Irrig Sci (2009) 27:393–400 DOI 10.1007/s00271-009-0154-3
ORIGINAL PAPER
Laboratory evaluation of dual-frequency multisensor capacitance probes to monitor soil water and salinity J. L. Starr · D. J. Timlin · P. M. Downey · I. R. McCann
Received: 8 July 2008 / Accepted: 16 March 2009 / Published online: 7 April 2009 © US Government 2009
Abstract Real-time information on salinity levels and transport of fertilizers are generally missing from soil proWle knowledge bases. A dual-frequency multisensor capacitance probe (MCP) is now commercially available, for sandy soils, to simultaneously monitor volumetric soil water content (VWC) measured as a percentage and salinity as a unitless volumetric ion content (VIC). The objectives of this research were to assess the relationship of salinity and water content with these dual-frequency MCPs under laboratory conditions, and assess its potential for Weld use in sandy soils of the mid-Atlantic region of the US. Water and salinity studies were conducted in two sand-Wlled PVC columns, 1.2 m long by 0.25 m ID. Each column was instrumented with ten dual-frequency capacitance sensors and two thermocouple temperature sensors. Four salinity levels were studied in the two columns using 0.5, 1, 2, and
Communicated by P. Waller. J. L. Starr (&) USDA-ARS Hydrology and Remote Sensing Lab, Bldg. 007, 10300 Baltimore Ave., BARC-W, Beltsville, MD 20705, USA e-mail:
[email protected] D. J. Timlin USDA ARS Crop Systems and Global Change Lab, Bldg. 001, Rm. 342, 10300 Baltimore Ave., BARC-W, Beltsville, MD 20705, USA P. M. Downey USDA-ARS Environmental Management and Byproduct Utilization Lab, Bldg. 007, 10300 Baltimore Ave., BARC-W, Beltsville, MD 20705, USA I. R. McCann Department of Bioresources Engineering, University of Delaware REC, Georgetown, DL 19947, USA
4 dSm¡1 NH4NO3 solutions. Water, salinity, and temperature readings were continuously recorded at 1-min intervals. The VIC values were found to be primarily qualitative, but combined with real-time VWC measures the probe could still be an important fertigation management tool to provide near-continuous real-time information on fertilizer penetration, spread and subsequent changes during crop growth.
Introduction Real-time measurement of soil water content is important in irrigation management. Instruments such as the neutron probe have been used for many years by the research and agricultural consulting community to measure volumetric soil water content. Similarly, low cost instruments such as tensiometers and electrical resistance sensors have been successfully used by growers to measure soil water potential for irrigation scheduling purposes. Sensors based on the dielectric properties of the surrounding soil matrix are a more recent development. Typically, such sensors either measure the capacitance of the soil and infer the water content based on the large diVerence between the permittivities of air and water (Starr and Paltineanu 2002), or they measure the travel time of electromagnetic pulses through the soil (Time domain reXectometry, TDR; Noborio 2001). Unlike resistance sensors or tensiometers, they do not require physical equilibration with the soil and so can respond instantaneously to changes in soil water content. Using capacitance sensors with data-loggers allows nearcontinuous measurement and observation of short- and long-term trends such as diurnal water uptake by roots. A useful format for sensors is in a multisensor probe in which a number of sensors are mounted vertically and
123
394
Fig. 1 Schematic of an EnviroSCAN® multisensor capacitance probe showing three sensors
installed in an access tube in the soil so that simultaneous measurements can be made at a number of depths in the proWle (Fig. 1). Commercial examples include EnviroSCAN®1 (Sentek Pty. Ltd., 77 Magill Rd, Stepney, South Australia 5069); Aquaspy® (Aquaspy Group Pty. Ltd., 16 Phillips St., Thebarton, South Australia 5031); ProWle Probe® (Delta-T Devices Ltd., 130 Low Rd., Burwell, Cambridge, UK CB25 0EJ); and Moisture Point® (Environmental Sensors Inc., 2071C Malaview Ave., Sidney, BC Canada V8L 5X6). Where the sensors measure capacitance, such as EnviroSCAN, they can be referred to as multisensor capacitance probes (MCPs). The use of EnviroSCAN MCPs for near-continuous real-time soil-proWle-water monitoring has been documented by Buss (1993); Fares and Alva (2000); Starr and Timlin (2004); McCann and Starr (2007). The MCPs and similar soil water monitoring systems can give critical information for managing irrigations to provide optimal soil water conditions for crop production with minimal environmental impact. The information consists of both individual measurements of soil water content and trends over depth and time. With the EnviroSCAN MCP, the manufacturer’s default calibration is best for sandy soils with low electrical conductivity (Kelleners et al. 2004a). Site-speciWc calibrations are required to obtain accurate volumetric soil water content (VWC) for many soils and site conditions (Paltineanu and Starr 1997; Kelleners et al. 2004a; Polyakov et al. 2005). Yet, these probes 1
Trade names are used in this publication to provide speciWc information. Mention of a trade name does not constitute a guarantee or warranty of the product or equipment by the USDA or an endorsement over other similar products.
