VZJ sign up for citetrack
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

Published online 16 November 2005
Published in Vadose Zone J 4:1201-1209 (2005)
DOI: 10.2136/vzj2005.0011
© 2005 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 ISI Web of Science (7)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by van der Velde, M.
Right arrow Articles by Clothier, B. E.
Right arrow Search for Related Content
PubMed
Right arrow Articles by van der Velde, M.
Right arrow Articles by Clothier, B. E.
GeoRef
Right arrow GeoRef Citation
Agricola
Right arrow Articles by van der Velde, M.
Right arrow Articles by Clothier, B. E.
Related Collections
Right arrow Vadose Zone Processes and Chemical Transport
Right arrow Recharge
Right arrow Field evaluation techniques

ORIGINAL RESEARCH

Evaluation of Drainage from Passive Suction and Nonsuction Flux Meters in a Volcanic Clay Soil under Tropical Conditions

M. van der Veldea,*, S. R. Greenb, G. W. Geec, M. Vancloostera and B. E. Clothierb

a Departement of Environmental Sciences and Land Use Planning, Université catholique de Louvain-la-Neuve (UCL), Louvain-la-Neuve, Belgium
b HortResearch, PB 11030, Palmerston North, New Zealand
c Battelle, Pacific Northwest Division, USA
* Corresponding author (vandervelde{at}geru.ucl.ac.be)



   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Root zone drainage measurements are needed to improve fertilizer management in areas where agriculture may be impacting groundwater supplies. We present results of field tests where drainage was measured with two types of suction (resolution of 0.16 and 1.6 mm tip–1) and a nonsuction (resolution of 0.22 mm tip–1) water flux meter (WFM). The soil was a microstructured weathered volcanic ash located on a coral atoll subject to intense rainfall and located in the Kingdom of Tonga. Our objectives were to evaluate water flux measurements by comparing them with (i) simple water balance estimates of cumulative fluxes, (ii) cumulative fluxes deduced from soil moisture content changes, and (iii) simulated fluxes using HYDRUS-1D. Soil hydraulic properties were obtained at five soil depths. During the 60-d evaluation period rainfall totaled 340 mm. The WFMs were installed in duplicate using disturbed soil. The consistency of the shape of the drainage curves measured with the WFMs, those derived from soil moisture changes, and those obtained with modeling led us to conclude that soil disturbance during WFM installation did not severely influence measurements. This was attributed to the strong microaggregation and disturbance introduced by plowing. Water balance and HYDRUS model estimates of drainage corresponded well with the measurement by nonsuction WFMs. Suction WFMs overestimated drainage, possibly due to flow convergence created by wick and divergence barrier lengths being not properly sized for the observed flow conditions. After the evaluation period some of the WFMs failed to respond. Nevertheless, flux meters are seen as promising tools to provide remote and continuous measurement of root zone drainage.

Abbreviations: C-WFM, capacitance water flux meter • DOY, day of year • ET0, reference crop evapotranspiration • SMC, soil moisture content • T-WFM, tipping bucket water flux meter • WFM, water flux meter • Z-WFM, nonsuction WFM with zero tension


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 CONCLUSIONS
 REFERENCES
 
DETAILED ESTIMATES of the water flux in a field soil are needed to understand the fate of water and solutes in soil. Water fluxes can be estimated using either indirect or direct methods. Indirect methods include the use of water balance calculations based on changes in measured soil water potential or soil water content, and the use of numerical modeling that requires soil physical and hydraulic properties as well as appropriate crop and climate data. Direct methods introduce an obstruction in the soil that is designed to intercept, to measure, and often to sample the drainage water. A distinction can be made between instruments without suction and those that introduce a suction to the surrounding soil. The greatest advantage of these devices is the possibility of simultaneously measuring the flux of water and solute in the field. Disadvantages include a certain level of disturbance when the instrument is installed and the limited soil volume that is sampled by the device.

The creation of a good contact between the soil and the top of the instrument is critical. Zero-tension lysimeters (Russel and Ewel, 1985; Zhu et al., 2002) collect drainage water only when the soil directly above them is saturated. Instruments that apply a tension or suction to the soil have the potential to more accurately simulate matric potentials in the soil. The suction is either controlled by the generation of a vacuum (Montgomery et al., 1987; van Grinsven et al., 1988; Brye et al., 1999; Barzegar et al., 2004; Kosugi and Katsuyama, 2004) or it is set by the height of a hanging water-column introduced by a wick in contact with the soil (Holder et al., 1991; Boll et al., 1992; Knutson and Selker, 1994; Brandi-Dohrn et al., 1996; Louie et al., 2000; Gee et al., 2002). Brye et al. (1999) developed the equilibrium tension lysimeter, where suction was manually adjusted according to the matric potential of adjacent soil directly above the lysimeter. Masarik et al. (2004) developed an automated version of the equilibrium tension lysimeter of Brye et al. (1999). Kosugi and Katsuyama (2004) developed an alternative lysimeter where the period of suction was controlled directly using the difference in matric potential between the natural and the sampled soil profile. Several field studies have been performed using these instruments. For example, Boll et al. (1997) used grid pan samplers to quantify the spatial distribution of water and solute flux with a temporal resolution dependent on the time of sampling.

