SSSAJ Grow Your Career with SSSA
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

Published online 20 September 2006
Published in Soil Sci Soc Am J 70:1860-1871 (2006)
DOI: 10.2136/sssaj2006.0009
© 2006 Soil Science Society of America
677 S. Segoe Rd., Madison, WI 53711 USA
This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF) Free
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Morari, F.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Morari, F.
GeoRef
Right arrow GeoRef Citation
Agricola
Right arrow Articles by Morari, F.
Related Collections
Right arrow Water Flow Models
Right arrow Recharge
Right arrow Soil Methods/Instrumentation

Soil Physics

Drainage Flux Measurement and Errors Associated with Automatic Tension-controlled Suction Plates

Francesco Morari*

Dipartimento di Agronomia Ambientale e Produzioni Vegetali, Agripolis, Universitá Di Padova, Viale dell'Università, 16 35020, Legnaro (Pd)–Italy

* Corresponding author (francesco.morari{at}unipd.it)


   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 MATERIAL AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Automatic control of high tension soil solution samplers (suction lysimeters) in response to the surrounding soil matric potential has been proposed as a new method to reduce the convergence and divergence fluxes around the lysimeters. It is important to evaluate the instruments' performance with automatically regulated suction because the efficiency of the control and its effects on the flux volumes could vary with the surrounding soil matric potential. An automatic equilibrium tension lysimeter (AETL) system was developed composed of 16 small-area (572 cm2) ceramic suction plates that were continuously kept in equilibrium by an automated control device. The automated control device consisted of a datalogger connected to electronic tensiometers and to an electric vacuum pump that was activated to regulate the ceramic plate suction. The system was evaluated by simulating its performance with the HYDRUS-2D finite-element model. The AETL system continuously kept the lysimeters' suction in equilibrium with the surrounding soil when the soil water matric potential head ranged from –306 to 0 cm H2O. Suction control was less effective during low drainage conditions (soil matric potential head less than –50 cm H2O), when it took more time to re-equilibrate the pressure heads. However analysis of the variability of drainage and model simulations showed that incorrect suction control in drier conditions had a limited effect on the collected volumes. Suction was much better controlled when drainage flux was high, allowing a correct estimate of drainage. Automatic control of suction plates is a valuable tool for estimating drainage fluxes. Fine control of suction is needed when drainage flux is high because even the slight overapplication of suction, which is typically done during field experiments to overcome porous plate resistance and ensure that a sample is collected, could produce marked overestimation of water drainage (>30%).

Abbreviations: AETL, automatic equilibrium tension lysimeter • AL, above lysimeter • BS, bulk soil • CV, coefficient of variation • GB, growth box • RSE, relative standard error • SE, standard error


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
MATERIAL AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
DIFFERENT METHODS have been proposed to measure the movement of water and pollutants through the vadose zone. Soil coring is relatively inexpensive (Rhoades, 1982) but is destructive and labor intensive, limiting its practical applications. Suction-cup lysimeters (high tension soil solution samplers) are frequently used in plots and the open field (e.g., Webster et al., 1993), because they are relatively simple to install and result in minimal soil disturbance during subsequent sampling (Lord and Shepherd, 1993). The use of these devices in structured soils with preferential flow can cause errors if flow bypasses the porous samplers (Barbee and Brown, 1986). Zero-tension lysimeters collect only gravitational water and require the formation of a saturated zone above the porous plate (Barbee and Brown, 1986). Only part of the naturally occurring drainage in a typical soil profile can be collected, because the prerequisite for saturation above these sampling devices may cause drainage water to bypass the porous plates (Jemison and Fox, 1992). Brown et al. (1986) introduced passive capillary samplers as an alternative for collecting soil water. These use self-priming wicks to produce a hanging water column that exerts suction on the soil above the lysimeters. It has been proved that passive capillary samplers have a better collecting efficiency than zero-tension lysimeters (Zhu et al., 2002) and suction cups (Barzegar et al., 2004).

Non-weighing tension pan lysimeters provide a means for collecting drainage water from a known cross-sectional area (Lentz and Kincaid, 2003). To operate, tension pan lysimeters depend on two conditions: (i) minimal discontinuities in the flow path of moving water and (ii) a driving force for water movement. Maintaining a capillary connections between the soil column and the surface of the lysimeters is fundamental to minimize ponding and preferential flow around the lysimeter (Brye et al., 1999). Appropriate suction is important for correctly measuring water and solute fluxes. Water-flow convergence or divergence around the porous plate is observed when in situ soil tension differs from the suction applied to fixed-tension samplers, which produces an over- or undermeasurement of the percolate volume (Cochran et al., 1970; van der Ploeg and Beese, 1977). Solute concentrations are also affected by applied suction (Hansen and Harris, 1975; Haines et al., 1982; Magid and Christensen, 1993). Haines et al. (1982) observed differences in water quality sampled by tension and zero tension lysimeters. They explained these discrepancies by hypothesizing that the efficiency of the two systems in sampling macropore and matrix flow differed. Magid and Christensen (1993) also observed differences in soil solutions sampled with and without tension. According to these authors zero-tension samples represented the flux concentrations, whereas tension samples were an approximation of resident concentrations.

