Preferential Flow

ASSESSING THE IMPACT AND MAGNITUDE OF MACROPORE FLOW ON WATER AND SOLUTE TRANSPORT BY USING WICK LYSIMETER

We have conducted extensive field and laboratory investigations using a variety of sampling methodologies. It was found that the effectiveness of each method largely depends and the soil type and that no one device is best under all experimental situations. However, the multi-segment wick lysimetry system better represented flow through the unsaturated zone when moisture and solutes were transported predominantly by matrix flow (e.g., in a loamy sand). On the other hand, in a clay soil, where the bulk of the transport was through macropores, both the gravity pan and wick pan lysimeters performed equally well (for further details see 'Field evaluation of wick and gravity pan samplers'). Porous cup samplers are not recommended for use in the unsaturated zone. These sampling devices spuriously indicated preferential solute movement in loamy soils and failed to register any solute flow in clay soils with macropores.

Using this apparatus, laboratory experiments were conducted on large undisturbed soil cores collected from basaltic agricultural soils at Grassmere, 40 km north of Warrnambool, in Victoria's Western District (Map 32 E 63 property 60 (CFA, 1989)). The purpose of this study was to illustrate the impact and magnitude of macropore flow on water and solute transport on farm soils typically used for agricultural purposes. The soil's characteristics down the profile were also determined. These characteristics include pH, moisture content, colour, organic carbon content, particle size and phosphorus retention indices. Two undisturbed soil cores (32 x 32 cm wide x 40 cm deep) were taken from a farm located at Grassmere. This property had not applied superphosphate for approximately 25 years and is used for cattle grazing. The area is situated on gently undulating basalt plains. The bedrock consists of olivine and iddingsite basalt, limburgite minor scoria and tuff. The soil above the bedrock consists of fine soil particles through to clay with buckshot inclusions and randomly distributed rocks varying in diameter from a few millimetres to approximately 20 cm. The soil cores were extracted early in May 1995, before the heavy winter rains commenced. The soil cores were carefully encased in wooden boxes. The interior of each box was lined with non-reactive polyurethane foam filler to prevent deterioration of the soil structure during transport and analysis. Irrigation was applied on the soil surface by using a purpose-built X-Y raster-scanning, drip-irrigation system which provided a constant and uniform application of water and solutes at the soil surface (see Figure 3c).

No significant trends in soil pH with depth were found. The pH ranged from 6.63 to 5.33. On the other hand, organic carbon content showed significant variation from point to point, indicating significant heterogeneity and a possible cause for variation in chemical adsorption. The organic carbon content ranged from 1.68 to 4.11%. There was no trend with depth, however. The depth to bedrock was approximately 50 to 55 cm. Particle size analysis indicated that the greatest proportion of fine particles (<300 mm) were found at depth approximately 50 cm, just above the bedrock. At the approximate depth of 10 cm, a large proportion (32%) of particles >1700 mm were found and associated with a lateritic buckshot layer. Regardless of the macropore structure of the soil, the particle distribution analysis indicated that preferential flow was likely. Phosphorus retention indices were averaged over four depth ranges, 0-5 cm, 5-15 cm, 30-40 cm, 45-55 cm and the respective Pmax (mg/kg) values corresponding to these depths were 18328, 6626, 18382, -25641. The mean Pmax was 19880. The Pmax values were calculated by fitting regressions to the Langmuir equation. The R2 values ranged from 99.8% to 100%. From the Freundlich curves, the values of k (an empirical constant measuring the number of P-sorption sites) were respectively 851 (for depth 0-5 cm), 933 (for 5-15 cm), 955 (for 30-40 cm), and 977 (for 45-55 cm); a slight increase with depth. Figures 1 illustrates the dynamic variation in soil-moisture patterns collected at the base of one of the cores over a two week period commencing late in February, 1996. The lighter shaded areas indicate high moisture content and the darker areas indicate lower moisture content. Fig. 1b (right) illustrates changes in the moisture distribution of the same core just 7 days after Fig. 1a (left). Clearly, the moisture distribution is not static nor uniform.

Figure 2 illustrates the average discharge collected for each wick-lysimeter over a 3 week period commencing late in February, 1996. The error bars are 95% confidence intervals and the discharge volumes were found to be statistically very highly variable (one-way ANOVA, p < 0.001). The figure demonstrates significant preferential percolation in the soil core.

Figure 3 illustrates the spatial variation in outflow patterns for the same soil core. Higher discharge patterns where found for the cells located in row S5 when compared to the rest of the core. In part, this may be attributed to "edge effects", i.e. water flowing preferentially faster down the side of the box at the boundary of the soil column and the foam-filler. At the start of the experiment, the soil core was level, however, by the completion of the experiment, the soil core was tilting slightly. Cells 1 to 5 located in row S1 were about 5 mm higher than their counterparts located in row S5. The tilting was caused by soil swelling during the experiment. The slight change in slope may have enhanced the edge effect at S5.

Figure 4 illustrates variation in peak and average nitrate concentration with time. Significant nitrate loss occurred just two days after the initial application (see Fig. 8). The correlation between daily nitrate mass and daily volume of water (totalled for all 25 wick-lysimeters) was surprisingly low (r2 = 0.1934). Also, for daily peak-concentration, no relationship with water flow was evident. The particular wick-lysimeter that contributed to the largest peak concentration for any particular day are indicated inside the bar chart on Figure 7. The peak concentration was not always found in the same lysimeter. Further, the daily peak concentration was sometimes found in low discharge cells (e.g., cell 1), and at other times in moderate discharge cells (e.g., cells 10, 11, and 18) and high flow cells (e.g., cells 20, 24, 25). Similar results were found for chloride loss.

Figure 5 illustrates the changing patterns in breakthrough concentrations for each lysimeter per day. It is interesting to observe differences in times and peak concentrations for each lysimeter. After 18 days, 100% of the applied nitrates and 70% of chloride was leached indicating little incorporation into the soil matrix. However, in the same time period less than 1% of the applied phosphates were leached from the column, indicating strong adsorption. Some wick-lysimeters showed early peak-concentrations of phosphates. However, the amounts of phosphates were only a small fraction of the total applied to the soil surface. In summary then, these results indicate considerable heterogeneity in water flow patterns and solute leaching. Preferential flow of nitrate and chloride is significant, phosphate transport by this mechanism was negligible. The processes controlling the transport and distribution of solutes in the soil may indeed be different to the processes controlling the rate and distribution of moisture. This must be taken into account in developing models of preferential solute flow.