Published online 16 May 2007
Published in Soil Sci Soc Am J 71:993-1002 (2007)
DOI: 10.2136/sssaj2006.0282
© 2007 Soil Science Society of America
677 S. Segoe Rd., Madison, WI 53711 USA
SOIL & WATER MANAGEMENT & CONSERVATION
Performance of a Capacitance-Type Soil Water Probe in a Well-Drained Sandy Soil
W. M. Bandaranayakea,*,
L. R. Parsonsa,
M. S. Borhanb and
J. D. Holetona
a Univ. of Florida, IFAS, Citrus Research and Education Center, 700 Experiment Station Rd., Lake Alfred, FL 33850
b Greenhouse and Processing Crop Res. Center, Agriculture and Agrifood Canada, Harrow, ON N0R 1G0, Canada
* Corresponding author (Wijeb{at}ufl.edu).
 |
ABSTRACT
|
|---|
Most soils in the Central Florida Ridge (CFR) area are Entisols that contain >95% sand, <3% clay, and <2% organic matter. Field capacity (
fc) is commonly 
0.08 m3 m–3. Therefore, accurate estimation of soil water content (v) is important in these soils. The objective of this study was to evaluate the performance of ECH2O probes when estimating v for scheduling irrigation in CFR soils. Probes were tested for (i) probe-to-probe output variability, (ii) soil volume sampled, (iii) sensitivity to salinity, temperature, and air pockets close to the sensor surface, (iv) pockets of very dry soil close to the sensor surface, and (v) performance after installation in the field. According to the calibration, a 1% change in water content corresponds to a probe output of 17 mV. Laboratory testing suggested that output variability from probe to probe can be a problem in these soils. The sampling volume of the probe was within 1.5 cm from either side of the sensor surface. Salinity induced during fertigation increased the output by about 200 mV, and for each 1°C drop in temperature, the sensor output dropped by 2.3 mV. When the bulk density was changed from 1.56 to 0.94 Mg m–3, the output decreased by 3.5 MV for each 1% drop in air-filled porosity. When very dry soil lenses with <0.01 m3 m–3 v were associated with the probe surface, the probe failed to sense the wet soil even 1 cm away from the sensor surface. Sensor failure was common due to water leaking into the circuit when sealing material deteriorated or casing material was damaged by insects. These issues need to be addressed before the probes can be considered reliable to estimate v or used in automated irrigation.
Abbreviations: CFR, Central Florida Ridge
|
INTRODUCTION
|
|---|
There are >251500 ha of citrus in Florida, and much of it is grown on sandy soils. Rainfall is typically low during the spring flowering period. Water stress at flowering can reduce yield, so irrigation is important. Well-drained, sandy soils extend north and south through the central part of the Florida peninsula called the Central Florida Ridge. A typical soil in the CFR is Candler fine sand (hyperthermic, uncoated Typic Quartzipsamment). Obreza et al. (1997) found that the soil water characteristics (v vs. matric potential [hm]) of Candler soil measured at two different sites were similar. According to them, the v values at 5 kPa hm (fc) and at 1500 kPa hm (v at the permanent wilting point [wp]) were 0.085 and 0.02 m3 m–3, respectively (Morgan et al., 2001). Therefore, there was only 0.065 m3 m–3 of plant-available water (AW).
To measure v successfully with this small amount of AW, reliable and precise equipment is needed. For example, a reduction in soil water content as small as 0.01 m3 m–3 represents a depletion in AW of 15% (Morgan et al., 1999). Frequent irrigations in this sandy soil can prevent water deficits during the critical flowering and fruit-set times, but could lead to unnecessary overirrigation. Overirrigation not only adds to production costs, but also wastes water and potentially pollutes groundwater. Depending on tree size, time of year, and tree spacing, mature trees can use from 30 to 300 L (8–80 gallons) of water per day (Parsons and Morgan, 2004). Because of this wide variation in crop water requirement, it is important to monitor soil water depletion for appropriate irrigation scheduling to avoid over- or underirrigation. An accurate, low-cost probe could benefit growers by helping them improve irrigation practices. The ECH2O EC-20 probe (Decagon Devices, Pullman, WA) is a relatively low-cost soil moisture probe that may prove useful in improving irrigation management (Borhan et al., 2004).
To monitor soil water depletion reliably, soil water measuring equipment needs to provide accurate values with time. For a water content of 8% v/v, this probe would read between 7.8 and 8.2% depending on the accuracy of calibration and the type of soil (www.decagon.com/echo/calibration.html; verified 19 Feb. 2007). The ECH2O soil water probe measures the dielectric permittivity or capacitance of the surrounding soil medium (Kelleners et al., 2005), and the final output from the sensor is a millivolt value that can be converted to a volumetric water content using a calibration equation.