The functioning of fixed-tension lysimeters can be enhanced by extending side walls above the porous plate (Corey et al., 1982; Cepuder et al., 1992). The soil volume confined in the sampler is thought to reduce the influence of varying soil water potential on water flow through the sampler opening (Corey et al., 1982). This device, although simple, may promote preferential flow down the side of the soil column when dry and disturb lateral water movement. To address the problems associated with fixed-tension lysimeters, recent studies have proposed controlling the suction for extracting unsaturated soil water by referring to soil water matric potentials observed in the surrounding natural soil profile. The suction can be adjusted to a target value decided from the matric potential observations in the natural soil profile (Lentz and Kincaid, 2003; Siemens and Kaupenjohann, 2004; Ciglasch et al., 2005), or the suction control is set so that the soil matric potential immediately above the porous lysimeter should be similar to the matric potential at the same depth in the natural soil (Brye et al., 1999; Kosugi and Katsuyama, 2004; Masarik et al., 2004). In controlled-tension lysimeters, the suction can be adjusted manually or automatically. In the system developed by Brye et al. (1999) lysimeter suction was manually adjusted on the basis of the soil water potential measured by heat-dissipation sensors. Masarik et al. (2004) improved the lysimeters originally constructed by Brye et al. (1999) by developing an automated control system. The performances of this system were equivalent to those of the manually adjusted equilibrium tension lysimeter (Masarik et al., 2004). Large-surface lysimeters were used, which are preferable, because variability in drainage measurements decreases with increasing lysimeter size (Radulovich and Sollins, 1987). However, the installation of large-surface lysimeters is problematic in stony soil or small experimental plots. To overcome porous plate resistance, Masarik et al. (2004) set the suction inside the lysimeters at 2 kPa more than the soil-water matric potential in the surrounding soil. As a consequence, the soil-water matric potential above the lysimeter was also constantly slightly less (<2 kPa) than the soil-water matric potential in the bulk soil. The authors made no attempt to estimate potential errors in drainage flux due to this additional suction. The automatic control system developed by Kosugi and Katsuyama (2004) adjusted the water extraction period by applying a strong vacuum instead of controlling the suction inside the lysimeters. Through these controls a dynamic equilibrium was achieved with periods of undersuctions that followed alternately periods of oversuctions. Also these authors made no attempt to estimate the errors in drainage flux.

I hypothesize that the magnitude of the errors varies according to the gradient created by the lysimeter suction and soil-water potential conditions. When soil-water potential is high (i.e., near saturation), even a slight constant matric potential gradient could lead to significant errors in drainage measurement. It is not generally possible to evaluate a new lysimeter system by comparing it with accurate lysimeters installed in the same type of environment. However, a good evaluation of lysimeter performance can be achieved by using mathematical models that can simulate the functioning of the lysimeter in field conditions with a proper time-space scale. Advanced mathematical models such as HYDRUS (Simunek et al., 1999), are capable of simulating complex unsaturated water flow dynamics like those observed in the soil surrounding equilibrium tension lysimeters (Mertens et al., 2005; Ciglasch et al., 2005).

The overall objective of this study was to design, test, and evaluate the performance of an automated equilibrium system composed of small-area tension lysimeters. Testing and evaluation consisted of field experiments and mathematical simulation of the field results.


   MATERIAL AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIAL AND METHODS
RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
The Automatic Equilibrium Tension Lysimeter System: Design And Installation
This research was part of a 3-yr experiment (October 1999–September 2002) to identify appropriate cattle slurry application rates on silage maize (Zea mays L.) (Borin et al., 2003). The 21 x 34 m study site was on the University of Padova's Experimental Farm, at Legnaro, NE Italy. It was conducted in 48 20-yr-old growth boxes (GBs; 2 x 2 m on a side x 1.3 m deep) with concrete walls buried in the field and open bases to allow water drainage. The soil was an Oxyaquic Eutrudept, coarse-silty, mixed, mesic (Soil Survey Staff, 1998), containing 13% clay, 52% silt, and 34% sand, with a pH of 7.8. Organic C content was 0.96% in the top 46 cm and decreased to 0.47% at a depth of 73 to 101 cm.

