Simultaneous use of newly adopted simple sensors for continuous measurements of soil moisture and salinity.
Salinity caused by the presence of shallow saline groundwater is widely acknowledged to adversely affect production of the lands around the world, especially in arid and semi-arid regions (Kandiah 1990). The upward evaporative flux from a shallow saline water table results in the accumulation of salts at or near the soil surface. In investigating the sustainability of agricultural systems in arid and semi-arid regions, it is essential to evaluate accurately rates of evaporation and salt accumulation. In experimental studies of saline soils in these regions, methods available to monitor salinity and moisture content with fine spatial resolution are limited. This is because, in arid an semi-arid regions, evaporative demand of the air is greater than the ability of the soil to conduct water; surface layers of the soil profile become very dry and also saline due to transport of salt from the saline water table. In this way, high salinities and a wide range of moisture content (i.e. from saturation near the water table to almost air-dry in the surface layers) occur. Several methods are used to determine soil water content or matric potential, but none of them is completely satisfactory (Campbell and Mulla 1990). The gravimetric method is not desirable because of the time and labour needed and because it is destructive.
This paper describes the design, development, and testing of 2 instruments to monitor continuously water and salinity profiles in a soil column. These were thermal conductivity and 4(multi)-electrode resistivity probes to monitor soil water content and salinity, respectively. Both probes are commercially available but are too large for detailed measurements. Therefore, they were designed, manufactured, and adapted to the experimental set-up to measure water content and salinity with a fine spatial resolution. The construction, calibration, and performance of the probes are described. The accuracy, reliability, stability, and variability were tested over a wide range of volumetric water contents, at high salinities, at different temperatures, and for 2 different soil types (clay loam and sandy loam).
Construction of the instruments
Four-electrode probe construction
Each probe consisted of a plastic tube, 35 mm in diameter and 310 mm long, having 30 stainless steel rings spaced equally (10 mm from center to center) along the probe. Construction details for the 4-electrode probe are given in Fig. 1. A current is applied by an ammeter across the outer pair of any 4 adjacent electrodes and the corresponding voltage drop is measured by a voltmeter between the inner pair. This voltage drop is converted to the soil bulk electrical conductivity (E[C.sub.a]) and then soil solution electrical conductivity (E[C.sub.w]) as explained below (see Theory section).
[FIGURE 1 OMITTED]
Thermal conductivity probe construction
Several designs have been described in publications referred to by Wechsler et al. (1965). The probe described by Fritton et al. (1974) was relatively easy to build and used readily available materials. Therefore, it was adopted as shown in Fig. 2.
[FIGURE 2 OMITTED]
It consisted of a heater element, a thermocouple, and a protective covering. A K-type commercial thermocouple was used (RS components) to measure temperature.
A multirange microammeter was used to measure the voltage supplied, and the temperature of the thermocouple was read by a digital thermometer to an accuracy of 0.1 K. Determination of volumetric water content from the temperature measurements is described in the Theory section.
Theory for 4-electrode probe measurements
Determination of E[C.sub.w] from E[C.sub.a] measurements using a 4-electrode probe has been studied extensively by Nadler and Frenkel (1980), Nadler (1981), and Rhoades et al. (1989). Here, the following procedure was used.
By employing an appropriate cell constant, c, it is possible to determine the specific electrical conductivity from the resistance and temperature measurements (American Public Health Association 1985) as:
(1) E[C.sub.25] = 1000c/R[1 + 0.0191(T-25)]
where E[C.sub.25] is electrical conductivity of the sample (dS/m) at 25[degrees]C; c is cell constant (1/m), which represents the specific resistance measured by the electrodes; T is temperature of the measurements ([degrees]C); R is measured resistance of the sample ([OMEGA]), which may be expressed as R = V/I; V is potential difference between 2 inner electrodes (V); and I is current (A).
