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


     

Published online 16 November 2005
Published in Vadose Zone J 4:1037-1047 (2005)
DOI: 10.2136/vzj2004.0138
© 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 ISI Web of Science (1)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Plauborg, F.
Right arrow Articles by Lærke, P. E.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Plauborg, F.
Right arrow Articles by Lærke, P. E.
GeoRef
Right arrow GeoRef Citation
Agricola
Right arrow Articles by Plauborg, F.
Right arrow Articles by Lærke, P. E.
Related Collections
Right arrow Vadose Zone Processes and Chemical Transport
Right arrow Water Content
Right arrow Field evaluation techniques

SPECIAL SECTION: SOIL WATER SENSING

In Situ Comparison of Three Dielectric Soil Moisture Sensors in Drip Irrigated Sandy Soils

Finn Plauborg*, Bo V. Iversen and Poul E. Lærke

Danish Institute of Agricultural Sciences, Department of Agroecology, Research Centre Foulum, P.O. Box 50, DK-8830 Tjele, Denmark
* Corresponding author (Finn.Plauborg{at}agrsci.dk)

Received 24 September 2004.



   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 PERSPECTIVES
 REFERENCES
 
The number of different sensors available for measuring soil water content has increased since the introduction of time domain reflectometry (TDR). In this study the performances of the CS616 (Campbell Scientific, Ltd., Shepshed, UK) sensor and the Aquaflex (Streat Instruments., Ltd, Christchurch, NZ) sensor were compared with TDR using both vertically and horizontally installed sensors. It was found that the CS616 manufacturer's standard calibration needed to be linearly transformed, y = 0.59x + 0.01 (m3 m–3), to obtain accurate measurements in a sandy soil with horizontally installed probes. In two different soils the standard calibration performed better, and smaller corrections were found, y = 0.87x – 0.01 for a sandy loam and y = 0.96x – 0.001 (m3 m–3) for a sandy loam with a larger clay content, respectively. The CS616 sensor was most likely affected by electrical conductivity at 1.6 dS m–1 in the soil solution when measuring in drip-fertigated potatoes (Solanum tuberosum L.). In the period of fertigation, the sensor overestimated the soil water content in the sandy soil by 0.01 to 0.11 m3 m–3 compared with TDR measurements. The dynamic response of the vertically installed sensor to changes in soil water content was shown to be good, and the sensor may be useful for assessing threshold values in water content for the start and end of irrigation. The performance of the Aquaflex sensor was investigated in the sandy soil only, and the sensor was found to reflect the dynamics of soil water content well. However, the manufacturer's standard calibration underestimated the soil water content in the order of 0.10 m3 m–3 and even showed negative values.

Abbreviations: EC, electrical conductivity • TDR, time domain reflectometry • TDT, time domain transmission • VWC, volumetric water content


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 PERSPECTIVES
 REFERENCES
 
SINCE the introduction of TDR for nondestructive determination of soil water content (Topp et al., 1980), many publications have shown the method to be useful in various applications (e.g., Topp et al., 1982; Topp and Davis, 1985; Wraith and Baker, 1991; Baker and Spaans, 1994; Plauborg, 1995). Jacobsen and Schjønning (1993) checked the Topp calibration (Topp et al., 1980) for Danish soils using measurements taken with the Trace system I, model 6050XI, Soil Moisture Equipment Corp. (Santa Barbara, CA), and found that the Topp calibration is adequate for water contents below 0.18 m3 m–3 volumetric water content (VWC), but that it tends to overestimate slightly at larger water contents. Jacobsen and Schjønning (1993) stressed that when using the Soil Moisture system it is not possible to check how the TDR traces are interpreted. Despite wide acceptance of TDR, it is still thought to be too expensive for operational use by farmers (e.g., for the timing of irrigation). Other similar techniques, where soil water content is assessed from measures of changes in the medium dielectric properties, seem more attractive in terms of costs per sensor when only a few sensors are needed, as is the case with capacitance methods such as the ThetaProbe (Delta-T Devices, UK) and ECH2O-probe (Decagon Devices, Inc., Pullman, WA, USA), multiple reflection of electromagnetic pulses such as the CS615 Water Content Reflectometer (Campbell Scientific), and time domain transmission (TDT) as used in the Aquaflex sensor. However, some of these sensors tend to have lower accuracy than TDR. Walker et al. (2004) found the CS615 to give VWC greater than soil porosity when using the manufacturer's standard calibration. The Aquaflex sensor showed a good response over time, but overestimated soil water content by up to 0.30 m3 m–3 VWC when the factory calibration for a silt loam was used (Leib et al., 2003). As technology advances, improvements will continuously be brought to the market, making it relevant from time to time to investigate new or improved methods to obtain more information than is available from the technical specifications. The CS615 has recently been replaced by an improved version, the CS616 (Campbell Scientific, 2002). In this study we investigated the manufacturer's standard calibration and sensor-to-sensor variability of the CS616 sensor, the performance of the Aquaflex sensor, and the usefulness of the sensors in drip irrigated potatoes.