A common aspect of capacitance-type sensors is that once they are installed in the soil, they can be affected by air gaps, texture, temperature, and salinity (Baumhardt et al., 2000). The magnitude of such effects may depend on the design of the probe, too. The manufacturer indicates that the ECH2O probe has a comparatively low sensitivity to soil salinity and temperature. Soil should be carefully packed during installation to avoid air gaps. Problems due to soil variation can be avoided by calibrating the probe for each texture separately and subsequently using the calibration equation relevant to a particular texture.
In an experiment that evaluated a high frequency, low volume approach to irrigation control, Schroder et al. (2005) designed an automatic system including ECH2O sensors that determined the appropriate soil water level for starting irrigation. This system produced the most effective control compared with tensiometer-based, historical evapotranspiration (ET) data-based, or farmer-based systems. In another study, Norikane et al. (2005) compared ECH2O probes with a temperature and moisture acquisition system and tensiometers. They concluded that ECH2O sensors monitored evaporative loss from a soil medium better than the other systems. Although not specific, they implied the need for further studies with respect to ECH2O probes.
The ECH2O sensors need to be evaluated in the sandy soils of the CFR because of the high precision needed to detect soil water depletion in such a narrow range of available water. Therefore, the objectives of this study were to determine (i) probe-to-probe signal variability, (ii) response of the sensors to fertilizer-induced salinity, (iii) the soil volume sampled by the sensors, (iv) probe sensitivity to pockets of air or dry soil, (v) response to soil temperature variations, and (iv) performance of the probes after installation in the field.
|
MATERIALS AND METHODS
|
|---|
Laboratory Evaluation
Laboratory evaluations consisted of six tests to verify probe-to-probe variation and the effect of some field conditions on probe response including the following: (i) probe-to-probe signal variability, (ii) the effect of fertilizer-induced salinity, (iii) the area of influence to changes in soil water (sampling volume), (iv) the effect of a dry soil layer close to the sensor surface, (v) the effect of macropores (large voids) or micropores close to the sensor surface, and (vi) the effect of soil temperature. The correct functionality of each probe and the probe-to-probe variability (Test 1) was tested by measuring the millivolt reading of each individual probe in air and tap water separately. Test 2 was conducted using a soil column (25.7-cm diameter and 60-cm height) (Fig. 1a). Tests 3, 4, 5, and 6 were conducted using a plastic box shown in Fig. 1b that measured 28.5 by 10.5 by 5.0 cm. Test 5 was also conducted using water in 1-L cylinders at different temperatures.
View larger version (43K):
[in this window]
[in a new window]
|
Fig. 1. The ECH2O performance laboratory test setup: (a) laboratory setup using a soil column to evaluate fertigation and the salinity effect on probe performance; and (b) rectangular plastic box of 28 by 10 by 5 cm used in Laboratory Tests 3, 4, 5, and 6 and setup that changed the thickness and distance to the dry soil layer from the probe surface.
|
|
Test 1
The output values were measured in air and water for each probe from a group of 14 new probes. This was done during four consecutive days. To take the air value, a probe was hung freely in air and the readings were recorded every minute for 4 min using a hand-held ECH2O Check soil moisture monitor (Decagon Devices). Similarly, the probes were immersed in 1-L jars filled with tap water and readings were taken every minute for 4 min for each of the 14 probes.
Test 2
To evaluate the salinity effect from fertilizer on probe readings, four ECH2O probes were installed in a column packed with soil to a bulk density of 1.5 Mg m–3 (Fig. 1a). Two ECH2O probes were placed 12 cm and two 27 cm below the soil surface (15 cm below the first set of probes). During fertigation, fertilizer was mixed with irrigation water at the rate of 1 L of fertilizer to 4 L of irrigation water. Solution fertilizer (9:0:7.7 N–P–K) was sprayed on the soil surface at a rate equivalent to 1.018 L per tree (302 L ha–1) and diluted to the same concentration that reached the field soil surface. After applying the fertilizer solution to the soil, several pulses of water were sprayed on the surface from time to time to move the fertilizer close to the sensor. Once the fertilizer reached the sensor surface (as indicated by the readings of the top sensors), the water input was regulated to a constant 10.8 L h–1 flow rate. The probe output value was compared with a control treatment that was run without fertilizer. The leachate from the bottom of the soil column (Fig. 1a) was collected during the experiment and tested for the total dissolved solids using an Oakton TDSTestr 1 (Oakton Instruments, Vernon Hills, IL), which measures within a range of 0 to 1.99 g L–1.
