Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Calibrating electromagnetic short soil water sensors

Calibrating electromagnetic short soil water sensors J. Hydrol. Hydromech., 58, 2010, 2, 114­125 DOI: 10.2478/v10098-010-0011 WILLIBALD LOISKANDL1), GRAEME D. BUCHAN2), WOLFGANG SOKOL1), VILIAM NOVAK3), MARGARITA HIMMELBAUER1) 1) University of Natural Resources and Applied Life Sciences, Institute of Hydraulics and Rural Water Management, Muthgasse 18, A-1190 Vienna, Austria. 2) Soil & Physical Sciences Department, Lincoln University, Canterbury, New Zealand. 3) Institute of Hydrology, Slovak Academy of Sciences, Racianska 75, 831 02 Bratislava, Slovak Republic. Correspondance: W. Loiskandl, Mailto: loiskandl@mail.boku.ac.at The use of electromagnetic (EM) soil moisture probes is proliferating rapidly, in two broad domains: in field and laboratory research; and in strongly practical applications such as irrigation scheduling in farms or horticultural enterprises, and hydrological monitoring. Numerous commercial EM probes are available for measurement of volumetric water content (v), spanning a range of measurement principles, and of probe dimensions and sensing volumes. However probe calibration (i.e. the relationship of actual v to probe electrical output) can shift, often substantially, with variations in parameters such as soil texture, organic matter content, wetness range, electrical conductivity and temperature. Hence a single-valued, manufacturer-supplied calibration function is often inadequate, forcing the user to seek an application-specific calibration. The purpose of this paper is to describe systematic procedures which probe users can use to check or re-determine the calibration of their selected probe(s). Given the wide diversity of operating principles and designs of commercially-available EM probes, we illustrate these procedures with results from our own calibrations of five different short probes (length of 5 to 20 cm). Users are strongly recommended to undertake such calibration checks, which provide both pre-use experience, and more reliable in-use data. KEY WORDS: Soil Moisture Probes, Electromagnetic Probes, TDR, Calibration. Willibald Loiskandl, Graeme D. Buchan, Wolfgang Sokol, Viliam Novak, Margarita Himmelbauer: KALIBRÁCIA ELEKTROMAGNETICKÝCH SNÍMACOV VLHKOSTI PÔDY. J. Hydrol. Hydromech., 58, 2010, 2; 29 lit., 8 obr., 2 tab. Pouzívanie elektromagnetických (EM) snímacov vlhkosti pôdy sa rýchlo rozsiruje tak v terénnom výskume, ako aj v laboratóriu. Sú pouzívané v praktických aplikáciách ako je riadenie závlah na farmách a záhradách, ako aj v hydrologickom monitoringu. Pre meranie vlhkosti pôdy (v) sú dostupné pocetné typy komercných EM snímacov, zalozených na viacerých princípoch merania a snímace majú rozdielnu vekos snímaných objemov pôdy. Kalibracné krivky takýchto snímacov (t.j. závislosti medzi reálnou vlhkosou pôdy v a elektrickým výstupom snímac sa môzu posúva ­ niekedy podstatne ­ a to v závislosti od rozdielnych parametrov pôdy, ako je jej textúra, obsah organických látok, rozsah vlhkostí, elektrická vodivos a teplota. Z toho vyplýva, ze jednoznacná kalibracná krivka, dodávaná výrobcom je casto neadekvátna, co núti uzívatea snímac kalibrova v specifických podmienkach. Cieom tohto príspevku je opísa procedúry, ktoré môzu by pouzité uzívatemi pri rekalibrácii vybraných typov snímacov. Berúc do úvahy sirokú paletu princípov EM snímacov, ilustrujeme tieto procedúry výsledkami vlastných kalibracných testov na piatich typoch krátkych snímacov (dzka od 5 do 20 cm). Uzívateom odporúcame rekalibráciu komercných snímacov, ktorými získajú predbezné skúsenosti a spoahlivejsie výsledky pri meraní vlhkosti pôdy. KÚCOVÉ SLOVÁ: snímace vlhkosti pôdy, elektromagnetické snímace, TDR, kalibrácia. Introduction For many applications of soil water sensors, it is insufficient to use a sensor in the field without pre114 vious laboratory tests and a critical evaluation of the manufacturer's calibration. One aspect often overlooked is that the obtained value of volumetric water content (v) is derived from a surrogate measured electromagnetic (EM) parameter. The relationship between v and the parameter is central to the calibration procedure. Most common is the Topp relation between v and relative permittivity r (Topp et al., 1980), valid for a wide range of applications. However, in general, standard calibrations are good only for the conditions they are derived for (e.g. for a sand) and over a limited water content range. EM signal propagation is affected by several soil physical properties, especially colloid (clay and organic matter) contents, electrical conductivity, bulk density and temperature (Jacobsen & Schjonning, 1993; Malicki et al., 1996; O'Connor and Dowding, 1999; Gong et al., 2003; Yuanshi et al., 2003). The authors have personal experiences of being consulted to explain obviously wrong readings (e.g., at one extreme, v >100 % for a saturated clay soil). Results of other authors also clearly demonstrate the need for adjusting or testing the validity of standard calibrations (e.g. Groves and Rose, 2004; Blanc and Dick, 2003). Thus users should not rely uncritically on manufacturers' or standard calibrations, and also recognise that reading accuracy is determined mainly by the specific field soil conditions, and not only by the signal accuracy of the recording system. When installing sensors, it takes little extra effort to take check soil samples for gravimetric analyses. It is also recommended to take samples over a wide range of soil wetness. Key criteria for sensor selection include: ease of use; reliability; accuracy; range of application; automatic recording capability; sensing volume; the risk of disturbance errors (e.g. due to installation procedure, or in stony soils); and cost. Accuracy is especially critical, and relies on good calibration. Manufacturer's or standard accuracy specifications often pertain to a specific electronic system, and to measurement under very controlled conditions (e.g. in a sand), but need not be valid for the user's conditions, especially in non-standard applications like in soils high in colloids (clay or organic matter, e.g. Burgess et al., 2006), or saline soils, or in materials like organic waste. This paper presents procedures for calibration of EM short soil moisture probes, and illustrates them with results for five commercial probes. Calibration methods Accurate calibration is complicated by the need to reproduce conditions similar to those in the final application, and avoid uncontrolled variations in v in the calibration soil sample. To ease and improve calibration, we recommend and describe here an automatic procedure, based on Young et al. (1997) and as applied by Loiskandl et al. (2003). However this method has limitations. The most important constraint is the sensor rod length. Hence we distinguish here between short and long sensors. While there is no exact dividing point, for practical applications short sensors are considered to be 20 cm, and typically 10 cm or shorter. This paper focuses on calibration of short sensors. Calibration of short sensors For short sensors, a soil sample in a cylinder containing the probe is placed on a weighing scale. Water is supplied by dynamic upward infiltration by an adjustable pump (Fig. 1). The weight and sensor reading are recorded simultaneously, providing paired values of average v (determined gravimetrically) and sensor reading. The soil sample height and sensor length should be the same to ensure equal averaging. Because