Previous PageTable of ContentsNext Page

Hydraulics of rill initiation on a low-slope sandy soil

Tatard, L.1; O. Planchon1; G. Nord2; D. Favis-Mortlock3; J. Wainwright4;
N. Silvera5; O. Ribolzi5 and M. Esteves6

Keywords: rainfall simulation, water erosion, erosion models, rill, Senegal

Abstract

Most sandy soils in West Africa have low slopes, typically of less than one percent. They however are prone to water erosion because of their susceptibility to crusting and their weak cohesion. This study presents a rainfall-simulation experiment. It consisted of a 2 hrs rainfall event at a constant intensity of 75 mm h-1, on a smooth bare soil with a 1 percent slope with the objective of forming a rill. After the rill was formed, soil elevation was measured at a horizontal resolution of 2.5 cm. Flow velocities were measured at 62 locations on the plot with the Salt Velocity Gauge technology, an automated, miniaturized device based on salt tracing. Measured velocities ranged from 0.006 m s-1 to 0.26m s -1. Three hydrological models were tested using these experimental data and their ability to simulate the velocity fields was studied. The first model solved the Saint-Venant equations in 2D. The second model used a kinematic wave in 1D in the slope direction coupled with a 2D flow-routing algorithm. The third model involved an empirical runoff algorithm close to the diffusion wave equation in 2D. The Darcy-Weisbach friction factor was calibrated in all cases. The comparison of simulated to observed velocities indicated that the full Saint-Venant equation gave better results than either the kinematic or the diffusion-wave equation. This result is attributed to the low slope angle of the plot, which is in part attributed to the fact that at low slopes, the local variations of water depth are of the same order of magnitude to that of variations in soil elevation. All models underestimate the velocity in the rill and overestimate velocities in the interril area. These results demonstrate that the Darcy-Weisbach friction factor used in the models should vary with the Reynolds number while all models considered it constant.

Introduction

Rill erosion is a major contributor to sediment removal from agricultural fields. Other erosion processes such as interrill erosion, splash erosion or tillage erosion preferentially lead to translocation of soil within the field (cf. Parsons et al., 2004). Thus, rill erosion has not only a crucial importance for on-site effects of erosion, but also for off-site effects and environmental concerns. It is therefore critical to understand the occurrence of dynamics of flow in rills and their controls on the pattern and timing of erosion.

Rill density has been thoroughly measured and related to landscape properties (e.g. Desmet and Govers, 1997). Hydraulics in rills has also been studied (Gimenez and Govers, 2001; Govers, 1992; Nearing et al., 1997). These studies have highlighted a number of interactions between flow and bed roughness. A first kind of interaction involves grain roughness and Reynolds number Re (Re = 4.u.r/ν), where u is average velocity, r is hydraulic radius and ν is the fluid kinematic viscosity; This formula is classically used to estimate Re in shallow free surface flows; e.g. Savat, 1980, Gilley et al., 1990, Abrahams et al. 1995, Pilotti and Menduni, 1997). Grain roughness does not affect flow velocity the same way in laminar and turbulent flow. As a result, the friction factor ff in the Darcy-Weisbach equation decreases with increasing values of Re. This interaction was shown experimentally by Nearing et al. (1997) on bare sandy soils, where grain size is small with regard to flow depth. At greater grain size and/or when natural vegetation interacts with the flow, the relationship between ff and Re is even more complex, as shown by Abrahams et al. (1995).

A second type of interaction involves channel roughness and Froude number Fr ( Fr = u/ g.h ,where u is the average velocity, g is gravitational acceleration and h is flow depth). Grant (1997) provided the first assessment of such an interaction on high-gradient alluvial channels. He showed that Fr could not be higher than unity over long distances or long periods of time. The same interaction was demonstrated by Gimenez and Govers (2001) on eroding rills. Finally, Gimenez et al. (2004) hypothesized that critical flow was a necessary condition for rill initiation. Interaction with Fr lies on the development of small hydraulic jumps along the rill when the flow velocity is critical. At the jump location, localized erosion occurs due to turbulence in the jump, with the result of eroding the channel, enlarging it and finally lowering the average velocity above the critical speed.

This knowledge of rill initiation and develop­ment is however underexploited in most erosion models. Rill density is either predefined (one rill per metre for WEPP: Gilley et al. 1988) or as an input parameter (Siepel et al., 2002). A few models are aimed at dynamically developing a rill network (RillGrow model, Favis-Mortlock et al., 2000; PSEM_2D model, Nord and Esteves, in press) However, these models use uniform friction factors. Finally little attention had been paid to the reliability of Re and Fr values simulated by erosion models.

This study presents a rainfall-simulation experiment carried out at Thies, Senegal (14º45′43′′N, 16º53′16′′W), on a 40-m2 plot with sandy soil and low slope (1%). Flow velocity was measured at 58 individual points on the plot with a miniaturized version of the salt velocity gauge (SVG) technology (Planchon et al. 2005). SVG is an automated salt-tracing technique which provides reliable point velocity data over a wide range of flow speeds and with no lower limit on flow depth. Measured velocities have been compared with simulated data from three models and the consequences on the simulation of Re and Fr are assessed and discussed. The results allow us to draw some research perspectives for the modelling of rill initiation.

Material and methods

The new generation of SVG

The SVG technology has been presented in Planchon et al. (2005). It consists of injecting salty brine into the flow and recording the conductivity peak simultaneously at two locations downstream. A new generation of SVG has been developed for this experiment. Each conductivity sensor consisted of two aluminium pins spaced 1-cm apart, which allowed for measuring the velocity of a narrow flow path. The inter-probe distance was 10cm. The flow velocity was calculated by fitting a 1D convection-dispersion model for velocity and dispersion coefficients (Eq. 1). Hayami (1951), reported by Henderson (1966), gave Eq. 2 as the solution of Eq. 1 when C(0,t) is the Dirac function, i.e. injection is instantaneous. Eq. 3 describes the least-squares sum that is minimized in the model used by the SVG.

Management of Tropical
Sandy Soils for Sustainable
Agriculture

where C is salt concentration (g. l-1); C1 is measured at the upper probe; C2 is measured at the down probe; t is time (s); x is length (m); V is flow velocity (m.s-1); D is dispersion (m2.s-1); * is the convolution product; C is Hayami’s solution from Eq. 2 with x being the inter-probe distance, i.e. 0.1m; a and b are coefficients that account for salt losses between the two probes (due to infiltration or other cause); ssq is the quadratic sum that is minimized by fitting V, D, a and b for each pair of peaks.

The new generation of SVG requires two operators. The first is located on the plot to place the probes in the measuring locations and to do the injections manually. The brine was coloured with potassium permanganate to allow for visual control of the tracing process. Four probes were multiplexed to the datalogger, allowing four locations to be measured simultaneously. The second operator was at the computer. At a given signal, the data acquisition was triggered when the first operator injected the brine a few centimetres upstream of the probes. Conductivity was measured at 200 Hz during 2.5s and the model was then automatically fitted to the data. The measurement was replicated until clear peaks were seen on the software graphical interface and the model gave satisfactory results.

Rainfall-simulation experiment

The rainfall-simulation site was located at Thies, Senegal. The plot was 10m long by 4m wide, with a 1% slope, and sandy soil (1% clay, 7% silt, 43% fine sand, 49% coarse sand). The rainfall simulator was as described by Esteves et al. (2000a). It allowed for rainfall at constant intensity of 70mm .hr-1 in average. In order to limit wind effects, which may cause noticeable variations of rainfall intensity, simulations were carried out at a maximum wind speed of 1 m.s-1. Six tipping-bucket rain gauges with electronic recording were placed along the plot borders for monitoring the actual rainfall intensity. The flow discharge was collected in a trough and alternately directed, via a 4-inch flexible hose, into two 150-litre cylindrical buckets, one being filled while the other was drained. The volume in the filling bucket was monitored by recording the rise of a float. The resolution of this apparatus was 2.5litres. The typical flow discharge at steady state was 0.5l .s-1.

On day 1 of the experiment, a wetting rainfall of 20mm was applied and the plot was manually ploughed to a depth of 50cm. The surface was then raked in order to form a slight V shape, with 1% slope longitudinally and 1% slope towards the median axis of the plot. The purpose of the V shape was to prevent a rill from forming by the edge of the plot.

The experiment detailed in this article was held on day 7. The days before, a total of six hours of rainfall had already been applied on the plot for others experiments that are not reported in this article. The consequence of these successive experiments was an already ‘old’ surface with a well organized flow pattern. The longitudinal slope had evolved from straight to slightly concave (Figure1) with thick sand deposits in the concave downstream part.

Days 6 and 8 (i.e. the day before the experiment, and the day after) were used to carry out microrelief measurements. The relief-meter was the same as described by Planchon et al. (2001). It consists of a vertical rod with a sensor at the end that detects the soil surface. Stepper motors allow the apparatus to move in small increments in all directions. The horizontal resolution is 2.5cm transversally to the plot and 5cm longitudinally . The vertical precision is 0.5mm. With a maximum acquisition rate of 1.6 point. s-1, the 16,000 measured points of the entire plot required a full day.