The number of field studies reporting the direct measurement of water flux with a high temporal resolution is still quite limited. To our knowledge, no studies have reported the performance of different types of WFMs in microstructured and weathered volcanic free-draining tropical clay soils. Furthermore, given the practical nature of our research, we sought to test the reliability and robustness of a low cost system that could operate without the need of maintenance and without much human interference. The harsh tropical conditions and a free-draining clay soil posed specific challenges to the devices. The principal objective was to evaluate the performance of the WFMs, and this was achieved by comparing the water flux measurements with (i) potential drainage estimated from a simple water balance, (ii) soil moisture content changes derived from calibrated soil moisture measurements, and (iii) one-dimensional modeling of water movement using measured soil properties.

The research was part of an applied research project. Within the project (http://www.croppro.alterra.nl, verified 6 Sept. 2005) we wished to develop sustainable strategies for managing agrichemicals on the island of Tongatapu. In 1987, a squash (Cucurbita maxima Duchesne) industry emerged on Tongatapu that solely exports to the Japanese market during an off-season niche period in November. At a national level, the squash industry has accounted for about 40% of the total export revenue and about 80% of export revenue derived from agriculture during the last 10 yr. With the export of the squash, a significant increase in the importation and use of agrichemicals has occurred. The potential for leaching of applied agrichemicals has raised awareness of the threat of contamination of Tongatapu's water resources. Knowledge on the flux of water and solutes can lead to better application strategies for fertilizers and pesticides, resulting in both economic and ecological benefits.


   MATERIALS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS
 DISCUSSION
 CONCLUSIONS
 REFERENCES
 
The experiments were performed on the island of Tongatapu (175°12'W, 21°08'S), the main island of the Kingdom of Tonga, located in the South Pacific Ocean. The island is a flat raised coral atoll with a highest elevation of 60 m above sea level (Furness and Helu, 1993). The limestone base of the island consists of permeable limestone that has a saturated hydraulic conductivity of approximately 0.015 m s–1 (Hunt, 1979). Groundwater is present as freshwater lenses that float on the saltwater in the limestone aquifer. Tongatapu has no surface water. The groundwater is interconnected to a relatively enclosed sea-waree lagoon, and a mix of ground and sea water can be seen seeping into the lagoon at low tide. Soil depth on the island generally decreases from about 6 m in the west to about 0.5 m in the east (Cowie et al., 1991). The soil sits directly on top of the permeable limestone. Any water and solutes that have passed through the soil are likely to flow downward through the limestone toward the freshwater lenses.

Experimental Site
An experimental site was established at the Vaini Agricultural Research Station during the 2002 growing season of squash. A meteorological station was installed at the experimental site to measure global radiation, relative humidity, air temperature, wind speed, and rainfall. Incoming global radiation was measured with a silicon-cell pyranometer (SKS 1110, Skye Instruments, Llandrindod Wells, Powys, UK). Air temperature and relative humidity were measured with a Vaisala (HMP45A, Vaisala, Vantaa, Finland) probe. Wind speed was measured with a sensitive three-cup anemometer (R30, Vector Instruments, Rhyl, UK), and a tipping-spoon rain gauge (Rain-O-Matic, Pronamic, Silkeborg, Denmark) was used to monitor rainfall. Data were recorded every 15 s and 30-min averages were stored on a data logger (CR10X, Campbell Scientific, Logan, UT). The instruments were mounted on a mast in the middle of the field, at a height of 2 m above the ground. Daily meteorological data were used to calculate the FAO reference crop evapotranspiration, ET0, following Allen et al. (1998).