this calibration procedure imposes an axial moisture gradient along the sensor, it will strictly succeed only for probes which have a linear relation between the output signal and actual v, across the range of v occurring in the sample. A simple test of linearity is to partly submerge the sensor rod in water, e.g. 50 % immersion should give a v of 50 %. Fig. 2 shows results for two of the probes described below: Hydra-probe: a linear relationship was found over the whole rod length; and Theta probe: the relation is not linear, but is nearly linear for the main range of r encountered in practice. Soil permittivity can be represented using a three-phase concept (Woodhead et al., 2003): r = (Vi i ) 1/ , where phase i (= s, w, a for solids, water, air) has volume fraction Vi and permittivity i. Hence (1) r = [ v + (1 ­ ) + ( ­ v ) a ] 1/ , (2) w s where is porosity, v ­ volumetric water content (range 0 to 1) and ­ a shape factor. Sensor Data logger Soil sample in cylinder Data storage and conversion software 2 x RS 232 Balance RS 232 Interface continuous supply Fig. 1. Set-up for short sensor calibration (example with Hydra-probe). The sample is progressively wetted by upward infiltration of pumped water. Obr. 1. Zariadenie pre kalibráciu krátkeho snímaca (s príkladom snímaca Hydr. Vzorka je postupne navlhcovaná infiltráciou tlakovej vody zospodu. Fig. 2. Test of linearity: relationship between probe length submersed in water and probe output dielectric permittivity, r. HydraProbe response is linear, while Theta-Probe becomes non-linear at high water content. Obr. 2. Test linearity; vzah medzi dzkou vzorky ponorenej do vody a výstupom r (dielektrická permitivit. Snímac Hydra ­ závislos je lineárna, Theta snímac vykazuje pri vyssích vlhkostiach nelineárnu závislos. For probe submergence in water, = 1 and also = 1. Hence r = [v w + (1 ­ v ) a ] . With w = 81 and a = 1 this gives the linear relation r = (80v + 1) . Calibration of long sensors (3) For long sensors, the same procedure can be applied, but factors limiting its application are as follows. 1. Any sensor leads to different spatial averaging, compared to the gravimetric method. 2. For 116 samples with low hydraulic conductivity (especially clays or compact silts), the time to establish a range of water contents is very long, or even impractical. 3. The length in combination with the lateral zone of influence leads to a heavy soil sample, requiring a balance weighing this load with high accuracy. Therefore other procedures are recommended, e.g. using a re-packed column with pre-wetted soil. Field calibration is also possible, but takes greater time and effort to cover a sufficient range of v (e.g. Whalley et al., 2004). Further, if the sensor is installed for continuous monitoring, samples must be taken near the probe but outside its sensing volume. TRIME-Probe Overview of the five calibrated types of short sensors Since 1990, a wide range of EM soil moisture systems have appeared. For reviews see Topp and Ferre (2002), Evett (2000; 2005), Charlesworth (2000), Robinson et al. (2003). Here we focus on the five short sensors listed in Tab. 1. For the user, most important is a clear definition of what is measured and its relation to soil moisture. Hence Tab. 1 shows the EM principles of the sensors, and Tab. 2 the manufacturers' conversion relations. Calibration results for the five selected sensors For the above sensors, the standard calibrations are analysed here and compared to measurement results. Also summarized are the records or logging procedures used. The TRIME TDR system gives only v, but is user-friendly. Data transfer is very reliable and, due to the field bus/RS232 system, possible over long distances. User calibration is catered for, and requires special software (IMKO SMCAL Version 2.1, 1992). Independently measured v are compared with TRIME-Probe values and the polynomial coefficients (Tab. 2) are optimised, then programmed into the probe. Thus each probe can be adjusted to specific conditions, although calibrated sensors are then not easily interchanged. To analyse the standard function, a column filled with sand was used. SM-CAL proved to be a very valuable tool to obtain good results. First, the probe was calibrated (i.e. the polynomial coefficients were optimised) in glass beads. This calibration turned out to be very suitable for the sand. Fig. 3 shows results for 11 different v values, starting at about 45 % and progressing to dry conditions. This calibration was then used in a soil-organic material mix. The larger deviations observed clearly demonstrate the need to obtain a medium-specific calibration (Fig. 3). T a b l e 1. Operating principles and data conversion methods of the five types of sensors. T a b u k a 1. Princípy funkcie (measuring concept) a metódy konverzie údajov piatich typov snímacov. Sensor type/supplier TRIME-Probe/ IMKO LOM/RS EASY TEST ThetaProbe/Delta T Devices Measuring concept Travel time of a pulse along wave guides (Fundinger et al., 1995). Signal attenuation is reduced by coating the rods with a non-conductive material. is converted to v in an electronic device using inbuilt coefficients. As for the TRIME, but uncoated sensor rods. A single-pole-n-throw high frequency multiplexer connects sensors to a recording unit. Software converts data and transfers data to the recording unit. The Theta-Probe (ML2x) uses a simplified standing wave measurement to determine the impedance of a 4-rod sensing array (Delta-T Devices, 1999). The standing wave originates from a difference of the internal transmission line impedance and the impedance formed by the array embedded in soil. Measures the complex dielectric constant, and resolves the capacitive and conductive parts of electrical response. The capacitive part is most indicative of soil moisture, while the conductive part reflects mainly soil conductivity (Vitel Inc, 1994). Measures dielectric permittivity and provides a voltage output easily integrated into data logging systems (Decagon, 2002; 2008). Sensors with probe lengths of 20 cm (EC-20) and 5 cm (EC-5) are tested. Hydra-Probe/Vitel. Inc. ECH2O Probes, Decagon Devices, Inc. 4 deviation 3 2 0,50 1 Vol.% -0,60 0 -1 -2 -4 1 2 3 4 5 6 Reading 7 8 9 10 11 -0,70 -0,60 0,70 0,70 -0,30 -0,40 + 3 Vol.% 4 deviation 3 2,10 2 1 0 -1 -0,50 -1,40 SAND - 3 Vol.% -2 -2,00 -4 ,10 1 ,00 2 3 4 5 -1,00 3,50 3,20 1,70 1,60 + 3Vol.% 0,90 0 -0,90 ORGANIC MATERIAL - 3 Vol.