The lysimeters were installed in 16 GBs. All 16 GBs used in this study received the same volume of liquid in slurry and supplemental water applications during the study. Before sowing, the top soil layer (0–20 cm) was manually tilled to incorporate the fertilizers and prepare the seedbed. In 2002, the silage maize was sown on April 18 at a density of 7 plants m–2 and harvested on August 30.

Automatic Equilibrium Tension Lysimeter System
The AETL system consisted of the 16 small-area lysimeters connected to a central vacuum unit by means of 2-mm polyamide pipes (Fig. 1 ). The lysimeters were 0.75-mm thick porous ceramic plates (Soilmoisture Equipment Corp., Santa Barbara, CA) with a surface of 572 cm2 (radius 13.5 cm), air-entry suction of 50 kPa ({approx}510 cm H2O), and saturated hydraulic conductivity of 1.25 x 10–5 cm s–1 {approx}1.08 cm d–1. The central control unit was located near the plots in a small insulated wood building. The control unit was composed of an electric vacuum pump (0.37 kW), a vacuum tank (50 L), two 5-L bottles to act as water traps, and 16 pairs of 1-L bottles (one pair per lysimeter) to collect water samples (Fig. 1). The vacuum lines to each lysimeter were individually valved so that defective lysimeters could be isolated. The vacuum lines were protected by PVC (polyvinyl chloride) pipe. Two isolated pipelines were created to apply the same number of vacuum regimes.


Figure 1
View larger version (25K):
[in this window]
[in a new window]
  		Fig. 1. Diagram of a portion of the automatic equilibrium tension lysimeter (AETL) system, including electronic tensiometers (E-tensiometers), lysimeters, central collection unit, and automatic vacuum controller.

 
The lysimeter system was installed in spring 2000. The 16 ceramic plates were washed with 2 L of 1 M hydrochloric acid followed by 5 L of distilled water (Neary and Tomassini, 1985), then installed in the GBs at a depth of 90 cm by digging a 70-cm2 hole 100 cm deep. This was the minimum size possible to allow plate installation while limiting soil disturbance. On one side of this hole, a small cavity, approx. 30 cm wide x 5 cm high x 40 cm deep (into the side of the original hole), was excavated with a specially designed tool at a depth of 95 cm. To ensure continuity between soil and ceramic plate, the cavity was filled with soil slurry and the plate was inserted horizontally, forcing the excess soil slurry out. The soil column extending from the plate up to the soil surface thus remained undisturbed. Wooden wedges were used to firmly press and hold the plates against the soil mass. Each plate was then connected to a vacuum/sample line and the holes were backfilled with soil, to regain the original profile. A valve connection just above the GBs allowed the ceramic plate to be rewet by siphoning water into the sampling tube from a container on the surface. This rewetting was necessary to re-establish hydraulic conductivity with the soil above the ceramic plate when soil water potential exceeded the air-entry potential of the porous plate. Rewetting was done in late summer 2000 and 2001.

Three 30-cm long TDR probes (MP-917, Environmental Sensors, Inc., Victoria, BC, Canada) were then inserted at a 45° angle in each of the boxes containing lysimeters. The probes were installed in the undisturbed portion of the soil profile to give integrated measurement of soil moisture in the 0- to 20-, 25- to 45-, and 60- to 80-cm horizons. The MP-917 uses remote shorting diodes in a combination of differential detection techniques to assure correct identification of the waveform (Hook et al., 1992). The experiment used a single-segment probe, which had only one diode at the top. The TDR, calibrated in the laboratory, had an accuracy of within 1.5%. Weekly measurements were taken manually.

Automatic Vacuum Controller
The automated control system was developed over a period of 3 yr and only during the final year of the experiment (2002) was it used and evaluated in a fully automated mode. Before this, the AETL system was regulated manually, with the ceramic plates' suction being adjusted every 1 or 2 d to match that of the surrounding soil.

During the first 2 yr, drainage from the lysimeters was highly variable. Therefore only one vacuum regime was applied to the 16 lysimeters by monitoring the soil water potential in the soil around three of them. Soil water potential was measured with tensiometers (model 2100F, Soilmoisture Equipment Corp., Santa Barbara, CA), which were modified by replacing the vacuum dial gauge with calibrated pressure transducer (output 0–1000 mV). The tensiometer porous cups were 6 mm in diameter and 2.5 cm long and attached to the tensiometer body via a 1.8-m length of polyethylene water-filled tubing (Fig. 1). The tubing was formed by an outer tube (0.24 mm in diam.) and an inner vent tube (0.08 mm in diam.). Tubing water volume was 5.6 cm3 while the water reservoir in the tensiometer's main body had a volume of 12 cm3.