Soil
The soil on the island was classified as typic Argiudoll, very-fine, halloysitic and isohyperthermic (Cowie et al., 1991). Halloysite is the dominant clay mineral (>90%). Two characteristic layers, each associated with different volcanic deposits can generally be identified on the island. Often a shallow (<50 mm) buried soil can be found between those layers. Perched water tables may occur at the transition between the young and old ash deposits. The soil at the experimental station is part of the Vaini soil series (Cowie et al., 1991). Total soil depth is approximately 2.8 m. Groundwater is found at a depth of about 20 m in the limestone aquifer. The soil surface is flat, and no surface ponding or runoff was observed. The transition between the younger and the older clay layer occurs at around 90 to 100 cm (Cowie, 1980). The soil has a clay content from about 70% at the top of the profile to 90% at the 1-m depth. Soft, weathered lapilli occur throughout the whole soil profile. The dry bulk density of the soil ranged from 0.5 to 1.2 mg m–3, and the pH ranged between 6 and 7.2. The organic C content in the top 18 cm was about 3.4% and decreased to 0.8% at a depth of 28 to 43 cm (Cowie et al., 1991). A strong microaggregation was observed in the soil and this was explained by interactions of the halloysitic clay minerals, organic components, and iron oxides (Trangmar, 1992). A large observation pit was excavated in a representative area of the field. Water retention was determined at the 15-, 30- (in triplicate), 60-, 90- and 120-cm depths (in duplicate) (Fig. 1) . Undisturbed cores (20-mm height, 50-mm diameter) were collected to measure water retention at 0.1, 0.5, 3.0, 6.0 kPa using Haines' apparatus and at 113.0 kPa using a pressure plate. Bulk soil obtained at the same depth was used for the determination of water retention at 1513 kPa using a pressure plate. The van Genuchten (1980) parameters associated with the soil's water retention properties are presented in Table 1.



View larger version (27K):
[in this window]
[in a new window]
  		Fig. 1. Measured water retention characteristics and the corresponding van Genuchten (1980) curves obtained for five soil depths at the field site.

 

View this table:
[in this window]
[in a new window]
  		Table 1. van Genuchten (1980) parameters and saturated hydraulic conductivity for 5 soil depths.

 
The following strategy was chosen to fit the curves. Saturated and residual water content were set so that they reflect the measured water content at, respectively, 10–1 and 1.5 x 103 kPa. The {alpha} and n parameters were obtained by means of least square regression (m = 1 – 1/n).

The soil retains a lot of water even at very low soil matric potentials. The values of the parameters are similar to the water retention properties determined in three other profiles at the island (data not shown) and as measured by Cowie (1980) and Cowie et al. (1991). This led us to consider that, for this exercise, the values of the parameters determined for five layers were representative for the soil above the WFMs. Hydraulic conductivity (measured at a head of –10 mm) was measured in triplicate using a tension disc infiltrometer at the same depths in the observation pit. These were linearly extrapolated to obtain a value of the soil's saturated hydraulic conductivity (Ksat). The values for Ksat indicate that water transport through the structured clay is rapid. Similar soil hydraulic properties have been observed in other soils from volcanic origin. Dorel et al. (2000) described a volcanic soil hydrodynamically behaving as a sand at matric potentials greater than –300 kPa and as clay below –1550 kPa. The microaggregation associated with this effect was illustrated by Miyamoto et al. (2003). Microaggregated clay soils derived from volcanic ash exhibiting this behavior are commonly referred to as pseudo-silts or pseudo-sands.

Water Flux Meters Design and Installation
A schematic figure of the WFMs is shown in Fig. 2 . All WFMs were installed with the top of the funnel at a depth of 1 m (Fig. 2, vertical exaggeration of funnel height). We used two types of passive-capillary wick (Gee et al., 2002, 2003) and one type of nonsuction WFM. The WFMs were designed to minimize divergence of flow. Measurements were performed in duplicate. Installation of the WFMs should ideally be done by inserting the WFMs into the ground and placing an undisturbed soil column on top of the device. Unfortunately the tillage practices and the reworked heavy clay soil did not allow for an easy insertion of the tube into the soil to create an undisturbed core. Therefore the soil above the WFMs was refilled and repacked to a bulk density similar to the surrounding field soil following tillage. Given the microaggregation of the soil and the principal objective of the study—evaluation of the WFMs, the interpretation of the WFM measurements is believed to remain sound in these conditions. Two soil pits close to the observation pit were excavated, and a WFM of each type was installed into each pit. The two pits were repacked and the surface leveled with respect to the surrounding soil surface. The microstructure of the soil in the flux meter resembled that of the surrounding tilled soil. After the first rain event, a minor depression above one hole was leveled off. The WFMs are equipped with a pipe that extends from the top of the funnel to minimize divergence. Gee et al. (2004) showed that a WFM in a clay soil should perform well at high fluxes, while at low fluxes divergence may occur. The collection efficiency increased with increasing height of the pipe. The suction WFMs had a collection area of 314 cm2, and the nonsuction WFMs had a collection area of 779 cm2. Because of the larger collection area the pipe height could be lower (Gee et al., 2004). Below each instrument a cavity was excavated and refilled with sand to accommodate drainage water from the devices.



View larger version (29K):
[in this window]
[in a new window]
  		Fig. 2. Schematic cross section (exaggerated horizontal scale) of the three water flux meter (WFM) types installed in the soil. A soil solution sample can be extracted from each device (with Z-WFM, nonsuction water flux meter; C-WFM, capacitance water flux meter; and T-WFM, tipping-bucket water flux meter).

