SSSAJ Journal of Natural Resources and Life Sciences Education
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

Published online 25 January 2008
Published in Soil Sci Soc Am J 72:320-330 (2008)
DOI: 10.2136/sssaj2007.0042
© 2008 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 An erratum has been published
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 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 Google Scholar
Google Scholar
Right arrow Articles by Eguchi, S.
Right arrow Articles by Hasegawa, S.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Eguchi, S.
Right arrow Articles by Hasegawa, S.
GeoRef
Right arrow GeoRef Citation
Agricola
Right arrow Articles by Eguchi, S.
Right arrow Articles by Hasegawa, S.
Related Collections
Right arrow Hydraulic Conductivity
Right arrow Preferential Flow
Right arrow Soil Hydrology

SOIL PHYSICS

Determination and Characterization of Preferential Water Flow in Unsaturated Subsoil of Andisol

Sadao Eguchia,* and Shuichi Hasegawab

a National Institute for Agro-Environ. Sciences, Tsukuba, Ibaraki 305-8604, Japan
b Graduate School of Agriculture, Hokkaido Univ., Sapporo, Hokkaido 060-8589, Japan

* Corresponding author (sadao{at}affrc.go.jp).


   ABSTRACT
TOP
 NOTES
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Experimental approaches for directly determining preferential water flow in field soils usually require artificial interruption of the soil matrix's water flow. This interruption may change the quantitative as well as the qualitative nature of the preferential water flow, particularly in soils such as Andisols where matrix water flow is the dominant process of drainage. To overcome this problem, we applied both the one-dimensional form of Darcy's equation and the soil water balance method, during and shortly after each rain event, for characterizing as well as determining the preferential water flow across a depth of 1 m in an Andisol. The volumetric water contents and pressure potentials at different soil depths were monitored by time domain reflectometry and tensiometry during a 7-yr period. Preferential water flow was detected only two to seven times per year; nevertheless, it accounted for 16 to 27% of the annual total drainage. These flows occurred under macroscopically homogeneous soil water conditions, with pressure potentials mostly below the air-entry pressure in the subsoil. The in situ unsaturated hydraulic conductivity at the depth of 1 m, determined to apparently explain the total drainage during a preferential water flow event, varied over more than one order of magnitude, with little change in water content, and the maximum value reached the saturated hydraulic conductivity. This indicates that preferential water flow occurred mainly through the largest macropores at this depth due to pore-scale physical nonequilibrium. Macropore-mediated preferential water flow and solute transport should be considered in unsaturated subsoil of Andisols.

Abbreviations: TDR, time domain reflectometry


   INTRODUCTION
TOP
 NOTES
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Preferential water flow is the process in which water moves along preferred pathways through the soil profile while bypassing a large part of the soil matrix. The volume of water passing across a horizontal cross-sectional area per unit of time is therefore significantly larger than that predicted by the ideal one-dimensional vertical Darcy's equation using the area-averaged hydraulic conductivity and hydraulic gradient. At times, this phenomenon results in the rapid penetration of suspended as well as dissolved contaminants into underlying groundwater bodies (White, 1985; Bouma, 1991).

Preferential movement of water occurs not only through the so-called macropores of shrinkage cracks (Bouma and Dekker, 1978; Bronswijk, 1988), interaggregate pore space (Rao et al., 1980; Cote et al., 1999), root channels (Ishiguro, 1991; Li and Ghodrati, 1994), earthworm burrows (Zachmann et al., 1987; Edwards et al., 1989), and fractures in rocks (Dahan et al., 1998; Pruess, 1999), but also through the soil matrices with macroscopic hydraulic heterogeneity (Nielsen et al., 1973; Kung, 1990; Roth, 1995) or water repellency (Ritsema et al., 1993; Dekker and Ritsema, 1996). These two types of preferential water flow—macropore mediated and matrix mediated—should be distinguished separately. They may, however, simultaneously occur in a field soil, and may create different passages to groundwater environments. The physical, chemical, and biological interactions of water, solutes, and colloids with the soil matrix in preferential water flow pathways are likely to be much less in the former type than in the latter; consequently, the former may contribute to a great extent to the rapid invasion of dissolved and suspended contaminants to underlying groundwater bodies. Therefore, for elucidating nonideal flows and transport processes in the vadose zone, it is essential to distinguish between these two types of preferential water flow or to determine whether or not macropore-mediated preferential water flow occurs.