123
Irrig Sci (2009) 27:393–400
are commonly used by irrigation managers without sitespeciWc calibrations because their immediate needs are qualitative, i.e., to see relative in situ changes in VWC via graphical trends in soil proWle water dynamics due to soil water storage, plant water uptake and changing climatic conditions aVecting evapotranspiration (ET) in their irrigated Welds, and thereby determine irrigation timing and amounts. The ability to simultaneously monitor both water and solutes in soil proWles would provide a powerful day-to-day management tool for maintaining optimal growing conditions with minimal adverse impact on the environment. Until recently, time domain reXectometry (TDR) has been the primary dielectric-based tool available that can monitor both soil water and salinity with the same probe (Mallants et al. 1996; Das et al. 1999; Noborio 2001). However, short cable lengths and degraded waveforms with increasing salinity are limitations to Weld applications of TDR. Concurrent salinity and VWC measurements from capacitance sensors would be a valuable addition to the soil proWle knowledge base where real-time information on salinity levels and transport of fertilizers can be critical for eYcient and economic crop production (Walker et al. 1999). Measurement of water content by capacitance sensors, however, is overestimated when soil salinity levels are high in soil (Kelleners et al. 2004b). Often there is little information on soil salinity in Weld water studies, so the use of dual-frequency capacitance sensors could indicate when and where the capacitance sensor water measurements are likely to be overestimated. Dual-frequency sensor A dual-frequency MCP that simultaneously monitors soil water and solute content is commercially available from Sentek (the manufacturer of the EnviroSCAN) under the trade name TriSCAN®. These sensors use a high frequency (fH > 100 MHz) response to measure VWC and a low-frequency (fL < 27 MHz) response from which the manufacturer’s proprietary formula automatically derives2 the soil’s volumetric ion content3 (VIC). The fL takes advantage of the fact that capacitance sensors are more sensitive to salinity at induction frequencies lower than 50 MHz (Campbell 1990). The VIC is considered proprietary information by the manufacturer; hence it is only described as a numerical value that relates to soil electrical conductivity (EC) but is not directly interchangeable with it (Buss et al. 2004). This dual-frequency capacitance probe is optimized for sand to 2
The raw fL response values and the formula for calculating VIC are not made available to the user. 3 “Content” not concentration. VIC is unitless as provided by the manufacturer.
Irrig Sci (2009) 27:393–400
395
sandy loam soils that are prone to fertilizer leaching. Structurally, this dual-frequency TriSCAN probe is identical to the single-frequency EnviroSCAN probe (Fig. 1), with capacitance sensors consisting of one or more pairs of 50.5 mm diameter cylindrical metal rings separated by a nonconducting plastic ring and mounted on a support rod that is inserted into a previously installed PVC access tube (Paltineanu and Starr 1997). The TriSCAN probe has been described in detail elsewhere (Watson et al. 1995; Dalton et al. 2004; Buss et al. 2004). The objectives for this research were to assess the relationship of salinity and water content with dual-frequency MCPs under laboratory conditions and assess their potential for Weld use in coarse-textured soils of the mid-Atlantic Region of the US.