Two tensiometers were installed in three of the 16 GBs. Their porous ceramic cups were placed just above (1 cm) the ceramic plates and in the adjacent 50 cm of bulk soil at a depth of approximately 90 cm. To reduce soil disturbance during installation, the tensiometers were first inserted in a steel probe and then pushed obliquely into the soil profile down to the defined depth. To avoid preferential flow, the first 5-cm soil layer around the pipes was sealed with bentonite clay. The water body tube of the tensiometer with the transducer was initially housed in an aboveground insulated sample box, but to further reduce the effect of air temperature on the tension readings, it was later placed underground, still in the box, at a depth of 15 cm.

In this study, soil water potentials are described by the water head (1 cm {approx} 0.098 kPa). The AETL system was controlled by a datalogger (Model CR7, Campbell Scientific). The datalogger recorded soil water potential as measured by the six tensiometer-mounted pressure transducers and controlled two electronic valves that connected the samplers to the vacuum supply lines (Fig. 1). By controlling the vacuum supply period (Kosugi and Katsuyama, 2004), the dataloggers maintained equilibrium between soil water potential above and around the ceramic plates. Every 10 min the datalogger: measured, averaged and compared the matric pressure measured individually by the three tensiometers at the above-plate and bulk soil positions. If the average pressure head above the plate was 3 cm higher than the pressure head in the bulk soil around the plates, the datalogger increased plate tension by opening vacuum line flow valves for 1 min. A maximum vacuum of –408 cm was applied. By averaging all tensiometers values from a given position the control program allowed defective electronic tensiometers to be excluded. The AETL system was not designed to correct vacuum values if they exceeded the soil-tension-based target. As with the control system developed by Lentz and Kincaid (2003), soil water flows to the samplers were enough to cause sampler system pressures to decline over time, allowing a quasi-equilibrium to be maintained in the soil water potential. In theory, the program stops the automatic suction when average matric pressure head are lower than –510 cm. However, in the first 2 yr it was found that the drainage flux was too low to collect reliable volumes when the pressure head of the bulk soil was below –306 cm. I therefore raised the limit to –306 cm to stop automatic suction by closing the line valves. Suction units were sampled every day; the water volumes collected in the 16 pairs of 1-L bottles by mean of the vacuum lines were measured and percolation samples were frozen until laboratory analysis.

Characterization of Drainage Collected by Means of the AETL System
The following description of the AETL system evaluation refers only to the final year of the study when the AETL was fully operational. To verify that the AETL system maintained soil water potential equilibrium between the ceramic plates and the soil during the drainage season, a detailed analysis was performed based on a temporal resolution of 10 min during two 2-d periods. The first during a period of heavy drainage (May 28–29), when the lower soil layers were almost saturated and the second during a period with less drainage (May 20–21), which was characterized as an intermediate phase of soil water redistribution.

The variability of daily drainage was then analyzed which allowed me to determine the number of lysimeters needed for a required level of accuracy expressed in terms of relative standard error (RSE):

Formula 1[1]
where SE is standard error and Formula 1 the daily drainage mean.

The relationship between RSE and number of lysimeters was obtained with the equation:

Formula 2[2]
where n is the sample size, and a and b are the fitted parameters of the relationship between variance (s2) and Formula 2, calculated as follows:

Formula 3[3]
Equation [3] was fitted using the daily volumes collected from the lysimeters during the drainage season. The influence of percolation volume on the RSE was then evaluated by varying Formula 3 in Eq. [2] (Snedecor and Cochran, 1980).

Numerical Simulation of the AETL System
The system was also evaluated by simulating its performance with the HYDRUS-2D (Simunek et al., 1999) finite-element model. Although the system functioned spatially in three dimensions, assumed was the axial symmetry for isotropic soils (Fig. 2 ), which reduced the Richards equation to two dimensions (Inoue et al., 1998). For rigid porous media, this can be written as:

Formula 4[4]
where C is the soil water capacity (cm–1), h is the pressure head (cm), K the unsaturated hydraulic conductivity (cm s–1), r the radial coordinate (cm), and t is time (s). The equation was solved by HYDRUS-2D with the Galerkin finite element method based on the mass conservative iterative scheme proposed by Celia et al. (1990). Initial conditions for which Eq. [4] was solved can be defined as:

