Published online 26 May 2006
Published in Vadose Zone J 5:751-756 (2006)
DOI: 10.2136/vzj2005.0114
© 2006 Soil Science Society of America
677 S. Segoe Rd., Madison, WI 53711 USA
ORIGINAL RESEARCH
Comparing Ambient Temperature Effects on Heat Pulse and Time Domain Reflectometry Soil Water Content Measurements
Ole K. Olmansona and
Tyson E. Ochsnerb,*
a Dep. of Soil, Water, and Climate, Univ. of Minnesota, St. Paul, MN 55108
b USDA-ARS, Soil and Water Management Research Unit, St. Paul, MN 55108
* Corresponding author (ochsner{at}umn.edu)
This work supported in part by the National Science Foundation under Grant no. 0337553.
1 Mention of trade names or commercial products in this article is solely for the purpose of providing specific information and does not imply recommendation or endorsement by the USDA. 
Received 16 September 2005.
 |
ABSTRACT
|
|---|
Time domain reflectometry (TDR) and the heat pulse method are both used to measure soil water content. Changes in ambient temperature have been shown to affect TDR measurements, but less is known about the behavior of heat pulse sensors in response to changes in temperature. This study directly measured and compared the temperature sensitivity of the TDR and heat pulse methods. Both methods were used to estimate water content in silt loam and sand at two fixed water contents across a wide temperature range. An increase in temperature led to an increase in measured water content in most cases. Across the 40°C temperature range, changes in measured water content were generally 0.04 m3 m–3 or less for both methods. Weighted linear regression showed that in these soils the heat pulse method exhibited greater temperature sensitivity than the TDR method, although the differences were not statistically significant. A previously proposed correction for the temperature sensitivity of the TDR method produced mixed results. The temperature sensitivity of the heat pulse method was attributed to the changes in the density and specific heat of water and specific heat of soil with respect to temperature. When the changes in these parameters were accounted for, the temperature sensitivity was eliminated in three out of four cases.
Abbreviations: TDR, time domain reflectometry • TTDR, thermo-time domain reflectometry
|
INTRODUCTION
|
|---|
SOIL MOISTURE data are fundamentally important for a wide variety of agricultural, engineering, and hydrological applications. For example, in the agricultural community, properly gauging soil moisture facilitates efficient irrigation. In the geotechnical field, strict moisture requirements demand that an accurate measure of soil moisture is performed and reported. Measurements of soil moisture are also required in many branches of the environmental sciences. Because of these needs there has been a great deal of research on methods for measuring soil moisture.
Topp et al. (1980) showed that TDR can be used to effectively determine soil volumetric water content (
). This technique has been extensively developed and widely used since then. Others have shown that heat pulse methods may be used to arrive at an estimation for soil water content (Bristow et al., 1993; Campbell et al., 1991). A growing number of researchers have implemented this approach (Heitman et al., 2003; Song et al., 1998; Tarara and Ham, 1997).
Several researchers have investigated the influence of temperature on TDR water content measurements. It has been shown that substantial temperature influences exist for some soil types and water contents. Pepin et al. (1995) found that the apparent dielectric permittivity decreased as temperature increased in loam and saturated peat at volumetric water contents >0.29 m3 m–3 and saw no significant effect on sand under similar conditions. Wraith and Or (1999) demonstrated that the dielectric permittivity of silt loam increased with increasing temperatures at low moisture but decreased at higher moisture levels. Similarly, Logsdon (2000) found that estimated water content increased when the temperature was increased from 6.5 to 23.5°C for silt loam and sand up to a moisture content of 0.5 m3 m–3.
Little is known about the effects of ambient temperature on water content measurement using heat pulse sensors. Bristow et al. (1994a) found that over a temperature range from 19 to 88°C measured heat capacity of Clayton sand rose from 1.055 x 106 to 1.190 x 106 J m–3 °C–1. This increase in heat capacity would correspond to a 0.032 m3 m–3 increase in measured water content. Other work has shown no significant effect of change in ambient temperature (Bristow et al., 1994b). A field study by Heitman et al. (2003) showed diurnal cycling of measured soil moisture, with the lowest readings occurring when soil temperature was at the daily minimum and highest readings occurring when soil temperature was at the daily maximum. In this case, the authors attributed the cycling to measurement error.