   MATERIALS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS AND DISCUSSION
 CONCLUSIONS
 PERSPECTIVES
 REFERENCES
 
In this study TDR, which is based on relating changes in permittivity of the soil to changes in soil water content, is used as the reference method for comparisons with CS616 and Aquaflex measurements. The TDR setup included both manual and automatic readings with the use of the Tektronix 1502C cable tester (Textronic Inc., Beaverton, OR, USA). Manual readings were taken with both vertically and horizontally installed two-rod TDR probes. The vertically installed probes had a rod length of 30 cm, rod separation of 5 cm, and a rod diameter of 6 mm. The horizontally installed probes had a rod length of 30 cm, rod separation distance of 5 cm, and rod diameter of 4 mm. Details of the probe design and wiring of the 1:4 balun were given by Thomsen (1994). The automatic readings were sampled with an improved version of the original PC-based multiplexing system (Thomsen and Thomsen, 1994) using horizontally installed TDR three-rod probes (Zegelin et al., 1989) with a length of 30 cm, rod separation distance of 2.5 cm, and rod diameter of 4 mm. The widely accepted Topp calibration (Topp et al., 1980) was used to calculate volumetric soil water content from the TDR measurements. The interpretation of traces from both the two-rod and three-rod probe was performed using the mathematical software developed by Thomsen (1994). Despite the findings of Jacobsen and Schjønning (1993), the Topp calibration was used for the Danish soils included in the present study. We think the suggested improvement of the Topp calibration for these soils (Jacobsen and Schjønning, 1993) may be questionable because it was based on measurements with the Soil Moisture system, in which trace interpretation is not accessible to the user.

The CS616 sensor makes use of multiple reflections of high frequency pulses traveling back and forth along a 30-cm two-rod sensor. The specifications of the CS616 (Campbell Scientific, 2002) state the sensor-to-sensor variability to be ±0.005 m3 m–3 VWC in dry soil and up to ±0.015 m3 m–3 in saturated soil. The precision is listed as 0.0005 m3 m–3 VWC and the accuracy as ±0.025 m3 m–3 VWC when using the factory standard, nonlinear calibration. According to Campbell Scientific (2002), this calibration should provide accurate VWC in mineral soils with bulk electrical conductivity (EC) <0.5 dS m–1, bulk density <1.55 g cm–3, and clay content <30%. In our study, readings from the CS616 sensors were sampled using a CR10X data logger (Campbell Scientific, Ltd.).

The Aquaflex sensor is based on TDT and is available both as a self-sustained system (built-in data logger) and as a system that requires control from an external data logger. The sensor is supplied with calibrations for two groups of soil texture (one group including sand, sandy loam, loam, silt, and silt loam and another group including clay and clay loam) and is designed to measure the average soil water content along a 300-cm-long flat ribbon cable. The sensor responds to the dielectric properties of the surrounding soil in a volume of about 6 L. Streat Instruments lists the precision and accuracy as ±0.005 m3 m–3 and ±0.02 m3 m–3 VWC, respectively. Both types of sensors were tested in the present study, and the CR10X data logger was used as the external data logger for one of them.

Description of the Studied Sites
Investigations were performed at the semi-field facilities at Research Centre Foulum (56°30' N, 9°34' E) and in the field at Jyndevad Research Station (54°9' N, 9°12' E). The semi-field facilities consist of lysimeters (1 m2) and basins (4.32 m2), both 1.5 m in depth. Three different soils were used for the experiments at the semi-field facilities: Jyndevad soil (Orthic Haplohumod), Foulum soil (Typic Hapludult), and Rønhave soil (Typic Agrudalf). The three soils were packed to natural layering and bulk density in 1994 and have been planted with common agricultural crops in a crop rotation every year. At Jyndevad Research Station, experimental plots (480 m2) were used for the investigations. The physical properties of the studied soils can be seen in Table 1.


View this table:
[in this window]
[in a new window]
  		Table 1. Physical properties of the studied soils.