% 6 Reading Fig. 3. Deviations (v in %) of TRIME readings for two medium types: sand and soil-organic material mix, as obtained by Dreiseitel (1997). Calibration parameters were obtained for glass beads, and gave good results for the sand. For the soil-organic mix, the larger deviations demonstrate the need to obtain a medium-specific calibration. Obr. 3. Odchýlka (v v %) snímaca TRIME pre dva typy prostredí: piesok, zmes organického materiálu, poda Dreiseitela (1997). Kalibracné parametre boli získané pre sklené guôcky, s dobrými výsledkami pre piesok. Väcsie odchýlky pre organickú pôdu demonstrujú potrebu získania osobitnej kalibracnej krivky. Laboratory operated meter for recording soil moisture, LOM/RS This TDR equipment was originally designed for controlling long-term laboratory experiments on soil columns, and can be combined with mini tensiometers to provide paired v and suction data for a retention curve (Malicki et al., 1996). The system is controlled by a PC. The supplied software operates the system and performs data conversion (r to v) using three equations with fixed coefficients (Tab. 2). Nevertheless, the conversion functions gave an almost linear relationship to water content (Fig. 4). To evaluate sensor response in different dry to wet conditions, tests in water and in loam soil were performed. Tests in water: 1. The uniformity of a set of 20 sensors was tested fully submerged in water; 2. To test for linearity, the sensors were fixed and plates of 5.1 mm thickness were placed under the cup (inside diameter of 5 cm), giving stepwise increase of the submerged depth. The tests were carried out at a constant temperature of 20oC and with the probes maintained in a vertical position. Triplicate readings were taken. The results are shown in Fig. 4. LOM/RS probes gave a very unfavourable response to the linearity test in water, with reasonable results only for submergence > 50 %. The most probable cause for this is a secon118 dary reflection occurring at the interface between air and water when sensors are submerged less than the resolution length of the system. By contrast, using the r values estimated with the three-phase concept (Eq. (3)) gave results very close to the theoretical 1 : 1 line in Fig. 4. In order to the test the sensors before their use in a pot experiment (Himmelbauer and Loiskandl, 2007), measurements were carried out in sand and a loam soil (38 % sand, 41 % silt, 21 % clay). The results in loam soil are presented here. Soil samples were packed into 6 cm high cylinders with inside diameter of 5 cm to a bulk density of 1.2 g cm. The measurements were conducted in air-dry soil (hygroscopic water content of 0.02 m3 m), at 0.5 saturation (0.27 m3 m), at 0.75 saturation (0.41 m3 m), and in a nearly saturated soil (0.51 m3 m). The estimated total porosity was 0.54, but in practice the maximum v achieved was slightly below that. As before, triplicate TDR measurements were taken. The gravimetric method supplied the reference values. As can be seen on Fig. 5, the mean offset of the readings (of 20 sensors) from the reference values remained below 5 % for all levels of saturation tested, i.e. 1 % ± 1.6 for the dry soil, ­3.6 % ± 2.1 for the 0.5 saturation, 2.2 % ± 1.7 for the 0.75 saturation and 4 % ± 2.8 for the completely saturated soil. The poorer accuracy at the high water content was probably a result of Calibrating electromagnetic short soil water sensors T a b l e 2. Electrical outputs and manufacturer equations for conversion to v, for the five types of sensors. T a b u k a 2. Elektrický výstup a rovnice pre konverziu na v (kalibracné rovnice) dodané výrobcom, pre pä typov snímacov. Sensor type TRIME LOM/RS EASY TEST Output Digital/Fieldbus IMP232-Micronet Digital/RS232 Manufacturer calibration equations Polynomial, order selectable to max order of 5 v = 0 if r < 1.48733 (special communication = 0.106387 * - 0.158247 if < 6 v r r protocol) v = 1 - (1 - 0.106387 * 6 - 0.158247) * (9 - r ) / 3 if r 6 Voltage V (0­1V) Theta-probe v = [1.1 + 4.4V ] - a0 a1 or v = Hydra-probe 4 Voltages (0­ 5V)/Permittivity 1.07 + 6.4V - 6.4V 2 + 4,7V 3 - a0 a1 v = C0 + C1 r + C2 r 2 + C3 r 3 sand: C0 = -8.63, C1 = 3.216, C2 = -9.54*10-2, C3 = 1.579*10 silt: C0 = -13.04, C1 = 3.819, C2 = -9.129*10-2, C3 = 7.342*10-4 clay: C0 = -20.93, C1 = 6.553, C2 = -0.246, C3 = 3.241*10-2 ECH2O probes: EC-20 EC-5 Voltage (250­1000mV) v = b1V - b0 for mineral soils: b1 = 0.695, b0 = 0.29 b1 = 1.19, b0 = 0.401 r ­ real dielectric constant V ­ output voltage a0, a1, b0, b1 ­ calibration coefficients v ­ volumetric water content Ci ­ polynomial constants ­ square wave period the vertical probe position and some reflection effects. Nevertheless, the reading deviations from the means are of the order of the accuracy stated in the manual (Fig. 5). For horizontal installation, as typical in pot experiments, more favourable conditions can be expected, since at a certain depth a homogenous moisture distribution is easier achieved than in a vertical direction. For exact measurements, a medium-specific calibration is highly recommendable. Theta probe For the Theta probe, the amplitude of a difference voltage V (Tab. 2) gives its relative impedance, hence r and thus a measure of v. The data logger needs conversion parameters, either from the manufacturer's calibration or from soil-specific calibration. The calibration method proposed above may be used if linearity of response is assumed for an output signal range 0 ­ 1 Volt, equivalent to v of 0 to 0.5 m3 m. In view of the strong non-linearity of the Theta probe at high v (Fig. 2), instead of submerging in water, it was progressively inserted in a cylinder of saturated coarse (1 mm) sand. The sensor was fixed vertically and a mobile platform was raised stepwise (Fig. 6). This method is discussed by Quinones and Ruelle (2001), who found good agreement (for Campbell Scientific CS615 probes) between progressively inserting the probe, discontinuous mixing of soil with water, and wetting the column from the top with a dripper. Fig. 6 shows two test results. The saturated v according to the added water was 0.36 and 0.34 m3 m for tests 1 and 2 respectively. Corresponding calculated values from the Theta probe were 0.36 m3 m for both tests. r was calculated with the third order polynomial (Tab. 2) and for conversion to v the general calibration mineral soil coefficients were used. If non-linearity of the output signal requires a different calibration, this may be done in the traditional way using homogenously wetted soil samples (Miller and Gaskin, 1999). Paired measurements of gravimetric water content w and output voltage V are needed, and V for dry and wet soils are used to 119 1,2 1,2 e?r m Volumetric water content, m 60 Permittivity 0,6 40 0,4 20 0,6 change of function at r = 36 0,4 0,0 0 20 40 60 Permittivity 0,0 0,0 0,4 0,6 1,0 Submerged rod length Fig. 4. LOM/RS probes performance: Manufacturer's conversion functions (cited in Tab. 2); Test of linearity: probe submergence in water; r ­ line represents the permittivity calculated using Eq. (3), v calc ­ the volumetric water content derived from the manufacturer's equations using the calculated r, v meas ­ the measured volumetric water content (LOM readings). Obr. 4. Správanie sa snímacov OM/RS: Kalibracná krivka výrobcu (citovaná v tab. 2); Test linearity: snímac ponorený vo vode; r ­ priamka reprezentuje permitivitu vypocítanú pomocou rov. (3), v calc ­ objemová vlhkos získaná pomocou kalibracnej krivky výrobcu s vypocítaným r, v meas ­ meraná objemová vlhkos (LOM). 