When the probes are placed in a moist soil, E[C.sub.a] is measured. E[C.sub.a] depends upon the soil moisture content ([theta]), E[C.sub.w], and the volume conductance contribution of the surface conductance of the clay, E[C.sub.s]. Therefore, a correction has to be applied to the measured E[C.sub.a] because the desired electrical conductivity is that of the soil solution (E[C.sub.w]). The E[C.sub.a] - E[C.sub.w] relationship is described by Nadler (1982) as:
(2) E[C.sub.a] = E[C.sub.w]/[F.sub.s] + E[C.sub.s]
where [F.sub.s] is a shape factor of the clay at saturation defined as a geometric term related to the porosity ([phi]) by Burger (1919) and Nadler (1982). It is expressed as:
(3) [F.sub.s] = 1 + k 1 - [phi]/[phi]
where k is the shape factor that depends on the ratio of the axes of the particles. It can be calculated if [F.sub.s] is known for water-saturated soil conditions. Having obtained k, [F-[theta]] relationships (for unsaturated conditions) may be obtained through representing the porosity term by volumetric water content. Porosity, [phi], is equal to the volumetric water content at saturation. Therefore, water contents below saturation can be represented as 'effective porosity for electric conductance' (Nadler 1982). Thus:
(4) [F.sub.c] = 1 + k 1 - [theta]/[theta]
where [F.sub.c] is the F factor for a range of high water contents where Eqn 4 applies. The limits of this range were discussed by Nadler (1982).
Theory for thermal conductivity probe
Thermal conductivity measurements are based upon the rate at which heat is dissipated from the heating element and are directly affected by the amount of moisture in the soil or porous medium around the heat source (Shaw and Baver 1939; Bloodworth and Page 1975). Water is a better heat conductor than air. More heat will be dissipated as the water content increases in the soil. This will result in a temperature rise in the soil around the probe (Fredlung 1992).
To measure the water content, the thermal conductivity of the surrounding soil, which is calculated from the temperature measurements, must be calibrated against the water content, [theta]. The theory to calculate thermal conductivity from temperature measurements was given by Jackson and Taylor (1986). The temperature differences during the time of cooling depend on the thermal conductivity as:
(5) T - [T.sub.0] = (q/4[pi]K)(a + lnt)
where [T.sub.0] is original temperature before applying current ([degrees]C); T is temperature observed during cooling time after stopping the heating current ([degrees]C); q is amount of heat produced (W); K is thermal conductivity of soil (W/mK); a is a constant independent of time; and t is time (s)
A plot is made of [DELTA]T = (T - [T.sub.0]) v. lnt. For large values of t, a plot of [DELTA]T = (T - [T.sub.0]) v. lnt will produce a straight line. The thermal conductivity, K, is then calculated from the slope of this line. The theoretical slope, S = [DELTA](T - [T.sub.0])/[DELTA]lnt, may be derived from Eqn 5 as:
(6) S = q/4[pi]K
The amount of heat 'q' that is produced when a current 'I' passes through the resistor of resistance 'R' is given by:
(7) q = [I.sup.2]R
Using Ohm's law the resistance in the equation is found as R = V/I, where V is the voltage. Substituting [I.sup.2]R for q and rearranging Eqn 6 yields:
(8) K = 0.0796 [I.sup.2] R/S
Four-electrode probe calibration
First, the cell constant c was determined through the standard method presented in American Public Health Association (1985). Following this procedure, the probe was placed in a standard potassium chloride (KCl) solution (0.01 N = 0.745 g/L = 1.165 dS/m). By measuring the resistance for each set of 4 electrodes and noting the temperature, the cell constant for every switch position was separately calculated from Eqn 1. Then the probes were calibrated by placing them in solutions of salts (NaCl), which varied in concentrations from distilled water to 70 dS/m. The aim of choosing such a wide range of salinity, up to 70 dS/m, was to investigate the range in which the probe performed. A calibration curve for each electrode group or switch position (for solution only) was obtained.
Using the specific cell constants, the performance of the probes was first tested to measure the salinity of a solution. There was a very good linear correlation ([R.sup.2] = 0.98) between electrical conductivity E[C.sub.w] and salt concentration C, such that E[C.sub.w] = a + bC. Fig. 3 represents the calibration curve of the probes before (a = -0.143 [+ or -] 0.561, b = 94.50 [+ or -] 2.865, and [R.sup.2]] = 0.98) and after the experiment (a = 0.132 [+ or -] 0.355, b = 91.30 [+ or -] 1.817, and [R.sup.2] = 0.98). Statistical analyses showed that there was no significant difference between the calibration curves at 0.95 confidence level before and after the experiments. Therefore, it can be concluded that the calibration curves can be reproduced accurately and are stable.
[FIGURE 3 OMITTED]
Thermal conductivity probes calibration
The thermal conductivity probes were calibrated at constant temperature. Similar procedures have been reported by Konukcu (1997).