 
In 2003, the semi-field facilities at Research Centre Foulum (lysimeters and basins) were planted with oat (Avena sativa L.) on 7 April. The crop emerged 17 April and was fertilized 23 April with 154 kg mineral N ha–1 (1-5-12 NPK) and 3 June with 60 kg mineral N ha–1 (potassium ammonium phosphate). The crop was harvested 25 August. During the experiment, the soils in the semi-field facilities were exposed to natural rainfall, but an additional irrigation of 30 mm was given to the soils on 9 October. Precipitation was measured hourly at the meteorological station at Foulum located about 500 m from the site.

In 2004 the basins with Foulum and Jyndevad soils in the semi-field experiments at Research Centre Foulum were planted with potatoes. The experiments were established 22 April by planting potatoes at a depth of 8 cm with a density of 40000 plants ha–1, a row distance of 0.75 m and a distance between plants of 0.33 m. Stages of crop development are given in Table 2. Before planting potatoes, the soils were fertilized with approximately 55 kg N ha–1 (20 Mg pig slurry ha–1). The soils were ridged in a two-step process, first to a height of about 25 cm, measured from top to bottom of the ridge. Drip lines (NetaFim RAM, Tel Aviv, Israel; 1.2 L h–1) were placed centrally on the ridge and a final ridging with 10 cm soil then covered the drip lines. In this process of ridging, the pig slurry was distributed evenly in the ridge above the potato tuber. The drip lines were pressure compensated, thus maintaining a constant discharge over a wide range of pressures. The distance between drip points was 25 cm for the Jyndevad soil and 50 cm for the Foulum soil. The timing and amount of irrigation via the drip lines were based on thresholds of plant available soil water as calculated with a simple water balance model. Two treatments with different levels of irrigation (Jyndevad W1 and W2) were established in the Jyndevad basins. Jyndevad W1 received the full amount of irrigation, whereas Jyndevad W2 received only 80% of the amount of water proposed by the water balance model. The Foulum basin (Foulum W1) received the full amount of irrigation proposed by the model. Additional N ({approx}81 kg N ha–1) was given through the drip lines (fertigation) using a simple schedule. Nearly one-half the amount was given 1 wk after crop emergence; the rest was given in equal amounts 4 to 9 wk after emergence (Tables 2 and 3). An automatic roof was used to exclude natural precipitation. Instead, an overhead irrigation system was used to supply the basins with 12 mm of artificial precipitation per week.


View this table:
[in this window]
[in a new window]
  		Table 2. Crop development and fertigation of potatoes at the studied sites during the period of measurement.

 

View this table:
[in this window]
[in a new window]
  		Table 3. Description of the studied sites.

 
At Jyndevad Research Station in 2004, potatoes were planted 19 April according to the same technique as described above for the 2004 semi-field experiment at Research Centre Foulum. Distance between drip points was 25 cm, and the simple water balance model was used to control the time and amount of irrigation. The N need was estimated from a C/N model (still under development), which resulted in 11 N-fertigated applications, giving a total of approximately 71 kg mineral N ha–1 (Tables 2 and 3). The soil was exposed to natural precipitation that was measured hourly at the meteorological station at Jyndevad located about 300 m from the plot.

The study of soil moisture sensors in the present work was performed in two steps. First the CS616 sensor-to-sensor variability, response, and calibration were tested. Subsequently, the performances of the CS616 sensor and the Aquaflex sensor were compared with the TDR sensor.

CS616 Sensor-to-Sensor Variability, Response, and Calibrations
The tests were performed at the semi-field facilities at Research Centre Foulum in 2003. To study the CS616 sensor-to-sensor variability and sensor response, five sensors were installed vertically from the soil surface on 20 April in one Jyndevad lysimeter (Table 4). Sensors were separated with a distance of 20 cm from each other; the output from the sensors was recorded hourly.


View this table:
[in this window]
[in a new window]
  		Table 4. Installation of sensors at the studied sites.

 
To test the factory standard calibration of the CS616 sensor, three CS616 and three two-rod TDR sensors were installed horizontally in each of the three soils (Jyndevad, Foulum, and Rønhave) on 25 September in the basins (Table 4). In the three soils, the three CS616 and the three TDR sensors were horizontally pushed, using a spirit level to ensure horizontal alignment, into the soil wall from two narrow trenches (one for CS616 and one for TDR) at the 15-cm depth. The distance between each sensor was 15 cm, and the distance between the two trenches was 30 cm. Outputs from the TDR sensors were measured manually two to three times each week, whereas outputs from the CS616 sensors were recorded hourly. At the time of installation the soils were covered with stubble from the oat crop.