Several experimental approaches have revealed the hydrologic importance and water quality characteristics of the macropore-mediated preferential water flow. Approaches for collecting or determining preferential water flow through macropores in field soils include the use of tension-free or suction drainage lysimeters (Shaffer et al., 1979; Barbee and Brown, 1986; Jardine et al., 1990), isolated large undisturbed soil cores (Bouma et al., 1981; Germann and Beven, 1981; Greco, 2002), and drainage water samplers connected to each macropore or fracture on the vertical or horizontal soil walls (Edwards et al., 1989; Tsuboyama et al., 1994; Dahan et al., 1998); however, these methods artificially interrupt or change the in situ matrix water flow through the soil continuum due to the presence of large instrument walls or air-faced soil walls. This interruption may severely, and constantly, cause changes in the hydraulic head of the soil matrix surrounding macropores near sampling devices, which can significantly affect the quantitative as well as qualitative nature of the preferential water flow (Phillips et al., 1995). In particular, in soils where the matrix water flow described by Darcy's equation is the dominant process of drainage, any interruption or change in the in situ matrix water flow may severely affect the nonuniform water flow phenomena. In other approaches using tracers, such as conservative solutes (Hendriks et al., 1999) or stable or radio isotopes of water (Gazis and Feng, 2004), it is possible to directly detect and characterize the in situ preferential water flow; however, the amount of the preferential water flow passing across a certain depth of field soil in each event cannot be quantified unless additional hydrologic studies are performed. At present, no field methods are available for directly determining the preferential water flow without interrupting the in situ matrix water flow.

Examination of the applicability of the one-dimensional form of Darcy's equation using the average unsaturated hydraulic conductivity and hydraulic gradient with reference to the soil water balance can be useful to infer the presence of preferential water flow (Hasegawa et al., 1994). In this approach, preferential water flow is defined as the residue of the soil water balance. Hasegawa and Sakayori (2000) first applied both the one-dimensional form of Darcy's equation and the soil water balance method for directly determining the quantity of preferential water flow in the subsoil of an Andisol during and shortly after a heavy rain event. A major advantage of this method is that it makes possible the experimental determination of preferential water flow across a certain depth of the field soil without significantly interrupting the in situ matrix water flow by using time domain reflectometry (TDR) probes and tensiometers. They successfully determined both the preferential water flow and matrix water flow across a depth of 1 m. Furthermore, they revealed that the matrix water flow is the dominant process of drainage in the subsoil of this Andisol, and that preferential water flows occurred only several times per year (Hasegawa and Sakayori, 2000; Hasegawa and Eguchi, 2002; Eguchi, 2006). However, the characteristics of the preferential water flows have not been examined, that is, through which pathways—soil matrix or macropores—and under which soil water conditions these flows occur.

The objective of this study was to characterize and determine the preferential water flow to a depth of 1 m in an Andisol under field conditions by applying both the one-dimensional form of Darcy's equation and the soil water balance method. We discuss whether or not the macropore-mediated preferential water flow occurred in the unsaturated subsoil of this soil. We used long-term field data of volumetric water contents and pressure potentials at different depths of the soil during the 7-yr period of 1997 to 2003. This long-term study was necessary to accumulate a sufficient number of preferential water flow events to discuss the general characteristics of this flow.


   MATERIALS AND METHODS
TOP
 NOTES
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Study Site
The study was conducted in an experimental field at the National Institute for Agro-Environmental Sciences (36°01' N, 140°07' E) located at an altitude of 24 m in Tsukuba-Inashiki Upland, Tsukuba, Ibaraki, Japan. The annual average precipitation and temperature during 1971 to 2000 were 1236 mm and 13.5°C, respectively (Japan Meteorological Agency, 2006). In winter, from December to February, the daily minimum temperature was frequently below 0°C, and the soil surface was frozen to a depth of <5 cm.

The soil is an Andisol (Hydric Hapludand) derived from volcanic ash. Table 1 shows some selected physical and chemical properties of the soil. The soil was divided into three layers: surface soil (0–20 cm), plow sole (20–30 cm), and subsoil (30–120 cm). The soil texture is light clay to heavy clay, and >80% of the clay fraction is composed of allophane and other poorly ordered materials (Watanabe, 1984). In contrast to other light- or heavy-clay textured soils, this soil was characterized by low dry bulk densities of 0.55 to 0.81 Mg m–3, high porosities of 0.69 to 0.80 m3 m–3, and high hydraulic conductivities of 3 · 10–6 to 1 · 10–4 m s–1 at and near saturation (Table 1). No cracks were observed in the soil profile. A large number of cylindrical macropores with diameters of 1 to 2 mm formed by decayed plant roots are visible in the subsoil. Soft x-ray image analyses (Narioka et al., 2000; Iwata and Narioka, 2002) have revealed that the vertical cylindrical macropores in the subsoil have relatively low tortuosity and are connected with each other by horizontal cylindrical macropores. The network of these macropores appears to contribute to the relatively high air-entry pressure of the subsoil, which ranges from –8 to 0 cm (Table 1). The water table usually fluctuates around a depth of 2 m. The unconfined aquifer is underlain by a heavy-clay textured, low-permeability layer at a depth of approximately 2.6 m.


View this table:
[in this window]
[in a new window]

  		Table 1. Selected physical and chemical properties of the soil.

 
Sweet corn (Zea mays L.) and Chinese cabbage (Brassica pekinensis Rupr.) were grown every year in summer and autumn, respectively. In winter, either of two catch crops—winter wheat (Triticum aestivum L.) or Italian ryegrass (Lolium multiflorum Lam.)—was grown. The catch crop was cut in spring before it matured, and it was incorporated into the soil as green manure.