 
The suction WFMs consisted of the overlying pipe that rested on a funnel filled with soil. The suction WFMs were each equipped with two prepared (Knutson et al., 1993) intertwined wicks (no. 1381, Pepperell Braiding Company, Pepperell, MA), each with a diameter of 12.5 mm, to control the suction. The wick has a Ksat of 1168 cm h–1. The inside area of the funnel cone is occupied by carefully separated and aligned fiberglass wick material of 10-cm length. The remaining 50 cm of the wick goes through the funnel neck downward, where it ends just above the measurement device. The advantage of using a wick instead of a vacuum to create suction is that no additional equipment or energy is used for operation of the measurement. Diatomaceaous earth was placed in the bottom of the funnel, above the wick strands, to prevent soil filtering into the device. The two types of suction WFMs are of similar design but use different techniques to measure the drainage water flux. The first type, described by (Gee et al., 2003), is referred to here as the tipping bucket water flux meter (T-WFM). This device uses a mini-tipping bucket (Rain-O-Matic, Pronamic) to measure the flow out of the bottom of the wick. The tipping bucket was calibrated at 5.0 mL tip–1, which equates to a resolution of 0.16 mm tip–1. The second type of suction WFM is commercially available as a Gee Passive Capillary Lysimeter or Drain Gauge (Decagon Devices, Pullman, WA). We will refer to this device as the capacitance water flux meter (C-WFM). Drainage water is accumulated in a siphon chamber that automatically empties at a volume that is predetermined. An ECHO-type capacitance sensor, similar to that of Masarik et al. (2004) measures the water depth in the siphon chamber. The instrument autosiphons at approximately 50 mL, which equates to a resolution of 1.6 mm tip–1. However, the resolution can be increased if the water level in the instrument is directly measured and recorded. The nonsuction WFM with zero tension is referred to as the Z-WFM. Drainage through the Z-WFM was measured with a tipping bucket that has an accuracy of 17.4 mL tip–1, which equates to 0.22 mm tip–1. The Z-WFMs have a 20-µm nylon mesh to filter out soil particles transported with the drainage water; the mesh also prevents deep roots from entering the device. The sum of the tips measured by the WFMs were stored every 30 min by the CR10X data logger. Before installation, the operation of each WFM was checked with a number of calibration pulses at high intensity. The Z-WFMs and the T-WFMs both functioned well during these simulated events. The autosiphoned volume of the C-WFMs was determined to be 43 mL. For this study, only the siphon events of the C-WFMs were recorded because we were not convinced that a higher accuracy would be needed during high-intensity drainage events. A simple edge-detection algorithm was used to determine each time a volume was autosiphoned from the collection chamber. The performance of the WFMs was evaluated during a 2-mo period between Day of Year (DOY) 200 and 260.

Soil Moisture Content Measurements
Soil moisture content (SMC) was measured using a number of water content reflectometer probes (CS616, Campbell Scientific). Measurements of SMC were done once every 30 min. Two horizontal sets (30, 60, 90, and 120 cm) and two vertical probe sets (0–30, 30–60, 60–90, and 90–120 cm) were installed in the volcanic ash soil during the previous year. CS616 probes are known to exhibit a calibration that is sensitive to the soil's clay content. Because the soil on Tongatapu has such a high clay content (>70%), which increases with depth, it was felt prudent to check the calibration of the CS616 probes. A field calibration of the CS616 probes was done with soil moisture contents that were gravimetrically determined using rings of a known volume. The SMC was determined by placing all samples into an oven (>100°C) for 48 h. Separate calibration curves for the CS616 probes (Fig. 3) were determined for the clay soil of the upper part of the profile, and for the deeper clay (70 cm). A good linear relationship (r2 = 0.94 and 0.79 for the top and the deeper part of the clay soil, respectively) was found between measured and actual SMC. However, the calibration for both layers was found to differ markedly from the 1:1 line (Fig. 3).



View larger version (19K):
[in this window]
[in a new window]
  		Fig. 3. Calibration curve for CS616 probes in the weathered volcanic ash soil of Tongatapu (SMC denotes soil moisture content). The calibration had a r2 of 0.94 for the top clay soil and a r2 of 0.79 for the deeper clay soil.

 
Uncalibrated, the error in SMC would be about 0.15 cm3 cm–3 at the wet end for the deep clay layer and approximately 0.05 cm3 cm–3 for the top clay soil. The difference between the two layers is likely related to the lower soil organic mater, higher clay content, and other structural properties of the clay (Miyamoto et al., 2003) that become more pronounced deeper in the profile. Similar studies in volcanic soils (Veldkamp and O'Brien, 2000) have previously shown errors of up to 0.15 m3 m–3 in SMC using uncalibrated CS615 probes. The highest saturated SMC measured by the calibrated CS616 probes was 0.62 cm3 cm–3. Compared with the water retention curves obtained at 60 and 90 cm (Fig. 1), this is low. The high saturated and residual water content obtained at 60 and 90 cm are not unusual compared with Cowie (1980), Cowie et al. (1991), and other water retention data we have obtained for the same soil type elsewhere on the island. We attribute the difference to the natural variability in the soil profile (Cowie, 1980).