The experiment on day 7 consisted of a 2h15′-long continuous rainfall at constant rainfall intensity (69 mm.h-1 on average). After the discharge had stabilized, flow velocity was measured at 72 locations with three to six replications, which led to a total of 348 individual velocity measurements. Among this set, 122 individual measurements, covering 68 locations, have been selected for further analysis. The other data were discarded for various reasons: in particular because of the poor quality of either one of the two conductivity peaks or poor quality of the modelled peaks.

At the end of the experiment, a series of digital pictures of the plot were taken from a height of 6 metres above the plot. The pictures have been mounted in a single file and geometrically corrected so that each pixel corresponds to one square millimetre in the field. The resulting image can be combined with a DEM to calculate virtual pictures. Figure1 shows one of these views with the relief magnified ten times and the colour contrast enhanced. The native soil appears in black (its natural colour is a yellowish light brown). White and reddish colours correspond to various types of sand deposits.

Management of Tropical
Sandy Soils for Sustainable
Agriculture

Figure. 1. Location of the velocity measurements showed on a virtual picture of the plot. Vertical axis has been magnified ten times. Colour contrast has been enhanced. The native soil appears in black. White and reddish colours correspond to various types of sand deposits

The models

PSEM 2D (Plot Soil-Erosion Model 2D; Nord and Esteves, in press; Esteves et al., 2000b) is a soil erosion model dedicated to small experimental plots, typically of less than 100m 2.

Overland flow is described by the depth-averaged two-dimensional unsteady flow equations commonly referred to as the Saint-Venant equations (Zhang and Cundy, 1989). The friction slopes are approximated using the Darcy-Weisbach equation (Eq. 4.) derived for uniform steady flow. The second-order explicit scheme of MacCormack (1969) is used for solving the overland flow equations. Infiltration is computed at each node using a Green-Ampt model (Green and Ampt, 1911).

Management of Tropical
Sandy Soils for Sustainable
Agriculture

where ff is the calibrated Darcy-Weisbach friction factor. A constant value is assuming during the simulation.

NCF (New Conceptual Framework; Parsons et al., 1997, Wainwright et al., 1999) is a flexible model that can be used for experimental plots as well as small watersheds. The hydraulics consists of solving the kinematic wave equation in 1D along the flow direction derived from a DEM which depressions have been previously filled (using the algorithm from Planchon and Darboux, 2001). The kinematic wave simplification uses the continuity equation from Eq. 5, together with the Darcy-Weisbach equation in one dimension (Eq. 6). The numerical scheme used with this model is the Euler simple backward difference (Scoging, 1992).

Management of Tropical
Sandy Soils for Sustainable
Agriculture

where g is gravitational acceleration (m.s-2), d is flow depth (m), q is unit discharge (m2s -1) and s is the slope (m.m-1).

The flow is routed from each cell to one of its four adjacent cells in a finite difference grid using a topographically based algorithm based on the greatest difference in altitude of the cells. Overland flow is generated as Hortonian (infiltration excess) runoff by determining the difference between the rainfall and infiltration rate. The latter is predicted using the Smith-Parlange model with modifications to allow runon infiltration and temporally variable rainfall.

RillGrow2 (Favis-Mortlock, 1998; Favis-Mortlock and Boardman, 2000) is a model dedicated to the numerical simulation of emerging rill patterns. Space is discreticized at very small scale so that any cell is supposed to be entirely inside, or entirely outside a rill. Each cell is eroding independently to each other. Cells lower while eroding. Eroding cells thus attract more water flow, subsequently increasing the erosive power of the rill. Because of its high computational needs, applications of RillGrow2 are limited to experimental plots of a few tens of square metres.

RillGrow2 hydraulics consists of calculating a ‘potential flow velocity’ with a Manning-type equation, based on the water depth: u = w.d.Rn, where u is ‘potential flow velocity’, w is an empirical roughness coefficient, d is water depth, R is the hydraulic gradient and n = 0.5. The RillGrow2 numerical scheme is unique in soil-erosion modelling: at each time step, the model checks a single cell, chosen at random, and processes it. The check consists of calculating u and determining if outflow is possible from this cell. If the answer is yes, an outlet cell is chosen among eight neighbours according to the steepest descent of the free surface. The required amount of water is then passed from the source cell to the destination cell in order to level the free surface between the two cells. This procedure is then repeated until all cells have been chosen at the particular time step.

Results

Surface-feature patterns

Figure2 shows the left bank of the rill viewed from downstream. The coloured arrows represent the flow velocities as computed by the best model result (to be detailed below). At this point of the result presentation, the model results are used as a convenient illustration of the various flow conditions on the plot and their relation to surface features.

Table 1 summarizes the qualitative information detailed in this section.

Location A represents a high point with a convex soil surface. No visible flow could be seen and the model actually predicts a flow velocity lower than 0.02 m.s-1. Because of their higher position, these locations were sediment sources with regards to splash erosion: sediments occasionally splashed onto these areas were sooner or later splashed back to the lower areas. The soil surface at A had there fore the colour of the native soil, which is a yellowish light brown.

Management of Tropical
Sandy Soils for Sustainable
Agriculture

Figure 2. Detail of the left bank of the rill viewed from downstream. Light blue crosses show velocity measurement locations. Coloured arrows are modelled flow velocity. Capital letters show the four surface features that develop on the plot (comments in text). Axis labels are in mm

Table 1. Qualitative information on flow condition deduced from field observation during rainfall and from surface feature description after the experiment

Location

Surface 
feature

Velocity
(m.s-1)

Turbu-
lence

Flow
 regime

A Native soil,
light brown
<0.02

Laminar

 
B Discontinuous
sand deposit
0.02 <0.05    
C Continuous
reddish sand
deposit
01 <0.2   Sub- critical
D Continuous
white sand
deposit with
crossed wavy features
>0.2

Turbulent 

Super­critical

Location B represents the first visible flow. It is characterized by small undulating furrows, ~10 mm wide and 2mm deep. Flow velocity was still slow (~0.05 m.s-1 according to the model). The transport capacity was subsequently negligible. However, uneven sand grains could be observed in these tiny channels, slowly creeping downstream until a raindrop hit them and splashed them away. The only turbulence in the flow was caused by raindrop impacts.

At location C, a well established stream was flowing at 0.1 to 0.15 m.s-1. The soil surface was covered by a continuous layer of reddish sand that was slowly creeping downstream. Flow was turbulent, but still subcritical. Turbulent flow could easily be determined from the observation of the fate of the coloured brine. In laminar flow, the brine left the injection point very gradually, thus forming a long colourer tail. This fate indicates a vanishing flow velocity at the bottom of the flow, which is typical to laminar flows. In turbulent flows, the tracer left the injection point in a fraction of seconds, indicating a very sharp vertical velocity profile that did not allow the tracer to ‘stick’ to the soil surface, as it did in laminar conditions.

Location D is characterized by white sand deposits with crossed wavy features typical to supercritical flow. The white colour of the sand indicates that the sand grains were washed up by turbulence until all clay and organic particles had detached. These field observations indicate that the flow was certainly turbulent and supercritical. Modelled as well as measured velocities were all above 0.2m .s-1.

Qualitative results

RillGrow2 used 5-cm cells in order to follow its requirement that a given cell should be entirely inside or entirely outside a flow path. NCF used 50-cm cells for the opposite reason: because the 1-D hydraulics does not allow for lateral flow movement, NCF requires that the same flow path will not be divided into multiple cells, otherwise the modelled free surface may be unrealistic, which leads in practice to numerical instabilities. PSEM_2D used 10-cm cells, which was the smallest cell size the model could simulate without numerical oscillations. Only RillGrow2 was able to run the raw DEM. Both PSEM_2D and NCF needed a smoothed and depression-free DEM.

Each model was calibrated from the hydrograph. The infiltration parameters were calibrated from the total runoff and the steady infiltration rate. The friction factor was calibrated from the hydrograph rise (Figure3).

Management of Tropical
Sandy Soils for Sustainable
Agriculture

Figure 3. Picture of the plot (with contrast magnified) compared to the velocity, Re and Fr maps predicted by the models. Calibrations were done from the hydrograph

The velocity field from PSEM2D is very similar to what can be estimated from the picture at the left of the figure. One can notice for example the location of the predicted maximum velocity. It corresponds to the white sands at the centre of the plot, which we interpreted as a mark of supercritical flow. The pattern from NCF is similar to PSEM, with a noticeable loss of precision due to the coarser grid. RillGrow2 predicts a wide area of high velocity in the bottom part of the plot which corresponds fairly well to the concave area of reddish sands deposits that can be seen either on Figures 3 and 1.