The container used for calibration was a plastic box, 70 mm deep, 200 mm wide, and 250 mm long. The bottom of the box was provided with as many small holes as possible. Six holes of 10 mm diameter were drilled along the long wall of the box, through which the probes were inserted. Air-dry clay loam and sandy loam soils, sieved through a 2-mm mesh, were packed into the box as uniformly as possible to a bulk density of 1.3 g/[cm.sup.3]. The container was then placed in water overnight to allow the soil to become saturated from the bottom. To make the measurements, the heating wires were connected to the power supplier (0.25 A, 4 V) and the thermocouple wires were connected to the digital thermometer. When the current was switched on, the temperature around the heater element increased. The first reading was taken in the saturated soil. The current was supplied for 5 min and then disconnected. When the heating current was switched off, the temperature decreased. The temperature decrease was recorded every 3 s for 5 min. Next, the container was removed from the water, the holes at the bottom were closed, and the container was weighed to obtain the saturated water content. Then the holes were re-opened to allow the water to drain. In the following days, the holes were all kept closed and after one night of evaporation, the soil surface was covered and water distribution within the soil profile was allowed to re-equilibrate. The soil was kept covered during the measurements and weighed after each measurement to obtain the water content. The calibration process was repeated for as many days as necessary to bring the soil to the air-dry state through evaporation of water from the surface. No artificial means were used to speed up the soil drying during calibration. Following this procedure, the calibration curves were produced for different soil type, temperature, and salinity.
Figure 4 shows the calibration curves of the thermal conductivity probes for sandy loam and clay loam soil. There was a linear relationship ([R.sup.2] = 0.98 and 0.96 for sandy loam and clay loam soils, respectively) between volumetric water content [theta] and thermal conductivity K obtained from Eqn 8. The general form of the thermal conductivity K v. water content [theta] curve is 'S'-shaped (Hillel 1980). However, Fig. 4 demonstrates that a linear relationship between K and [theta] can be used to calculate moisture content from temperature measurements. A linear relation between thermal conductivity and water content was also obtained for a silty loam soil by Al Nakshabandi and Kohnke (1964), Camillo and Gurney (1986), and Kasubuchi and Hasegawa (1994). Ochsner et al. (2001) showed that K of a medium-textured soil at 20[degrees]C could be accurately described as a linear function of [theta] with a correlation coefficient ([R.sup.2]) of 0.93. Although Tarnawski and Gori (2002) stated that K and [theta] showed a nonlinear behavior, they also used a linear relationship because of the difficulties to model this nonlinearity.
[FIGURE 4 OMITTED]
The temperature dependence of the probes was tested by calibrating them at 2 different temperatures, 18 and 38[degrees]C (a = 0.021 [+ or -] 0.080, b = 0.852 [+ or -] 0.269, and [R.sup.2] = 0.97) (Fig. 5). Although there was no statistically significant difference between the curves, there was a small displacement of the calibration curves towards greater K values when measured at 38[degrees]C compared with those measured at 18[degrees]C. Tarnawski et al. (2000) also reported a poorer agreement between K and [theta] at greater temperatures after calibration at 30, 50, 70, and 90[degrees]C.
[FIGURE 5 OMITTED]
To asses the variability of the calibration curves under saline conditions, the probes were also calibrated with the saline (25 dS/m) soil. Fig. 6 indicates that the measurements were unaffected by salinity (a = 0.0077 [+ or -] 0.0046, b = 0.857 [+ or -] 0.0015, and [R.sup.2] = 0.98). There was no statistically significant difference between the curves at 0.95 confidence level.
[FIGURE 6 OMITTED]
To test the reproducibility of the calibration curve, the probes were recalibrated at the same external temperature (18[degrees]C) after being used in the flow experiment for a period of 286 days. Fig. 7 shows the calibration curves of the thermal conductivity probes before and after the experiment. There was no statistically significant change from the original calibration curve (a = 0.0049 [+ or -] 0.0099, b = 0.831 [+ or -] 0.032, and [R.sup.2] = 0.97).
[FIGURE 7 OMITTED]
Eighteen probes were calibrated simultaneously. There was no significant difference between the probes. Therefore, a single calibration curve for each soil type was produced between water content and thermal conductivity. The linear relationship was K = a + b[theta]. However, an inverse relation, such that [theta] = (K- a)/b, with a = -0.0128 [+ or -] 0.0103 and b = 0.828 [+ or -] 0.034 for sandy loam soil and a = 0.0175 [+ or -] 0.0094 and b = 0.717 [+ or -] 0.043 for clay loam soil, was used to infer [theta] from measurements of K estimated from Eqn 8. According to Fig. 4, thermal conductivity ranged from 0.07 to 0.41 W/m.K for the volumetric water content from 0.09 to 0.49 [m.sup.3]/[m.sup.3] for sandy loam soil and from 0.07 to 0.29 W/m.K for the volumetric water content from 0.13 to 0.44 [m.sup.3]/[m.sup.3] for clay loam soil. Considering the water content ranges and soil types, the thermal conductivity values seem a realistic estimate (Hillel 1980; Abu-Hamdeh et al. 2000).