Test 3
Soil sampling volume, or the volume of soil detected by probes during changes in v, was measured using a 28 by 10 by 5 cm rectangular plastic box shown in Fig. 1b. Two kilograms of air-dried soil was mixed thoroughly with 27 mL of tap water to bring the soil to 0.02 m3 m–3 v. This wet soil was packed in the plastic box to 1.5 Mg m–3 bulk density. After packing, an ECH2O probe was installed in the middle along the length of the box and connected to a CR-10X datalogger (Campbell Scientific, Logan, UT). The logger was programmed to measure the output from the ECH2O sensors. Three milliliters of fertilizer solution (9:0:7.47 N–P–K), the same concentration used in Test 2, was introduced evenly with a spray bottle along the length of the probe, 4 cm away from the sensor surface. Four milliliters of tap water was sprayed in the same area after the introduction of fertilizer to distribute the fertilizer and water in the soil. The same amount of fertilizer and water was applied with the same procedure on the other side of the sensor blade. This 14 mL of water and fertilizer on both sides of the sensor blade can bring the soil close to fc at 4 cm away from the sensor blade. During fertilizer and water introduction to soil at 4 cm away in the plastic box, the rest of the soil (which was at 0.02 m3 m–3 v in the box) was separated with a thin plastic sheet to prevent mixing. Data were recorded at 1-min intervals for 30 min. If there was no change in output, the fertilizer and water application was advanced by 0.5 cm toward the sensor blade. The procedure was repeated progressively on either side of the probe until a clear change in output was noticed. A fertilizer solution was used since the soluble salts could improve the output signal very sharply due to the salinity effect (Test 1). When the output first appeared to change, the distance from the sensor surface was noted and used to estimate the volume of soil sensitive to v changes (sampling volume). The measurements were repeated with three different sensors.
Test 4
This test was set up to evaluate probe response to a very dry soil layer (lens) located at different positions or distances from the probe. The plastic box was packed with air-dried soil (v 0.005 m3 m–3) and a probe was installed in the center of the box. Wet soil (v 0.08 m3 m–3) was introduced at decreasing distances along the length (Fig. 1b) and sensor output was measured for different thicknesses of dry soil associated with the sensor surface. The wet soil was then replaced with dry soil at different locations along the probe length to simulate dry soil layers at different depths when a probe is installed vertically close to the soil surface. The change in output was recorded when dry soil was located at different locations along the probe length.
Test 5
To study the macropore effect on probe readings, soil cores were removed along the length of the probe 0 to 0.2 cm away from the sensor surface using a 0.56-cm-diameter core sampler to a depth of 4 cm (volume of a single core is 1 cm3) and a 1.74-cm-diameter sampler to a depth of 2 cm (volume of a single core is 4.75 cm3) from soil (0.08 m3 m–3 v) packed to 1.5 Mg m–3 bulk density in the plastic box (Fig. 1b). Sensor response was measured by changing the size of the cores, the number of cores, and the position of cores in relation to the probe length. To study the micropore effect on probe reading, a known weight and v of a soil was compacted to known volumes using calibrated marks in the sliding panels (Fig. 2a). Sensor output was measured using three probes in response to changes in compaction level. Using the following equations, an estimate of the total porosity (St, m3 m–3) and air-filled pore space (Sa, m3 m–3) was made using the bulk density (
b, Mg m–3), mineral density (p, Mg m–3), and water content (v, m3 m–3) of the soil after compaction (Danielson and Sutherland, 1986). In calculations, the p was assumed to be 2.65 Mg m–3.
 | [(1)] |
 | [(2)] |
View larger version (79K):
[in this window]
[in a new window]
|
Fig. 2. The ECH2O performance field observations: (a) field installation of probes at the two study sites; and (b) linear regression between the probe output and the measured soil water content.
|
|
Test 6
To evaluate the temperature effect on probe readings, 2 kg of air-dried soil was mixed thoroughly with 106.7 mL of tap water to bring the soil to 0.08 m3 m–3 v. This wet soil was packed in the plastic box (shown in Fig. 1b) to 1.5 Mg m–3 bulk density. After partial packing, two ECH2O probes and one temperature probe (Campbell Scientific CS 109) were installed parallel to each other in the box, and these sensors were connected to a CR-10X datalogger. The datalogger was programmed to measure the output from the ECH2O sensors and the soil temperature. The plastic box with the sensor–datalogger setup was placed in a temperature-controlled chamber, and air temperature in the chamber was dropped from 37.2°C (99°F) to 7.2°C (45°F). Sensor output was recorded with temperature changes. The temperature effect on probe output was also estimated by inserting the ECH2O probe into a cylinder containing water that ranged from 10 to 40°C. This temperature range approximates the daily or seasonal soil water temperature fluctuations close to the soil surface.