Time Domain Reflectometry and Tensiometry
Time domain reflectometry and tensiometry were used to monitor volumetric water contents and pressure potentials at different depths of the soil (Hasegawa and Sakayori, 2000). Figure 1 shows the layout of the TDR probes and tensiometers that were installed in the field.


Figure 1
View larger version (44K):
[in this window]
[in a new window]

  		Fig. 1. Schematic diagram of the field installation of time domain reflectometry (TDR) probes and tensiometers (left) and vertical cross-section of the tensiometer (right). The triplicate vertical TDR probes and the tensiometers are numbered 1, 2, and 3, and the duplicate horizontal TDR probes are labeled a and b to distinguish between their spatial positions. Units are in centimeters.

 
The volumetric water contents from the soil surface to depths of 30 ({theta}0–30) and 100 cm ({theta}0–100) and at a depth of 100 cm ({theta}100) were determined by TDR (Fig. 1, left). Two parallel stainless rods, 5 mm in diameter, 5 cm apart, and 30 and 100 cm long, were used as TDR probes. A pair of TDR probes with lengths of 30 and 100 cm was inserted vertically into the soil to depths of 30 and 100 cm, approximately 15 cm apart. Three pairs of such probes were installed in a row at intervals of 1 m. Two 30-cm-long TDR probes were horizontally inserted at a depth of 100 cm from the lateral wall of two soil pits located 2 m apart. The pits were backfilled following the installation. The eight TDR probes—six vertical and two horizontal—were connected to a metallic TDR cable tester (Model 1502B, Tektronix, Beaverton, OR), which was connected to a data logger (Model 21X, Campbell Scientific, Logan, UT) with a programmable read-only-memory (PROM). The travel time of a pulsed electromagnetic signal was determined by using the digitized waveform temporarily recorded in PROM and used for calculating the apparent dielectric constant, according to the method of Baker and Allmaras (1990). As the universal Topp's equation (Topp et al., 1980) underestimates the water content of volcanic ash soils (Hatano et al., 1995), a specific calibration curve was made for the relation between the soil water content directly measured by the oven-drying method using the soil core samples taken under dry to wet field soil conditions and that calculated by the Topp's equation (Hasegawa, 1997). The storage water to a depth of 1 m (S0–100) was determined from the {theta}0–100 values. When the rainfall intensity was <10 mm h–1 and the initial S0–100 value was below field capacity, the amount of the cumulative rainfall up to 40 mm coincided with the change in storage water to a depth of 1 m, {Delta}S0–100 (Hasegawa, 1997), indicating that the {theta}0–100 value can be determined by the 100-cm-long TDR probes.

The pressure potential heads at depths of 90 ({psi}90) and 110 cm ({psi}110) were measured by tensiometry (Fig. 1, left). A ceramic porous cup with a diameter and length of 18 and 60 mm, respectively, was glued to a polyvinyl chloride tube and was used as a tensiometer. A pair of tensiometers positioned 20 cm apart was inserted vertically into the soil to depths of 90 and 110 cm. Three such pairs of tensiometers were installed approximately 15 cm apart from the three vertical TDR probes installed to the depth of 1 m. A pressure transducer (Model VPRN, Valcom Co., Osaka, Japan) was used to measure the air pressure inside the tensiometer tube. The water level inside of the tensiometer was set to 15 cm below the soil surface (Fig. 1, right) to prevent water from freezing in winter as well as to reduce the effect of air temperature fluctuations. As the water level gradually decreased with time (a few millimeters per day), the water in the tube was replenished every 2 to 3 wk through a hole in the metallic septum (Fig. 1, right), which was closed with a screw lid during the measurements. The water level was determined from the volume of replenished water. The change in the water level during the measurement periods was estimated by linear interpolation and was taken into account for determining the pressure potential values.

The surface of the soil around which the TDR probes and tensiometers were installed was maintained flat without constructing any furrows to reduce the effect of microrelief-induced nonuniform surface infiltration (Hasegawa, 1997). The TDR probes and tensiometers were installed at the correct depths with a precision of ±1 mm with reference to the average ground surface level. The precipitation was measured with a tipping-bucket rain gauge (Model TE525MM, Campbell Scientific). All the measurements were conducted every 30 min and were recorded in the data logger.

Combining Water Content–Pressure Potential and Hydraulic Conductivity–Pressure Potential Relations of Undisturbed Soil Cores with Different Dry Bulk Densities
The unsaturated hydraulic conductivity (K) was measured by the steady-state method (Richards and Moore, 1952; Hasegawa and Sakayori, 2000) using triplicate undisturbed soil cores, 4 cm high and 10 cm in diameter, that were taken at a depth of 1 m. The dry bulk densities of the triplicate samples were 0.549, 0.564, and 0.583 Mg m–3. The value of K was measured during two and a half cycles of drying and wetting processes in the range of –200 cm to near saturation. The K value at a given pressure potential head ({psi}) tended to increase with the dry bulk density ({rho}d) (Fig. 2a ).