The probes were tested during experiments on simultaneous upward movement of salt and water in soil columns containing shallow (300 mm) saline (4 dS/m) water tables. The columns were placed in an evaporation chamber which provided a high evaporative demand (16.3 [+ or -] 0.8 mm/day) at constant temperature (32 [+ or -] 2[degrees]C). Sandy loam (porosity 49.3% and clay content 12%) and clay loam (porosity 44.6% and clay content 30.2%) soils, sieved through a 2-mm mesh, were used. The soil columns were cylindrical PVC tubes of 200 mm internal diameter. There were 6 columns, 2 instrumented with 4-electrode and thermal conductivity probes and 4 consisting of 16 rings (15 x 20 mm + 1 x 50 mm = 350 mm) for destructive sampling but containing dummy probes to ensure similar flow patterns to the instrumented soil columns. Silicon sealant and electrical tape were used to bind the rings together to build a segmented soil column. The bases of the columns were closed by a plastic sheet.
The bottom 50 mm of each soil column was filled with gravel grains after inserting the thermal conductivity probes horizontally at 16 specified depths down the soil columns (at 10 mm and every 20 mm afterwards) and placing the salinity probes vertically into the columns. Air-dry soils were then packed into the soil columns as uniformly as possible in narrow layers to a bulk density of 1.3 g/[cm.sup.3] for both soil types. To saturate the soil, distilled water was introduced into the soil columns from the bottom to avoid any entrapped air. After saturating the soil, water was allowed to drain while keeping the soil surface covered. The soil columns were then placed in the evaporation chamber and Mariotte bottles filled with saline water connected to maintain the watertable at 300 mm below the soil surface. Soil columns were left for one night to achieve equilibrium conditions in water and temperature profiles before allowing the evaporation from the soil surface. The measurement circuit was placed outside the chamber. The probes were connected to the measurement circuit by extension wires through a hole in the wall of the evaporation chamber. Mariotte bottles were also placed outside the chamber.
Since the evaporative demand was greater than the ability of the soil to transmit the water, the topsoil became dry and also salt accumulated close to the soil surface. In this way, a wide range of moisture contents and high salinities, as experienced in arid and semi-arid regions, were achieved within the soil columns to test the performance of the probes.
Readings from the thermal conductivity and salinity probes were taken 2 times, i.e. 4 and 36 days from the start of the experiment. The pre-cut soil columns were also sectioned after 4 and 36 days. Gravimetric water content was measured by drying at 105[degrees]C for 24 h. The salt content of the samples was obtained from the measurements of the electrical conductivity of 1:10 of the solution extracted from the soil samples.
To calculate the E[C.sub.w] from the resistance measurements, E[C.sub.a] was computed from Eqn 1 and then the E[C.sub.w] was determined from Eqn 2.
To obtain the corresponding water content from the temperature reading, the temperature decrease was plotted against the logarithm of cumulative time and the slope of the line, which was directly related to the water content, was calculated. Then the thermal conductivity was determined from the slope by using Eqn 8. Finally, water content was computed from the calibration curves developed between water content and thermal conductivity, K.