The Re predictions follow approximately the same pattern as the flow velocity. However, according to the threshold of 2000 commonly used for the transition between laminar and turbulent flows, the spatial extension of turbulent flow is underestimated with regard to observations made by eye during the velocity measurement (as explained in the previous section). Fr is even more problematic since no pattern at all is predicted by PSEM_2D or NCF while the pattern predicted by RillGrow2 does not fit to the field observations reported in Table 1.

Models results: Comparison with measured velocity

Figure 4 shows the modelled velocity vs the observed ones. All models have a better fit at low velocities than at higher ones. PSEM_2D and NCF slightly overestimate the low velocity and strongly underestimate the high ones. RillGrow2 simulates very well the slowest flows (i.e. v<0 .05m .s-1) and underestimates the other cases. Results from NCF are not exactly comparable to measured data because while measured velocities are point data, the model results represent a 0.25 m2 cell. Localized maxima or minima cannot be expected to figure in NCF results.

Modelling the interaction between friction factor and flow conditions

The velocity modelled by PSEM_2D (Figure 4) fits Eq. 8, which can be used to estimate ff1, the true value of ff at each cell. This is done by solving, at each cell, the set of equations 8 to 11 for ff1, which Eq. 12 gives the solution. Eq. 9 states the unit discharge at the cell location will not change after ff is corrected from ff0 to ff1. Eq.10 and 1 1 are the Darcy-Weisbach equation before and after correction, respectively.

Management of Tropical
Sandy Soils for Sustainable
Agriculture

where a = 0.5; b = 0.28; ff0 = 0.26; (h0, v0) are flow depth and flow velocity read at a given cell in the PSEM_2D results shown in Figures3 and 4; v1 is the observed velocity; h1 is the corresponding flow depth according to the modelled unit discharge.

ff1 was calculated from Eq. 12 at each cell. The resulting map was then smoothed to prevent the model from producing numerical instabilities. A threshold of ff<2 was finally applied to account for inconsistent velocities predicted at very small water depths, whereas the value of this threshold proved to have little influence on the final result. PSEM_2D is the only model which allows for spatially non uniform values of ff. It was therefore used for validation. Figure5 shows the resulting maps for V, Re and Fr. The Fr >1 limit is in fair agreement with the white sands that has been interpreted as the mark of supercritical flow. The Re >2,000 limit is wider (albeit still limited to the central channel). Figure 6 shows the graph of modelled vs observed velocity. Results are scattered around the one-one line, which was the expected result of the use of Eq. 12. Figures 5 and 6 show that the results with varying ff are far more realistic, and closer to the field observations, than those obtained from homogeneous ff.

Management of Tropical
Sandy Soils for Sustainable
Agriculture

Figure 4. Modelled velocity against measured values

Figure7 relates ff with Re in logarithmic co-ordinates. It shows that ff is high at low Reynolds numbers and decreases with increasing Re. This is the same kind of relationship as previously obtained by Nearing et al. (1997). Figure8 compares the two experiments. Each line was drawn inside the data limits of the corresponding experiment. Results from Nearing et al. (1997) are therefore extended from rill flow and high Re values to interrill flow at the transition between laminar and turbulent conditions.

Management of Tropical
Sandy Soils for Sustainable
Agriculture

Figure 5. PSEM_2D results with ff calculated from Eq. 12: output maps compared with the picture of the plot

Management of Tropical
Sandy Soils for Sustainable
Agriculture

Figure 6. PSEM_2D modelled velocity with ff calculated from Eq. 12

Management of Tropical
Sandy Soils for Sustainable
Agriculture

Figure 7. Relationship between Friction factor ff and Reynolds number Re simulated by PSEM_2D with friction factor calculated from Eq. 12

Management of Tropical
Sandy Soils for Sustainable
Agriculture

Figure 8. Relationship between Friction factor ff and Reynolds number Re: comparison with results from Nearing et al. (1997)

Discussion and Conclusion

The SVG technology has allowed flow-velocity measurement in a wide range of flow speeds (from .006m .s-1 to 0.27.s-1 in this experiment). The use of salty brine as a tracer makes SVG suitable for measuring very shallow flows. The only limitation was the probe size, which was 1-cm wide and 10-cm long. Thanks to this technology, we were able to measure velocity in a wide variety of flow conditions, from unconcentrated to concentrated in a small rill, from laminar to turbulent, and from subcritical to supercritical. The data obtained have been used to test three hydrological models (PSEM_2D, NCF, and RillGrow2) which were very different to each other, having only in common the use of a Manning/Darcy-Weisbach-type hydraulic equation with a constant, homogeneous friction factor. The main results were the followings:

These results lead to the following conclusions:

Acknowledgements

This work was granted by the RIDES project, an ECCO research program. The new miniaturized version of SVG has been developed in a collaborative project between The Institut de Recherche pour le Développement (IRD) and the USDA-ARS National Soil Erosion Laboratory (NSERL). The authors want to thank Dr. Chi Hua Huang, from NSERL, for his support and helpful advices in the development of SVG. The rainfall simulations have been conducted by Kokou Abotsi, from IRD Dakar, Senegal.

References

Abrahams, A.D., Anthony, J.P. and Wainwright, J., 1995. Effects of vegetation change on interrill runoff and erosion, Walnut-Gulch, Southern Arizona. Geomorphology, 13, 37-48.

Desmet, P.J.J. and Govers, G., 1997. Two-dimensional modelling of the within-field variation in rill and gully geometry and location related to topography. Catena, 29, 283-306.

Esteves, M., Planchon, O., Lapetite, J.M., Silvera, N. and Cadet, P., 2000a. The ‘EMIRE’ large rainfall simulator: Design and field testing. Earth Surface Processes and Landforms, 25, 681-690.

Esteves, M., Faucher, X., Galle, S., Vauclin, M., 2000b. Overland flow and infiltration modelling for small plots during unsteady rain: numerical results versus observed values. Journal of Hydrology 228, 265-282.

Favis-Mortlock, D., 1998. A self-organizing dynamic systems approach to the simulation of rill initiation and development on hillslopes. Computers & Geosciences, 24(4), 353-372.

Favis-Mortlock, D.T., Boardman, J., Parsons, A.J. and Lascelles, B., 2000. Emergence and erosion: a model for rill initiation and development. Hydrological Processes, 14, 2173-2205.

Gilley, J.E., L.J. Lane     , J.M. Laflen, H.D. Nicks and W.J. Rawls. 1988. USDA-water erosion prediction project: New generation erosion prediction technology. In: Modeling, Agricultural, Forest, and Rangeland Hydrology. Symposium Proceedings. Pub. 07-88:260-263. Available from Am. Soc. Agric. Eng., St. Joseph, MI.

Gilley, J.E., Kottwitz, E.R., Simanton, J.R., 1990. Hydraulic characteristics of rills. Trans. Am. Soc. Agric. Eng. 33, 1900-1906.

Gimenez, R. and Govers, G., 2001. Interaction between bed roughness and flow hydraulics in eroding rills. Water Resources Research, 37, 791-799.

Gimenez R., Planchon O., Silvera N., Govers G. 2004. Longitudinal velocity patterns and bed morphology interaction in a rill. Earth Surface Proccesses and Landforms 29, 105-114.

Govers, G., 1992. Relationship between discharge, velocity and flow area for rills eroding loose, nonlayered materials. Earth Surface Processes and Landforms, 17, 515-528.

Grant, G.E., 1997. Critical flow constrains flow hydraulics in mobile-bed streams: A new hypothesis. Water Resources Research, 33, 349-358.

Green, W.H. and Ampt, G.A. 1911. Studies on soil physics: 1, flow of air and water through soils. Journal Agric. Sciences 4, 1-24.

Hayami S. 1951. On the Propagation of Flood Waves. Disaster Prevention Research Institute, Kyoto University, Japan, Bulletin no. 1.

Henderson, F.M. 1966. Open Channel Flow. New York: Macmillan Company.

MacCormack, R.W. 1969. The effect of viscosity in hypervelocity impact cratering. New-York, American Institute of Aeronautics and Astronautics.

Nearing, M.A., Norton, L.D., Bulgakov, D.A., Larionov, G.A., West, L.T. and Dontsova, K.M., 1997. Hydraulics and erosion in eroding rills. Water Resources Research, 33, 865-876.

Nord, G. and M. Esteves. in press. PSEM_2D: a physically-based model of erosion processes at the plot scale. Water Resources Research.

Parsons, A.J., Wainwright, J., Abrahams, A.D. and Simanton, J.R., 1997. Distributed dynamic modelling of interrill overland flow. Hydrological Processes, 11(14), 1833-1859.

Parsons, A.J., Wainwright, J., Powell, D.M., Kaduk, J. and Brazier, R.E. 2004. A conceptual model for under­standing and predicting erosion by water, Earth Surface Processes and Landforms 29, 1293-1302.

Pilotti, M. and Menduni, G. 1997. Application of lattice gas techniques to the study of sediment erosion and transport caused by laminar sheetflow. Earth Surface Processes and Landforms, 22, 885-893.