Materials and methods Water and salinity studies were conducted in two sandWlled PVC columns. Each 1.2 m long by 0.25 m ID column had a 1.25 cm thick PVC bottom plate with a free-drainage port just above it. Washed and air-dry sieved sand (Table 1) was weighed and carefully packed in 5-cm increments. Column-average bulk densities were 1.61 and 1.63 Mg m¡3 with standard deviations of 0.052 and 0.022 Mg m¡3. After packing the sand in the two columns, a specially constructed access-tube holding rig, similar to that reported by Paltineanu and Starr (1997), was temporarily attached to the top of each column to center the tube in the column during installation. An inward tapered metal cutting ring was attached to the bottom of each access tube to facilitate tube installation and a tight contact between the access tube and the sand. As the access tube was pushed through the holding rig into the dry sand, the sand was vacuum-removed from inside the tube. After inserting each access tube the tube-bottom was made watertight with a rubber compression plug.
Fig. 2 Schematic view of column instrumented with a TriSCAN® multisensor dual-sensor capacitance probe placed inside a PVC tube, and two thermocouples. The outXow through the top plate during bottom-up saturation came through four equally spaced 3.2 mm plastic tubes
A recessed groove was previously cut into the top of each column to hold a clear acrylic top plate to minimize evaporative losses. A 5-cm diameter center hole was cut into each top plate allowing the access tube to extend above the column. In addition, four small holes were drilled into the top plates, midway between the access tube and the column wall and equally spaced around the column. Tightly Wtting 3.2 mm plastic tubing was inserted in these holes to collect column eZuent and monitor its salinity during bottom-Wlling. A schematic of an instrumented column is shown in Fig. 2.
Table 1 Sand particle size distribution ASTMa Sieve (#) 4
Retained (%) 0.2
Size (mm)
USDA classiWcationb
4.75
Fine gravel
8
1.3
2.36
Fine gravel
16
8.5
1.18
Very coarse sand
30
42.7
0.60
Coarse sand
50
40.0
0.30
Medium sand
100
6.0
0.15
Fine sand
200
1.1
0.075
Very Wne sand
a b
American Society for testing and materials Anonymous (1999)
Sensor normalization Prior to installing the access tubes in the sand columns, ten dual-frequency capacitance sensors were placed in 10-cm increments on each sensor holding rod. High frequency readings in air (Fa) and in 22°C deionized water (Fw) were then recorded for each sensor so that all subsequent high frequency sensor readings in “soil” (Fs) could be normalized through the scaled frequency (SF) relationship: SF = (Fa ¡ Fs)/(Fa ¡ Fw) as described by Paltineanu and Starr (1997). For this laboratory study the manufacturer’s default calibration for water content was used: SF = 0.1957
123
396
Irrig Sci (2009) 27:393–400
0.404 + 0.02852, where is VWC expressed as a percentage. TriSCAN’s VWC and VIC values were recorded every minute at all ten sensor depths with a Campbell ScientiWc data-logger. Column temperature Temperature was continuously recorded with thermocouples that were inserted horizontally through the columns and into the sand at 15- and 85-cm depths, shown as T15 and T85 in Fig. 2. Column treatments Four salinity levels were studied using 0.5, 1, 2, and 4 dSm¡1 NH4NO3 solutions that were stored in large carboys next to the columns. These salinities are typical of values that may be observed under non-saline sandy soil conditions in the mid-Atlantic region, USA. Electrical conductivity of the column inXuent and eZuent solution was monitored with an Orion model 160 conductivity meter (Orion Research Incorporated, Boston, MA). The columns were sequentially saturated (from low to high salinity) at each salinity concentration by pumping the required NH4NO3 solution through the bottom port (Fig. 2), with a peristaltic pump until the column-top eZuent was greater than 99% of the inXuent concentration. The columns were then allowed to drain through the bottom port until TriSCAN’s VWC and VIC readings were stable at each sensor depth, which required about 4 days.