Result and discussion
The performance of the probes to measure unsaturated soil salinity was tested in the flow experiments. The upward movement of saline water from shallow water tables and subsequent evaporation from the soil surface caused accumulation of salt in the upper part of the soil profile (up to 5 cm). Also, this part of the soil profile became very dry due to the high evaporative demand of the air achieved in the flow experiments. Therefore, salt concentration in the solution further increased while the water content decreased. Fig. 8 shows the results from gravimetric sampling and the 4-electrode probes on days 4 and 36 for sandy and clay loam soil. There were no significant differences between the two methods for both soil types. Salt concentrations measured by the salinity probes were as accurate as the destructive method for both soil types on day 4 when the salt concentrations in solution were 4-25 dS/m and the water contents of both soil profiles were >0.20 [m.sup.3]/[m.sup.3]. The salt concentration of the soil solution in the upper part of the soil profile increased to about 60 dS/m and the water content decreased to 0.16 and 0.09 [m.sup.3]/[m.sup.3] for clay loam and sandy loam soil, respectively, on day 36. Under this extreme condition, the measured values by 4-electrode probe in this part of the soil column deviated, up to 2-fold, from the values by destructive method. The deviation in the measurement of E[C.sub.w] by the probe may be attributed to the decreasing water content of the same volume of soil. The measurement of a solution EC up to 70 dS/m as shown in Fig. 3 without any discrepancies proves the assumption that the deviation is due to the lower water content of the soil. The water contents at which the salinity measurements started deviating were around 0.10 and 0.22 [m.sup.3]/[m.sup.3] for sandy loam and clay loam soil, respectively (cf. Figs 8 and 9). The lower limit of the soils' water content at which the salinity measurements by the probe were valid was very close to the water content of the soils at wilting point. Corresponding water contents of a sandy loam and clay loam soil at the wilting point were 0.07 and 0.24 [m.sup.3]/[m.sup.3], respectively, which were quite close to the given values above.
[FIGURES 8-9 OMITTED]
The 4-electrode probes gave better results with the clay loam soil over the salinity range. Excluding the upper part of the soil column where the water content fell below the wilting point, E[C.sub.w] calculated by 4-electrode and E[C.sub.w] measured by extraction are in reasonably good agreement ([+ or -] 15% and [+ or -] 27% for clay loam and sandy loam, respectively). Nadler et al. (1984) also reported +20% error, while Fredman (1998) indicated errors up to [+ or -] 25% for electrolyte concentration for a typical of irrigated soil, increasing with increasing distribution of pore size.
Thermal conductivity probes
The salinity effect in dry soil on the measurements of water content was tested well in the flow experiments conducted in the evaporation chamber. To test the performance of the probes to measure water content under saline condition, moisture contents measured using thermal conductivity probes were compared with the gravimetric method in Fig. 9 at different times during the experiment, viz. 4 and 36 days. The measurements covered a wide range of water contents, from saturation to 0.16 [m.sup.3]/[m.sup.3] for clay loam and to 0.09 [m.sup.3]/[m.sup.3] for sandy loam soil, with great sensitivity ([R.sup.2] > 0.95) and were unaffected by salt accumulation in the soil profile.
The instruments monitor distribution of soil water and salinity, providing fine spatial resolution (1-cm intervals) and continuous, non-destructive measurements. Both methods are rapid, simple, and cheap, and probes can be manufactured and adapted easily to any experimental set-up.
Thermal-conductivity probes measured water content over a wide range and results were unaffected by salt accumulation. Soil thermal properties, which are important in many agricultural, engineering, and meteorological applications, can also be determined by thermal conductivity probe.
The 4-electrode probes provided reliable measurements ([R.sup.2] > 0.95) of the salinity of the soil solution for the range relevant to agricultural application. However, the accuracy of the probe decreased with the decreases in the water content of the same volume soil after a certain point. This is the limitation of this method. The lower limit of water content at which the salinity measurements by the probe were valid was around the water content at wilting point.
Both types of probes are stable during extended use of totally 285 days, showing statistically no drift from the original calibration curves. A single calibration curve can be used for thermal conductivity probes for a specific soil texture, but for the 4-electrode probes, a specific cell constant must be determined for each group of 4 electrodes. This is the limitation of thermal conductivity probe. Commercially available thermal conductivity probes are surrounded by a ceramic material that must be in equilibrium with the surrounding soil in water content during the calibration, and bringing them into equilibrium, especially in the dry range, takes a long time.
Thermal conductivity probes also record the soil temperature simultaneously with the water content, both of which are required to estimate E[C.sub.w] from soil E[C.sub.a]. Hence, using thermal-conductivity and 4-electrode probes simultaneously for salinity research has great advantage over other techniques.
Both probes can also be used to detect the phase transition (i.e. evaporation front) accurately from direct measurements of water content and salinity profiles because the probes are small in size and provide fine spatial resolution.
The authors acknowledge Dr A. Nadler from Institute of Soil Water and Environment, The Volcani Center, in Israel, for helping in the analysis of EC data.
Abu-Hamdeh NH, Reeder RC, Khdair AI Al-Jalil HF (2000) Thermal conductivity of disturbed soils under laboratory conditions. Transactions of the ASAE 43(4), 855-860.
Al-Nakshabandi G, Kohnke H (1964) Thermal conductivity and diffusivity of soils as related to moisture tension and other physical properties. Agricultural Meteorology 2, 271-279.