The total amount of water for soil layers between 30 and 60, 60 and 90, and 90 and 120 cm was summed to provide the average water content between 30 and 120 cm. Evaporation and plant water uptake were considered to occur only in the top 30 cm, since very few roots were encountered below 30 cm during the evaluation period. This was also confirmed by deeper soil moisture content measurements that registered almost no plant water uptake (<5 mm during the evaluation period) deeper than 30 cm. Ignoring the top 30 cm of soil, changes in stored water were attributed to either infiltration or drainage. Thus, infiltration was calculated by simply summing up the positive differences in total stored water. Likewise, drainage was calculated by summing the negative differences.

Modeling with HYDRUS
Soil water movement was modeled using the HYDRUS-1D numerical code that solves Richards' equation (Simunek et al., 2003) using the van Genuchten–Mualem model (van Genuchten, 1980; Mualem, 1976). We used the soil hydraulic properties listed in Table 1, although the soil disturbance likely led to a greater homogeneity of the soil physical and hydraulic properties directly above the WFMs. The upper boundary condition for water flow was prescribed by the rainfall rate, measured in millimeters per 30 min. Evapotranspiration was taken as daily ET0 equally distributed over 30-min periods between 0730 and 1930 h. The lower boundary condition at 100 cm was set as a seepage face and as a constant matric potential of –50 cm to mimic the capillary suction applied to the soil by the wick material. Initial soil matric potential was set equal for the whole soil profile at –50 cm because the soil was initially quite wet due to recent rainfall.

Land Preparation
The experimental plot was prepared by three ploughs. The top 30 cm of the soil was effectively turned over by the plowing and resulted in a rough soil surface consisting of large clumps of soil and loose aggregates. Thereafter mounds of 15 cm high were prepared manually at a spacing of 1.5 by 1.5 m. Mounds were also made above the two pits where the WFMs were installed. One pit was located in the middle of four mounds while the other pit was located under one mound. One replicate of each WFM type was thus located in the pit underneath the middle of four mounds, and the other replicate was placed in the pit directly beneath one mound. Within each pit, each individual WFM was thus configured differently with respect to the mounds on the surface. Some 120 g of NPK fertilizer, on average, was added and mixed into each soil mound. Squash seeds were planted on 25 July (DOY 206). The soil remained bare until about DOY 220, when the squash started to emerge. After about 6 wk a further side dressing of 15 g of urea was applied to each mound.


   RESULTS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Climatological Measurements
Daily values of rainfall and ET0 reflect the occurrence of heavy rain events, which are typical in this tropical climate (Fig. 4) . Large rainfall events exceeding 100 mm can stretch over several days, with the appearance of almost continuous rain (e.g., DOY 232–233).



View larger version (23K):
[in this window]
[in a new window]
  		Fig. 4. Daily rain and FAO crop reference evapotranspiration values for the evaluation period.

 
During the evaluation period a total of four rain events (see Fig. 4, starting on DOY 203, 224, 232, and 235) occurred that triggered flow through the nonsuction WFMs. The nonsuction WFMs did not register flow beyond DOY 240, as there were no extreme rainfall events and plant cover became more established. Highest rainfall intensity was measured at 35.6 mm h–1 in a 30-min period. The largest amount of rain during a single storm event was 120.8 mm on DOY 232 and 233 (Fig. 4). In the rain-free periods between storm events, ET0 was typical for a tropical climate (Allen et al., 1998) and ranged from 1 to 5 mm d–1. Independent estimates of evaporation and transpiration obtained in the field using a combination of microlysimeters and sapflow measurements in the squash (van der Velde et al., 2005) indicated losses measured at about 126 mm while cumulative ET0 amounted to 197 mm. Cumulative rain equaled 343 mm. This does not include the 35.4 mm that fell between the day of installation DOY 192 and DOY 200; it is assumed that this rain brought the soil close to saturation (as indicated by SMC measurements).

WFM Performance
Measured drainage divided by potential drainage is a measure of the collection efficiency of the WFMs (Table 2). Potential drainage was calculated with a simplified water balance. It was derived as the difference between measured rain (343 mm) and measured ET (126) between DOY 200 and 260, giving cumulative potential drainage of 217 mm.


View this table:
[in this window]
[in a new window]
  		Table 2. Drainage and collection efficiency of the water flux meters (WFM). Collection efficiency was calculated with potential drainage estimated from a simple water balance at 217 mm. Each WFM was tested in duplicate. T-WFM 1 was not functioning properly.{dagger}

 
The Z-WFMs collected the lowest drainage amounts. The Z-WFMs collected 92 and 129% of the water balance estimation of drainage. No consistent relation between total drainage and location in one of the two pits was observed. Z-WFM1 collected a higher cumulative drainage than Z-WFM2, while C-WFM1 collected a lower cumulative drainage than C-WFM2. The surprising result was that the suction WFMs collected more drainage than rain—we will discuss this below.