Figure 2
View larger version (12K):
[in this window]
[in a new window]

  		Fig. 2. The (a) unsaturated hydraulic conductivities and (b) volumetric water contents of the triplicate undisturbed soil cores taken at a depth of 1 m with dry bulk densities of 0.549 (low {rho}d), 0.564 (middle {rho}d), and 0.583 Mg m–3 (high {rho}d) as a function of pressure potential heads during two and a half cycles of drying and wetting processes. Figure (b) also shows all the in situ scanning curves observed in the field for the relation between the volumetric water content at a depth of 1 m ({theta}100) and the average of the pressure potential heads at depths of 90 ({psi}90) and 110 cm ({psi}110) during 1997 to 2003.

 
The soil water characteristic curve was determined by the suction plate method using the same triplicate undisturbed soil cores that were used for the K measurement. The relation between the volumetric water content ({theta}) and {psi} was obtained during two and a half cycles of the drying and wetting processes in the range of –200 cm to near saturation. The {theta} value at a given {psi} tended to increase with {rho}d (Fig. 2b).

The K–{psi} (Fig. 2a) and {theta}–{psi} relations (Fig. 2b) were comparable with the results obtained from a previous study in this field (Hasegawa et al., 1994). In addition, the {theta}–{psi} relation measured in the laboratory under equilibrium conditions was mostly consistent with all the scanning curves between {theta}100 and the average of {psi}90 and {psi}110 in the field during the 7-yr period under nonequilibrium conditions (Fig. 2b). These facts imply that the K–{psi} and {theta}–{psi} relations obtained in the laboratory measurements by using the undisturbed soil cores can be regarded as being representative of the field and can be applied to describe the field subsoil hydraulic behavior.

The relation between K and {theta} (Fig. 3 ) was obtained by combining the K–{psi} (Fig. 2a) and {theta}–{psi} relations (Fig. 2b). As the {theta}–{psi} relation near saturation in the region {psi} > –5 cm, which corresponded to the region {theta} > 0.76 m3 m–3 (Fig. 3), was not measured, the K values near saturation in the region {theta} > 0.76 m3 m–3 were not plotted in Fig. 3. The saturated hydraulic conductivities that were determined by several researchers (Sakayori et al., 1998; Narioka et al., 2000; Iwata and Narioka, 2002) using undisturbed soil cores 5 cm in diameter and 5.1 cm high taken from around a depth of 1 m in this field were also plotted against the average porosities (Fig. 3).


Figure 3
View larger version (19K):
[in this window]
[in a new window]

  		Fig. 3. Unsaturated hydraulic conductivities (K) of the triplicate undisturbed soil cores taken at a depth of 1 m with dry bulk densities of 0.549 (low {rho}d), 0.564 (middle {rho}d), and 0.583 Mg m–3 (high {rho}d) as a function of volumetric water content during two and a half cycles of drying and wetting processes. The average saturated hydraulic conductivities (Sakayori et al., 1998; Iwata and Narioka, 2002) and the range of the maximum and minimum saturated hydraulic conductivities (Narioka et al., 2000) of the undisturbed soil cores taken around a depth of 1 m in this field are also plotted against the average soil porosity. The solid line is the fitted curve representing the average unsaturated hydraulic conductivity, Kmean (Eq. [1]). The range from 1/3Kmean to 3Kmean corresponds to the varying range of almost all the measured unsaturated hydraulic conductivities.

 
No hysteresis was observed in the K–{theta} relation (Fig. 3) during the drying and wetting processes, whereas it was obvious in the K–{psi} (Fig. 2a) and {theta}–{psi} relations (Fig. 2b). Moreover, the K–{theta} relation (Fig. 3) appears to be independent of {rho}d, while both the K–{psi} (Fig. 2a) and {theta}–{psi} relations (Fig. 2b) are dependent on {rho}d, indicating that the K value at the depth of 1 m can be expressed as a unique function of {theta}100 as measured by TDR. Therefore, we used the K–{theta} relation to determine the value of K in the field.

An empirical equation of the linear superposition of a logistic curve and a logarithmic curve was used to describe the K–{theta} relation (Fig. 3).

Formula 1[1]
where Kmean is the "fitted" average hydraulic conductivity of the undisturbed soil cores taken at the depth of 1 m, Kmax is the logarithmic average (3.16 · 10–5 m s–1) of the saturated hydraulic conductivity values shown in Fig. 3. Furthermore, Kmin, m, n, a, and b are the fitting parameters determined by a numerical iteration scheme, such that Kmin = 10–20, m = 61, n = 1.5, a = 7 x 10–9, and b = 5. Almost all the measured K values are distributed within the range from 1/3Kmean to 3Kmean (Fig. 3).

Matrix Water Flow and Preferential Water Flow
Matrix water flow was defined as the water flowing through the soil matrix that can be described by the one-dimensional form of Darcy's equation using the measured average hydraulic gradient and average unsaturated hydraulic conductivity, Kmean (Eq. [1]), of the undisturbed soil cores. The matrix water flow across a depth of 1 m, qm, defined to be positive vertically downward, was calculated by

Formula 2[2]
where Kmean({theta}100) is the average unsaturated hydraulic conductivity at a depth of 1 m (Eq. [1]) as a function of the volumetric water content, {theta}100, measured by TDR; {psi}90 and {psi}110 are the pressure potential heads measured by tensiometers at depths of 90 and 110 cm, respectively; and L is the distance between the depths of 90 and 110 cm. We assumed that the crop root zone that was effective for water uptake was limited up to the depth of 90 cm.