Ideally, water content sensors would be insensitive to ambient temperature. Investigating sensor response to ambient temperature aids in the selection of appropriate sensors and helps ensure accurate water content measurements. Furthermore, once the temperature response of a sensor is quantified, it may be possible to develop a method for correcting the readings to reduce the effect of temperature. The objective of this study was to measure and to directly compare the temperature sensitivity of the TDR and heat pulse methods.
|
MATERIALS AND METHODS
|
|---|
Sensors
Time domain reflectometry waveguides and heat pulse sensors are physically similar devices, typically consisting of two or three rods or needles that can be inserted into the soil. Because of these similarities, TDR and heat pulse technologies can be combined into a single sensor (Ren et al., 2003a). These hybrid devices are called thermo-time domain reflectometry (TTDR) sensors. Since TTDR sensors can provide heat pulse and TDR water content measurements at the same location simultaneously, they are well suited for comparing temperature effects on the heat pulse and TDR methods.
We used a three-needle design for the TTDR sensors, with a needle length of 4 cm and a 0.6-cm spacing. The outer needles contained 40 gauge type E (cromel-constantan) thermocouple wires for measuring temperature rise. These needles were also used as an extension of the coaxial shield for the purposes of TDR measurement. The middle needle contained a heater constructed from two loops of 38 gauge Nichrome 80 wire. In addition, the middle needle was soldered to the center conductor of the coaxial cable. A two-part casting resin was used for the sensor head (CR-600, Micro-Mark, Berkeley Heights, NJ).1
Data Acquisition System
The hardware used for data acquisition included a data logger (CR23X), a thermocouple multiplexer (AM25T), a time domain reflectometer (TDR 100), and a TDR multiplexer (SMDX50), all from Campbell Scientific, Inc., Logan, UT. A direct current relay was used for switching power to the heater needles for a duration of 8 s. Power was provided using a 12-V marine battery.
Calibration and Validation
All sensors were calibrated and validated before the experiments. A temperature controlled water bath (Poly Science 9112, Niles, IL) was used to validate the thermocouples. All thermocouples matched the water bath temperature setting to within ±0.22°C over the range from 5 to 45°C. The standard deviation of 50 readings on each thermocouple at each temperature was always <0.085°C.
To calibrate needle spacings, the sensors were then suspended in agar-stabilized water (6 g L–1), and heat pulse readings were collected. These spacings were then validated by obtaining heat pulse readings in saturated quartz sand with a known bulk density. The volumetric heat capacity (C) was calculated from the heat pulse data by
 | [1] |
where q is the average heating power (W m–1), to is the duration of heating (s), e is the base of the natural logarithm, r is the needle spacing (m), 
Tm is the maximum temperature increase (K) and 
is to/tm where tm is the time that the maximum temperature increase occurred (Knight and Kluitenberg, 2004). The measured heat capacity from Eq. [1] was compared with theoretical predictions of heat capacity found using
 | [2] |
where 
b is the bulk density (Mg m–3), cs is the specific heat of the solids (0.73 MJ Mg–1 K–1 for quartz sand), cw is the specific heat of the water (4.1818 MJ Mg–1 K–1 at 20°C) and g is the gravimetric water content (Mg Mg–1) (Kluitenberg, 2002). The heat capacity values from the heat pulse method were accurate to within ±4.6%, and the coefficient of variation for 10 heat capacity measurements from each sensor was 2.2% or less.
The TDR mode of the sensor was calibrated using the following relationship for apparent refractive index (na)
 | [3] |
where La is the apparent length of the sensor in a medium as determined from the TDR waveform, Lo is travel distance of the signal in the sensor head, and Le is the effective length of the exposed portion of the sensor needles. Waveforms were acquired in air and water to determine Lo and Le for each sensor (Heimovaara, 1993). Validation consisted of obtaining 10 waveforms in acetone and comparing measured na with the known refractive index. The measured values were accurate to within ±3.3% and the coefficient of variation for a single sensor was not more than 1.7%.