American Public Health Association (1985) 'Standard method for water and waste water examination.' 16th edn (WPCF/AWWA: Portland, OR/Denver, CO))
Bloodworth ME, Page JB (1975) Use of thermistors for the measurement of soil moisture and temperature, Soil Science Society of America Proceedings 21, 11-15.
Burger HC (1919) Das Leitvermogen verdunnter mischkristalfreien Legiefungen. (In English) Phys. Z. 20, 73-75.
Campbell GS, Mulla DJ (1990) Measurement of soil moisture content and potential. In 'Irrigation of agricultural crops--Agronomy Monograph No. 30'. (Eds BA Stewart, DR Mielsen) pp. 127-142. (ASA: Madison WI)
Camillo PJ, Gurney RJ (1986) A resistance parameter for bare-soil evaporation models. Soil Science 141, 95-105.
Fritton DG, Busscher WJ, Alpert JE (1974) An inexpensive but durable thermal conductivity probe for field use. Soil Science Society of America Proceedings 38, 854-855.
Fredlund DG (1992) Background, theory, and research related to the use of thermal conductivity sensors for matric suction measurement. In 'Advances in soil physical properties: bringing theory into practice'. (Eds GC Topp et al.) Soil Science Society of America Proceeding, Special Publication No. 30 (SSSA: Madison, WI)
Fredman SP (1998). Simulation of potential error in determining soil salinity from measured apparent electrical conductivity. Soil Science Society of America Journal 62, 593-599.
Hillel D (1980) 'Fundamentals of soil physics.' (Academic Press Inc.: London)
Jackson RD, Taylor SA (1986) Thermal conductivity and diffusivity. In 'Methods of soil analysis: Part I. Physical and mineralogical methods'. 2nd edn (Ed. A Klute) Agronomy Monograph No. 9. (SSSA: Madison, WI)
Kandiah A (Ed.) (1990) Water, soil and crop management relating to the use of saline water. FAO Expert Consultation. AGL/MISC/16/90.
Kasubuchi T, Hasegawa S (1994). Measurements of spatial average of field soil water content by the long heat probe method. Soil Science and Plant Nutrition 40, 565-571.
Konukcu F (1997) Upward transport of water and salt from shallow saline watertables. PhD thesis. Newcastle University, UK.
Nadler A (1981) Field application of the four-electrode technique for determining soil solution conductivity. Soil Science Society of America Journal 45, 30-34.
Nadler A (1982) Estimating the soil water dependence of the electrical conductivity soil solution/electrical conductivity bulk soil ratio. Soil Science Society of America Journal 46, 722-726.
Nadler A and Frenkel H (1980) Determination of soil solution electrical conductivity from bulk soil electrical conductivity measurements by the four-electrode method. Soil Science Society of America Journal 44, 1216-1221.
Nadler A, Frenkel H, Mantel A (1984) Applicability of the four-probe technique under extremely variable water contents and salinity distribution. Soil Science Society of America Journal 48(6), 1258-1261.
Ochsner TE, Horton R, Ren TH (2001). A new perspective on soil thermal properties. Soil Science Society of America Journal 65, 1641-1647.
Rhoades JD, Waggoner BL, Shouse PJ, Alves WJ (1989) Determining soil salinity from soil and soil paste electrical conductivities: Sensitivity analysis of models. Soil Science Society of America Journal 53, 1368-1374.
Shaw BT and Baver LD (1939) An electrothermal method for following moisture changes of soil in situ. Soil Science Society of America Proceedings 4, 78-83.
Tarnawski VR, Gori F (2002). Enhancement of the cubic cell soil thermal conductivity model. International Journal of Energy Research 26(2), 143-157.
Tarnawski Vr, Leong VH, Bristol KL (2000). Developing a tempertaure dependent Kersten function for soil thermal conductivity. International Journal of Energy Research 24, 1335-1350.
Wechsler A E, Glaser PE, McConnel KR (1965) Methods of laboratory and field measurements of thermal conductivity of soils. Cold Regions Research and Engineering Laboratory, Hanover, New Hampshire, Special Report No. 82.
Manuscript received 26 June 2001, accepted 23 September 2002
F. Konukcu (A,B), A. Istanbulluoglu (A), I. Kocaman (A)
(A) Trakya University, Tekirdag Agricultural Faculty, Department of Agricultural Construction and Irrigation, 590030 Tekirdag, Turkey
(B) Corresponding author: e-mail: firstname.lastname@example.org