3 Volumetric water content(LOM readings) LOM, m m v, m m 0.4 0.3 0.2 0.1 0 0.0 0.1 0.2 0.3 0.4 0.5 3 grav, m m 0.6 air dry 0.5sat 0.75sat saturation 1:1 line 0.7 Volumetric water content, m Sensor number mean, air dry mean, 0.5 sat mean, 0.75 sat mean, saturation Fig. 5. Tests of a set of 20 LOM/RS probes in a loam soil: LOM readings plotted against the gravimetrical estimations of the volumetric water content; Scatter of individual probe measurements against the mean value of the four tested water contents: in a nearly saturated soil, at 0.75 saturation, at 0.5 saturation and in an air-dry soil. Obr. 5. Testy 20 snímacov LOM/RS v hlinitej pôde: hodnoty odcítané z LOM-u v závislosti od objemovej vlhkosti pôdy; Rozptyl meraní jednotlivých snímacov v závislosti od priemernej hodnoty styroch vlhkostí pôdy, pri 0.75 nasýtenia, pri 0.5 nasýtenia a pre vzduchosuchú pôdu. 1,0 qcvc calc vm meas qm 1,0 Calibrating electromagnetic short soil water sensors Fig. 6. Theta probe performance: Set up for progressive insertion of a sensor in a coarse sand sample; Proof of linearity for recommended water content range. Obr. 6. Snímac Theta; zariadenie na ponorenie snímaca do hrubozrnného piesku, dôkaz linearity pre odporúcaný rozsah vlhkostí. calculate 0 ( a0) and W respectively. These then determine, in combination with the corresponding w, the coefficient a1 = W - 0 . W (4) Hydra probe The Hydra-probe has four output voltage channels. Permittivity r is calculated from three voltages, similar to a small network analyser for determining the components of impedance. The fourth voltage supplies the temperature, T, from a sensor in the probe head. Software transforms the raw voltages to the real and imaginary components of the complex permittivity , plus T, v and salinity. Further, temperature corrected values are given. This broad spectrum of soil information is very useful for calibration, but data logging is more laborious. v is calculated via a polynomial (Tab. 2). The supplier's software provides coefficients for three soil types (sand, silt and clay, see Fig. 7). A graph of r versus v enables judgement of the range of applicability of the standard values. Fig. 7 shows an example of our specific calibration results for organic waste material. It is obvious that standard polynomials are only useful for v < 0.5 m3 m. The function for silt shows more realistic values for the higher r, though a change from 30 to 50 is accompanied by almost no change in v. If the standard coefficients are unsatisfactory, new ones are relatively easily determined with the calibration procedure described above. For soil specific calibration it is important to determine the full saturation, so that no extrapolation is needed over the full measuring range. A polynomial of order > 3 is generally not justified. For municipal solid waste, a nearly linear relation was found using all data points of five test runs (Fig. 7). Using e.g. only one test run would give a different polynomial. MSW is an aggressive medium, which may progressively damage a probe, causing drift in calibration and requiring calibration checks. ECH2O probes Most EM sensors perform better in high-sand soils, so that calibration is particularly important for the ECH2O probe. According to the factory manual (Decagon, 2002; 2008), the ECH2O probe's voltage output is highly proportional to v, thus the calibration is a simple linear relation. The probes have low sensitivity to temperature, but output is a function of the supply voltage. For a 2.5 V excitation, equation constants are provided (Tab. 2). Two types of 121 probes were examined: EC-20 and EC-5 with probe lengths of about 20 and 5 cm. The linearity test for the EC-20 by submerging in water showed a fairly linear probe response, resulting in an appropriate linear regression (Fig. 8). The sensor appears to be very suitable for our calibration procedure, so media-specific calibration is easily performed. From the tests in a sand column (grain size 0.2­0.63 mm) with a controlled 2.5 V supply, use of the manufacturer's calibration substantially underestimated the actual water content (Fig. 8). The User's Manual (Campbell, 2005) does state that the standard calibration is not accurate for soils with high sand or salt content. However the large difference established (see Fig. 8) between the slope of our calibration curve (b1 = = 2.054, b0 = 0.834, R2 = 0.988) and the manufacturer's curve (b1 = 0.695, b0 = 0.29) is a clear example supporting our general argument of the need for user specific calibration. EC-5 probe has a different two-prong design than EC-20. The EC-5 measurement principle is the same as for EC-20, but the higher measurement frequency allows measurements of water content up to 100% (Decagon, 2008). Therefore, the accuracy is higher than for the EC-20 and at least 0.03 m3 m for almost all soil types with electrical conductivity (salinity) up to 8 dS m-1. The linearity test in water, however, showed a rather non-linear probe response for the whole range measured, but only at lower water contents (Fig. 8). The second test in glass bead columns (Dragonit 25, grain size 0.45 mm) was performed in three replications. The samples were prepared in a similar way as for the LOM/RS tests. The results also show a non-linear response relative to the reference water content, approximating the shape of the curve in the submerging test (Fig. 8). By contrast the manufacturer's equation between water content and voltage output is linear. Thus the calibration procedure using glass beads may not be the most appropriate in the case of the EC-5 probe, possibly because the large pores would result in a system with a `capillary front' between saturated material below and unsaturated material above. 0,6 0,4 fitted polynom Rep 1 Rep 2 Rep 3 Rep 4 Rep 5 0,0 0 10 20 30 40 50 60 Dielectric constant r Fig. 7. Hydra probe sensors: The manufacturer-supplied relationships of r versus v [%] for sand, silt and clay; the coefficients for the polynomials are cited in Tab. 2; Calibration for organic material from a municipal solid waste site. The polynomial was fitted to data averaged over five replicate samples (v expressed as a fraction). Obr. 7. Snímace Hydra; výrobcom dodaná závislos medzi r a v [%] pre piesok, prach a hlinu; koeficienty polynómov sú v tab. 2, kalibrácia pre organický materiál z verejnej skládky odpadov. Charakteristiky polynómu boli spriemerované z piatich opakovaní (v je vyjadrené v castiach z vlhkosti vodou nasýtenej pôdy). Fig. 8. ECH2O probe calibrations for: EC-20 probe submergence in water (test of linearity), and in sand; EC-5 probe submergence test in water and triplicate measurements in glass beads. Obr. 8. Kalibrácia snímaca ECH2O pre: EC-20 ponorenú vo vode (test linearity) a v piesku; pre EC-5 ponorený vo vode a meranie v sklených guôckach s trojnásobným opakovaním. Conclusions Measuring water content of composite media such as soils, composts or wastes presents a challenge for many applications. EM probe users should be aware that, for a given volumetric water content, probe output can vary, often strongly, with properties such as texture, organic matter content, electrical conductivity and temperature; and hence that a single manufacturer-supplied calibration may be unreliable, or may only have been derived for a limited range of wetness. Hence users should have transparent, easily implemented calibration procedures. We have presented above simple, easily performed lab calibration procedures which provide good insight into system performance, and have illustrated them for five short-length probes. Soil-specific lab tests are strongly recommended for both equipment familiarisation, and for testing sensor applicability for a specific task. Field calibration is sometimes essential. Each EM sensor has its own optimum range of application, the limits of which have to be carefully assessed. Plotting the signal-to-v conversion function is an easy method to identify the sensitivity and the best range of a sensor's application. List of symbols a0, a1, b0, b1 ­ calibration coefficients, ­ polynomial constants, Ci i ­ subscript denoting phase (i = s, w, a for solids, water, air), TDR ­ Time-Domain Reflectometry, ­ volume fraction of phase (inclusion) i, Vi V ­ output voltage, ­ shape factor, r ­ relative dielectric permittivity (real part), i ­ relative permittivity of phase i (i = a, w, s for air, water, solids, respectively), ­ porosity, v, w ­ volumetric, gravimetric water content, ­ pulse or square wave period. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Journal of Hydrology and Hydromechanics de Gruyter

Loading next page...