Field Evaluation
Experiment Sites
Probes were calibrated in the field (Hamlin grove) by measuring the output against known v values. Volumetric water content (v) was determined using soil cores of 5.3-cm diameter and 6.0 cm long extracted with a bulk density sampler and oven dried at 104°C for 24 h. The estimated gravimetric water content (Gardner, 1986) was then multiplied by the dry bulk density of each core sample. We verified this calibration curve against a calibration that used normalized output values. Normalizing of output was done using the sensor output in air (mVair) and water (mVwater).
 | [(3)] |
ECH2O soil moisture probes were installed at two sites where orange trees [Citrus sinensis (L.) Osbeck] were irrigated with microsprinklers. The soil at both locations was an excessively drained Candler fine sand. Site 1 had Valencia oranges, was located in Lake Alfred, FL (28°06'23'' N, 81°42'48'' W), and had three treatments: (i) irrigated, (ii) unirrigated, and (iii) no irrigation and no infiltration of water from rainfall. Infiltration of water was prevented in Treatment 3 by placing a water-excluding fabric (Tyvek) under the tree canopy. These treatments were imposed from mid-November to the middle of March (mid-fall to early spring). During the rest of the year, all treatments were irrigated equally. The Valencia tree spacing was 7.5 by 3 m, and trees were about 10 yr old. The ECH2O probes in the Valencia grove were used to monitor soil water status under the imposed stress levels. Five ECH2O probes were installed horizontally in the soil profile (Fig. 2a) in the tree line close to the canopy drip line at 10-, 20-, 30-, 50-, and 90-cm depths in each of the three treatments and were directly connected to an EM5 datalogger (Decagon Devices). Probe placement was facilitated by excavating a 40-cm-wide trench to a depth of 1 m about 30 cm away from the center of the tree row. The plastic blade of the probe was strong enough to push into the sandy soil by reaching the desired depth via the trench. When insertion of the probe was difficult due to roots, a sharp metal blade was used to cut roots and make an incision wide enough to insert the probe. Data were recorded at 1-h intervals and downloaded biweekly for processing.
Site 2 was a Hamlin orange grove assigned to an irrigation scheduling study that consisted of different irrigation schedules as treatments in a Water Conserv II reclaimed water testing project 40 km west of Orlando, FL (28°28'20'' N, 81°38'50'' W) (Parsons et al., 2001). The Hamlin orange trees were about 20 yr old and were planted at a 7.5- by 4.5-m spacing. In the Hamlin grove, five ECH2O sensors were installed under a selected treatment tree in each of six treatments (Fig. 2a). The probes were installed vertically at three depths in the soil profile in the tree line close to the tree canopy drip line about 1.2 m away from the microsprinkler. Two sensors were centered at depths of 15 and 50 cm and a single sensor was centered at 90 cm (Fig. 2a). The 90-cm probe was used to monitor water movement below the main root zone. The five probes from each treatment were connected by a 19 AWG telephone wire to a CR-10X datalogger via a multiplexer (Campbell Scientific, Logan, UT). The excitation voltage of sensors at both locations was set at 2.5 V. The datalogger was programmed to trigger irrigation when the mean output value measured by the upper four sensors (15- and 50-cm depths) reached a predetermined value. The results from the field calibration were used to interpret the output values in v units or to determine the average millivolt drop required to start an irrigation in each treatment. The v units will be expressed as cubic meters per cubic meter. Data collected at 1-h intervals were downloaded biweekly.
The distribution pattern of water in the sensor area was studied using 15 catch cans under each treatment tree. Five cans each were placed in a radial pattern at distances of 0.5, 1.0, and 1.5 m from the tree trunk. Microsprinklers were operated for exactly 1 h and the volume of water collected from each can was recorded and converted to a linear scale (millimeters) using the cross-sectional area of the can.
|
RESULTS AND DISCUSSION
|
|---|
The calibration of probes produced a linear relationship between probe output and v with a coefficient of determination (r2) of 0.85 (Fig. 2b). Using normalized output values in the calibration did not improve the r2 of the regression curve (data not shown). According to the calibration curve, the output corresponding to fc (0.085 m3 m–3) was 450 mV, and wp (0.02 m3 m–3) was 350 mV. Therefore, the value within the available water content range is equivalent to 100 mV. Hence, a 1% change in the water content could be interpreted as a 17-mV change in probe output. This determination assumes a linear calibration.
Laboratory Observations
Signal Variability from Probe to Probe
Probe output measured in air and water appeared to be a constant for a particular probe and varied from probe to probe. When output values were measured on different days, however, probe values showed day-to-day variability. Table 1 indicates the mean of four readings per day on four consecutive days: the maximum, the minimum, and the standard deviation of output values for each probe. Day-to-day variability in water temperature and conductivity and air temperature and humidity may be the quality factors that influence output values in water and air. The difference in output values between air and water for a particular probe also varied from day to day (Table 1), indicating that the water and air quality parameters mentioned above have no direct relationship. The probe-to-probe variability and the noninterdependency in air and water values may be the reason why normalizing of output using air and water values did not improve the prediction accuracy of the calibration curve.