Planchon O., Silvera N., Gimenez R., Favis-Mortlock D., Wainwright J., Lebissonnais Y., Govers G. 2005. Estimation of flow velocity in a rill using an automated salt-tracing gauge. Earth Surface Proccesses and Landforms. 30, 833-844.

Planchon, O. and Darboux, F., 2001. A fast, simple and versatile algorithm to fill the depressions of digital elevation models. Catena, 46(2-3), 159-176.

Planchon, O., Esteves, M., Silvera, N. and Lapetite, J.M., 2001. Microrelief induced by tillage: measurement and modelling of Surface Storage Capacity. Catena, 46(2-3), 141-157.

Savat, J., 1980. Resistance to flow in rough supercritical sheet flow. Earth Surface Processes, 103-122.

Scoging, H. 1992. Modelling overland-flow hydrology for dynamic hydraulics. Overland Flow Hydraulics and Erosion Mechanics. A.J. Parsons and A.D. Abrahams. London, UCL Press, 89-103.

Siepel, A.C., Steenhuis, T.S., Rose, C.W., Parlange, J.Y. and McIsaac, G.F., 2002. A simplified hillslope erosion model with vegetation elements for practical applications. Journal of Hydrology, 258, 111-121.

Wainwright, J., Parsons, A.J. and Abrahams, A.D., 1999. Field and computer simulation experiments on the formation of desert pavement. Earth Surface Proccesses and Landforms. 24: 1025-1037.

Zhang, W. and T.W. Cundy. 1989. Modelling of two dimensional overland flow. Water Resources Research 25, 2019-2035.


1 IRD, BioEMCo, Bat EGER Aile C, 78850 Thiverval- Grignon, France. Olivier.Planchon@Grignon.Inra.fr
2 LTHE, Grenoble, France.
3 School of Geography, Queen’s University Belfast, Belfast, UK.
4 Sheffield Centre for International Drylands Research, Department of Geography, University of Sheffield, Sheffield, S10 2TN, UK.
5 IRD, Vientiane, Lao PDR.

The co-composting of waste bentonites from the processing of vegetable oil and
its affect on selected soil properties of a light textured sand

Soda, W.1; A.D. Noble1; S. Suzuki2; R. Simmons1; L. Sindhusen3 and S. Bhuthorndharaj3

Keywords: Co-composted waste acid bentonite, waste bentonite, soil amelioration, degraded soils

Abstract

Waste acid bentonite is a byproduct from vegetable oil bleaching that is both acidic (pH <3.0) and water repellent (hydrophobic). These materials are currently disposed in landfills and are an environmental hazard due to the aforementioned properties. A study was undertaken using three different sources of waste oil bentonites collected from processing plants within the Bangkok metropolitan area. These wastes included soybean oil bentonite (SB), palm oil bentonite (PB) and rice bran oil bentonite (RB), each of which was co-composted with rice husk, rice husk ash, and chicken litter in order to eliminate their acid reactivity and hydrophobic nature. The chemical and physical characteristics of acid activated bentonites before and after bleaching and the co-composted materials after addition to a degraded light textured soil were assessed and are reported herein. The organic carbon (OC) content, pH, exchangeable cations and cation exchange capacity (CEC) of the waste oil bentonites increased significantly after the co-composting phase. In addition, the hydrophobic nature of these materials as measured using the Water Drop Penetration Test (WDPT) decreased from 10,800 seconds to 16-80 seconds after composting. Furthermore, when these composted materials were incorporated into a degraded light textured sandy soil positive impacts to soil physical attributes in terms of specific surface area, total porosity and available water content for crop growth were observed. The results from this study demonstrate the positive impact of the waste products when modified through composting on the physical and chemical properties of a light textured sandy soil.

Introduction

Sandy light textured soils are common throughout the globe and it is estimated that they cover approximately 900 million hectares (Driessen et al., 2001). Whilst these soil types (Arenosols) predominate the arid and desert regions of the world, these soils occur extensively within the semi- and humid-tropics and support the livelihoods of large populations. Within Thailand, the upland regions of the Northeast (Isaan) occupy approximately one third of the entire land area of Thailand. The soils in this region are dominated by light textured soils with clay contents of <15%. These soils are aeolian in origin and have formed under a rainfall regime that exceeds 1,200 mm per annum (Lesturgez, 2005). In addition, these soils are predominantly acid in reaction with low soil organic matter (Kheoruenromne and Suddhiprakarn, 1998). Under their native Dipterocarp forests, these soils are highly productive and support large amounts of biomass in association with closed and highly efficient nutrient cycling. However, when these soils are cleared for agricultural production a rapid decline in fertility occurs in association with a loss in soil organic matter, which is reflected in a concomitant decrease in cation exchange capacity (CEC) (Noble et al., 2000). Further, the combination of inappropriate land management and over-exploitation of these limited natural resources has resulted in significant soil degradation. Hence these degraded soils are often regarded as marginal for crop production.

An important step in addressing the low productivity of these degraded soils is to address the fundamental problem of diminished nutrient holding capacity (as indicated by CEC) associated with a decline in soil organic matter. A possible approach to addressing the aforementioned issue is through the application of natural materials (Noble et al., 2004; Noble et al., 2001) or industrial waste products that are easily available to resource poor farmers.

Bentonite is a 2:1 layer silicate containing smectite minerals, usually montmorillonite. It is used in a large number of applications ranging from foundry moulds, drilling muds, pet litter adsorbent to stock feed supplementation. The annual production is estimated to be 8 million tonnes (Mt) (Virta, 2001). Bentonite is also used to remove a variety of impurities in the vegetable oil industry including phosphotides, fatty acids, gums, and trace metals. This results in the production of light-colored and stable oils acceptable to consumers (Foletto, et al., 2002). Acid activated calcium-bentonites are the preferred form of clay used as an absorbing agent during the decolorizing, clarification and refining process in the manufacture of vegetable oil. Acid activated clays are produced by treating calcium bentonites with mineral acids (i.e. hydrochloric or sulphuric acids). In their natural non-activated state bentonites have a limited absorbing capacity. However, when bentonites are acid activated as a result of treatment with hot mineral acid solutions, hydrogen ions attack the aluminosilicate layers via the inter layer spacing (Taylor and Jenkins, 1987 in Foletto et al., 2002; Falaras, et al., 1999). This alters the structure, chemical composition and physical properties of the clay and significantly increases the adsorption capacity (Mokaya et al., 1993 in Foletto et al., 2002).

The reuse of waste oil bentonites is an area of potential opportunity for cost savings in the oil processing industry that would have the added benefit of protecting the environment. The application of either modified or naturally occurring bentonites have been shown to increase the soils cation exchange properties and increase crop productivity (Noble, et al., 2001; 2004). However, a significant problem in the re-use of waste acid activated bentonites from vegetable oil processing is their hydrophobic nature due to the presence of fats and oils on the surfaces of the spent clay as well as their acidic nature. This reduces the ability of these materials to adsorb nutrients and water. Croker et al. (2004) applied waste oil bentonite to a degraded light textured soil at rates up to 40 t ha-1 and observed increases in CEC from 0.6 to 1.9 cmolc kg-1. However, due to nutrient insufficiencies (nitrogen and potassium) and the acidic nature of these materials, overall productivity of a maize test crop was limited.

The objectives of the study were (i) to determine whether composting of waste oil bentonite with rice husk, rice husk ashes and chicken litter, could remediate the hydrophobic and acidic nature of the waste bentonite, and (ii) to assess the potential of these composted materials on soil properties.

Materials and Methods

Chemical characterization of acid activated bentonites before and after bleaching:

Samples of the raw acid activated bentonite prior to and post bleaching were obtained from three oil processing companies (Morakot, King and Tip) in the Bangkok metropolitan region, Thailand. Samples were air dried and passed through a 2-mm mesh sieve. Soil electrical conductivity (EC) and pH were measured in water and 0.01 M CaCl2 at a clay:solution ratio of 1:5. Exchangeable cations (calcium (Ca2+), magnesium (Mg2+), potassium (K+) and sodium (Na+)) and cation exchange capacity (CEC) were determined using 1 M NH4-acetate buffered at pH 7.0 (Rayment and Higginson, 1992). Organic carbon was determined by wet oxidation using the Walkley and Black method as modified by Rayment and Higginson (1992) and total N by Kjeldahl steam distillation. In addition, the co-compositing materials, namely chicken litter, rice husk and rice husk ash were similarly assessed.

Co-composting of waste acid bentonite with chicken litter, rice husk and rice husk ash:

A total of 9 composting treatments were imposed (Table 1). Ratios of the different components making up the composting material were mixed air dry. In addition, 4, 5, or 6 kg of dolomitic lime was applied to each treatment (1-9) in order to neutralize the residual acidity associated with the acid activated bentonite as based on the amount of chicken litter added. The mixed treatments were then placed in concrete composting bins with drainage holes placed at the base of bins to avoid excess water at the bottom of the bins. The amount of water added to each bin varied depending on the treatment. However, all bins that received water had the same consistency. A further set of control treatments (10-12) were assessed where the waste acid bentonites were placed in composting bin with no additions. These treatments were subject to the same mixing routines as previously described. Temperature within the composting bins was monitored on a daily basis at 3 points within the pile. The compost materials were turned every 2 weeks and water re-applied to maintain the desired moisture. Samples from each of the bins were collected on three occasions during the composting period, air dried, subjected to chemical analysis and water repellency assessment. Water Drop Penetration Time (WDPT) was assessed on all samples at each sampling date using the methodology of Bisdom et al. (1993).