 
/lp/de-gruyter/calibrating-electromagnetic-short-soil-water-sensors-sKMhdedErR
Publisher
de Gruyter
Copyright
Copyright © 2010 by the
ISSN
0042-790X
DOI
10.2478/v10098-010-0011-3
Publisher site
See Article on Publisher Site

Abstract

J. Hydrol. Hydromech., 58, 2010, 2, 114­125 DOI: 10.2478/v10098-010-0011 WILLIBALD LOISKANDL1), GRAEME D. BUCHAN2), WOLFGANG SOKOL1), VILIAM NOVAK3), MARGARITA HIMMELBAUER1) 1) University of Natural Resources and Applied Life Sciences, Institute of Hydraulics and Rural Water Management, Muthgasse 18, A-1190 Vienna, Austria. 2) Soil & Physical Sciences Department, Lincoln University, Canterbury, New Zealand. 3) Institute of Hydrology, Slovak Academy of Sciences, Racianska 75, 831 02 Bratislava, Slovak Republic. Correspondance: W. Loiskandl, Mailto: loiskandl@mail.boku.ac.at The use of electromagnetic (EM) soil moisture probes is proliferating rapidly, in two broad domains: in field and laboratory research; and in strongly practical applications such as irrigation scheduling in farms or horticultural enterprises, and hydrological monitoring. Numerous commercial EM probes are available for measurement of volumetric water content (v), spanning a range of measurement principles, and of probe dimensions and sensing volumes. However probe calibration (i.e. the relationship of actual v to probe electrical output) can shift, often substantially, with variations in parameters such as soil texture, organic matter content, wetness range, electrical conductivity and temperature. Hence a single-valued, manufacturer-supplied calibration function is often inadequate, forcing the user to seek an application-specific calibration. The purpose of this paper is to describe systematic procedures which probe users can use to check or re-determine the calibration of their selected probe(s). Given the wide diversity of operating principles and designs of commercially-available EM probes, we illustrate these procedures with results from our own calibrations of five different short probes (length of 5 to 20 cm). Users are strongly recommended to undertake such calibration checks, which provide both pre-use experience, and more reliable in-use data. KEY WORDS: Soil Moisture Probes, Electromagnetic Probes, TDR, Calibration. Willibald Loiskandl, Graeme D. Buchan, Wolfgang Sokol, Viliam Novak, Margarita Himmelbauer: KALIBRÁCIA ELEKTROMAGNETICKÝCH SNÍMACOV VLHKOSTI PÔDY. J. Hydrol. Hydromech., 58, 2010, 2; 29 lit., 8 obr., 2 tab. Pouzívanie elektromagnetických (EM) snímacov vlhkosti pôdy sa rýchlo rozsiruje tak v terénnom výskume, ako aj v laboratóriu. Sú pouzívané v praktických aplikáciách ako je riadenie závlah na farmách a záhradách, ako aj v hydrologickom monitoringu. Pre meranie vlhkosti pôdy (v) sú dostupné pocetné typy komercných EM snímacov, zalozených na viacerých princípoch merania a snímace majú rozdielnu vekos snímaných objemov pôdy. Kalibracné krivky takýchto snímacov (t.j. závislosti medzi reálnou vlhkosou pôdy v a elektrickým výstupom snímac sa môzu posúva ­ niekedy podstatne ­ a to v závislosti od rozdielnych parametrov pôdy, ako je jej textúra, obsah organických látok, rozsah vlhkostí, elektrická vodivos a teplota. Z toho vyplýva, ze jednoznacná kalibracná krivka, dodávaná výrobcom je casto neadekvátna, co núti uzívatea snímac kalibrova v specifických podmienkach. Cieom tohto príspevku je opísa procedúry, ktoré môzu by pouzité uzívatemi pri rekalibrácii vybraných typov snímacov. Berúc do úvahy sirokú paletu princípov EM snímacov, ilustrujeme tieto procedúry výsledkami vlastných kalibracných testov na piatich typoch krátkych snímacov (dzka od 5 do 20 cm). Uzívateom odporúcame rekalibráciu komercných snímacov, ktorými získajú predbezné skúsenosti a spoahlivejsie výsledky pri meraní vlhkosti pôdy. KÚCOVÉ SLOVÁ: snímace vlhkosti pôdy, elektromagnetické snímace, TDR, kalibrácia. Introduction For many applications of soil water sensors, it is insufficient to use a sensor in the field without pre114 vious laboratory tests and a critical evaluation of the manufacturer's calibration. One aspect often overlooked is that the obtained value of volumetric water content (v) is derived from a surrogate measured electromagnetic (EM) parameter. The relationship between v and the parameter is central to the calibration procedure. Most common is the Topp relation between v and relative permittivity r (Topp et al., 1980), valid for a wide range of applications. However, in general, standard calibrations are good only for the conditions they are derived for (e.g. for a sand) and over a limited water content range. EM signal propagation is affected by several soil physical properties, especially colloid (clay and organic matter) contents, electrical conductivity, bulk density and temperature (Jacobsen & Schjonning, 1993; Malicki et al., 1996; O'Connor and Dowding, 1999; Gong et al., 2003; Yuanshi et al., 2003). The authors have personal experiences of being consulted to explain obviously wrong readings (e.g., at one extreme, v >100 % for a saturated clay soil). Results of other authors also clearly demonstrate the need for adjusting or testing the validity of standard calibrations (e.g. Groves and Rose, 2004; Blanc and Dick, 2003). Thus users should not rely uncritically on manufacturers' or standard calibrations, and also recognise that reading accuracy is determined mainly by the specific field soil conditions, and not only by the signal accuracy of the recording system. When installing sensors, it takes little extra effort to take check soil samples for gravimetric analyses. It is also recommended to take samples over a wide range of soil wetness. Key criteria for sensor selection include: ease of use; reliability; accuracy; range of application; automatic recording capability; sensing volume; the risk of disturbance errors (e.g. due to installation procedure, or in stony soils); and cost. Accuracy is especially critical, and relies on good calibration. Manufacturer's or standard accuracy specifications often pertain to a specific electronic system, and to measurement under very controlled conditions (e.g. in a sand), but need not be valid for the user's conditions, especially in non-standard applications like in soils high in colloids (clay or organic matter, e.g. Burgess et al., 2006), or saline soils, or in materials like organic waste. This paper presents procedures for calibration of EM short soil moisture probes, and illustrates them with results for five commercial probes. Calibration methods Accurate calibration is complicated by the need to reproduce conditions similar to those in the final application, and avoid uncontrolled variations in v in the calibration soil sample. To ease and improve calibration, we recommend and describe here an automatic procedure, based on Young et al. (1997) and as applied by Loiskandl et al. (2003). However this method has limitations. The most important constraint is the sensor rod length. Hence we distinguish here between short and long sensors. While there is no exact dividing point, for practical applications short sensors are considered to be 20 cm, and typically 10 cm or shorter. This paper focuses on calibration of short sensors. Calibration of short sensors For short sensors, a soil sample in a cylinder containing the probe is placed on a weighing scale. Water is supplied by dynamic upward infiltration by an adjustable pump (Fig. 1). The weight and sensor reading are recorded simultaneously, providing paired values of average v (determined gravimetrically) and sensor reading. The soil sample height and