View this table:
[in this window]
[in a new window]
|
Table 1. The mean, minimum, and maximum sensor output in water and in air measured during four consecutive days with four readings per day, the difference between air and water values, and the standard deviation for each of 14 different ECH2O soil water probes.
|
|
Effect of Fertilizer-Induced Salinity
The salinity effect from fertilizer applications increased the output by about 200 mV. Under nonsaline conditions, the output value from probes was 600 mV at saturation, and this value was increased to 800 mV at both 12- and 27-cm depths in the column after introducing fertilizer with water into the soil column (Fig. 3a). This high output will increase the estimated v substantially. The salinity effect on probe reading is crucial in automated, sensor-dependent irrigation because high output from probes can lead to underirrigation due to a false indication of high soil water content. When water was flushed through the column, the high concentration of total dissolved solids reached the normal tap water level after 1 h (Fig. 3b).
View larger version (31K):
[in this window]
[in a new window]
|
Fig. 3. Results of the ECH2O performance laboratory tests: (a) sensor response to salinity during fertigation as tested with a soil column; and (b) concentration of dissolved solids (fertilizer) in leaching water with time (minutes) as tested with a soil column.
|
|
Sampling Volume of the Probe (Area of Influence)
The probe could not sense soil wetting or salinity from fertilizer salts when the wetting phase was 2 cm away from the sensor surface. Sensors were closely monitored (1-min intervals), and sufficient time (30 min) was allowed for the wet soil to remain at that distance from the sensors. Since the added water was calculated to bring the soil near fc, we assumed there was no free drainage from the bottom of the plastic box toward the sensors. After 1 min, the probe detected when wet soil and salts were between 1 and 2 cm from the sensor surface (Table 2). Calculations using these values indicated that the sampling volume of soil for a probe was between 128 and 256 cm3. Other tests (Lab Test 3) showed that the upper 1/3 of the probe surface was more sensitive to soil water than the rest of the probe surface.
Effect of a Thin, Dry Soil Layer at the Sensor Surface
Tests using the plastic box indicated that when a very dry soil layer about 0.6 cm thick around the sensor surface was present, the sensor reading was equivalent to the water content of that dry soil layer. The probe could not identify wet soil beyond that point (Fig. 4a). The variability of soil wetting by microsprinklers during irrigation can affect the water distribution pattern in the subsurface in this sandy soil (Table 3). Hence, a dry soil layer can develop close to the probe surface, preventing the probe from reading the correct overall soil water content. Also, it is possible for a dry soil layer to develop horizontally. A dry layer can occur close to the soil surface between irrigations. The surface soil can get very dry due to surface evaporation and water uptake by upper roots. When a probe was installed vertically close to the soil surface, the drying effect on the upper third of the probe changed the reading by about 42 mV (Fig. 4b), which is equivalent to about a 0.025 m3 m–3 v drop. This excessive drop may change the v value that represents the average water content of the soil layer corresponding to the probe length.
View larger version (28K):
[in this window]
[in a new window]
|
Fig. 4. Results of the ECH2O performance laboratory tests: (a) sensor response when a dry soil layer surrounds the probe surface; and (b) sensor output difference when probes are installed vertically in a constructed soil column, when a dry soil layer surrounds the probe surface at different depths.
|
|
View this table:
[in this window]
[in a new window]
|
Table 3. The effect of size, number, and position of large pores (macropores) on sensor response when pores are located at the lower 1/3 of the probe (L), middle 1/3 of the probe (M), and upper 1/3 of the probe (U).
|
|
Effect of Pore Space Close to the Sensor Surface
Contrary to what was expected, the effects of large pores on sensor readings were not as great as that of small pores. With 18 holes of 0.54-cm diameter (6.4% of total voids in relation to the sampling volume) along the sensor surface, the sensor output difference obtained was 13 mV. With eight holes of 1.74-cm diameter (19.8% total voids), the sensor output difference obtained was 16 mV (Table 3). The output difference when microsize pores were changed was greater. With 59% air-filled porosity (0.94 Mg m–3 bulk density created by loose compaction) compared with 35% porosity at 1.56 Mg m–3 bulk density, the output difference was 72 mV (Table 4). Loose compaction is an oxymoron. Many of these voids were probably in contact with the probe, so it was able to "see" them. The output value was positively correlated with bulk density in this soil. A regression between bulk density of the packed soil and the probe output indicated an exponential relationship [y = 227exp(0.33x), r2 = 0.85]. Thus, if the soil is loosely packed, it will indicate a lower v than the actual v and a higher v when the same soil is compacted.