Table 1. Composted treatment combinations imposed on three source of acid waste bentonite with rice husk, rice husk ash and chicken litter

Treatment Code

Treatment description

Ratio*

1. SB01

Waste soybean oil bentonite + Rice husk + Rice husk ash + Chicken litter + water.

1:1:1:1
2. SB02

Waste soybean oil bentonite + Rice husk + Rice husk ash + Chicken litter + water

1:1:1:2
3. SB03

Waste soybean oil bentonite + Rice husk + Rice husk ash + Chicken litter + water

1:1:1:3
4. PB01

Waste palm oil bentonite + Rice husk + Rice husk ash + Chicken litter + water.

1:1:1:1
5. PB02

Waste palm oil bentonite + Rice husk + Rice husk ash + Chicken litter + water.

1:1:1:2
6. PB03

Waste palm oil bentonite + Rice husk + Rice husk ash + Chicken litter + water.

1:1:1:3
7. RB01

Waste rice bran oil bentonites + Rice husk + Rice husk ash + Chicken litter + water.

1:1:1:1
8. RB02

Waste rice bran oil bentonites + Rice husk + Rice husk ash + Chicken litter + water.

1:1:1:2
9. RB03

Waste rice bran oil bentonites + Rice husk + Rice husk ash + Chicken litter + water.

1:1:1:3
10. P-SB

Rare waste soybean oil bentonite

 
11. P-PB

Rare waste palm oil bentonite

 
12. P-RB

Rare waste rice bran oil bentonite

 

* Ratio of waste oil bentonite: rice husk: rice husk ash: chicken litter mixed on a volumetric basis.

Pot experiment

A pot experiment was established in the Kasetsart University Central Laboratory greenhouse, Bangkok, Thailand. The twelve composted materials generated from the first phase of the study were evaluated (Table 1). To undertake the experiment, over 400 kg of degraded light textured sandy soil (0-15 cm) was collected from an experimental site established at the Animal Nutrition Development Station, Department of Livestock, Chiang Yuen, Mahasarakham Province, Northeast Thailand. The soil was classified as a Satuk series or isohyperthermic Oxic Paleustult (LDD, 1993; Soil Survey Staff, 1990). The soil was air dried, sieved to pass a 2-mm sieve and thoroughly mixed to ensure homogeneity. All laboratory analysis was undertaken at the Land Development Department, Soil Chemistry Laboratory. The soil used in this study was acid in reaction (pHCa 4.04) with a low CEC (1.76 cmolc kg-1) and organic carbon content (0.5%). The values indicate the low nutrient supplying capacity of the test soil as evidenced through comparison with the composted materials. In addition, the gravimetric soil moisture content of the air dried soil used in the current study was 0.01 kg kg-1 (water/oven dry soil). With respect to the texture of the soil, sand, silt, and clay content were 834, 55, 111 g kg-1, respectively.

The design of the experiment was a 4 x 12 + 1 incomplete factorial design consisting of 4 application rates of 0, 15, 30, 60, and 120 t ha-1, 12 sources of materials and a single control replicated 3 times. The resultant 147 pots were placed in a complete randomized block design on six benches within an evaporatively cooled greenhouse. Soil chemical properties were determined on all samples using the methodology previously described.

Water retention and porosity

In order to determine changes in the water retention and porosity of soils associated with the application of co-composted (PB, SB, and RB) and non-composted (P-PB, P-SB, and P-RB) waste oil bentonites, rates equivalent to 0, 15, 30, 60, and 120 t ha-1 were applied to 3 kg of the air dried soil, thoroughly mixed and placed in a free draining PVC pot. Annually, the Northeast of Thailand experiences several rainfall events >50 mm (Noble et al., 2004). Therefore, to simulate such an event, deionized water was applied to the soil surface and allowed to drain freely through the pot at room temperature for 3 days. At the conclusion of this equilibration period, three soil samples were collected from each pot using a stainless steel cylindrical core (47 mm in diameter and 30 mm in height). The soil cores were capillary-saturated using deionized water and the matric potential of samples was adjusted to -10 kPa (approximately equivalent to the field capacity θfc) by the suction table method. Thereafter the volumetric moisture content and bulk density of the samples were determined by oven drying at 110ºC for 24 hrs. The total porosity of the samples was derived from the bulk density and particle density of the soil using the following relationship;

Management of Tropical
Sandy Soils for Sustainable
Agriculture

where, θp is total porosity (m3 m-3), ρd is bulk density (Mg m-3) and Gs is particle density (assumed to be 2.65 Mg m-3).

Using soil samples collected from the pots (following the removal of the soil cores), the matric potential of the soil was measured over the range of -6.1 MPa to -0.2 MPa during the desorption process from saturation using the freezing point depression method (Suzuki, 2004). This facilitated the deter­mination of the volumetric moisture content at the permanent wilting point (i.e. -1.5 MPa in matric potential, θpwp). The available water content for crop growth (θawc) was derived from the difference between θfc and θpwp.

Statistical analysis

The analysis of variance (ANOVA) was used to estimate the significance of treatment effects using GENSTAT 5 release 3.22 statistic program.

Results and Discussion

Chemical characterization of acid activated bentonites before and after bleaching

For brevity, the results and discussion are confined to soil chemical and physical changes. Although detailed assessment of the composting process and productivity of two consecutive maize crops on the amended soils was conducted, the reader is referred to Soda et al. (2005) for further details.

Selected chemical attributes of the acid activated bentonites before and after the bleaching process of the three contrasting vegetable oil bentonites along with materials used in the composting process are presented in Table 2. As expected, the initial pHw of the acid activated bentonites was low (range: 3.10-4.13) thereby enabling these materials to absorb fats and oils during the bleaching process. After undergoing the oil bleaching processing, the pHw of the waste soybean oil (SB) decreased slightly whilst the palm (PB) and rice bran (RB) waste oil bentonites increased slightly (Table 2). The largest changes in chemical characteristics of the acid activated bentonite associated with bleaching was a dramatic decline in exchangeable Ca2+ in all materials and in exchangeable Mg 2+ in the soybean waste oil bentonite (Table 2). Contrasting this, there was an increase in exchangeable Mg2+ in the waste rice bran oil bentonite after the bleaching process whilst the waste palm oil bentonite remained unchanged. It is of note that the EC in the bentonites before and after processing was relatively high this being due to a high level of dissolved salts that are not associated with the exchange complex. Confirmation of the high level of soluble salts can be found if one compares the CEC as measured in buffered ammonium acetate with the sum of exchangeable bases, the latter being considerably larger than the former (Table 2). Both exchangeable Na+ and K+ were low although for all samples exchangeable Na+ increased with bleaching and exchangeable K+ increased in the case of the rice bran oil waste bentonite (Table 2).

The absorption of fats and oils associated with the bleaching of vegetable oils resulted in an increase in the organic carbon content of the waste oil bentonites (Table 2). In general the organic carbon content increased by the same amount regardless of the oil source. Total N content on the both the acid activated and bleached spent bentonite remained unchanged after processing except for the palm oil bentonite where the N content increased from 0.06% to 0.13% with bleaching (Table 2).

Table 2. Selective chemical properties of acid activated bentonite before and after bleaching of selected vegetable oils, chicken litter, rice husk and rice husk ash

Source

pHw

pHCa

EC

CEC

Exch.
Ca2+

Exch.
Mg2+

Exch.
Na+

Exch.
K+

Sum
bases

Total N

OC

WDPT

(dS m-1)

   

cmolc kg-1

   

(%)

(%)

(sec)

Acid activated soybean bentonite

3.26 3.24 5.35 51.04 120.70 22.39 0.78 0.69 144.56 0.05 0.25

nd

Waste soybean oil bentonite

2.98 2.97 5.06 47.46 28.08 4.88 2.73 0.44 36.13 0.05 36.13

25

Acid activated palm bentonite

4.13 4.26 3.49 52.13 97.97 18.26 0.49 0.65 117.37 0.06 0.21

nd

Waste palm oil bentonite

4.46 4.48 2.38 42.58 69.24 17.51 1.72 0.60 89.07 0.13 23.88

>10,800

Acid activated rice bentonites

3.10 3.06 3.26 35.60 88.84 3.54 0.62 0.77 93.77 0.06 0.15

nd

Waste rice oil bentonite

3.60 3.54 1.40 19.18 9.90 21.21 2.82 2.76 36.69 0.05 21.16

9,000

Chicken litter

6.84 6.85 23.80 50.58 69.14 36.02 25.47 77.31 207.94 2.55 18.44

nd

Rice husk 6.03 5.54 1.58 1.25 12.50 18.03 1.33 1.86 33.72 0.29 43.78

nd

Rice husk ash

8.02 7.37 3.12 6.72 4.09 5.22 12.42 5.36 27.09 0.07 5.09

nd

The other components used in the composting process, namely chicken litter, rice husk and rice husk ash had variable nutrient compositions (Table 2). As expected, rice husk pH increased with ashing, this being associated with the formation of oxides, and the chicken litter had a neutral to slightly acid reactivity (Table 2). Exchangeable cations in the rice based materials were low when compared to the chicken litter. Of importance with respect to the chicken litter is the elevated K+ level (77 cmolc kg-1). It is of note that most of the cations extracted from the chicken litter, rice husk and rice husk ash were in a soluble form as evidenced by the elevated EC and difference between CEC and sum of bases (Table 2). This implies that these nutrients are readily available for plant uptake as well as being potentially subject to leaching (Table 2). Chicken litter had the highest total N content (2.55%) compared to all materials and hence would be an effective source of N in the composting process.