Formula 5[5]
where r = R (50 cm) denotes the radius of flow domain, the ordinates z = 0 and z = –90 cm are placed on the soil surface and at the bottom of the lysimeter (Fig. 2), hi is the initial pressure head, and t0 corresponds to the beginning of the simulated periods. The z-coordinate coincided with the vertical axis of symmetry. To describe the flow domain in the AETL system, lower boundary conditions were differentiated according to the pressure head above and adjacent to the lysimeters (Fig. 2):

Formula 6[6]

Formula 7[7]
where hls and hbs are the variable pressure head measured in the soil above the lysimeters and in the adjacent bulk soil, respectively, and tend corresponds to the end of the simulated periods. In solving Eq. [4], unsaturated hydraulic properties were defined by the models of van Genuchten (1980) and Mualem (1976).


Figure 2
View larger version (63K):
[in this window]
[in a new window]
  		Fig. 2. Sketch of the domain simulated by HYDRUS-2D.

 
HYDRUS-2D was used to simulate AETL functioning in the two 2-d periods of heavy and light drainage. The purpose of the simulation was to verify that the drainage volumes collected with the lysimeters estimated theoretical drainage volumes below the lower boundary. The model was applied to each of the three boxes containing the electronic tensiometers (pressure transducers). The variable lower boundary conditions (hbs and hls) were specified on a 10-min time scale. To estimate total drainage in the boxes not influenced by the presence of lysimeters, I imposed only the pressure head hbs as lower boundary condition. No-flux boundary conditions were applied to both vertical sides of the flow domain. The domain size was large enough to justify the assumption of a zero flux boundary condition along the lateral boundaries.

Hourly rainfall data and potential evapotranspiration rates calculated according Allen et al. (1998) were used as atmospheric boundary conditions. Root water uptake was calculated by applying the plant water stress response function proposed by Feddes et al. (1978). Weather data were measured by the automatic weather station located about 100 m from the GBs. Initial water content conditions were specified, using data measured by TDR. The water retention properties were determined at the beginning of the experiment from undisturbed soil cores (eight replicates), by desorption using suction tables (Romano et al., 2002) for low-range potentials and a pressure plate apparatus (Dane and Hopmans, 2002) for mid- and high-range potentials. Saturated hydraulic conductivity was measured with constant and falling-head methods (Klute and Dirksen, 1986). RETC software (van Genuchten et al., 1991) was used to fit Eq. [8] to the measured retention curve data. Soil hydraulic parameters are reported in Table 1.


View this table:
[in this window]
[in a new window]
  		Table 1. Soil hydraulic parameters used for the HYDRUS-2D simulations.

 
A trial and error calibration was performed with data collected from 14th to 16th May to make the model representative of the soil water dynamics observed in the two 2-d periods. During calibration, simulated saturated hydraulic conductivity was adjusted to fit daily simulated drainage to measured drainage. HYDRUS-2D was then validated with drainage volumes measured in the two 2-d periods.

HYDRUS-2D was also applied to evaluate potential errors arising from inappropriate suctions. The lower lysimeter boundary conditions were set at ±10, ±30, ±50, and –100% of the average bulk soil pressure heads measured during the heavy drainage phase. This period was selected because errors made during heavy drainage have the greatest influence on mass water balance. Positive percentages indicate undersuction: when this is applied, the pressure head in soil above the plates is higher than pressure head in the bulk soil; the contrary for negative percentages. A zero-tension lysimeter was also simulated imposing seepage face boundary conditions (+100%). AETL system error was calculated in terms of relative error between simulated volumes collected by the lysimeters and simulated volumes without the effect of the lysimeters.


   RESULTS AND DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 MATERIAL AND METHODS
 RESULTS AND DISCUSSION
CONCLUSIONS
 REFERENCES
 
Automatic control of the equilibrium tension lysimeter system operated from the beginning of March 2002 to the end of June 2002. The previous winter had been dry, with only 60 mm of rainfall in 3 mo. The soil water content was low and relatively uniform, ranging from 0.17 cm3 cm–3 in the top layer (0–20 cm) to 0.2 cm3 cm–3 in the bottom layer (60–80 cm).

To test the AETL system, four 25-mm irrigations were applied to simulate rainfall events at the end of February (Fig. 3 ). From April to the beginning of June 330 mm of rain fell. Most of the drainage occurred during this period and was collected by the AETL system (Fig. 3). Drainage ceased at the end of June due to evapotranspiration of the maize crop, which exceeded the summer rainfall. No irrigation was applied. No further drainage volumes were collected.