View this table:
[in this window]
[in a new window]
|
Table 4. The effect of micropores on sensor response. The water content of the soil was held constant at 0.06 m3 m–3.
|
|
Effect of Soil Temperature
Diurnal and seasonal temperature fluctuations can be substantial, especially if radiation penetrates to the soil through the citrus canopy. Temperature-controlled chamber tests using soil with a known constant v in a plastic box indicated that when soil temperature changed from 37.2°C (99°F) to 7.2°C (45°F), the output from two sensors changed from 490 to 420 mV (70-mV difference) in Probe 1 and 466 to 405 mV (61-mV difference) in Probe 2 (Fig. 5a). Figure 5b indicates that soil temperature is linearly correlated to the probe output with an r2 value of 0.94. Also, a shift in the curve is indicated at about 25°C (room temperature). Capacitance of the probes and the permittivity is dominated by the dielectric constant of soil water compared with that of other soil components. If the shift were in the opposite direction, then it could be explained with the known relationships of the dielectric constant of water and permittivity against temperature. With increasing temperatures, the dielectric constant of water and the permittivity decreases (Lide, 2000, p. 6-3 and 6-15). Also, the tendency is for soil water to decrease by evaporation when the temperature increases. Therefore, neither of these two factors explains the positive shift in the curve with the increase in temperature.
View larger version (22K):
[in this window]
[in a new window]
|
Fig. 5. Results of the ECH2O performance laboratory tests: (a) sensor response when the soil is subjected to different temperatures in a control chamber; and (b) linear regression between probe output and the measured soil temperature.
|
|
According to the field calibration, this change due to temperature is equivalent to a 4.9% change in water content in Probe 1 and 4.4% in Probe 2. On average, this change is equivalent to about 50% of available water in this sandy soil. When probes were tested in water by changing the water temperature from 10 to 40°C, the probe response to changing temperatures indicated a similar trend but with a lesser effect. The mean output from four sensors changed from 953 to 994 mV, a 41-mV difference (data not shown). Soil temperature is closely related to the temperature of the soil water associated with it.
Field Performance
Normal Sensor Response
If sensors installed in the soil operate normally, output should indicate events such as irrigation or rainfall, redistribution and drainage, and soil water depletion resulting from daily water uptake. This information can be useful to ascertain v at (i) saturation (s), (ii) field capacity (fc), and (iii) wilting point (wp), along with (iv) the redistribution pattern of soil water, and (v) possible drainage below the root zone. This could be used to decide the time and amount of irrigation. The maximum s possible is 0.43 m3 m–3 (equivalent to the porosity of the soil), fc = 0.085 m3 m–3, and wp = 0.02 m3 m–3 for this soil. Figure 6a indicates sensor response when the soil receives water from irrigation or rainfall, and Fig. 6b indicates the response from the sensors when water inputs were excluded using Tyvek during February 2005. Sensors at the15-cm depth (Fig. 6a) responded well to water inputs by an increase in v to 0.17 or 0.19 m3 m–3 with irrigation or rainfall and then a decrease with time to about 0.06 m3 m–3 before the next irrigation. Saturated hydraulic conductivity is >0.6 m h–1 for this soil (Paramasivam et al., 2001), which is very high. Rainfall intensity or sprinkler discharge rates are not high enough for the soil to reach full saturation. It is clear that from 23 to 28 February, v did not drop to 0.06 m3 m–3 because of the distributed rainfall. The probes were in an irrigated treatment (irrigation was done when soil water was depleted to about 0.06 m3 m–3). Sensors indicated that the 50- to 90-cm deep soil did not get sufficient time to dry to 0.06 m3 m–3 during irrigations. Also, no response to irrigation on 11 February and a very low response on 17 February mean that irrigations were not excessive. The response was clear after a well-distributed rainfall on 27 February, indicating that the subsurface soil was wetted above fc. When the sensor response at 15 cm was compared with those at 50 and 90 cm, it is obvious that the soil water use at deeper depths was not as intense as at 15 cm. Probes may indicate a wetter soil, however, if a thin layer of soil close to the sensor surface is relatively wetter than the rest of the area. Figure 6b indicates that the Tyvek cover prevented infiltration from rainfall on 27 Dec. 2005. While soil water at 10 cm deep appeared to be the lowest, all sensors indicated a gradual decrease during the month until 24 February. After that day, there was an indication of a slight increase by the sensors at 20-, 30-, and 50-cm depths, probably due to a redistribution of soil water after the successive rainfalls (in spite of infiltration exclusion). It is also apparent that, despite the removal of the Tyvek cover on the morning of 16 March, allowing infiltration to occur (Fig. 7a), water didn't get close to the sensor area during the rest of that month. Only the probe at 90 cm deep indicated a response. The water from rainfall may have bypassed the shallower sensors due to dry soil lenses, which may have developed close to those sensors during the time when infiltration was excluded.