Results and discussion TriSCAN’s VWC distributions with sensor depth for columns 1 and 2 are shown in Fig. 3. One curve shows the distribution of water content just before drainage following bottom-Wlled column-saturation; and one at 3 days of column-drainage. The linear decrease in “saturated” VWC with decreasing sensor depth may partially result from increasing water aeration as the rising water table displaced pore-air during saturation. The quadratic rise in VWC with sensor depth at the 3-day drainage time reXects the increasing water content with depth. Thus the drainage port becomes the bottom boundary open to the atmosphere where gravity-drainage ends. The drainage port at 120 cm approximates a water table when the columns are drained resulting in matric potentials ranging from h + ¡110 cm at the 10-cm depth to h + ¡20 cm at the 100-cm depth. The negative sign indicates that the water is under tension, i.e., at less than saturated water content. This is a potential energy due to capillary forces (when expressed as cm of water pressure), in this simple case (e.g., without ET) the
123
Fig. 3 Volumetric water contents (VWC) versus sensor depth under saturated and 3-day drainage water contents. Columns 1 and 2 are represented by solid and open symbols, respectively
height above the bottom of the cylinder represents the negative pressure head because the soil is open to the atmosphere at this bottom boundary. Drainage experiments following bottom-Wlled saturation at four salinities A graphic illustration of TriSCAN’s VWC and VIC temporal distributions at one depth (30-cm), across all salinity treatments is shown in Fig. 4. The salinity treatment associated with each saturation-drainage is shown near the top of the VIC displays. The initial VIC and VWC values are at the water saturation for the Wrst salinity treatment (0.5 dSm¡1). The horizontal line is drawn at the drained 0.5 dSm¡1 treatment to provide a baseline for the impact of salinity level on drained water content. The Wrst 4.0 dSm¡1 treatment was aborted early, due to a 4-day loss of ambient temperature control in the laboratory that resulted in the 11°C temperature drop and increase from May 12 to 16 and the corresponding drop and increase in VIC values. An approximate temperature-corrected change in VIC is shown by the hand-drawn dashed VIC-line. Due to the unplanned temperature drop, the aborted 4 dSm¡1 salinity treatment was immediately repeated, which made it possible to have a shorter time to reach saturation and maximum salinity values. Figure 4 shows a salinity eVect on the measured VWC, with apparent saturated VWC increasing from 31% at 0.5 dSm¡1 to 39% at 4.0 dSm¡1. However, there was little salinity eVect on the drained VWC values, which approached asymptotic drainage values of »10.5% across all four salinities. In contrast the fL response to increasing salinity shows increased VIC values with increasing salinity treatments under both saturated and drained conditions. At greater salinities increasingly skewed apparent water
Irrig Sci (2009) 27:393–400
397
Fig. 4 Logged VWC and volumetric ion content (VIC) trend lines at the 30-cm sensor depth, across four NH4NO3 saturation– drainage treatments, and temperature measured at 15-cm depth. Horizontal line is drawn at 0.5 dS/m drainage and vertical lines are drawn from intersection of the horizontal line and VWC drainage lines. The date-time is shown at the top of the vertical lines and values shown at each intersected trend line
contents may be anticipated under both saturated and drained VWC conditions, especially at low frequencies (fL) (Campbell 2007; Kelleners et al. 2004b). This low-frequency response enabled the manufacturer to develop this dual-frequency capacitance probe to simultaneously monitor water and salinity levels (Watson et al. 1995). The VIC variation with depth and between columns was more variable at saturation than when drained at all four salinities (Fig. 5). The diVerences in sensor response between the two columns at the same depths were quite large when the soil was at saturation but became small when drained, especially toward the top of the columns where water contents were lower (Fig. 3). The higher VIC values at the bottom of the columns (70–100 cm) resulted from the higher, near-saturated water contents at the bottom of the columns (Fig. 3). Note that the VIC values were still more variable between columns at the bottom of the columns after drainage. This variability, from a practical standpoint, may not be important under Weld conditions since the soil is more likely to exist in a near-drained state than near-saturated. The rapid drop in VIC with decreasing VWC (Fig. 4) was due to a decreasing salt mass on the soil’s unit volume basis with saline water being replaced by air during drainage. Sample linear regression Wts of VIC versus four salinities at six drainage times are shown for one sensor in Fig. 6. All the linear coeYcients of determination for both columns from 10- to 80-cm depths are graphically shown in Fig. 7. These results were strongly aVected by: depth due to increasing VWC with increasing depth (Fig. 3); and by