Changes in chemical attributes with composting

Changes in chemical characteristics after the 84 days of the composting phase are presented in Table 3. The pHw of the co-composted spent bentonites increased with composting when compared to the original waste bentonites (Tables 2 and 3), these increases being associated with the addition of the rice husk ash, chicken litter and dolomitic lime. The mean pH of the co-composted spent bentonite increased to 7.18 compared to an initial mean pH of 3.68. Significant increases in pH within an individual spent bentonite treatment were observed with increasing additions of chicken litter (Table 3). For example, with the ratio of constituents increasing from 1:1:1:1 to 1:1:1:3 in the case of spent palm oil bentonite, pHw increased from 6.95 to 7.72 (Table 3). Associated with the increase in pH of the co-composted materials there was an increase in EC of the materials due to the mineralization of organic constituents with the concomitant release of cations (Table 3). K+ concentrations increased with increasing additions of chicken litter resulting in its dominance on the exchange complex in the co-composted treatments (Table 3). Contrasting this, organic carbon declined from the initial values observed in the individual constituents, suggesting losses of CO2 associated with the mineralization of organic matter. In general, the quality of the compost material with respect to available nutrients increased in those treatments receiving chicken litter, rice husk and rice husk ash. Calcium and Mg2+ concentrations increased along with the CEC (Table 3). A cursory assessment of CEC and the sum of basic cations indicates that a significant proportion of the exchangeable cations are not associated with the exchange complex and are therefore in a soluble and mobile form.

Treatments that did not receive additions of rice husk, rice husk ash and chicken litter did not undergo significant changes in chemical properties (Tables 2 and 3). They remained acidic with little change in their chemical characteristics from their original state (Tables 2 and 3).

One of the main reasons for undertaking co-composting with readily available agri-waste products was to reduce the hydrophobic nature of the waste oil bentonites associated with the deposition of oils and fats on the surface of these materials. The WDPT for the waste materials prior to and after 84 days co-composting are presented in Tables 2 and 3. Prior to the co-composting process the WDPT ranged from 25 to >10,800 seconds for the soybean and palm oil wastes respectively (Table 2). Through the co-composting process the WDPT declined significantly (Tables 2 and 3). With increasing additions of chicken litter to the compost mix there was a significant increase in the WDPT (Table 3) although these values were substantially lower than the original waste bentonites (Table 2). In those treatments that did not undergo co-composting, the WDPT either declined from their original values in the case of the rice oil or remained the same in the case of the soybean and palm oil (Table 3).

Table 3. Selective chemical characters of co-composted treatments 84 days after the initiation of the composting process

Treatment

pHW

pHCa

EC

CEC

Exch. Ca2+ Exch. Mg2+ Exch. Na+

Exch. K+

Total N

OC

WDPT

(mS cm-1)

   

(cmolc kg-1)

 

(%)

(%)

(sec)

SB01 6.45 6.32 7.59 39.68 44.95 30.13 10.20 23.57 0.67 12.92

16.0

SB02 6.79 6.65 9.97 39.17 37.12 29.54 13.75 36.26 0.97 13.67

22.0

SB03 6.96 6.83 11.59 45.60 35.00 27.35 14.86 46.62 1.06 15.33

28.2

PB01 6.95 6.73 5.75 45.01 15.55 19.61 11.31 28.49 0.82 11.56

20.6

PB02 7.14 6.89 7.85 51.90 19.29 17.42 7.80 41.44 1.06 12.17

33.7

PB03 7.72 7.49 8.71 51.02 20.50 17.84 16.19 50.50 1.23 12.84

30.1

RB01 7.39 7.32 7.05 30.78 25.81 19.19 10.20 23.57 0.92 13.40

47.3

RB02 7.50 7.38 9.79 37.81 26.71 20.37 14.19 40.14 1.10 14.36

66.9

RB03 7.76 7.58 11.24 43.76 30.10 18.77 16.63 50.50 1.21 15.19

79.7

P-SB 2.99 2.96 5.17 42.44 46.76 14.98 1.48 0.60 0.07 22.05

24.1

P-PB 4.47 4.30 2.45 42.30 8.74 15.07 2.44 1.83 0.11 27.42

>10,800

P-RB 3.02 2.91 1.62 23.48 15.86 3.79 2.19 0.38 0.06 24.36

23.5

LSD0.05

0.04 0.03 0.13 2.67 7.64 2.06 0.67 3.32 0.13 1.89

3.4

The decline in the hydrophobicity of the waste rice oil bentonite after 84 days may in part be attributed to a rise in pile temperature observed in this treatment (Soda et al., 2005).

Changes in soil chemical attributes following the application of composted bentonite wastes to a degraded light textured soil.

Selected chemical properties of soils collected at the conclusion of the cropping phase are presented in Table 4. It is clearly evident that increasing rates of application of the co-composted waste oil bentonites resulted in a significant increase in soil pH (Table 4). The greatest increase in pHCa (a more meaningful measure of soil reactivity than pHW as it negates the influence of soluble salt on pH) was observed in the RB03 treatment (pH 6.64) applied at 120 t ha-1 (Table 4). In addition, increasing the ratio of chicken litter in the original co-composted mix resulted in significantly greater increases in pH at any given application rate. Contrasting this, the application of the non-composted waste bentonite treatments (P-SB, P-PB and P-RB) had a mixed effect on the soil reactivity (Table 4). The waste soybean oil bentonite resulted in a significant decline in soil pH with increasing application rates, this being associated with the very acid (pHCa 2.98) nature of this material (Table 2). The waste rice bran oil bentonite had a variable effect on pH whilst the waste palm oil bentonite significantly increased pH, this again being a reflection of the overall pH of this material (Tables 3 and 4).

The electrical conductivity (EC) of the 1:5 extract ranged from 0.003 dS m-1 in the un-amended soil to 0.853 dS m-1 in the case of the non co-composted waste soybean oil bentonite (P-SB) at a rate of 120 t ha-1 (Table 4). The range in EC values do not exceed the threshold value associated with sensitivity to salinity in maize of 1.5-3.0 dS m-1 as published by Mass (1990). However, these relatively high values indicate a high level of soluble salts that are not associated with the soil exchange complex. A plot of ECEC (∑ Ca2+ + Mg2+ + Na+ + K+) versus CEC measured in an ammonium acetate buffered system clearly demonstrates the

presence of a significant proportion of the measured cations being in a ‘soluble’ form (Figure 1). Intuitively, if all of the measured cations were held on the exchange complex, then a plot of ECEC versus CEC would cause the coordinates for each of the samples to fall very close to the 1:1 line. The results suggest that a significant proportion of the extracted cations are not associated with the exchange complex and are therefore subject to leaching loss. The fact that there is still a large amount of ‘soluble’ cations at the conclusion of the study may indicate either a lack of significant leaching events during the growth of the crop and/or the mineralization of root material after the harvesting of the final crop.

Measurement of CEC undertaken in a buffered system effectively allows an assessment of changes in exchange capacity that are not confounded by the generation of charge associated with pH. Increasing rates of treatment application resulted in significant increases in the amount of CEC generated with the greatest increase in CEC being observed with the application of non co-composted waste palm oil bentonite (P-PB) at a rate of 120 t ha-1 (Table 4). As one of the main reasons for applying these materials is to increase the CEC of these degraded soils, it is clearly evident from the results that this has been achieved. Furthermore, by developing CEC response functions for each of the materials evaluated, an estimate of the amount of material to be applied to achieve a predetermined target CEC can be made.

Management of Tropical
Sandy Soils for Sustainable
Agriculture

Figure 1. Relationship between measured CEC and the sum of exchange cations for soils treated with co-composted materials at the conclusion of two cropping cycles with maize

As expected, increasing additions of co-composted waste oil bentonites resulted in a significant increase in the organic carbon content of the soils (Table 4). The greatest increases were observed in the non co-composted waste oil bentonite treatments (P-SB, P-PB and P-RB). Increases in organic carbon would have contributed to the observed increases in CEC and would be of benefit to the overall quality of the soil.