View larger version (35K):
[in this window]
[in a new window]
|
Fig. 6. Soil water content measured with five ECH2O probes against time from 1 to 28 February in (a) Treatment 3 in the Hamlin block; and (b) in the no-irrigation and no-rainwater treatment in the Valencia block.
|
|
View larger version (34K):
[in this window]
[in a new window]
|
Fig. 7. Soil water content measured with five ECH2O probes against time in (a) the no-irrigation and no-rainwater treatment in the Valencia block from 1 to 30 March, and (b) Treatment 6 in the Hamlin block from 3 to 19 May, including sensor response to fertigation.
|
|
The sensor response after fertigation (Fig. 7b) differed from the normal response. Sensors at 15 cm deep responded to induced salinity during fertigation by increasing the output noticeably. This response was similar to the response we noticed during laboratory tests. Sensors at 50 and 90 cm deep, however, did not respond similarly. This was possibly due to dry soil layers close to the sensor surface that prevented a correct reading (Fig. 4a). This treatment was a drier treatment, with no water applied other than rainfall. Dry soil layers (lenses) appear to be a common problem in drier treatments in this sandy soil. Similar to the laboratory test, the maximum response to fertigation was about 800 mV (equivalent to 0.33 m3 m–3 v). The sensors indicated a higher value until the fertilizer was completely washed away from the soil after the 2.4 cm of rainfall on 11 May. If there were no rainfall, the false high v values would have delayed automatic irrigation until a manual irrigation would wash off the soluble salts (Fig. 3b). The distribution pattern of irrigation water close to where sensors were installed indicated that in one treatment (Treatment 3), the sensor area received 32% less water than the mean microsprinkler discharge rate when compared with other treatments (Table 5). The lack of response of sensors at deeper depths was partly due to uneven distribution of water from the microsprinklers.
View this table:
[in this window]
[in a new window]
|
Table 5. Distribution of irrigation water within an area where ECH2O probes were installed, irrigation study, Hamlin block.
|
|
Soil temperature fluctuations also can affect the output from sensors. Laboratory studies indicated that this effect can be crucial in automating irrigation using these sensors. Figure 8 shows the output from sensors during the day and night. Sensors 15 cm deep showed the greatest diurnal response. Sensors at 50 cm also indicated a delayed response. Day and night temperature difference in 1 to 3 January indicates that the difference at the 15-cm depth could be as much as 150 mV. Due to the delayed effect, the mean value from the 15- and 50-cm sensors ranged between 453 and 367 mV (<100mV; Fig. 8). This output fluctuation due to the temperature effect possibly can disrupt an irrigation schedule in a treatment.
View larger version (37K):
[in this window]
[in a new window]
|
Fig. 8. Sensor response to day and night soil temperature fluctuations in Treatment 1 between 1 and 3 January in the Hamlin site.
|
|
Occasionally, some sensors in the field unexpectedly failed to produce an output even though they were in the normal soil water range. Some other sensors completely failed to operate for no reason. Some probes remained inactive for a short period and then became active. Water leakage into the sensor circuit appeared to be a major concern, because some of the sensors that failed to operate in the field worked well after removal and storage under dry conditions. To avoid problems from such anomalous readings, the datalogger was programmed to discard values below 340 mV (corresponding to v = 0 m3 m–3) and above 1000 mV. Another isolated problem was damage from animals and insects (e.g., animals chewed the lead wire). These sensors will need further improvement if they are to be used to automatically control an irrigation system.
|
CONCLUSIONS
|
|---|
The probe tests indicated that the maximum sampling soil volume is about 256 cm3 and this soil volume extends along the 20-cm length on either side of the sensor blade to a distance of 1 to 2 cm. Dry soil lenses close to the sensor surface can disrupt the correct response of the probes and may not depict the normal wetting and drying rate of the bulk soil. In such situations, there may be a tendency to underestimate the average soil water content. The probe-to-probe output variation is considerable. The sensor output in air and water is not a constant and depends on quality aspects such as temperature and humidity in the case of air, and temperature and total dissolved solids in the case of water. These quality parameters of air and water are not interdependent. Therefore, output values in air and water are not useful to normalize the output values when trying to minimize probe-to-probe variation. It is questionable whether the above factors and the considerable probe-to-probe variability would allow these probes to be used in automation of irrigation in any soil.
Salinity due to fertilization increased probe output by about 200 mV (equivalent to an apparent change in v from 0.08 to 0.20 m3 m–3) until irrigation or rainfall removed the salts. If irrigation is triggered by sensor response, then manual irrigation a few days after fertigation is essential to flush the salts. The effect of soil temperature on output was considerable on probes installed close to the surface.
The ECH2O EC-20 soil moisture probe is easy to install, relatively low cost, and requires little or no maintenance. It does not require special skills to operate and can collect data at regular intervals if used with a basic datalogger such as the EM5. Installation of this probe is not a problem in this sandy soil. If the soil is not packed to the correct bulk density around the sensor area, however, a 0.1 Mg m–3 decrease in the bulk density close to the sensor surface (possibly due to loose packing) can decrease the output by 14 mV. One advantage of sandy soil is that the blade-shaped probe can be easily inserted into it without substantially affecting bulk density; however, insertion will not be easy if the soil contains stones, gravel, or thick roots. Macropores caused by animal burrowing and dead root channel development are unlikely in this sandy soil. If macropores did develop, our results indicated that the output may not change greatly or disrupt v estimations. In general, the performance of these probes was much better under laboratory conditions than under field conditions.