Soil Water Content Using CS616, Aquaflex, and TDR Sensors in Drip and Fertigated Potatoes
The investigation was performed in 2004, both in the semi-field facility at Research Centre Foulum (Jyndevad and Foulum soils) and in the field plot at Jyndevad Research Station (Jyndevad soil). The soils were established with potatoes and drip lines as described above.

On 21 April at Jyndevad Research Station, three CS616 sensors and one Aquaflex sensor, which was connected to the CR10X, were installed horizontally in the middle of the same potato ridge 17 cm from the top of the ridge and parallel to the row (Table 4). Two CS616 sensors were centered above a mother tuber. The third sensor was centered between two mother tubers (Fig. 1a) . On 12 May, two two-rod TDR sensors were installed at the same depth parallel to the row, and two rows away, each above a mother tuber (Fig. 1a). At installation, the sensors were pushed into the soil wall from a narrow trench using a spirit level to ensure horizontal alignment. Half the soil (to the middle of the ridge) was removed to install the Aquaflex 3-m ribbon centrally in the ridge about 2 m from the CS616 sensors. The removed soil was repacked against the ribbon to ensure good soil contact. The CS616 and the Aquaflex sensors were recorded every 10 min, whereas the TDR sensors were recorded manually two to three times a week.



View larger version (17K):
[in this window]
[in a new window]
  		Fig. 1. Installation of TDR and CS616 sensors in drip irrigated potatoes. In (a) experimental plots at Jyndevad Research Station, and (b) Jyndevad soil and (c) Foulum soil in the semi-field basins at Research Centre Foulum. Position of mother tubers and drip points is shown as well.

 
In the semi-field facility at Research Centre Foulum, two Jyndevad basins and one Foulum basin were instrumented with TDR and CS616 sensors (Table 4). On 2 June in the two Jyndevad basins (Jyndevad W1 and W2), a CS616 sensor was installed in each basin in the middle of the ridge, parallel to the row, 17 cm from the top of the ridge and immediately above a mother tuber (Fig. 1b). In the Foulum basin (Foulum W1) on the same date the CS616 sensor was installed 25 cm from the top of the potato ridge perpendicular to the row centered between two drip points (Fig. 1c). Earlier, on 10 May in the potato ridge next to the ridge used for installation of the CS616 sensor, the same number of TDR sensors was installed at corresponding positions (Fig. 1b and 1c). The Aquaflex sensor was installed in Jyndevad W2 using the same installation technique and position as at Jyndevad Research Station (described above). However, because of the length of the Aquaflex ribbon, two adjacent basins receiving the same amount of irrigation and fertigation were used for the installation. The logging interval of the CS616 sensor and the Aquaflex sensor was 10 min. The logging interval of the TDR sensors was 30 min. Since the PC-based TDR multiplexing system took approximately 25 min to read the TDR sensors during a single reading period, a moving average of 3 h of the readings of the three types of sensors was calculated to make a comparison possible.


   RESULTS AND DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
CONCLUSIONS
 PERSPECTIVES
 REFERENCES
 
CS616 Sensor-to-Sensor Variability, Response, and Calibrations
Figure 2 shows the results of the vertically installed sensors in the Jyndevad soil in the lysimeter at Research Centre Foulum. Also shown in the figure is the precipitation during the period of measurement. The five CS616 sensors showed the same response, but the differences between sensors were generally higher than the sensor-to-sensor variability stated by Campbell Scientific (2002). The maximum differences among sensors were 0.04 m3 m–3 VWC. These differences were especially pronounced in the first weeks of measurements. Later, a decrease in the differences was found as soil water content decreased. The decrease in difference may be due to the stated relation between sensor-to-sensor variability and water content (Campbell Scientific, 2002), but may also be caused by reduced variations in water flow. In general, the sensors reacted well to changes in soil water content caused by the precipitation events when using the factory standard calibration. However, further studies are needed to check if vertically installed sensors (measuring across both gradients in soil temperature and soil water content) can accurately measure the absolute changes in soil water content.



View larger version (22K):
[in this window]
[in a new window]
  		Fig. 2. Measurements of volumetric water content from vertically installed CS616 sensors in the Jyndevad soil in the lysimeter at Research Centre Foulum. FC, field capacity (50 hPa); WP, wilting point (15500 hPa).