sensor length should be the same to ensure equal averaging. Because this calibration procedure imposes an axial moisture gradient along the sensor, it will strictly succeed only for probes which have a linear relation between the output signal and actual v, across the range of v occurring in the sample. A simple test of linearity is to partly submerge the sensor rod in water, e.g. 50 % immersion should give a v of 50 %. Fig. 2 shows results for two of the probes described below: Hydra-probe: a linear relationship was found over the whole rod length; and Theta probe: the relation is not linear, but is nearly linear for the main range of r encountered in practice. Soil permittivity can be represented using a three-phase concept (Woodhead et al., 2003): r = (Vi i ) 1/ , where phase i (= s, w, a for solids, water, air) has volume fraction Vi and permittivity i. Hence (1) r = [ v + (1 ­ ) + ( ­ v ) a ] 1/ , (2) w s where is porosity, v ­ volumetric water content (range 0 to 1) and ­ a shape factor. Sensor Data logger Soil sample in cylinder Data storage and conversion software 2 x RS 232 Balance RS 232 Interface continuous supply Fig. 1. Set-up for short sensor calibration (example with Hydra-probe). The sample is progressively wetted by upward infiltration of pumped water. Obr. 1. Zariadenie pre kalibráciu krátkeho snímaca (s príkladom snímaca Hydr. Vzorka je postupne navlhcovaná infiltráciou tlakovej vody zospodu. Fig. 2. Test of linearity: relationship between probe length submersed in water and probe output dielectric permittivity, r. HydraProbe response is linear, while Theta-Probe becomes non-linear at high water content. Obr. 2. Test linearity; vzah medzi dzkou vzorky ponorenej do vody a výstupom r (dielektrická permitivit. Snímac Hydra ­ závislos je lineárna, Theta snímac vykazuje pri vyssích vlhkostiach nelineárnu závislos. For probe submergence in water, = 1 and also = 1. Hence r = [v w + (1 ­ v ) a ] . With w = 81 and a = 1 this gives the linear relation r = (80v + 1) . Calibration of long sensors (3) For long sensors, the same procedure can be applied, but factors limiting its application are as follows. 1. Any sensor leads to different spatial averaging, compared to the gravimetric method. 2. For 116 samples with low hydraulic conductivity (especially clays or compact silts), the time to establish a range of water contents is very long, or even impractical. 3. The length in combination with the lateral zone of influence leads to a heavy soil sample, requiring a balance weighing this load with high accuracy. Therefore other procedures are recommended, e.g. using a re-packed column with pre-wetted soil. Field calibration is also possible, but takes greater time and effort to cover a sufficient range of v (e.g. Whalley et al., 2004). Further, if the sensor is installed for continuous monitoring, samples must be taken near the probe but outside its sensing volume. TRIME-Probe Overview of the five calibrated types of short sensors Since 1990, a wide range of EM soil moisture systems have appeared. For reviews see Topp and Ferre (2002), Evett (2000; 2005), Charlesworth (2000), Robinson et al. (2003). Here we focus on the five short sensors listed in Tab. 1. For the user, most important is a clear definition of what is measured and its relation to soil moisture. Hence Tab. 1 shows the EM principles of the sensors, and Tab. 2 the manufacturers' conversion relations. Calibration results for the five selected sensors For the above sensors, the standard calibrations are analysed here and compared to measurement results. Also summarized are the records or logging procedures used. The TRIME TDR system gives only v, but is user-friendly. Data transfer is very reliable and, due to the field bus/RS232 system, possible over long distances. User calibration is catered for, and requires special software (IMKO SMCAL Version 2.1, 1992). Independently measured v are compared with TRIME-Probe values and the polynomial coefficients (Tab. 2) are optimised, then programmed into the probe. Thus each probe can be adjusted to specific conditions, although calibrated sensors are then not easily interchanged. To analyse the standard function, a column filled with sand was used. SM-CAL proved to be a very valuable tool to obtain good results. First, the probe was calibrated (i.e. the polynomial coefficients were optimised) in glass beads. This calibration turned out to be very suitable for the sand. Fig. 3 shows results for 11 different v values, starting at about 45 % and progressing to dry conditions. This calibration was then used in a soil-organic material mix. The larger deviations observed clearly demonstrate the need to obtain a medium-specific calibration (Fig. 3). T a b l e 1. Operating principles and data conversion methods of the five types of sensors. T a b u k a 1. Princípy funkcie (measuring concept) a metódy konverzie údajov piatich typov snímacov. Sensor type/supplier TRIME-Probe/ IMKO LOM/RS EASY TEST ThetaProbe/Delta T Devices Measuring concept Travel time of a pulse along wave guides (Fundinger et al., 1995). Signal attenuation is reduced by coating the rods with a non-conductive material. is converted to v in an electronic device using inbuilt coefficients. As for the TRIME, but uncoated sensor rods. A single-pole-n-throw high frequency multiplexer connects sensors to a recording unit. Software converts data and transfers data to the recording unit. The Theta-Probe (ML2x) uses a simplified standing wave measurement to determine the impedance of a 4-rod sensing array (Delta-T Devices, 1999). The standing wave originates from a difference of the internal transmission line impedance and the impedance formed by the array embedded in soil. Measures the complex dielectric constant, and resolves the capacitive and conductive parts of electrical response. The capacitive part is most indicative of soil moisture, while the conductive part reflects mainly soil conductivity (Vitel Inc, 1994). Measures dielectric permittivity and provides a voltage output easily integrated into data logging systems (Decagon, 2002; 2008). Sensors with probe lengths of 20 cm (EC-20) and 5 cm (EC-5) are tested. Hydra-Probe/Vitel. Inc. ECH2O Probes, Decagon Devices, Inc. 4 deviation 3 2 0,50 1 Vol.% -0,60 0 -1 -2 -4 1 2 3 4 5 6 Reading 7 8 9 10 11 -0,70 -0,60 0,70 0,70 -0,30 -0,40 + 3 Vol.% 4 deviation 3 2,10 2 1 0 -1 -0,50 -1,40 SAND - 3 Vol.% -2 -2,00 -4 ,10 1 ,00 2 3 4 5 -1,00 3,50 3,20 1,70 1,60 + 3Vol.% 0,90 0 -0,90 ORGANIC MATERIAL - 3 Vol.