Table 4. Selective chemical properties of oils treated with co-composted and non co-composted waste oil bentonites at the conclusion of the study

Treatment

Rate of
application
(t ha-1)

pHW

pHCa

EC

CEC

Ca2+

Mg2+

K+

OC

WDPT end

WDPT start

(dS m-1)

 

(cmolc kg-1)

 

(%)

(sec)

(sec)

Control 0 5.13 4.02 0.003 1.59 0.51 0.14 0.04 0.39 0.16

0.16

SB01 15 5.56 4.89 0.070 2.12 1.26 0.38 0.03 0.44 0.18

0.19

SB01 30 6.36 5.74 0.127 2.17 1.84 0.63 0.04 0.46 0.19

0.80

SB01 60 6.81 6.19 0.214 2.77 2.82 0.82 0.07 0.57 0.24

0.20

SB01 120 6.91 6.40 0.431 3.02 4.36 1.43 0.39 0.70 0.26

25.00

SB02 15 6.26 5.46 0.064 1.96 1.33 0.41 0.04 0.41 0.19

0.20

SB02 30 6.75 6.07 0.106 2.17 2.14 0.65 0.04 0.51 0.19

0.20

SB02 60 7.02 6.40 0.197 2.56 3.10 1.05 0.17 0.59 0.21

0.21

SB02 120 7.10 6.59 0.376 3.00 4.76 1.68 0.61 0.68 0.29

0.26

SB03 15 6.32 5.52 0.064 2.09 1.40 0.30 0.04 0.44 0.19

0.20

SB03 30 6.79 6.16 0.106 2.36 2.04 0.50 0.06 0.48 0.19

0.21

SB03 60 7.17 6.45 0.186 3.12 3.30 0.93 0.21 0.56 0.19

0.22

SB03 120 7.15 6.57 0.367 3.16 4.97 1.43 0.67 0.70 0.34

0.23

PB01 15 6.49 5.59 0.045 2.01 1.29 0.36 0.04 0.39 0.19

0.23

PB01 30 6.94 6.08 0.065 2.32 1.84 0.58 0.05 0.46 0.19

0.24

PB01 60 7.13 6.38 0.099 3.05 2.72 0.99 0.09 0.51 0.20

0.25

PB01 120 7.17 6.46 0.175 3.97 3.81 1.48 0.33 0.68 1.47

0.28

PB02 15 6.63 5.75 0.048 2.04 1.28 0.37 0.04 0.38 0.20

0.23

PB02 30 7.03 6.15 0.069 2.42 2.01 0.57 0.05 0.46 0.20

0.26

PB02 60 7.21 6.38 0.126 2.64 3.01 0.87 0.18 0.39 0.18

0.27

PB02 120 7.26 6.52 0.230 3.18 4.57 1.69 0.66 0.69 1.76

0.30

PB03 15 6.62 5.85 0.049 2.21 1.40 0.49 0.05 0.44 0.19

0.29

PB03 30 7.15 6.30 0.074 2.57 2.06 0.72 0.07 0.47 0.18

0.28

PB03 60 7.29 6.49 0.129 2.97 3.12 1.08 0.18 0.52 0.18

0.35

PB03 120 7.39 6.57 0.227 3.46 4.26 1.77 0.72 0.67 0.47

0.35

LSD0.05   0.14 0.09 0.033 0.48 0.29 0.16 0.04 0.09 9.64

7.18

Treatment

Rate of
application
(t ha-1)

pHW

pHCa

EC

CEC

Ca2+

Mg2+

K+

OC

WDPT end

WDPT start

(dS m-1)

 

(cmolc kg-1)

 

(%)

(sec)

(sec)

Control 0 5.13 4.02 0.003 1.59 0.51 0.14 0.04 0.39 0.16

0.16

RB01 15 6.48 5.11 0.062 2.21 1.93 0.49 0.05 0.43 0.19

0.22

RB01 30 6.90 5.98 0.081 2.51 1.74 0.64 0.05 0.46 0.49

0.25

RB01 60 9.99 6.27 0.128 2.63 2.41 0.82 0.07 0.44 0.39

0.24

RB01 120 6.99 6.45 0.266 3.33 4.07 1.24 0.31 0.67 0.30

0.27

RB02 15 6.40 5.45 0.060 2.02 1.35 0.45 0.05 0.42 0.18

0.25

RB02 30 6.77 6.09 0.095 2.40 2.08 0.64 0.05 0.42 0.23

0.26

RB02 60 7.15 6.47 0.156 2.66 3.02 0.90 0.17 0.53 0.30

0.27

RB02 120 7.31 6.58 0.280 3.04 4.11 1.34 0.51 0.65 0.22

0.30

RB03 15 6.55 5.66 0.061 2.30 1.57 0.46 0.05 0.43 0.19

0.30

RB03 30 7.03 6.25 0.083 2.23 2.02 0.60 0.06 0.46 0.21

0.27

RB03 60 7.29 6.51 0.148 2.53 3.25 0.95 0.22 0.55 0.27

0.30

RB03 120 7.42 6.64 0.270 2.90 4.31 1.54 0.57 0.70 0.19

0.42

P-SB 15 4.60 3.88 0.134 1.92 0.69 0.19 0.03 0.46 0.19

0.73

P-SB 30 4.26 3.78 0.245 1.97 1.00 0.29 0.03 0.57 0.19

1.00

P-SB 60 4.27 3.81 0.448 2.38 1.67 0.44 0.05 0.72 0.55

1.67

P-SB 120 3.82 3.53 0.853 4.02 3.16 0.81 0.14 1.08 17.73

3.00

P-PB 15 5.27 4.21 0.045 1.69 0.48 0.25 0.04 0.44 19.57

5.27

P-PB 30 5.24 4.32 0.068 2.03 0.81 0.38 0.06 0.56 74.50

154.02

P-PB 60 5.28 4.52 0.122 2.83 1.17 0.60 0.11 0.74 22.14

5,400

P-PB 120 5.01 4.51 0.286 4.32 1.61 0.71 0.15 1.18 4.17

6,300

P-RB 15 5.05 4.14 0.071 1.83 0.69 0.17 0.08 0.48 0.20

1.00

P-RB 30 4.91 4.17 0.129 2.05 0.96 0.23 0.04 0.58 0.73

3.30

P-RB 60 4.91 4.21 0.236 2.82 1.40 0.29 0.09 0.76 19.20

8.03

P-RB 120 4.40 3.99 0.520 3.20 2.40 0.41 0.15 1.26 20.20

10.37

LSD0.05

  0.14 0.09 0.033 0.48 0.29 0.16 0.04 0.09 9.64

7.18

Water penetration into those soils treated with the co-composted bentonites was rapid (Table 4). Clearly the co-composting process and time significantly reduced the hydrophobicity of these materials. As noted previously in the case of the non co-composted waste palm oil bentonite, the hydrophobic nature of the material remained unaltered (>10,800 sec) between the start and end of the composting phase (Tables 2 and 3). However, incorporation into the soil resulted in a significant decrease in water repellency indicating the potential role of inherent soil microbial populations in the breakdown of oils and fats (Table 4).

Changes in soil physical attributes

Total porosity

For all co-composted treatments at application rates of 15 and 30 t ha-1, no significant increases in the total porosity (θp) were observed. In contrast, increases in θp were observed for the SB01, SB03, PB03, RB02, and RB03 co-composted waste oil bentonite treatments at application rates of 60 t ha-1 as compared with the respective 15 and 30 t ha-1 treatments. Further, all co-composted treatments of waste oil bentonite (SB, PB and RB) applied at 120 t ha-1 resulted in significant increases in θp as compared with the respective 15, 30 and 60 t ha-1 treatments (Figure 2). No increase in the total porosity was observed for the non-composted P-PB, P-SB, and P-RB waste oil bentonite treatments.

Soil moisture content at the permanent wilting point

For all co-composted and non co-composted waste oil bentonites, the soil moisture content at the permanent wilting point (θpwp) slightly but pro­portionally increased with increased application rates from 15 to 120 t ha-1 even though the observed increase in moisture content was <0.08 m3 m-3 (Figure 3).

Available water content

Figure 4 represents changes in the available water content for crop growth (θawc) associated with the application of the co-composted and non co-composted waste oil bentonite treatments. Available water content for crop growth for all treatments had a large deviation between samples (LSD0.05 >0.026). However, increasing application rates of RB03 resulted in a proportional increase in θawc. For the co-composted RB01, and RB02 co-composted waste oil bentonite treatments θawc increased slightly with increasing application rate from 15 to 120 t ha-1. Further, significant increases in θawc were observed between the 15 and 30 t ha-1 co-composted PB02 treatments.