Figure 3
View larger version (28K):
[in this window]
[in a new window]
  		Fig. 3. (a) Daily ET and cumulative rainfall + irrigation and drainage; (b) variation of soil water matric potential head above the lysimeters (AL) and in the adjacent bulk soil (BS), with daily precipitation and irrigation. Bars indicate the simulated 2-d periods.

 
At the beginning of March, differences and variability were high in the monitored soil pressure heads: the average pressure head was –220 cm in soil above the lysimeters (AL) with a standard error (SE) of ±76.0 cm and –340 cm in the adjacent bulk soil (BS) with SE of ±90.1 cm (Fig. 3). These differences were probably caused by the rewetting performed in September 2001 to re-establish hydraulic conductivity with the soil above the ceramic plate. No drainage occurred in March, preventing the AETL system from equilibrating the pressure head with the bulk soil.

The wetting front produced by simulated rainfalls reached the lysimeters 40 d after application, as was promptly evidenced by a sudden increase in the pressure heads: in 2 d the AL values rose from –162 to 2 cm and the BS from –256 to –10 cm (Fig. 3). In the next 4 d, the AETL was able to equilibrate the pressure heads and maintain a dynamic equilibrium during the forthcoming drainage period. At the end of June, when the pressure heads and drainage fluxes were low, equilibrium control was less effective. When the pressure head dropped below –306 cm, the AETL stopped vacuum pump operations.

To evaluate the AETL system functioning in detail, the results of automatic control are discussed on a 10-min time scale (Fig. 4 and 5) . The figures relate to the two 2-d periods discussed earlier: the heavy drainage period (May 28–29), when the bottom layer was almost saturated, and the period with less drainage (May 20–21), which was in an intermediate phase of soil water redistribution.


Figure 4
View larger version (27K):
[in this window]
[in a new window]
  		Fig. 4. (a) Average 10-min variation of the soil water matric potential above the lysimeters (AL) and in the adjacent bulk soil (BS), with (b) their absolute differences ({Delta}h) observed over a 2-d period of high drainage flux; (c) cumulative drainage volumes simulated by HYDRUS-2D. The arrows represent the moments the electronic valves opened to control suction; SE = standard error.

 

Figure 5
View larger version (28K):
[in this window]
[in a new window]
  		Fig. 5. (a) Average 10-min variation of the soil water matric potential above the lysimeters (AL) and in the adjacent bulk soil (BS), with (b) their absolute differences ({Delta}h) observed over a 2-d period of low drainage flux; (c) cumulative drainage volumes simulated by HYDRUS-2D. The arrows represent the moments the electronic valves opened to control suction; SE = standard error.

 
The effect of the automatic suction control is clearly detectable in Fig. 4. During the 2 d of heavy drainage, the BS pressure head decreased slowly from 0 to –20 cm; in the first 34 h the drainage water accumulated above the lysimeters, increasing the AL pressure head as a consequence. The AETL system reacted promptly, activating the vacuum pump operations with a frequency proportional to the drainage rate. The AETL system followed the BS pressure head trend, maintaining average differences close to 0, with only a small oversuction on the second day. Variability among the pairs of tensiometers was also generally slight (Fig. 4).

The AETL system was less effective in controlling the suction during the period with a lower drainage rate (Fig. 5). Vacuum pump operations were activated seven times in 2 d and it took longer to re-equilibrate the pressure heads after the oversuction pulse was applied by the system. This is evident on May 21 when the above lysimeter (AL) pressure head never reached the higher BS levels. The reserve of vacuum in the tank allowed the system to collect the drainage without the need of an extra suction pulse, but the drainage flux was too low to equilibrate the system. To be more effective in the precise control I could have shortened the scanning period of the tensiometers. Indeed Kosugi and Katsuyama (2004) demonstrated that a short monitoring period (3 s) along with a strong vacuum (–450 cm) is important for the precise control of the matric pressure in the sampling profile. Nevertheless, the trend was correctly followed by the system and the differences were generally as slight as the variability among the pairs of tensiometers (Fig. 5).

Diurnal fluctuations of tensiometric readings due to surface temperature changes emerged throughout the period during which the AETL system was operating. Increases in soil pressure head were registered during the warmer hours of the day. Due to the high soil water content that allowed water to move easily from the tensiometer to the soil (Warrick et al., 1998), the daily soil water pressure head fluctuations in the experiment were modest, {approx}5 cm with an average daily temperature of 19 ± 4°C. Nevertheless, temperature fluctuations had a marginal influence on the pressure head differences, and consequently on AETL system performances, because the response of each tensiometer to surface temperature changes was almost identical. Lentz and Kincaid (2003) encountered the same problems in tension measurements. They suggested the use of advanced tension design (Hubbell and Sisson, 1998) to avoid temperature-induced fluctuations.