% 6 Reading Fig. 3. Deviations (v in %) of TRIME readings for two medium types: sand and soil-organic material mix, as obtained by Dreiseitel (1997). Calibration parameters were obtained for glass beads, and gave good results for the sand. For the soil-organic mix, the larger deviations demonstrate the need to obtain a medium-specific calibration. Obr. 3. Odchýlka (v v %) snímaca TRIME pre dva typy prostredí: piesok, zmes organického materiálu, poda Dreiseitela (1997). Kalibracné parametre boli získané pre sklené guôcky, s dobrými výsledkami pre piesok. Väcsie odchýlky pre organickú pôdu demonstrujú potrebu získania osobitnej kalibracnej krivky. Laboratory operated meter for recording soil moisture, LOM/RS This TDR equipment was originally designed for controlling long-term laboratory experiments on soil columns, and can be combined with mini tensiometers to provide paired v and suction data for a retention curve (Malicki et al., 1996). The system is controlled by a PC. The supplied software operates the system and performs data conversion (r to v) using three equations with fixed coefficients (Tab. 2). Nevertheless, the conversion functions gave an almost linear relationship to water content (Fig. 4). To evaluate sensor response in different dry to wet conditions, tests in water and in loam soil were performed. Tests in water: 1. The uniformity of a set of 20 sensors was tested fully submerged in water; 2. To test for linearity, the sensors were fixed and plates of 5.1 mm thickness were placed under the cup (inside diameter of 5 cm), giving stepwise increase of the submerged depth. The tests were carried out at a constant temperature of 20oC and with the probes maintained in a vertical position. Triplicate readings were taken. The results are shown in Fig. 4. LOM/RS probes gave a very unfavourable response to the linearity test in water, with reasonable results only for submergence > 50 %. The most probable cause for this is a secon118 dary reflection occurring at the interface between air and water when sensors are submerged less than the resolution length of the system. By contrast, using the r values estimated with the three-phase concept (Eq. (3)) gave results very close to the theoretical 1 : 1 line in Fig. 4. In order to the test the sensors before their use in a pot experiment (Himmelbauer and Loiskandl, 2007), measurements were carried out in sand and a loam soil (38 % sand, 41 % silt, 21 % clay). The results in loam soil are presented here. Soil samples were packed into 6 cm high cylinders with inside diameter of 5 cm to a bulk density of 1.2 g cm. The measurements were conducted in air-dry soil (hygroscopic water content of 0.02 m3 m), at 0.5 saturation (0.27 m3 m), at 0.75 saturation (0.41 m3 m), and in a nearly saturated soil (0.51 m3 m). The estimated total porosity was 0.54, but in practice the maximum v achieved was slightly below that. As before, triplicate TDR measurements were taken. The gravimetric method supplied the reference values. As can be seen on Fig. 5, the mean offset of the readings (of 20 sensors) from the reference values remained below 5 % for all levels of saturation tested, i.e. 1 % ± 1.6 for the dry soil, ­3.6 % ± 2.1 for the 0.5 saturation, 2.2 % ± 1.7 for the 0.75 saturation and 4 % ± 2.8 for the completely saturated soil. The poorer accuracy at the high water content was probably a result of Calibrating electromagnetic short soil water sensors T a b l e 2. Electrical outputs and manufacturer equations for conversion to v, for the five types of sensors. T a b u k a 2. Elektrický výstup a rovnice pre konverziu na v (kalibracné rovnice) dodané výrobcom, pre pä typov snímacov. Sensor type TRIME LOM/RS EASY TEST Output Digital/Fieldbus IMP232-Micronet Digital/RS232 Manufacturer calibration equations Polynomial, order selectable to max order of 5 v = 0 if r < 1.48733 (special communication = 0.106387 * - 0.158247 if < 6 v r r protocol) v = 1 - (1 - 0.106387 * 6 - 0.158247) * (9 - r ) / 3 if r 6 Voltage V (0­1V) Theta-probe v = [1.1 + 4.4V ] - a0 a1 or v = Hydra-probe 4 Voltages (0­ 5V)/Permittivity 1.07 + 6.4V - 6.4V 2 + 4,7V 3 - a0 a1 v = C0 + C1 r + C2 r 2 + C3 r 3 sand: C0 = -8.63, C1 = 3.216, C2 = -9.54*10-2, C3 = 1.579*10 silt: C0 = -13.04, C1 = 3.819, C2 = -9.129*10-2, C3 = 7.342*10-4 clay: C0 = -20.93, C1 = 6.553, C2 = -0.246, C3 = 3.241*10-2 ECH2O probes: EC-20 EC-5 Voltage (250­1000mV) v = b1V - b0 for mineral soils: b1 = 0.695, b0 = 0.29 b1 = 1.19, b0 = 0.401 r ­ real dielectric constant V ­ output voltage a0, a1, b0, b1 ­ calibration coefficients v ­ volumetric water content Ci ­ polynomial constants ­ square wave period the vertical probe position and some reflection effects. Nevertheless, the reading deviations from the means are of the order of the accuracy stated in the manual (Fig. 5). For horizontal installation, as typical in pot experiments, more favourable conditions can be expected, since at a certain depth a homogenous moisture distribution is easier achieved than in a vertical direction. For exact measurements, a medium-specific calibration is highly recommendable. Theta probe For the Theta probe, the amplitude of a difference voltage V (Tab. 2) gives its relative impedance, hence r and thus a measure of v. The data logger needs conversion parameters, either from the manufacturer's calibration or from soil-specific calibration. The calibration method proposed above may be used if linearity of response is assumed for an output signal range 0 ­ 1 Volt, equivalent to v of 0 to 0.5 m3 m. In view of the strong non-linearity of the Theta probe at high v (Fig. 2), instead of submerging in water, it was progressively inserted in a cylinder of saturated coarse (1 mm) sand. The sensor was fixed vertically and a mobile platform was raised stepwise (Fig. 6). This method is discussed by Quinones and Ruelle (2001), who found good agreement (for Campbell Scientific CS615 probes) between progressively inserting the probe, discontinuous mixing of soil with water, and wetting the column from the top with a dripper. Fig. 6 shows two test results. The saturated v according to the added water was 0.36 and 0.34 m3 m for tests 1 and 2 respectively. Corresponding calculated values from the Theta probe were 0.36 m3 m for both tests. r was calculated with the third order polynomial (Tab. 2) and for conversion to v the general calibration mineral soil coefficients were used. If non-linearity of the output signal requires a different calibration, this may be done in the traditional way using homogenously wetted soil samples (Miller and Gaskin, 1999). Paired measurements of gravimetric water content w and output voltage V are needed, and V for dry and wet soils are used to 119 1,2 1,2 e?r m Volumetric water content, m 60 Permittivity 0,6 40 0,4 20 0,6 change of function at r = 36 0,4 0,0 0 20 40 60 Permittivity 0,0 0,0 0,4 0,6 1,0 Submerged rod length Fig. 4. LOM/RS probes performance: Manufacturer's conversion functions (cited in Tab. 2); Test of linearity: probe submergence in water; r ­ line represents the permittivity calculated using Eq. (3), v calc ­ the volumetric water content derived from the manufacturer's equations using the calculated r, v meas ­ the measured volumetric water content (LOM readings). Obr. 4. Správanie sa snímacov OM/RS: Kalibracná krivka výrobcu (citovaná v tab. 2); Test linearity: snímac ponorený vo vode; r ­ priamka reprezentuje permitivitu vypocítanú pomocou rov. (3), v calc ­ objemová vlhkos získaná pomocou kalibracnej krivky výrobcu s vypocítaným r, v meas ­ meraná objemová vlhkos (LOM). 