 
Figures 3a through 3c show data of VWC measured with the CS616 sensor (factory standard calibration) and TDR sensors installed horizontally at the 15-cm depth in the three different soils in the basins at Research Centre Foulum. The factory standard calibration of the CS616 sensor showed variable success in the different soils. A poor relation seems to exist between the TDR and CS616 sensor for the measurements in the Jyndevad soil (Fig. 3a), where the CS616 sensor underestimated the soil water content compared with the TDR sensors. For the Foulum soil (Fig. 3b) the agreement between the two sensor types seems to be slightly better. Large variations among sensors, however, were found for the CS616 sensor. This may be explained by a considerable small-scale heterogeneity of the topsoil. The results from the CS616 sensor measuring the highest water content matched the soil water contents measured by the TDR sensors well, even though the distance to the TDR sensors was more than 30 cm. For the Rønhave soil (Fig. 3c) the agreement between the two sensor types was good even though both types showed a relatively large intersensor variability. This large variability may also be explained by the small-scale heterogeneity of the topsoil.



View larger version (28K):
[in this window]
[in a new window]
  		Fig. 3. Measurements of volumetric water content from horizontally sensors (TDR and CS616) installed in a depth of 15 cm in the lysimeter at Research Centre Foulum: (a) Jyndevad soil, (b) Foulum soil, (c) Rønhave soil.

 
In general, the factory standard calibration of the CS616 sensor in this experiment was not satisfactory for soil containing a large amount of sand, and a user-derived calibration for this type of soil would be appropriate. The data (Fig. 3a–3c) from the TDR and CS616 sensors installed in the Jyndevad, Foulum, and Rønhave soil in the lysimeters were then used to derive new calibrations for the CS616 sensors (Fig. 4a–4c) . Although the range of soil water content was not satisfactory, especially for the Foulum and Rønhave soils, for the three soils there was a linear relation between water content measured by TDR and CS616. The regressions showed that for the Jyndevad and Foulum soils (Fig. 4a and 4b) a new calibration is needed to accurately measure the soil water content, whereas for the Rønhave soil (Fig. 4c) a new calibration is not needed. For all three soils, a more straightforward test of the calibration would have been to compare CS616 measurements with gravimetric soil water content sampled close to the sensors in the range wilting point to field capacity. However, we suggest that the procedure used in the present study may be adequate, even though the sensors were separated about 30 cm. Also, this type of calibration is nondestructive, allowing for multiple-point calibration, as was also pointed out by Chandler et al. (2004).



View larger version (14K):
[in this window]
[in a new window]
  		Fig. 4. New calibration of the CS616 sensor for (a) Jyndevad soil, (b) Foulum soil, (c) Rønhave soil. Mean values (n = 3) of VWC measured by the TDR sensors are plotted against mean values (n = 3) of VWC measured by the CS616 sensors. Error bars indicate ± 1 standard deviation.

 
Factors other than soil water content may affect the CS616 calibration. However, for the soils used in our study, Jacobsen and Schjønning (1993) found that the TDR measurements were only slightly affected by bulk density and texture. Persson and Berndtsson (1998) found minor effects of soil temperature and EC on TDR measurements. They found the correction factor –0.00269 m3 m–3 °C–1 for sandy soils, indicating a decrease in water content with increasing temperature. For finer textured soils this effect may be positive or negative depending on the bulk EC. Campbell Scientific (2002) discussed the same effects on the CS616 sensor and found a temperature correction factor of about –0.001 m3 m–3 °C–1 at a VWC of 0.30 m3 m–3 and –0.0001 m3 m–3 °C–1 at 0.12 m3 m–3 VWC, which more or less corresponds to the small temperature effects for the TDR sensor. We did not confirm these values for our soils. In addition, Campbell presented a correction of approximately –0.03 m3 m–3 for an increase of bulk density from 1.4 to 1.6 g cm–3. The new calibrations we developed in this study were probably not affected by these additional factors, as the effects of electric conductivity and soil temperature were assumed negligible at the time when measurements were taken. In the autumn after harvest, temperature fluctuations and the nutrition level in soil were low. In addition, the bulk densities of the soils (Table 1) were close to the value of 1.4 g cm–3 used for the CS616 standard calibration (Campbell Scientific, 2002).