The ECH2O probe manufacturer, Decagon Devices, has introduced a new set of probes called ECH2O EC-5 and TE-5 probes. These are improved devices and may not be as sensitive to salinity or temperature fluctuations as the EC-20 probes. These new probes are currently under investigation.
|
ACKNOWLEDGMENTS
|
|---|
We acknowledge the Southwest Florida Water Management District (SWFWMD) and the Florida Citrus Production Research Advisory Council (FCPRAC) for their financial support in conducting this study.
|
NOTES
|
|---|
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 August 14, 2006.
|
REFERENCES
|
|---|
- Baumhardt, R.L., R.J. Lascano, and S.R. Evett. 2000. Soil material, temperature, and salinity effects on calibration of multisensor capacitance probes. Soil Sci. Soc. Am. J. 64:1940–1946.[Abstract/Free Full Text]
- Borhan, M.S., L.R. Parsons, and W. Bandaranayake. 2004. Evaluation of a low cost capacitance ECH2O soil moisture sensor for citrus in a sandy soil. p. 447–454. In Int. Irrig. Show, 25th, Tampa Bay, FL. 14–16 Nov. 2004. The Irrig. Assoc., Falls Church, VA.
- Danielson, R.E., and P.L. Sutherland. 1986. Porocity. p. 443–461. In A. Klute (ed.) Methods of soil analysis. Part 1. Physical and mineralogical methods. SSSA Book Ser. 5. SSSA, Madison, WI.
- Gardner, W.H. 1986. Gravimetry with oven drying. p. 493–507. In A. Klute (ed.) Methods of soil analysis. Part 1. Physical and mineralogical methods. SSSA Book Ser. 5. SSSA, Madison, WI.
- Kelleners, T.J., D.A. Robins, P.J. Shouse, J.E. Ayars, and T.H. Skaggs. 2005. Frequency dependence of the complex permittivity and its impact on dielectric sensor calibration in soils. Soil Sci. Soc. Am. J. 69:67–76.[Abstract/Free Full Text]
- Lide, D.R. 2000. CRC handbook of chemistry and physics. 80th ed. CRC Press, Boca Raton, FL.
- Morgan, K.T., T.A. Obreza, T.A. Wheaton, and L.R. Parsons. 2001. Comparison of soil matric potential measurements using tensiometric and resistance methods. Proc. Soil Crop Sci. Soc. Fla. 61:63–66.
- Morgan, K.T., L.R. Parsons, T.A. Wheaton, D.J. Pitts, and T.A. Obreza. 1999. Field calibration of a capacitance water content probe in fine sand soils. Soil Sci. Soc. Am. J. 63:987–989.[Abstract/Free Full Text]
- Norikane, J.H., J.J. Prenger, D.T. Rouzan-Wheeldon, and H.G. Levine. 2005. A comparison of soil moisture sensors for space flight applications. Appl. Eng. Agric. 21:211–216.[Web of Science][Medline]
- Obreza, T.A., D.J. Pitts, L.R. Parsons, T.A. Wheaton, and K.T. Morgan. 1997. Soil water-holding characteristic affects citrus irrigation scheduling strategy. Proc. Fla. State Hortic. Soc. 110:36–39.
- Paramasivam, S., A.K. Alva, A. Fares, and K.S. Sajwan. 2001. Estimation of nitrate leaching in an Entisol under optimum citrus production. Soil Sci. Soc. Am. J. 65:914–921.[Abstract/Free Full Text]
- Parsons, L.R., and K.T. Morgan. 2004. Management of microsprinkler systems for Florida citrus. Available at edis.ifas.ufl.edu/HS204 (verified 19 Feb. 2007). Inst. of Food and Agric. Sci., Univ. of Florida, Gainesville.
- Parsons, L.R., T.A. Wheaton, and W.S. Castle. 2001. High application rates of reclaimed water benefit citrus tree growth and fruit production. HortScience 36:1273–1277.[Web of Science]
- Schroder, J., R. Munoz-Carpena, M. Dukes, and Y. Li. 2005. The development and testing of an automated drip fertigation system for sustainable agriculture. ASAE Paper no. 052239. In ASAE Annual Int. Meet., Tampa, FL. Am. Soc. Agric. Eng., St. Joseph, MI.
This article has been cited by other articles:
|
|
|
|
 
L. R. Parsons and W. M. Bandaranayake
Performance of a New Capacitance Soil Moisture Probe in a Sandy Soil
Soil Sci. Soc. Am. J.,
June 29, 2009;
73(4):
1378 - 1385.
[Abstract]
[Full Text]
[PDF]
|
|
|
TOP
ABSTRACT
INTRODUCTION