No technical problems with the WFM measurement components (i.e., tipping bucket, capacitance probe) occurred during the growing season. Some problems related to the installation procedure did occur with one of the T-WFMs, as it appeared to have not enough drainable volume below the unit to accommodate excess water passing through the meter. It seems that free water temporarily flooded the instrument at such times. A similar problem occurred once with one of the C-WFMs (DOY 232). Here an alternative explanation may well be that the autosiphoning chamber could not cope with the high fluxes. The C-WFMs and Z-WFMs had ceased to function after the growing season. In experimental sites in the USA some C-WFMs have been dug up and it was found that the ends of the siphon tube were rusting and subsequently rusted shut. We suspect that the same occurred under the humid tropical conditions of our study. We suspect that, similar to our experience with the Z-WFMs during the previous experimental season, the tipping buckets of the Z-WFMs rusted loose and came off, although this season they had been sealed with water-proof glue. Unlike the Z-WFMs, the tipping buckets of the T-WFMs are constructed with plastic and stainless-steel materials and are rust resistant.

Timing
The time course of the drainage provides information on the flow velocities through the soil. Here we compare the response of the WFMs for the four large drainage events, which began on DOY 203, 224, 232, and 235. All WFMs showed a fast response soon after the onset of rain. This was corroborated by rises in the soil water content recorded by CS616 probes at 90 and 120 cm. The time difference between the highest rain intensity and the peak drainage flux ranged from 1.5 to 7.5 h for the different instruments. A similar time delay was observed in the peak in SMC measured with the CS616 probes at 90 cm. Although the CS616 probes always registered the water movement a little earlier than the WFMs. Similarity between WFMs and the passage of the waterfront registered by the CS616 probes at 90 cm suggests that the saturated hydraulic properties of the repacked soil were similar to that of the natural soil profile. This can be explained by the microstructuring of the clay soil. Differences in the timing of first drainage are most likely related to the different amounts of water needed to restore the soil moisture content of the repacked soil above the WFMs. Once the soil in the WFMs had obtained similar soil moisture conditions, all WFMs responded in a similar fashion, with peaks occurring within 1.5 h of each other for the three remaining drainage events. There is also good agreement with the timing of the HYDRUS modeled flux. Both the response and the peak of the modeled flux were at the same time, within 30 min, of the response of the first responding WFM for the first three events (DOY 203, 223, and 232). The modeled flux for the last event (DOY 335) predicted the first response about 1 h earlier than the measured flux, while the predicted peak flux occurred at the same time as that measured.

The temporal pattern of cumulative infiltration and drainage deduced from SMC data between 30 and 120 cm is shown in Fig. 5 . One set of probes measured 163 mm cumulative infiltration and 147 mm cumulative drainage during the period (Fig. 5). The other set of probes also measured more cumulative infiltration (140 mm) than drainage (115 mm) during the period (data not shown). Storage changes where thus positive and on the order of 20 mm.



View larger version (16K):
[in this window]
[in a new window]
  		Fig. 5. Infiltration and drainage estimated from the combined measurement of soil moisture content by one set of calibrated CS616 probes placed vertically at depths of 30 to 60, 60 to 90, and 90 to 120 cm.

 
None of the measurements done with the CS616 indicated saturated SMC as high as suggested by the water retention curves obtained at 90 cm. We therefore chose to compare SMC measured at 90 cm and modeled SMC at the 91-cm depth (Fig. 6) .



View larger version (18K):
[in this window]
[in a new window]
  		Fig. 6. Soil moisture content measured at 90 cm and modeled at 91 cm. Soil moisture contents were measured with two calibrated CS616 probes.

 
The shapes of the curves are similar, and they clearly reflect the four largest drainage events. The measured SMCs appear to drain over a smaller range of SMCs. However, the variability in water retention properties in the field, especially introduced at the interface of the younger and older clays (Cowie, 1980), leads to uncertainties in the values of the modeled SMC. Additionally, given the uncertainty due to natural variability and the error introduced by the calibration of the measured SMCs, we cannot justifiably compare the modeled and measured SMC values with the current data set.

Water Flux
The drainage flux at 100 cm, measured by two WFMs and modeled with HYDRUS for a snapshot period of 3 d is shown in Fig. 7 . There is a good correspondence between measured and modeled drainage events. The rapid drop in the water flux indicates the importance of saturated transport in this soil under these tropical conditions. Beyond DOY 240 the Z-WFMs ceased to record any drainage events. During this time, the suction WFMs continued to measure drainage. The peak water flux ranged between 0.004 mm s–1 (14.4 mm h–1) for the model and 0.007 mm s–1 (25.4 mm h–1) for the suction WFMs. The peak and the tail of the drainage events are both higher under the suction WFM compared with the Z-WFM and the model predictions (Fig. 7). The agreement between the modeled water flux at 100 cm and the Z-WFM is good. The model also confirms that significant water movement occurred during the same time that drainage events were registered by the Z-WFMs. The current divergence-control dimensions of the Z-WFMs seem to be appropriate to measure the large drainage events in this soil in this climate.