The rapid water drainage that cannot be described by the one-dimensional form of Darcy's equation (Eq. [2]), or that cannot be sensed by the equipment (Fig. 1) to monitor the macroscopic soil water conditions every 30 min, was regarded as the preferential water flow. The cumulative preferential water flow across a depth of 1 m, Dp, can include both the macropore- and the matrix-mediated preferential water flows at this depth, and this value reflects any interaction between preferential water flow and matrix water flow that can take place within the overlying soil layers. The Dp value was determined during and shortly after each rain event by the water balance equation (Hasegawa and Sakayori, 2000), assuming that the evaporation from the soil surface, the root water uptake, and the net surface runoff were negligible:

Formula 3[3]
where P is the cumulative rainfall; Dm is the cumulative matrix water flow across the depth of 1 m; and {Delta}S0–100 is the change in the storage water in the 1-m soil layer. Whether a significant amount of the preferential water flow occurs was determined by using the standard error of the preferential water flow, SE[Dp]. The law of error propagation was applied to calculate the values of SE[Dp], SE[Dm], and SE[{Delta}S0–100]. In this calculation, the measurement accuracies of TDR and tensiometry, which were set to ±0.01 m3 m–3 and ±4 mm, respectively, as well as the variabilities within the duplicate or triplicate measurement values were taken into account. Unless the Dp value exceeded the SE[Dp] value, it was regarded as zero. Accordingly, the detection limit of Dp in this method corresponded to the value of SE[Dp], which varied mostly in the range of 8 to 9 mm during each event. The end time of the preferential water flow, tnDp, was defined as the time when the increase in Dp during the subsequent 3 h was lower than 1 mm, or when it was minimum after the end time of rainfall, tnP. After deciding that the occurrence of preferential water flow was significant in an event, the start time of the preferential water flow, tiDp, was defined as the time after which Dp > 1 mm.

In Situ Unsaturated Hydraulic Conductivity
The in situ unsaturated hydraulic conductivity, Kin-situ, during a preferential water flow event was used to infer the presence of macropore-mediated preferential water flow across the depth of 1 m. It was determined based on the one-dimensional form of Darcy's equation to apparently explain the total drainage to below the depth of 1 m during a preferential water flow event:

Formula 4[4]
where P{Delta}t is the cumulative precipitation, {Delta}S0–100,{Delta}t is the change in the storage water to the depth of 1 m, and J90–110,{Delta}t is the time-average hydraulic gradient between depths of 90 and 110 cm during a small time increment of {Delta}t. The {Delta}t value was set constant to 3 h. If the Kin-situ value under unsaturated soil water conditions reaches the saturated hydraulic conductivity values determined by using the undisturbed soil cores, the largest soil macropores at the depth of 1 m should be the main pathways of the preferential water flow to explain such a high Kin-situ value.

Ratio of Preferential Water Flow to Total Water Flow in Each Event
The ratio of the cumulative preferential water flow, Dp,event, to the cumulative total water flow, Devent, across the depth of 1 m during the period from the start time of rainfall, tiP, to the end time of the preferential water flow, tnDp, in each event was used to decide whether the amount of the preferential water flow is significant even though the spatial variability in the K value was taken into account. We assumed that the upper limit of the varying range of K is <3Kmean in Fig. 3 because all the measured K values are distributed below 3Kmean. If the Kin-situ value defined by Eq. [4] is <3Kmean throughout the concerned event, the value of Devent should be less than that of 3Dm,event. Therefore, either Devent/Dm,event >– 3 or Dp,event/Devent >– 0.67 can be used as a criterion for deciding whether the Dp,event value was higher than that explained by the one-dimensional form of Darcy's equation even though the highest K value, that is, 3Kmean, was used.


   RESULTS AND DISCUSSION
TOP
 NOTES
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
CONCLUSIONS
 REFERENCES
 
Preferential Water Flow Caused by Heavy Rain Event
Figures 4a and 4b represent the changes in the rainfall intensity and soil water balance for the 1-m layer during and shortly after the heavy rainfall event observed on 10 Oct. 2001. The rainfall amounted to 119.8 mm during a period of 20.5 h, with an average intensity of 5.83 mm h–1. Although local surface ponding was temporarily observed all over the field when the hourly rainfall exceeded 10 mm, larger scale horizontal movement of the surface water was not observed. The surface water disappeared soon after the hourly rainfall decreased below 10 mm. The cumulative matrix water flow until the end time of preferential water flow amounted to only 24.1 ± 2.9 (mean ± standard error) mm; on the other hand, the preferential water flow was 48.7 ± 9.2 mm, which is twice as large as the matrix water flow (Fig. 4b), indicating that the preferential water flow contributed to a great extent to the rapid drainage of water deep into the soil, which can consequently prevent the generation of surface runoff and soil erosion in this field. The amount of preferential water flow continued to increase even after the end time of rain, tnP. As a result, 46% of the preferential water flow in this event was attributed to the increase in the preferential water flow during the 19-h period after the rain ceased. In contrast, the amount of matrix water flow increased little after the end time of rain.