Soil
Two soil types were used for this study, sand (Hanlon series; coarse-loamy, mixed, superactive, mesic cumulic Hapludolls) and silt loam (Ida series; fine-silty, mixed, superactive, calcareous, vesic Typic Udorthents). Before use, the sand was air dried and sieved through a 2-mm sieve. The silt loam was air dried, passed through a soil grinder, and sieved to 2 mm. Some basic physical properties for each soil are shown in Table 1.
View this table:
[in this window]
[in a new window]
|
Table 1. Soil physical properties, organic matter (OM), bulk density (b) and specific heat (cs) for Hanlon sand and Ida silt loam.
|
|
Target soil water contents of 11 and 32% by volume were achieved using a 5-mmol solution of CaCl2 applied to the loose soil with a spray bottle and mixed with a spatula. The soil was placed in 7.6-cm-diameter, 7.6-cm-tall aluminum cylinders in three lifts and compacted to a consistent bulk density of 1.28 Mg m–3 for the Ida silt loam and 1.58 Mg m–3 for the Hanlon sand. These cylinders were then covered with plastic film. Finally, the sensors were inserted into the soil and sealed in place with paraffin wax. A total of six cylinders were used for each moisture level, three cylinders containing the Hanlon sand and three containing the Ida silt loam.
Temperature Environments
We placed the cylinders, sensors, and the data acquisition system into an incubator (VWR Scientific 2020, West Chester, PA). The ambient temperature environment was raised in 10° increments from 5 to 45°C. Ten heat pulse and TDR readings were taken at each temperature. Readings were obtained while the soil temperature was within ±1°C of the target ambient temperature.
To separate temperature effects on the data acquisition system from temperature effects on the soil and sensors, an additional experiment was done with the cylinders and sensors in the incubator and the data acquisition system outside at room temperature. In this experiment the water content for both soils was 0.11 m3 m–3.
Water Content Calculations
Volumetric water content was estimated in the test samples using the TDR and heat pulse methods. For the TDR method, the following calibration equation was used (Ferre and Topp, 2002)
 | [4] |
Water content for the heat pulse method was calculated from
 | [5] |
where w is the density of water (0.99823 Mg m–3 at 20°C). The specific heat of the solids was estimated by performing heat pulse measurements on oven dried samples of each soil (Ren et al., 2003b).
|
RESULTS
|
|---|
The measured water content values for each method across the range of temperatures were examined to determine and compare temperature sensitivity. Simple plots of water content versus temperature showed that in most cases an increase in temperature resulted in an increase in the water content estimate (Fig. 1
and 2)
. This was true for both TDR and heat pulse methods and for both soil types.
View larger version (12K):
[in this window]
[in a new window]
|
Fig. 1. Heat pulse and TDR soil moisture estimation in Ida silt loam and Hanlon sand with an actual water content of 0.11 m3 m–3. Lines represent weighted linear regression. Symbols represent the mean of three replicates, and error bars represent one standard deviation.
|
|
View larger version (12K):
[in this window]
[in a new window]
|
Fig. 2. Heat pulse and TDR soil moisture estimation in Ida silt loam with an actual water content of 0.36 m3 m–3 and Hanlon sand with an actual water content of 0.32 m3 m–3. Lines represent weighted linear regression. Symbols represent the mean of three replicates, and error bars represent one standard deviation.
|
|
To quantify the temperature sensitivity of each method, weighted linear regression was performed. The inverse of the data variance at each temperature was used as the weighting factor. At a water content of 11% the heat pulse method in silt loam showed the steepest slope (Table 2). The TDR method in silt loam, heat pulse method in sand, and TDR method in sand exhibited sequentially decreasing slopes with the slope of TDR in sand not significantly different from zero. The slope for the TDR method in sand was significantly lower than the slopes of both the TDR and heat pulse methods in the silt loam on the basis of 95% confidence intervals.