Soil Water Content in Drip Irrigated and Fertigated Potatoes
Measurements with CS616
Figure 5 shows the VWC measured with the CS616 sensor using the manufacturer's standard calibration and the VWC measured with the TDR sensor (left y axis). Also shown are differences from TDR values. Differences were calculated using mean values of the CS616 and TDR measurements (right y axis). The differences are shown for both the manufacturer's standard calibration and the new calibration. For April and May, the standard calibration shows an underestimation of the soil water content for all measurements, whereas the new calibration (shown in Fig. 4a) performed well during the same period. The overestimation in the rest of the growing season (the period of growth of new potato tubers) using the new calibration may most likely be explained by a stronger effect of an increased bulk electric conductivity on the CS616 than on the TDR measurements. The increase in bulk EC was caused by the level of fertigation. As an example, the application of 11.8 kg mineral N ha–1 on 12 July (Table 2) in a mix of ammonium nitrate and calcium ammonium nitrate diluted in tap water corresponding to 2.7 mm water, gives an EC of the irrigation solution equal to 1.6 dS m–1. According to Campbell Scientific (2002), a soil solution EC above 2 dS m–1 could affect the CS616 reading, leading to an overestimation of the measured VWC. Campbell Scientific (2002) presented calibration equations for soils with bulk electric conductivities of 0.4 and 0.75 dS m–1. From these equations one can calculate the overestimation as 0.02 and 0.05 m3 m–3 at soil water contents of 0.2 and 0.3 m3 m–3, respectively, if the high bulk electric conductivity was not accounted for. Similarly, Miyamoto and Maruyama (2004) found a large effect in the signal of the CS615 sensor when measuring in a heavily fertilized paddy field.



View larger version (22K):
[in this window]
[in a new window]
  		Fig. 5. Volumetric water content measured with the CS616 sensor using the standard calibration and VWC measured with TDR at Jyndevad Research Station (left y axis). Differences between TDR calculated from mean values of the CS616 and TDR measurements are shown at the top of the figure (right y axis). The differences are shown for both the manufacturer's standard calibration and the new calibration. FC, field capacity (50 hPa); WP, wilting point (15500 hPa). For precipitation and irrigation amounts, see Fig. 8.

 


View larger version (19K):
[in this window]
[in a new window]
  		Fig. 8. Measured volumetric water content by the Aquaflex sensor and TDR sensors installed in Jyndevad soil at Jyndevad Experimental Station. FC, field capacity (50 hPa); WP, wilting point (15500 hPa). The right y axis shows both precipitation and irrigation.

 
In addition to the effects of high EC on the CS616 measurements, some additional changes in the soil might have caused a stronger effect on the CS616 measurements than on the TDR measurements, such as the possible changes in soil bulk density caused by the enlargement of the potato tubers in the growing period of the new potato tubers and changes in water content of the medium caused by the increasing water content inside the growing potato tubers. After ridging, the unstructured sandy Jyndevad soil is expected to return to its natural bulk density (1.2–1.4 g cm–3). However, when the potatoes started to develop and grow around 5 June (Table 2), the soil bulk density of the soil is expected to increase due to the volumetric increase of the tubers. Because the growth of tubers did take place at the position of the soil moisture sensors it caused the potatoes to be close to and in contact with the sensors at the end of the growing season (Fig. 6) .



View larger version (124K):
[in this window]
[in a new window]
  		Fig. 6. Photos of CS616 sensors centered above a mother tuber at the time of potato maturity in the Jyndevad soil at Jyndevad Research Station.

 
Figures 7a and 7b show the comparison between the TDR sensor and the new calibration of the CS616 sensor using data from the Jyndevad basin in the semi-field facility at Research Centre Foulum. Figure 7a shows data from the basin receiving the maximum amount of irrigation proposed by the irrigation model (Jyndevad W1) and Fig. 7b shows data from the basin receiving 80% of the irrigation proposed by the irrigation model (Jyndevad W2). In general, both the TDR sensor and the CS616 sensor seemed to cover the water dynamics well. Sharp increases (peaks) in the soil water content are seen after each irrigation event, and less sharp increases are seen after each precipitation event. Compared with the TDR sensors, the new calibrated CS616 sensors predicted the soil water content well during the first half of the measurement period, after which the CS616 sensor started to overpredict the water content compared with the TDR sensors. The CS616 sensor installed in Jyndevad W2 (Fig. 7b) generally overpredicted the VWC more than the sensors installed in Jyndevad W1 (Fig. 7a), especially in the first half of the measurement period. Thereafter this trend was reversed, and an overprediction by the CS616 sensor was now most pronounced in Jyndevad W1. On average, the CS616 sensors overpredicted the VWC by approximately 0.01 m3 m–3 in Jyndevad W1 and by 0.03 m3 m–3 in Jyndevad W2 compared with the TDR sensors. The relatively high overprediction by the CS616 in Jyndevad W2 is probably related to the high nitrate levels resulting from the application of 30 kg N ha–1 shortly after emergence (Fig. 7, Table 2). The results imply that measurements with the CS616 sensor need to be further adjusted when measuring in a soil with a high EC (Campbell Scientific, 2002). Since the sensors in the Jyndevad basin were installed immediately below the drip points, they were highly affected by the application of fertilizer leading to a high EC in the soil. After the first fertigation, the differences between the CS616 and TDR decreased steadily until the end of June. For Jyndevad W1, VWC measured by the CS616 sensor was for a time even slightly lower than the TDR measurements. From the beginning of July the difference between the two types of sensors started to increase again for Jyndevad W1. Especially during irrigation events, large differences were seen between the two types of sensors (up to 0.11 m3 m–3). Between irrigation events, differences between the two types of sensors were lower ({approx}0.03 m3 m–3). For Jyndevad W2, relatively large differences between the two types of sensors were seen already from the beginning of July, especially during irrigation events (up to 0.08 m3 m–3). The resumed differences between the two types of sensors are probably explained by increased soil EC as a result of fertigation starting up again in the end of June. The reason why the difference between the two types was most pronounced in Jyndevad W2 in the first half of the measuring period is probably because this was the basin receiving the smallest amount of irrigation water, leading to a higher concentration of salts in the soil compared with Jyndevad W1. The larger difference between the two types of sensors at the end of July for the Jyndevad W1 compared with W2 are probably explained by a continuously higher degree of N transport in Jyndevad W1 from the drip lines to the depth where the sensor was installed. This led to a higher VWC measured by the CS616 sensor. In general, VWC measured by TDR and CS616 in both the Jyndevad W1 and W2 basins was between the wilting point (15500 hPa) and field capacity (50 hPa). Only when the soil was irrigated did the VWC measured by TDR increase to values slightly above field capacity, but VWC dropped immediately, to values below field capacity, when irrigation stopped. After irrigation, VWC measured by the CS616 censor increased to values markedly above field capacity, but dropped to below field capacity <24 h after irrigation stopped. Neither the CS616 nor the TDR sensors showed readings of VWC above the porosity of the soil (Table 1).