View larger version (18K):
[in this window]
[in a new window]
  		Fig. 7. Measured and HYDRUS-1D modeled water flux at 100 cm. Measurements are shown of a nonsuction water flux meter (Z-WFM 2) and a suction tipping-bucket water flux meter (T-WFM 2).

 
Drainage Volumes
The temporal pattern of cumulative measured and modeled drainage is presented in Fig. 8 . The model indicates a cumulative drainage of 235 mm, about 20 mm higher than the simple water balance estimate of 217 mm. This difference is attributed to the high intensity of the events and a mismatch in the initial moisture conditions of the modeled soil profile. Drainage derived from SMC between 30 and 120 cm (147 mm) was considerably lower than both the modeled and simple water balance estimate. This is mainly attributed to the fact that under near-saturated conditions there can be large drainage fluxes of water and yet very little change in SMC. There is a good agreement between the cumulative drainage measured with the Z-WFMs and the modeled drainage. We attribute the improved collection efficiency of the Z-WFMs (92 and 129%) to the use of a divergence pipe and specific climate and soil hydraulic properties during the evaluation period. Reported values are normally found to be about 10 to 58% (Zhu et al., 2002; Jemison and Fox, 1992).



View larger version (17K):
[in this window]
[in a new window]
  		Fig. 8. Cumulative rain, modeled drainage (HYDRUS-1D), and measured drainage and standard deviations for replicate nonsuction water flux meters (Z-WFM 1 and 2) and suction WFMs (T-WFM 2 and C-WFM 1 and 2).

 

   DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
CONCLUSIONS
 REFERENCES
 
The suction WFMs clearly overestimated drainage because the totals exceeded actual rainfall (Fig. 8). Periods when the Z-WFMs were responding can be used to calculate cumulative drainage for the suction WFM when the soil saturated. The suction WFMs collected about 350 to 360 mm of drainage during the periods when the soil directly above them is saturated. This is comparable to the total amount of rain that fell during the period. An additional 200 mm was captured by the suction WFMs during unsaturated flow. In contrast, in the HYDRUS model about 20 mm of drainage, from a total of 235 mm, occurred under unsaturated conditions. Other authors have had reasonable success with passive capillary samplers in humid climate conditions but also have seen oversampling of drainage with wick units. In western Oregon, USA, Louie et al. (2000) measured an average collection efficiency of 125% comparing recharge estimated by 30 passive capillary samplers to a water-balance estimate of recharge on annual basis.

Under the high flow rate conditions experienced here, the height of the divergence barrier appeared to significantly affect measured drainage. For the Z-WFM, which had a 20-cm divergence height, the rainfall regime and high porosity soil combined to provide the tension control needed to achieve a nearly exact match for capturing the correct drainage. In contrast, the 60-cm extension pipe for the wick units was apparently too tall for the extremely wet conditions and highly permeable soil. It appears that there was significant convergent flow in all of the wick units. Rimmer et al. (1995) showed that fine-textured soils require a large sampling area to create an undisturbed zone above the sampler. Due to the microaggregation of the clay soil studied here fast flow is possible in the near saturated range. We attribute the overestimation of the suction WFMs to the effect of the continuous suction (h = –50 cm) applied by the wick plus the additional 60 cm of divergence control. We suspect that the matric potentials in the soil were often above –50 cm (less negative) during the measurement period. This was confirmed by the modeled head at depths of 90 and 100 cm (data not shown). Predicted drainage increased only by about 20 mm when HYDRUS was run with a continuous matric potential of –50 cm as lower boundary condition. The suction seemed to have drawn soil water from a larger area above and around the suction WFMs, and one-dimensional modeling is not sufficient to explain these measurements. We are investigating reasons for the overestimation of the suction WFMs found in our study with more detailed two-dimensional modeling. However there are also other factors that may have influenced the measurement. The occurrence of macropore flow may have contributed to the higher drainage measurement by the suction WFMs, especially during such high percolation rates. Other possibilities include natural soil variability, some influence of the repacking of the soil above the WFMs, and preferential flow triggered by the overlying pipe. The soil surface was quite rough, and even imperceptible local depressions may have triggered funnel flow above the WFMs. The largest differences between the suction WFMs were observed during saturated drainage events. Interestingly, after this drainage, the scatter of observations done by the suction WFMs decreased, so it seems that the suction applied to the soil by the wick tends to even out the total drainage collected by the suction WFMs. This is likely due to different ratios of macro- and micropores in the soil directly above the meters. Different pore sizes will tend to drain under different tensions.