3 Volumetric water content(LOM readings) LOM, m m v, m m 0.4 0.3 0.2 0.1 0 0.0 0.1 0.2 0.3 0.4 0.5 3 grav, m m 0.6 air dry 0.5sat 0.75sat saturation 1:1 line 0.7 Volumetric water content, m Sensor number mean, air dry mean, 0.5 sat mean, 0.75 sat mean, saturation Fig. 5. Tests of a set of 20 LOM/RS probes in a loam soil: LOM readings plotted against the gravimetrical estimations of the volumetric water content; Scatter of individual probe measurements against the mean value of the four tested water contents: in a nearly saturated soil, at 0.75 saturation, at 0.5 saturation and in an air-dry soil. Obr. 5. Testy 20 snímacov LOM/RS v hlinitej pôde: hodnoty odcítané z LOM-u v závislosti od objemovej vlhkosti pôdy; Rozptyl meraní jednotlivých snímacov v závislosti od priemernej hodnoty styroch vlhkostí pôdy, pri 0.75 nasýtenia, pri 0.5 nasýtenia a pre vzduchosuchú pôdu. 1,0 qcvc calc vm meas qm 1,0 Calibrating electromagnetic short soil water sensors Fig. 6. Theta probe performance: Set up for progressive insertion of a sensor in a coarse sand sample; Proof of linearity for recommended water content range. Obr. 6. Snímac Theta; zariadenie na ponorenie snímaca do hrubozrnného piesku, dôkaz linearity pre odporúcaný rozsah vlhkostí. calculate 0 ( a0) and W respectively. These then determine, in combination with the corresponding w, the coefficient a1 = W - 0 . W (4) Hydra probe The Hydra-probe has four output voltage channels. Permittivity r is calculated from three voltages, similar to a small network analyser for determining the components of impedance. The fourth voltage supplies the temperature, T, from a sensor in the probe head. Software transforms the raw voltages to the real and imaginary components of the complex permittivity , plus T, v and salinity. Further, temperature corrected values are given. This broad spectrum of soil information is very useful for calibration, but data logging is more laborious. v is calculated via a polynomial (Tab. 2). The supplier's software provides coefficients for three soil types (sand, silt and clay, see Fig. 7). A graph of r versus v enables judgement of the range of applicability of the standard values. Fig. 7 shows an example of our specific calibration results for organic waste material. It is obvious that standard polynomials are only useful for v < 0.5 m3 m. The function for silt shows more realistic values for the higher r, though a change from 30 to 50 is accompanied by almost no change in v. If the standard coefficients are unsatisfactory, new ones are relatively easily determined with the calibration procedure described above. For soil specific calibration it is important to determine the full saturation, so that no extrapolation is needed over the full measuring range. A polynomial of order > 3 is generally not justified. For municipal solid waste, a nearly linear relation was found using all data points of five test runs (Fig. 7). Using e.g. only one test run would give a different polynomial. MSW is an aggressive medium, which may progressively damage a probe, causing drift in calibration and requiring calibration checks. ECH2O probes Most EM sensors perform better in high-sand soils, so that calibration is particularly important for the ECH2O probe. According to the factory manual (Decagon, 2002; 2008), the ECH2O probe's voltage output is highly proportional to v, thus the calibration is a simple linear relation. The probes have low sensitivity to temperature, but output is a function of the supply voltage. For a 2.5 V excitation, equation constants are provided (Tab. 2). Two types of 121 probes were examined: EC-20 and EC-5 with probe lengths of about 20 and 5 cm. The linearity test for the EC-20 by submerging in water showed a fairly linear probe response, resulting in an appropriate linear regression (Fig. 8). The sensor appears to be very suitable for our calibration procedure, so media-specific calibration is easily performed. From the tests in a sand column (grain size 0.2­0.63 mm) with a controlled 2.5 V supply, use of the manufacturer's calibration substantially underestimated the actual water content (Fig. 8). The User's Manual (Campbell, 2005) does state that the standard calibration is not accurate for soils with high sand or salt content. However the large difference established (see Fig. 8) between the slope of our calibration curve (b1 = = 2.054, b0 = 0.834, R2 = 0.988) and the manufacturer's curve (b1 = 0.695, b0 = 0.29) is a clear example supporting our general argument of the need for user specific calibration. EC-5 probe has a different two-prong design than EC-20. The EC-5 measurement principle is the same as for EC-20, but the higher measurement frequency allows measurements of water content up to 100% (Decagon, 2008). Therefore, the accuracy is higher than for the EC-20 and at least 0.03 m3 m for almost all soil types with electrical conductivity (salinity) up to 8 dS m-1. The linearity test in water, however, showed a rather non-linear probe response for the whole range measured, but only at lower water contents (Fig. 8). The second test in glass bead columns (Dragonit 25, grain size 0.45 mm) was performed in three replications. The samples were prepared in a similar way as for the LOM/RS tests. The results also show a non-linear response relative to the reference water content, approximating the shape of the curve in the submerging test (Fig. 8). By contrast the manufacturer's equation between water content and voltage output is linear. Thus the calibration procedure using glass beads may not be the most appropriate in the case of the EC-5 probe, possibly because the large pores would result in a system with a `capillary front' between saturated material below and unsaturated material above. 0,6 0,4 fitted polynom Rep 1 Rep 2 Rep 3 Rep 4 Rep 5 0,0 0 10 20 30 40 50 60 Dielectric constant r Fig. 7. Hydra probe sensors: The manufacturer-supplied relationships of r versus v [%] for sand, silt and clay; the coefficients for the polynomials are cited in Tab. 2; Calibration for organic material from a municipal solid waste site. The polynomial was fitted to data averaged over five replicate samples (v expressed as a fraction). Obr. 7. Snímace Hydra; výrobcom dodaná závislos medzi r a v [%] pre piesok, prach a hlinu; koeficienty polynómov sú v tab. 2, kalibrácia pre organický materiál z verejnej skládky odpadov. Charakteristiky polynómu boli spriemerované z piatich opakovaní (v je vyjadrené v castiach z vlhkosti vodou nasýtenej pôdy). Fig. 8. ECH2O probe calibrations for: EC-20 probe submergence in water (test of linearity), and in sand; EC-5 probe submergence test in water and triplicate measurements in glass beads. Obr. 8. Kalibrácia snímaca ECH2O pre: EC-20 ponorenú vo vode (test linearity) a v piesku; pre EC-5 ponorený vo vode a meranie v sklených guôckach s trojnásobným opakovaním. Conclusions Measuring water content of composite media such as soils, composts or wastes presents a challenge for many applications. EM probe users should be aware that, for a given volumetric water content, probe output can vary, often strongly, with properties such as texture, organic matter content, electrical conductivity and temperature; and hence that a single manufacturer-supplied calibration may be unreliable, or may only have been derived for a limited range of wetness. Hence users should have transparent, easily implemented calibration procedures. We have presented above simple, easily performed lab calibration procedures which provide good insight into system performance, and have illustrated them for five short-length probes. Soil-specific lab tests are strongly recommended for both equipment familiarisation, and for testing sensor applicability for a specific task. Field calibration is sometimes essential. Each EM sensor has its own optimum range of application, the limits of which have to be carefully assessed. Plotting the signal-to-v conversion function is an easy method to identify the sensitivity and the best range of a sensor's application. List of symbols a0, a1, b0, b1 ­ calibration coefficients, ­ polynomial constants, Ci i ­ subscript denoting phase (i = s, w, a for solids, water, air), TDR ­ Time-Domain Reflectometry, ­ volume fraction of phase (inclusion) i, Vi V ­ output voltage, ­ shape factor, r ­ relative dielectric permittivity (real part), i ­ relative permittivity of phase i (i = a, w, s for air, water, solids, respectively), ­ porosity, v, w ­ volumetric, gravimetric water content, ­ pulse or square wave period.

Journal

Journal of Hydrology and Hydromechanicsde Gruyter

Published: Jun 1, 2010

References