drainage time as shown by the increasing VWC-salinity correlation with increasing drainage time. These results suggest that interpreting VIC readings at each sensor will be most useful at or near its asymptotic drainage level. This is an advantage since most agricultural crops do not grow well at or near-saturated VWC. However, long drainage times (e.g., several days) may be impossible under many irrigation/fertigation management practices. In this case, comparative VIC readings following diVerent irrigation/ fertigation should be made at the same drained VWC. Even though VIC varied greatly with VWC it is desirable to be able to generalize the VIC-salinity relationship. Based on the 100-h drainage time data (Fig. 7), the VICsalinity relationship across the top Wve sensor depths is shown in Fig. 8a. Note that the drained VIC trends in Fig. 5 correspond to the correlation trends at 100 h in Fig. 7 in that both have small changes in the upper part of the column where the drainage is most complete (Fig. 3). Although the VIC-salinity correlation was high, the increasing VIC spread with increasing salinity is a concern (Fig. 8a). This may be due to the large range in VWC (8.9– 19.9%) with maximum values being quite far from the asymptotic drained VWC of 10.5% at the 30-cm sensor (Fig. 4). Thus, a second generalization was made by selecting a lower maximum water content value of 11.5%. The new correlation was essentially unchanged when using this smaller data subset (Fig. 8b), but it did result in much smaller VIC scatter at the higher salinities. Even though there is a seemingly complex interaction of VIC with VWC
123
398
Irrig Sci (2009) 27:393–400
Fig. 5 Volumetric ion contents (VIC) in two columns under saturated and drained water contents at four salinities (0.5, 1, 2, 4 dSm¡1 NH4NO3) distributions with column depth
Fig. 6 Sample VIC values at six drainage times versus four salinities (NH4NO3), with linear regression lines and correlation coeYcients. Sample data come from the 30-cm depth in column 2
(Fig. 5), Fig. 8b shows that a strong quantitative VIC-salinity relationship can also exist under lower VWC conditions, which suggests that this dual-frequency capacitance probe has potential for estimating in situ salinity levels under Weld conditions in sandy soils. Implications for Weld application Soil water content and salinity can be highly dynamic under Weld conditions, where the physico-mineralogic-chemical
123
characteristics of the soil vary in three-dimensional space, and where water movement within the soil proWle due to irrigation, rainfall and drainage together with crop uptake and leaching all aVect the solute concentration. These processes become even more complex when the irrigation system is used to apply soluble fertilizer (fertigation), and when it does not apply water in a relatively homogeneous manner over the entire Weld. A center-pivot irrigation system can be used to fertigate, where the fertilizer concentration in the irrigation water is relatively constant and is applied in a relatively uniform manner over the entire irrigated area. In contrast, drip irrigation can also be used to fertigate, but the fertilizer is injected into the water as a “pulse”, and the water is applied in an array of point sources that wet only a small fraction of the Weld. In both cases, an agricultural manager would want information on the relative uniformity and quantity of the resultant fertilizer distribution in the root zone. For example, did the fertilizer pulse move below the root zone, where is the center of the pulse within the root zone, etc.? This study shows that it is theoretically possible to measure relative changes in soil salinity on a soil volume basis due to the addition of a soluble fertilizer in sandy soils with low native salinity levels as commonly found in the humid northeastern US. Techniques need to be developed, however, to enable its successful use under the more complex conditions in the Weld. For example, as we saw in Fig. 4 the
Irrig Sci (2009) 27:393–400
Fig. 7 Linear coeYcients of determination of VIC values across four salinities (0.5, 1, 2, 4 dS/m¡1 NH4NO3) at six drainage times and eight sensor depths in two columns
VIC is sensitive to temperature variation. Under Weld conditions this kind of variation would be accentuated near the soil surface. Diurnal temperature change eVects could be minimized by measuring VIC at the same time each day, preferably in the early morning. Also, each fertigation regime and soil may have its own pattern of VIC response that could be used with laboratory soil analysis results to interpret VIC measurements with the goal of quantifying transport of applied fertilizers for a particular situation. In a qualitative sense, the probe could still be an important fertigation management tool to provide near-continuous real-time information on fertilizer penetration, spread and subsequent changes during crop growth.
Conclusions These laboratory studies were conducted to assess the relationship of salinity and water content with dual-sensor TriSCAN capacitance probes in a sandy soil. Our primary observations and conclusions are:
399
Fig. 8 Volumetric ion contents (VIC) versus four NH4NO3 salinity treatments at the top Wve sensor depths measured at volumetric water contents of a