Disturbance caused by installation did not appear to severely influence observations. We deduce this from the consistency of (i) the passage of the water front and the shape of the drainage curve measured with the WFMs and the soil moisture probes, (ii) the agreement between the measurements of the Z-WFMs and the modeling using soil hydraulic properties obtained from undisturbed soil, and (iii) the similarity of the shape of the drainage curve obtained with the suction WFMs and the model. This is probably due to the strong microaggregation observed in the soil and the inherent disturbance of the top soil introduced by plowing. This led us to conclude that a large fraction of the rain drains through the soil toward the limestone under saturated conditions. The Z-WFMs indicated that about 70% of the precipitation drained under these wet conditions. High drainage volumes have been reported for similar climates with similar soils. Duwig et al. (1998) attributed 64% of a 170-mm rainfall event to drainage. Russel and Ewel (1985) reported on drainage measurements during big storms in Costa Rica. They found that 84% of the annual drainage was collected during 2-wk storms. The Z-WFM units with a 20-cm extender pipe (acting as a relatively short divergence barrier) worked better than the wick units in optimizing the drainage collected during the 60-d test. Testing is required for better matching of the divergence-control dimensions for the wet soil conditions in these highly permeable soils. In this case we would expect that it would call for a shorter divergence barrier and wick combination.


   CONCLUSIONS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 CONCLUSIONS
REFERENCES
 
Water flux meters are a useful and low-cost tool for direct determination of water fluxes in field soils. The performance of the WFMs depends on their design criteria and operational conditions, including the soil and climate conditions. In this study we analyzed the performance of three WFM types (Z-WFM, T-WFM, and C-WFM) in tropical microstructured clayey soils. The microstructuring of this specific soil yields a pseudo-sand type of hydrodynamic behavior that seemed to allow the use of disturbed soil material in the evaluation analysis. Water flux meters are relatively easy to install and maintain and are therefore suitable for on-farm use. However, during this study only short-term success was achieved (3–12 mo). At other sites with temperate climatic conditions (USA) the instruments have been functioning continuously for several years (Gee et al., 2004). Soil moisture content storage changes deduced from SMC measurements were positive and accounted for about 10% of the total drained volume. Successful direct measurements of water flux at the 100-cm depth were achieved with nonsuction WFMs, and this was supported via modeling of soil water movement using HYDRUS. Overestimation of drainage flux with suction WFMs was attributed to convergence of the flow paths above the divergence pipe. We speculate that during this evaluation period the soil matric potential was higher than the tension applied by the wick. From an operational point of view, we also recommend providing sufficient "drainable" soil volume under the WFMs if they are used under tropical conditions characterized by intensive rainfall events. Disadvantages of the WFMs include the locality of the measurement and the relatively small sampling areas. The discrepancies between suction WFM measurements and model and alternative WFM designs for these conditions will be explored with two-dimensional modeling in ongoing research. The results presented here will be combined with nitrate and pesticide analyses of the drainage water to develop more sustainable agricultural practices on this Pacific island atoll.


   ACKNOWLEDGMENTS
 
We gratefully acknowledge the staff of the MAF Vaini Research Station for their help with the installation of the WFMs. The reviewers are kindly acknowledged for their valuable comments. We also express appreciation to Decagon Devices, Inc, Pullman, WA and Sledge Products, Dayton, OR for supplying the wick lysimeters for testing. This work is funded by the European Commission under the INCO-DEV Programme (ICFP500A4PRO2) and the New Zealand Agency for International Development as part of the Pacific Initiative for the Environment.


   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 CONCLUSIONS
 REFERENCES
 




This article has been cited by other articles:


Home page
 Vadose Zone JHome page
J. Sansoulet, Y.-M. Cabidoche, P. Cattan, S. Ruy, and J. Simunek
Spatially Distributed Water Fluxes in an Andisol under Banana Plants: Experiments and Three-Dimensional Modeling
Vadose Zone J., May 27, 2008; 7(2): 819 - 829.
[Abstract] [Full Text] [PDF]

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]

Home page
 Soil Sci.Home page
J. Mertens, J. Diels, J. Feyen, and J. Vanderborght
Numerical Analysis of Passive Capillary Wick Samplers prior to Field Installation
Soil Sci. Soc. Am. J., January 1, 2007; 71(1): 35 - 42.
[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 ISI Web of Science (7)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by van der Velde, M.
Right arrow Articles by Clothier, B. E.
Right arrow Search for Related Content
PubMed
Right arrow Articles by van der Velde, M.
Right arrow Articles by Clothier, B. E.
GeoRef
Right arrow GeoRef Citation
Agricola
Right arrow Articles by van der Velde, M.
Right arrow Articles by Clothier, B. E.
Related Collections
Right arrow Vadose Zone Processes and Chemical Transport
Right arrow Recharge
Right arrow Field evaluation techniques

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