Management of Tropical
Sandy Soils for Sustainable
Agriculture

Figure 2. The effect of co-composted and non co-composted waste oil bentonites on the total porosity of soils. Vertical bars are the LSD0.05 between treatment means

In addition, significant beneficial increases in θawc were observed when increasing the non co-composted P-PB application rate from 30 and 60 t ha-1. However, no significant increase in θawc was detected when increasing the application rate of non co-composted P-PB to 120 t ha-1.

Relationship between θpwp and nutrient holding capacity

The soil water retention at lower matric potentials (approximately <-100 kPa) predominantly depends on adsorptive forces between the soil solid surface components (including soil particles and organic matter) and the soil solution, hence it is significantly affected by the specific surface area of soils. The θpwp is fairly well correlated with the surface area of a soil and would represent, roughly, about 10 molecular layers of water if it were distributed uniformly over the solid surface of the soil (Hillel, 1998). This indicates that an increase in θpwp is associated with an increase in the specific surface area of the soil.

Management of Tropical
Sandy Soils for Sustainable
Agriculture

Figure 3. The effect of co-composted and non co-composted waste oil bentonites on soil moisture content at the permanent wilting point. Vertical bars are the LSD0.05 between treatment means

Management of Tropical
Sandy Soils for Sustainable
Agriculture

Figure 4. The effect of co-composted and non co-composted waste oil bentonites on the available soil moisture content for crop growth. Vertical bars are the LSD0.05 between treatment means

As stated above, the observed θpwp of the degraded light textured sandy soil evaluated was significantly correlated with the co-composted and non co-composted waste oil bentonite treatments (Figure 4). It is suggested here that the inferred increases in specific surface area (as indicated by increased θpwp) is due to the applied bentonite and organic matter. In addition, the application of bentonite and organic matter in the co- and non co-composted treatments resulted in significant increases in CEC.

Management of Tropical
Sandy Soils for Sustainable
Agriculture

Figure 5. A relationship between gravimetric soil moisture content at the permanent wilting point (x) and cation exchangeable capacity (y) of the light textured sandy soils after application of different ratios of co-composted and non co-composted waste oil bentonites

Figure 5 illustrates the highly significant relationship between θpwp and CEC (R2 = 0.779, P <0.01). Since negative surface charge is directly related to the specific surface area of clay minerals and organic matter (Jury et al., 1991), the results of this study indicate that increases in CEC of the test soil through the incorporation of co- and non co-composted waste oil bentonites were due primarily to an increase in the specific surface area (and associated negative charges) as derived from the bentonite and organic matter. This has major positive implications with regard to the nutrient holding capacity of the test soil.

Conclusion

Through composting waste oil bentonites with readily available agricultural byproducts, the chemical and physical characteristics of these materials have been drastically improved. In addition, the composted materials have had beneficial effects on enhancing the chemical and physical properties of a degraded light textured sandy soil. The application of composted materials significantly increased the pH, and altered the characteristics of the soil through an increase in the soil organic matter, clay content and their nutrient supplying capacity. Further, they have had a positive impact on soil physical attributes resulting in an increase in total porosity and available water content. It is of note that composted materials are not fertilizers and hence there will be a requirement for the application of nutrients commensurate with crop requirements.

The process of co-composting these materials will reduce their potential negative impact on the environment through the neutralization of their acid reactivity and a decline in their hydrophobicity. Such processing, whereby locally available farm waste products (i.e. rice husk, chicken litter) can be utilized to produce an excellent soil amendment, may suit a small business model. Such a development would turn an environmental hazardous waste into a high quality soil amendment that would have a retail value.

Acknowledgement

The research team of both IWMI and LDD would like to express their gratitude to the Division of Microbiology of the Thai Department of Agriculture for their help during the composting period, especially Mrs. Bhavana Likhananont who provided technical guidance. The contribution of the Thai Ruam Jai Vegetable Oil Company Limited (rice bran oil), the Industrial Enterprise Company Limited (palm oil), and the Morakot Industries Public Company Limited are also acknowledged for providing the raw materials of waste oil bentonite and acid activated bentonite.

References

Bisdom, E.B.A.; Dekker, A.L.; Schoute, J.F.T. 1993. Water repellency of sieve fractions from sandy soils and relationships with organic material and soil structure, Geoderma 56, 105-119.

Croker, J., Poss, R., Hartman, C. and Bhuthorndharaj, S. 2004. Effects of recycled bentonite addition on soil properties, plant growth and nutrient uptake in a tropical sandy soil. Plant and Soil, Vol. 267, No. 1-2: 155-163.

Driessen, P. ; Deckers, J.; Spaargaren, O.; Nachtergaele, F. 2001. Lecture notes on the major soils of the world, FAO World Soil Resources Report No. 94, 334 p.

Falaras, P.; Kovanis, I.; Lezou, F.; Seiragakis, G. 1999. Cottonseed oil bleaching by acid-activated montmorillonite, Clay Minerals, Mineralogy Society, Vol. 34, No. 2, 221-232(12).

Foletto, E.L.; Alves, C.C.A.; Sganzerla, L.R.; Porto, L.M. 2002. Regeneration and utilization of spent bleaching clay. Latin American Research Journal, Vol. 32, 205-208.

Hillel, D. 1998. Environmental Soil Physics, San Diego: Academic press, 129-172.

Imsamut, S.; Boonsoppan, B. 1999. Established soil series in the Northeast of Thailand. Reclassified according to soil taxonomy 1998, Report Department of Land Development (Thailand), 154 p.

Jury, W.A.; Gardner, W.R.; Gardner, W.H. 1991. The soil solid phase. In: Soil Physics, 5th ed,. New York, John Wiley & Sons, 1-33.

Kheoruenromme, I.; Suddhiprakarn, A.; Kanghae, P. 1998. Properties, environment and fertility capability of sandy soils in Northern Plateau, Thailand. Kasetsart Journal 32, 355-373.

LDD, 1993. Characterization of soil series established in the Northeastern Region and their interpretation for uses. Technical report no. 314, Land Development Department, Bangkok, Thailand.

Lesturgez, G. 2005. Densification des sols sableux sous culture mécanisee: cas du Nord-Est Thaïlandais. Thesis submitted to the University of Nancy in fulfilment of the requirements for the degree of Doctor of Philosophy in Geosciences, France.

Mass, E.V. 1990. Crop salt tolerance. In Agricultural salinity assessment and management, ed. K.K. Tanji, New York, NY: American Society of Civil Engineering, 262-304.

Matsuo, K. 2002. Development of upland cropping systems for crop-animal integrated farming systems in Northeast Thailand. Final report of the collaborative research project between Japan International Research Centre for Agricultural Sciences, Department of Agriculture and Khon Kaen University, Thailand, Japan International Research Centre for Agricultural Sciences (JIRCAS).

Noble, A.D.; Gillman, G.P.; Ruaysoongnern, S. 2000. A cation exchange index for assessing degradation of acid soil by further acidification under permanent agriculture in the tropics, European Journal of Soil Science, 51, 233-243.

Noble, A.D.; Gillman, G.P.; Nath, S.; Srivastava, R.J. 2001. Changes in the surface charge characteristics of degraded soils in the tropics through the addition of beneficiated bentonite, Australian Journal of Soil Research 39, 991-1001.

Noble, A.D.; Ruaysoongern, S. Penning de Vries, F.W.T.; Hartmann, C.; Webb, M.J. 2004. Enhancing the Agronomic Productivity of Degraded Soils in Northeast Thailand through Clay-based Interventions. In: Water and Agriculture, eds., V. Seng, E. Craswell, S. Fukai and K. Fischer, ACIAR Proceedings No. 116, 147-160.

Rayment G.E.; Higginson, F.R. 1992. Australian Laboratory Handbook of Soil and Water Chemical Methods, Melbourne, Inkata Press.

Smart, M.; Cozen, G.; Zarcinas, B.; Stevens, D.; Barry, G.; Cockley, T.; McLaughlin, M. 1998. CSIRO Land and Water’s Method Manual for ACIAR Project No. LWR1/1998/119, 33-34.

Soil Survey Staff. 1990. Keys to Soil Taxonomy, 4th ed. Virginia Polytechnic Institute and State University, Blacksburg, SMSS Technical Monograph No. 19.

Suzuki, S. 2004. Verification of freezing point depression method for measuring matric potential of soil water, Soil Science and Plant Nutrition 50, 1277-1280.

Suzuki, S.; Noble, A.D.; Ruaysoongnern, S.; Chinabut, N. 2005. Improvement in water holding capacity and structural stability of a sandy soil in Northeast Thailand, Arid Land Research and Management: Submitted.

Virta, R.L. 2001. Clays, U.S. Geological Survey, Mineral Commodity Summaries.

February 2000, Available on the USGS Internet site: http:// www.usgs.gov/.


1 International Water Management Institute, Southeast Asia Regional Office, P.O. Box 1025, Kasetsart University, Bangkok 10903, Thailand.
2 National Agricultural Research Center for Hokkaido Region Hitsujigaoka Toyohira-ku Sapporo 062-8555, Japan
3 Land Development Department, Chatuchak, Bangkok 10900, Thailand.

Previous PageTop of PageNext Page