Drainage water was collected on 34 d. The mean cumulative volume collected from the 16 suction plates was 196 mm, with a CV of 25% and SE of ±6.4 mm. Daily drainage was computed as average of all the 544 daily measurements (16 plates x 34 d); the mean value was 5.3 mm with a CV of 45% and SE of ±0.6 mm. For individual lysimeters, the CV for daily drainage ranged from 85 to 150%. The variability between the lysimeters was likely caused by the heterogeneous soil hydraulic conditions. Heterogeneity in the infiltration processes can be caused by the existence of preferential flow pathways and soil disturbance during porous plate installation (Kosugi and Katsuyama, 2004; Mertens et al., 2005).

Error Analysis
To determine the number of lysimeters required for a preestablished level of accuracy, all 16 lysimeters were considered as replicates. Equation [3] was fitted using the 34 pairs of mean and variance data calculated on the days of drainage (Fig. 6 ). Four curves showing the relationship between sample size and RSE were constructed applying Eq. [3] to four given daily drainages: 0.1, 0.5, 1, and 10 mm (Fig. 6). The differences between curves are minimal, indicating that variability was not affected by volume rate. In operational terms this means that the AETL reacted correctly to the different drainage flux conditions. The single curves are characterized by an exponential decrease of RSE, which asymptotically approaches 7% and almost stabilizes at between 10 and 20 lysimeters. These results suggest that a minimum number of 10 are required in our pedoclimatic condition to obtain a reliable estimate of the drainage mean.


Figure 6
View larger version (14K):
[in this window]
[in a new window]
  		Fig. 6. (a) Relationship between sample mean (Formula 7) and variance (s2) of the daily drainage collected during the survey period; (b) relationship between sample size (number of lysimeters) and relative standard error (RSE) of measured drainage for given values of mean daily drainage.

 
Following a different statistical procedure, Zhu et al. (2002) obtained similar results comparing the leachate collection efficiency of zero-tension pan and passive capillary fiberglass wick lysimeters. They found that nine wick and ten pan lysimeters were necessary to obtain ±20% error margin with 10% error probability.

HYDRUS-2D Simulations
After calibration, HYDRUS-2D was able to properly represent the functioning of the AETL system in the 2-d periods. There was a good agreement between the daily drainage collected by the AETL system and that simulated by HYDRUS above the lysimeters (AL). The slope of the linear regression between the measured and simulated data (Fig. 7 ) is close to 1 and the intercept is not significantly different from 0. The coefficient of determination is 0.99 (p < 0.01). A very small tendency to overestimate the values was observed in the low drainage period. In my opinion, this behavior could be a function of both the performance of HYDRUS-2D, and of the measured data. It is probable that during conditions of very low drainage, the precision of the AETL system decreased. Values of calibrated saturated hydraulic conductivity were higher than the measured ones (Table 1). Calibration of saturated hydraulic conductivity probably allowed representation of lysimeter conditions that couldn't be quantified by the laboratory analysis on the undisturbed samples, that is, the effect of soil repacking around the ceramic plates.


Figure 7
View larger version (18K):
[in this window]
[in a new window]
  		Fig. 7. Comparison between simulated and measured daily drainage in the 2-d periods in the three growth boxes (GB).

 
During the period of intense drainage the simulated and observed total AL drainage were 13.2 mm (SE ± 0.62) and 13.1 mm (SE ± 0.74), respectively. Simulated total drainage without the effect of the lysimeters (BS) was 13.3 mm (SE ± 2.5). Based on these results, the AETL system operated efficiently with an overcollection of only 0.1 mm, corresponding to an error of only 1%. Simulated results indicated that overcollection occurred at the end of the second day, when there was oversuction up to 8 cm. In contrast, during the first 36 h, HYDRUS-2D simulated undercollection, with BS volumes higher than AL volumes (Fig. 4). Application of a single vacuum regime reduced the variability in cumulative AL volumes, as shown by the SE values, even if the matric pressure data variability was very small (Fig. 4).

During the low drainage period, which was also characterized as an intermediate phase of soil water redistribution (Fig. 5), control of the matric potential was more difficult but, due to the dryer soil condition and lower hydraulic conductivity, the differences simulated in the percolation volumes were small. Total simulated AL and BS drainage was 0.80 mm (SE ± 0.15) and 0.77 mm (SE ± 0.36) respectively, with an overcollection of only 0.03 mm, corresponding to 3.5%. The observed collected volume was 0.7 (SE ± 0.15). Also in this case, the variability of AL simulated volumes is less than that of the BS volumes (Fig. 5).