View this table:
[in this window]
[in a new window]
|
Table 2. Weighted linear regression analysis of estimated water content versus ambient temperature for heat pulse and TDR methods.
|
|
At a water content of approximately 34% the heat pulse method in sand showed the steepest slope followed in decreasing order by the slopes of the heat pulse method in loam, the TDR method in sand, and the TDR method in loam. For this level of moisture the slopes for the TDR method in loam and in sand were not significantly different from zero. The slopes for the heat pulse method in both soils were significant.
For both moisture levels the heat pulse method showed the steepest positive slope with respect to increasing ambient temperature. The TDR method, especially in sand, was less sensitive to temperature increases.
Similar results were obtained when the data acquisition system was kept at a constant temperature. Slopes increased slightly for both soil types and measurement methods, but these increases were not statistically significant.
|
DISCUSSION
|
|---|
The influence of ambient temperature on water content estimation using the TDR method is to be expected. Our results conform well to the results of Or and Wraith (1999) who, for a gravimetric water content of 0.14 kg kg–1, showed an increase in measured soil moisture with increasing temperature for Brocko silt loam of about 0.001 m3 m–3 °C–1 (data approximated from Or and Wraith [1999], their Fig. 8). A primary determinant of the temperature influence on TDR is soil surface area. Water at the surface of soil particles is bound more strongly at lower temperatures and is released with increasing temperatures. Because the surface area of the silt loam (129.8 m2 g–1) is much higher than that of the sand (18.5 m2g–1), the effects of increased temperature are more pronounced in the silt loam. The lower surface area of the sand results in less bound water, and therefore TDR readings are less affected. Wraith and Or (1999) showed that sandy soils exhibit decreasing dielectric permittivity with increasing temperature regardless of the water content. This behavior arises from the temperature dependence of the dielectric permittivity of free water. In our case this trend was apparent at a water content of 0.11 m3 m–3. In contrast, at 0.32 m3 m–3 an increase in estimated water content with temperature was observed, but at both moisture levels the slopes were not significantly different from zero.
Or and Wraith (1999) provided a correction to account for the temperature influence on TDR sensors. We used their approach to adjust the original TDR data (Fig. 3
). Adjusted values for silt loam at both moisture levels resulted in overcorrections, while corrections for sand showed a slight improvement at low water content (Table 2). When the correction was applied to sand at high water content the effect from temperature became more pronounced. These mixed results suggest that more research is needed to develop reliable methods for correcting the influence of temperature on TDR sensors.
View larger version (11K):
[in this window]
[in a new window]
|
Fig. 3. Corrected and uncorrected water content estimates using the TDR method in Ida silt loam with an actual water content of 0.36 m3 m–3 and Hanlon sand with and actual water content of 0.32 m3 m–3. Lines represent weighted linear regression.
|
|
The heat pulse method was also clearly influenced by ambient temperature, and no correction for this influence is available in the literature. We note that the heat capacity and density of water have a slight, but well-documented temperature dependence (Weast, 1972), and the soil solid phase has a temperature dependent specific heat as well (Kay and Goit, 1975). We reanalyzed the data to see if performance of the heat pulse method could be improved by accounting for these influences.
The temperature dependence of the specific heat of a soil can be estimated by applying the third law of thermodynamics. This law implies that the specific heat of a crystalline solid should approach zero as the temperature approaches absolute zero (0 K) (Kay and Goit, 1975). Furthermore, the relationship between cs and temperature is known to be linear (Kay and Goit, 1975; Kluitenberg, 2002). The temperature dependencies of cs can therefore be approximated by
 | [6] |
where m is the slope and T is the absolute temperature. We divided the specific heat values in Table 1 by the temperature at which they were measured to obtain slope estimates of 0.00309 and 0.00327 MJ Mg–1 K–2 for the sand and silt loam, respectively.
Specific heat values from Eq. [6] were then used, along with the known temperature-dependent density and specific heat values of water, to recalculate the water content from the original heat pulse data. This greatly reduced the influence of ambient temperature (Fig. 4
). The correction technique led to weighted regression slopes that were not significantly different from zero in three of four cases (Table 2). The exception was at a water content of 0.11 m3 m–3 in the sand where a statistically significant overcorrection occurred. Still, the temperature dependence of the coefficients in Eq. [5] appears to adequately explain the temperature sensitivity of heat pulse water content estimates.