View larger version (30K):
[in this window]
[in a new window]
  		Fig. 7. Comparison between TDR sensor and the calibrated CS616 sensor from the basins in the semi-field facility at Research Centre Foulum: (a) Jyndevad (Jyndevad W1), (b) Jyndevad (Jyndevad W2), (c) Foulum (Foulum W1). FC, field capacity (50 and 100 hPa in Jyndevad soil and Foulum soil, respectively); WP, wilting point (15500 hPa). Black dots indicate dates of fertigation (Table 2).

 
Figure 7c shows the comparison between the TDR sensor and the new calibration of the CS616 sensor using data from the Foulum basin. Increases in the soil water content caused by irrigation events are less clear compared with the measurements in the Jyndevad basins (Fig. 7a and 7b). This is explained by the perpendicular installation of the sensors between the drip points, which then caused the sensors to be less affected by the water leaving the drip points. Also the lower depth of installation and the higher transport time of water due to the higher content of clay explain the less dynamic behavior of VWC during the period of measurement in the Foulum basin. The less significant decrease in water content at the start of growth of the potato tubers is probably explained by the fact that, unlike in the Jyndevad basin, sensors were not centered above a mother tuber. In general, the agreement between the two sensors was good. At the beginning of the period of measurement, the CS616 sensor largely overestimated the soil water content compared with the TDR sensor. This was probably caused by soil disturbance in connection with the installation of the CS616 sensor. In the first part of July, the CS616 sensor started to overpredict the water content slightly, with this overprediction increasing toward the end of the period of measurement where the water content was overpredicted by up to 0.04 m3 m–3. This overprediction may have been caused by variability in water flow caused by the increasing volume of potato tubers, but can also be explained by a higher degree of N transport from the drip lines to the depths where the sensors were installed. The overprediction of the CS616 sensor started at the time where a general increase in the VWC of the soil was observed, as measured by both types of sensors. This general increase is related to the start of the period when drip irrigation intensity was increased. The better performance of the CS616 sensor in the Foulum soil is probably also caused by the difference in the installation of the sensors compared with the sensors in the Jyndevad soil. In this position, the fertilizer from the drip points influenced the sensor less. The VWC as measured both by TDR and the CS616 sensor in the Foulum W1 basin was between the wilting point (15500 hPa) and field capacity (100 hPa) during the whole period of measurement. It was only in the beginning of the period that the values of VWC measured by the CS616 sensor were above field capacity.

