological processes back to preclearing conditions.

In many areas increased concentration of chloride beneath stands of saltbushes can become a problem (i.e. Barrectt-Lennard and Malcolm, 1999). However, in the environment studied here, the production system appears to be sustainable, subject to survival of the saltbush plants. Because this part of the landscape has a one-dimensional groundwater flow system (i.e. movement is predominantly up and down, not to the side) salt buildup in the root zone of the saltbush plants will be prevented. This may not be the case where saline groundwater from another area flows to the root zone and replaces the depleted watertable.

Barrectt-Lennard, E.G. and Malcolm, C.V., 1999, Increased concentration of chloride beneath stands of saltbushes (Atriplex species) suggest substantial use of groundwater: Australian Journal of Experimental Agriculture, 39, 949-55,

Ferdowsian, R., R. George, F. Lewis, D. McFarlane, R. Short, and R. Speed, 1996, The extent of dryland salinity in Western Australia, In: Conference Proceedings, 4th National Conference and Workshop on the Productive Use and Rehabilitation of Saline Lands, Albany, Western Australia, 25-30 March 1996, Promaco Conventions: Perth, Western Australia, PP. 89-97, 1996.

Ferdowsian, R., D. J. Pannell, C. McCarron, A. Ryder, and L. Crossing, 2001, Explaining Groundwater Hydrographs: Separating Atypical Rainfall Events from Time Trends: Australian Journal of Soil Research, 39, 861-875.

Northcote, K.H., 1979, A Factual Key for Recognition of Australian Soils: Rellim, Adelaide, 1979.