Figure 4
View larger version (19K):
[in this window]
[in a new window]

  		Fig. 4. Changes in the (a) 30-min rainfall, (b) soil water balance in the 1-m layer, (c) pressure potential heads at depths of 90 ({psi}90) and 110 cm ({psi}110), (d) hydraulic gradient between depths of 90 and 110 cm (J90–110), (e) volumetric water contents from the surface to depths of 30 ({theta}0–30), and 100 cm({theta}0–100), and (f) volumetric water content at a depth of 100 cm ({theta}100) during and shortly after the heavy rain event that started at 0330 h on 10 Oct. 2001 (P, cumulative precipitation; Dm, cumulative matrix water flow across a depth of 1 m; {Delta}S0–100, change in the storage water in the 1-m layer; Dp, cumulative preferential water flow across the depth of 1 m; tiP, start time of the rain; tiDp, start time of the preferential water flow; tDp>SE, time after which Dp > SE[Dp]; tnP, end time of the rain; tnDp, end time of the preferential water flow). The vertical bars in (b) represent the standard errors of Dp and Dm. Subscripts #1, #2, #3, #a, and #b in (c) to (f) represent the sensor numbers of the TDR probes and tensiometers, as shown in Fig. 1.

 
Macroscopically Homogeneous Soil Water Conditions during Preferential Water Flow Event
The macroscopic soil water conditions measured by TDR and tensiometry during the preferential water flow event were relatively homogeneous in the horizontal directions (Fig. 4c–4f). The differences in the pressure potential heads measured by the triplicate tensiometers installed at depths of 90 and 110 cm were mostly 1 to 5 cm (Fig. 4c), and those at the depth of 90 cm were usually less than those at the depth of 110 cm. The differences in the response time of the triplicate tensiometers for the arrival of the wetting front at the depths of 90 and 110 cm were less than 1 and 2 h, respectively (Fig. 4c). The lowest peaks of the triplicate hydraulic gradients between the depths of 90 and 110 cm (Fig. 4d) ranged from –3.0 to –2.6 m m–1 and were observed within 1 h. The maximum differences in the triplicate or duplicate measurement values of the volumetric water contents of {theta}0–30, {theta}0–100, and {theta}100 (Fig. 4e and 4f) were 0.02 to 0.03 m3 m–3, which was slightly higher than the TDR measurement accuracy of 0.01 m3 m–3. The volumetric water content at the depth of 100 cm (Fig. 4f) was more homogeneous under wet soil conditions than under dry soil conditions (Fig. 4f), although the preferential water flow occurred under wet soil conditions.

The fact that a significant amount of preferential water flow occurred under macroscopically homogeneous soil water conditions suggests that the preferential water flow was mainly a pore-scale phenomenon, which cannot be detected by the centimeter- to meter-scale TDR probes and tensiometers, and that the soil water balance method is very sensitive to the occurrence of preferential water flow.

Macropore-Mediated Preferential Water Flow under Unsaturated Conditions
The in situ unsaturated hydraulic conductivity, Kin-situ, determined by Eq. [4] during the preferential water flow event (Fig. 4) was up to two orders of magnitude higher than the K values measured by the steady-state method using undisturbed soil cores (Fig. 5 ). The maximum value of Kin-situ reached the saturated hydraulic conductivity of 10–5 to 10–4 m s–1 (Fig. 5), particularly after the end of rainfall when the hydraulic gradient across the depth of 1 m remained close to zero (Fig. 4d); this indicates that macropore-mediated preferential water flow should have occurred through the largest macropores at this depth.


Figure 5
View larger version (18K):
[in this window]
[in a new window]

  		Fig. 5. In situ unsaturated hydraulic conductivity Kin-situ as a function of the volumetric water content at a depth of 1 m, determined by Eq. [4], during the preferential water flow event shown in Fig. 4; Kmean is the average unsaturated hydraulic conductivity (Eq. [1]) of the undisturbed soil cores (Fig. 3). Specific times of the start time of the preferential water flow (tiDp), the time after which Dp > SE[Dp] (tDp>SE), the end time of the rain (tnP), and the end time of the preferential water flow (tnDp) are shown in Fig. 4. The vertical bars represent the standard error of Kin-situ.

 
Moreover, the Kin-situ values largely fluctuated over more than one order of magnitude after the end time of rainfall, although there was little change in the water content (Fig. 5), indicating that the water flow cannot be explained by the one-dimensional Darcian water flow even if any hypothetical unsaturated hydraulic conductivity is assumed as a unique function of the water content.