Measurements with Aquaflex
Figure 8 shows the measured VWC by the Aquaflex sensor and the two TDR sensors in the Jyndevad soil at Jyndevad Research Station. The plot shows a poor correspondence between the self-sustained Aquaflex sensor and TDR. The Aquaflex sensor underestimated the water content on the order of 0.1 m3 m–3 VWC and even showed negative values. Also the Aquaflex sensor controlled by an external data logger and installed in the Jyndevad basins at the semi-field facility at Research Centre Foulum showed a poor agreement with the TDR sensor (Fig. 9) . This Aquaflex sensor also showed negative values and underestimated the water content up to 0.09 m3 m–3 VWC. During the period of measurement there seemed to be a tendency for the Aquaflex sensor to measure decreasingly lower water contents. A unique calibration curve seemed therefore not to exist between the two types of sensors in the current study. Effects of EC and soil temperature on the measurements with the Aquaflex sensor may have caused this. According to Streat Instruments, the measurements of VWC with the self-sustained Aquaflex sensor are automatically corrected for these effects, but Streat Instruments do not provide guidance on how to correct these effects for the sensor controlled by an external data logger. The possible effects on TDR measurements of the increasing water content associated with the growth of new potato were not investigated in the present study, but it may possibly also add to the bias. Both TDR sensors were placed centrally in the potato root systems where also the new potato tubers develop, whereas the Aquaflex sensor sensed quite a lot of soil volume not affected by the new potato tubers. However, further investigations will be needed to explore these effects.



View larger version (31K):
[in this window]
[in a new window]
  		Fig. 9. Comparison of volumetric water content between TDR sensor and Aquaflex sensor controlled by the user supplied data logger from the Jyndevad basin (Jyndevad W2) in the semi-field facility at Research Centre Foulum.

 

   CONCLUSIONS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
PERSPECTIVES
 REFERENCES
 
The vertically installed CS616 sensors responded well to changes in soil water content caused by the precipitation events when using the factory standard calibration. Differences in VWC among five sensors were up to 0.04 m3 m–3 and were higher than stated by the manufacturer. When measuring with horizontally installed CS616 sensors in a sandy and a sandy loam soil the factory calibration was incorrect and a local calibration was needed to obtain an accurate reading of soil water content. However, the manufacturer's standard calibration performed well in a sandy clay loam soil. The results also showed, in agreement with the specification of the sensor, that the manufacturer's standard calibration was not applicable in soils having a high EC, which in our study was caused by the level of fertigation in drip irrigated potatoes. Possible effects of bulk density and growing potato tubers on the output from the CS616 sensor were less clear. Therefore, it may be concluded that for use in drip irrigated and fertigated potatoes an additional correction is needed to take into account effects of high EC. The Aquaflex sensor, both the self-sustained version and the version to be mounted on a user-supplied data logger, responded to changes in soil water content. However, the manufacturer's calibration underpredicted the soil water content in sandy soil on the order of 0.10 m3 m–3 VWC and even showed negative VWC values.


   PERSPECTIVES
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 PERSPECTIVES
REFERENCES
 
The use of a CS616 sensor measurement in irrigation scheduling looks promising, as the sensor seems to be useful for assessing threshold values of water content for start and end of irrigation. In the case of potato growing, the best position would be central in the rooting system (e.g., 5 cm above the mother tuber). However some interference may then be expected if several new, growing potato tubers touch the sensors rods. In some soils a placement below the mother tuber may also be useful, especially later in the growing period when the root system has developed below the mother tuber. However, if more accurate soil water content readings are needed, recalibrations are necessary for sandy and some sandy loam soils. Additional studies are needed to determine if different calibrations may be needed for horizontally and vertically installed sensors. In soils with high EC, an additional correction needs to be assessed and accounted for in the calibration, or it may be relevant to coat the sensors as proposed by Miyamoto and Maruyama (2004).

A more detailed study on the Aquaflex sensor using a higher number of TDR sensors in a row to sense approximately the same soil volume will be needed to assess local calibrations for various soils. In addition, the manufacturer should be urged to provide information on how to correct measurements with the Aquaflex sensor controlled by an external data logger for effects of EC and soil temperature.


   ACKNOWLEDGMENTS
 
This work was financed jointly by the FertOrgaNic (Improved organic fertiliser management for high nitrogen and water use efficiency and reduced pollution in crop systems), EU 5th Framework RTD project (QLK5-2002-01799) and The Danish Institute of Agricultural Sciences. We wish to thank Finn Christensen, Kaj Eskesen, Ove Edlefsen, and Chris Nielsen for technical assistance.


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





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 ISI Web of Science (1)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Plauborg, F.
Right arrow Articles by Lærke, P. E.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Plauborg, F.
Right arrow Articles by Lærke, P. E.
GeoRef
Right arrow GeoRef Citation
Agricola
Right arrow Articles by Plauborg, F.
Right arrow Articles by Lærke, P. E.
Related Collections
Right arrow Vadose Zone Processes and Chemical Transport
Right arrow Water Content
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