The effect of dryer soil conditions around the plate is also evident in the simulation of potential errors arising from the application of erroneous suctions (Fig. 8 ). Systematic errors in applied suction produced large errors in the percolation measurements. The effect is clearer in the case of undersuction: that is, with an lysimeter boundary set at +30%, simulated drained volume was 10.7 mm, with a relative error of –19.5%; with the same percentage of oversuction (–30%), simulated drainage was 14.6 mm, with a relative error of 7.5%. The under-collection is even more marked when seepage face boundary conditions are considered: only 3.4 mm of simulated drainage with a relative error of –74%. With an oversuction of 100%, the relative error is 30%. Probably the unsaturated conditions created around the lysimeter by oversuction decrease soil hydraulic conductivity and slow down water movement toward the ceramic plate, reducing the overcollection error. The fundamental role of soil hydraulic conductivity in controlling the extraction domain of ceramic cups has also been highlighted by Weihermüller et al. (2005). In the case of zero-tension boundary conditions, water collection by the lysimeter is simulated only in the first 10 h of the simulated period, when saturated conditions were observed. These findings are consistent with the low, usually <40%, collection efficiencies (water volume collected divided by percolating volume calculated from a water balance) of zero-tension lysimeters calculated by other authors (Haines et al., 1982; Radulovich and Sollins, 1987).


Figure 8
View larger version (26K):
[in this window]
[in a new window]
  		Fig. 8. Cumulative drainage simulated with HYDRUS-2D considering different variable LS (lysimeter) lower boundary conditions.

 

   CONCLUSIONS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIAL AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
REFERENCES
 
The results of this experiment are limited to only one AETL system and are specific for the soil at the research site and the rainfall conditions observed during the experimental period. The AETL system kept the suction of the lysimeters in equilibrium with the surrounding soil when the soil matric potential head ranged from –306 to 0 cm. Suction control was less effective during the lower drainage conditions (potential head less than –50 cm), when it took more time to re-equilibrate the pressure heads after the system applied an oversuction pulse. Error estimates of applying over- or under-suction were obtained simulating the AETL system with HYDRUS-2D. They are based on the model sensitivity to variation in lower boundary conditions. Simulated results showed that incorrect suction control in drier conditions had a limited effect on the collected volumes. Instead, when drainage flux is high, even the application of a small steady oversuction above the lysimeters to force the water entering through the porous plates, as is usually done in field experiments, could produce marked overestimation of water drainage (>30%). The application of variable vacuum regimes to address changing soil water potential could improve lysimeter system performance. The simulated results showed that the application of a single vacuum regime reduced the variability in the collected volumes even if the variability in matric pressure was very slight. However, application of different vacuum regimes might be detrimental to system costs if we consider that a large number of lysimeters had to be used to compensate for their small surface area. The AETL system is relatively economical and could easily be implemented in field conditions, especially where stony soils make large surface lysimeters difficult to install.

A further system refinement might be to insert a series of tensiometers along the profile to automatically verify the water potential gradient and recognize when a decrease in the soil pressure head of the bottom layer is due to downward water redistribution or to upward flux, in this last case stopping the automatic suction. This is especially important in some areas of NE Italy, where most rain falls during spring and summer when the water table is shallow.


   ACKNOWLEDGMENTS
 
I thank Dr. M. Borin for the suggestions given during the set up of AETL system and Dr. J. Simunek to modify HYDRUS-2D for considering the two different variable bottom boundary conditions. I acknowledge also Dr. C. Camarotto for his contribution in the program of the datalogger. Special thanks goes to Dr. G. Vellidis for reviewing the paper. Research carried out with financial support of CNR (National Project "Recycling the wastes of the agro-industrial system"). Project leader: Dr. Maurizio Borin.

Received for publication January 9, 2006.


   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 MATERIAL AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 




This article has been cited by other articles:


Home page
 Vadose Zone JHome page
J. Mertens, V. Tuts, J. Diels, J. Vanderborght, J. Feyen, and R. Merckx
Design and Testing of a Drop Counter for Use in Vadose Zone Water Samplers
Vadose Zone J., April 14, 2008; 7(2): 434 - 438.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF) Free
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Morari, F.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Morari, F.
GeoRef
Right arrow GeoRef Citation
Agricola
Right arrow Articles by Morari, F.
Related Collections
Right arrow Water Flow Models
Right arrow Recharge
Right arrow Soil Methods/Instrumentation

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
The SCI Journals Agronomy Journal Crop Science
Vadose Zone Journal Journal of Plant Registrations
Journal of Natural Resources
and Life Sciences Education		 Journal of
Environmental Quality