The macropore-mediated preferential water flow across the depth of 1 m occurred under unsaturated soil water conditions even below the soil's air-entry pressure at this depth. The pressure potential heads at the depths of 90 and 110 cm (Fig. 4c) during the preferential water flow event were mostly below the air-entry pressure of –8 to 0 cm (Table 1), although those measured by two of the triplicate tensiometers at the depth of 110 cm were temporarily comparable with the air-entry pressure. This indicates that the macroscopic pressure potential in the soil matrix was not sufficiently high to produce macropore water flow through the largest macropores at this depth if a physical equilibrium condition is assumed. Therefore, the macropore-mediated preferential water flow should have occurred under pore-scale physical nonequilibrium conditions. Some of the largest cylindrical macropores with diameters of approximately 2 mm and low tortuosity, which have been directly observed by soft x-ray image analyses (Narioka et al., 2000; Iwata and Narioka, 2002), may have contributed to this phenomenon by functioning as preferential water flow paths.

These results provide the evidence, although it is indirect, first demonstrating the presence of macropore-mediated preferential water flow in an unsaturated Andisol subsoil under field conditions. Macropore-mediated preferential water flow and solute transport should be considered in the subsoil even with pressure potentials below the air-entry pressure, even in soils such as Andisols where matrix water flow is the dominant process of drainage.

Soil Water Balance and Conditions in Each Event
The preferential water flow across the depth of 1 m was detected only two to seven times per year—in total, 26 times during the 7-yr period of 1997 to 2003 (Table 2 ). In most cases, 40 to 80% of the total drainage during each event, Devent, was attributed to preferential water flow, Dp,event, indicating that preferential water flow generally played a significant role in rapid drainage of water during and shortly after each rain event. In the nine events 2, 3, 6, 8, 10, 15, 16, 20, and 22 (Table 2), the Dp,event/Devent values were ≥0.67, indicating that the amounts of preferential water flow in these events were higher than those explained by the one-dimensional form of Darcy's equation even though the highest K value, that is, 3Kmean (Fig. 3), was used. In the remaining 15 events, the Dp,event/Devent values were <0.67, indicating that the Dp,event values were as small as can be apparently explained by only the macroscopic heterogeneity in the K value of the soil matrix.


View this table:
[in this window]
[in a new window]

  		Table 2. Summary of the 26 preferential water flow events observed during 1997–2003 in relation to the soil water balance in the 1-m layer, the maximum pressure potential heads at depths of 90 and 110 cm, and the maximum water content at a depth of 1 m.

 
Mostly, the preferential water flow occurred under unsaturated soil water conditions with pressure potentials below the air-entry pressure of –8 to 0 cm (Table 2), and it was not detected when the maximum pressure potential heads were below –40 cm. The maximum pressure potential heads at depths of 90 ({psi}90max) and 110 cm ({psi}110max) during each preferential water flow event were greater than or equal to –40 and –38 cm, respectively, and the maximum water content at the depth of 1 m ({theta}100max) was ≥0.685 m3 m–3. In most cases, when these conditions were satisfied on a daily average basis, preferential water flow occurred (Fig. 6d, 6e, and 6g ), and this flow was usually associated with higher peaks of the daily matrix water flow (Fig. 6f).


Figure 6
View larger version (19K):
[in this window]
[in a new window]

  		Fig. 6. Changes in the (a) daily precipitation (Pday), (b) daily average storage water to a depth of 100 cm (S0–100day), (c) daily average storage water to a depth of 30 cm (S0–30day), (d) daily average pressure potential heads at depths of 90 ({psi}90day) and 110 cm ({psi}110day), (e) daily average volumetric water content at a depth of 100 cm ({theta}100day), (f) daily matrix water flow across a depth of 1 m (Dm,day), and (g) daily preferential water flow across a depth of 1 m (Dp,day) during 1997 to 2003. The ranges of the maximum pressure potential heads at depths of 90 ({psi}90max) and 110 cm ({psi}110max) and that of the maximum water content at a depth of 100 cm ({theta}100max) observed during the 26 preferential water flow events (Table 2) are also shown in (d) and (e).

 
The peak values of the macroscopic soil water conditions around the depth of 1 m during each preferential water flow event appear to have been usually well captured by the 30-min-interval measurements. The standard deviations for {psi}90max, {psi}110max, and {theta}100max show relatively small values of 0.1 to 4.0 cm, 0.7 to 5.5 cm, and 0.001 to 0.017 m3 m–3, respectively (Table 2). This indicates that the duplicate or triplicate sensors usually captured the same peak values of each event. Furthermore, the {theta}100 values plotted against the ({psi}90 + {psi}110)/2 values distribute within a relatively narrow range (Fig. 2b) regardless of whether or not preferential water flow occurred. The {psi}90, {psi}110, and {theta}100 values appear to have not reached saturation except when the water table rose shallower than the depth of 1 m, which occurred twice—in July 1999 and July 2000 (Fig. 6d and 6e).

Effect of Preferential Water Flow on Annual Soil Water Balance
The annual preferential water flow, Dp,year, accounted for 16 to 27% of the annual total drainage (Table 3 ). The Dp,year value tended to increase with the annual precipitation; however, the relationship between the annual drainage ratio of the preferential water flow to the total water flow, Dp,year/Dyear, and other hydrologic components such as the annual precipitation, annual total drainage, and annual evapotranspiration (ETyear) were insignificant (Table 3). This indicates that whether or not annual preferential water flow contributed greatly to the annual deep drainage had little effect on the annual soil water balance although the preferential water flow was highly important in a short-term soil water balance (Table 2).


View this table:
[in this window]
[in a new window]

  		Table 3. Annual soil water balance for the 1-m layer during 1997 to 2003.

 
Matrix water flow was the dominant process of drainage in the subsoil of this Andisol throughout the 7-yr period. This reconfirms the results of previous studies showing that the soil is a matrix-water-flow dominant soil (Hasegawa and Sakayori, 2000; Hasegawa and Eguchi, 2002; Eguchi, 2006) based on a long-term soil hydrologic point of view. The matrix water flow amounted to 167 to 627 mm yr–1 and accounted for 73 to 84% of the total annual drainage to below a depth of 1 m (Table 3). The matrix water flow occurred mainly in the downward direction (Fig. 6f). Upward matrix water flow was also observed and determined to be 5 to 60 mm yr–1 with a tendency to increase with a decrease in annual precipitation (Table 3).

The ETyear value was relatively constant in the range of 726 to 793 mm, and the average was 751 ± 27 mm (mean ± standard deviation) (Table 3). This value is comparable with those calculated from the soil water balances for Andisols obtained under various climates and experimental conditions in Japan, e.g., 771 ± 71 mm (experimental period 1965–1967) by Matsushita et al. (1969), 776 ± 106 mm (1968–1970) by Fujishima et al. (1972), 721 ± 54 mm (1966–1975) by Kamimura (1977), 768 ± 99 mm (1967–1968) by Fujishima (1984), 792 ± 55 mm (1975–1977) by Fujishima (1984), 631 ± 120 mm (1986–1990) by Tsukamoto et al. (1993), and 862 ± 135 mm (values read from the figures; 1978–1985) by Kobayashi et al. (1995). This indicates that annual evapotranspiration in Andisols shows similar values under the limited range of humid, temperate climate conditions regardless of the other experimental conditions. Furthermore, this suggests that the ETyear value obtained in this study is reasonable, and that the K–{theta} relation defined by Eq. [1] is appropriate to determine the matrix water flow in the subsoil of this Andisol under field conditions. Therefore, the amount of preferential water flow determined by applying both the one-dimensional Darcy's equation (Eq. [2]) and the soil water balance method (Eq. [3]) during and shortly after each rain event appears to be reliable.


   CONCLUSIONS
TOP
 NOTES
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
REFERENCES
 
The experimental approaches for determining preferential water flow by applying both the one-dimensional form of Darcy's equation and the soil water balance method successfully revealed the qualitative as well as the quantitative nature of preferential water flow across a depth of 1 m in an Andisol under field conditions. The preferential water flow was detected only two to seven times per year; nevertheless, it accounted for 16 to 27% of the annual total drainage. The preferential water flow mainly through the largest macropores at the depth of 1 m occurred in an event under macroscopically homogeneous soil water conditions, with pressure potentials mostly below the air-entry pressure at this depth, due to pore-scale physical nonequilibrium. The results provide the first field evidence demonstrating the presence of macropore-mediated preferential water flow in unsaturated subsoil of an Andisol. The macropore-mediated preferential water flow and solute transport should be considered in unsaturated subsoil even with pressure potentials below the air-entry pressure and even in soils such as Andisols where matrix water flow is the dominant process of drainage.


   ACKNOWLEDGMENTS
 
We thank Dr. Tetsuhisa Miwa, National Institute for Agro-Environmental Sciences (NIAES), for his valuable comments and suggestions on the statistical treatments of the field data; Dr. Hidetaka Katou, Dr. Keiko Nakano, and Mr. Katsuhiro Suzuki, NIAES, for valuable discussions; Mrs. Hiromi Gouhara, NIAES, for her assistance in the field measurements and in the processing of the field data; and the entire staff of the technical support division of NIAES for their assistance in soil and crop management throughout the long-term field experiment.


   NOTES
TOP
 NOTES
ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
All rights reserved. No part of this periodical may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying, recording, or any information storage and retrieval system, without permission in writing from the publisher. Permission for printing and for reprinting the material contained herein has been obtained by the publisher.

Received for publication January 29, 2007.


   REFERENCES
TOP
 NOTES
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 





This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF) Free
Right arrow An erratum has been published
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 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 Google Scholar
Google Scholar
Right arrow Articles by Eguchi, S.
Right arrow Articles by Hasegawa, S.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Eguchi, S.
Right arrow Articles by Hasegawa, S.
GeoRef
Right arrow GeoRef Citation
Agricola
Right arrow Articles by Eguchi, S.
Right arrow Articles by Hasegawa, S.
Related Collections
Right arrow Hydraulic Conductivity
Right arrow Preferential Flow
Right arrow Soil Hydrology

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