View larger version (11K):
[in this window]
[in a new window]
|
Fig. 4. Corrected and uncorrected water content estimates using the heat pulse method in Ida silt loam with an actual water content of 0.36 m3 m–3 and Hanlon sand with and actual water content of 0.32 m3 m–3. Lines represent weighted linear regression.
|
|
There are other possible explanations. Perhaps bound water, either on the soil surface or trapped between clay layers, is released at higher ambient temperatures, increasing the soil heat capacity and leading to a higher water content estimate. Additionally, it is possible that latent heat transfer near the sensor may be a function of initial temperature. At higher temperatures, water is more easily vaporized; if this water vapor migrates beyond the thermocouple needles before condensing, heat would be lost from the system. This would lead to a lower temperature rise resulting in a higher apparent heat capacity and, therefore, a higher soil moisture estimation. However, we are led to conclude that the most straightforward explanation, the temperature dependence of cs, cw, and w, is probably the correct one.
The lack of any significant changes in temperature sensitivity when the data acquisition system was kept at a constant temperature indicates that neither the heat pulse nor the TDR method was much influenced by the temperature of the lead wires or the data acquisition system. It suggests that changes in ambient temperature either directly affect the sensor or the physical environment surrounding the sensor.
|
CONCLUSIONS
|
|---|
We found that an increase in ambient temperature usually caused increased soil moisture readings with both TDR and heat pulse methods regardless of moisture levels or soil types. The exception was the TDR measurements in sand at low water content. Similar trends were observed when the data acquisition system was isolated from the change in temperature. While it is important to recognize the effect of temperature, it should be noted that these measurements were performed across a temperature range of 40°C, a larger range than would be encountered in many field applications. Therefore, users of these sensors in the field would likely often experience small temperature effects, on the order of 0.01 to 0.02 m3 m–3.
Weighted linear regressions showed that the TDR method tended to be less sensitive to changes in ambient temperature than the heat pulse method, although the differences were not statistically significant. This implies that for these soils the TDR method may be a slightly more stable method of soil moisture measurement when temperature sensitivity is a concern. However, others have shown that the temperature effect on TDR is strongly soil dependent. Attempts at correcting the TDR data to remove the temperature sensitivity met with little success.
In contrast, the temperature effects on heat pulse measurements of soil moisture were eliminated in three of four cases by accounting for the temperature dependencies of the specific heat of the soil and the specific heat and density of water. A single heat pulse measurement in oven-dry soil at a known temperature is sufficient to determine the temperature dependence of specific heat for that soil. This makes it relatively simple to correct the ambient temperature effects on heat pulse water content estimates.
|
ACKNOWLEDGMENTS
|
|---|
We thank Debra Palmquist, USDA-ARS, Peoria, IL for her assistance with statistical analysis and Todd Schumacher, USDA-ARS, St. Paul, MN for his advice during the data collection process. We also express gratitude toward the anonymous reviewers for their valuable questions and comments.
|
REFERENCES
|
|---|
- Bristow, K.L., G.S. Campbell, and K. Calissendorff. 1993. Test of a heat-pulse probe for measuring changes in soil water content. Soil Sci. Soc. Am. J. 57:930–934.[Abstract/Free Full Text]
- Bristow, K.L., G.J. Kluitenberg, and R. Horton. 1994a. Measurement of soil thermal properties with a dual-probe heat-pulse technique. Soil Sci. Soc. Am. J. 58:1288–1294.[Abstract/Free Full Text]
- Bristow, K.L., R.D. White, and G.J. Kluitenberg. 1994b. Comparison of single and dual probes for measuring soil thermal properties with transient heating. Aust. J. Soil Res. 32:447–464.[CrossRef]
- Campbell, G.S., C. Calissendorff, and J.H. Williams. 1991. Probe for measuring soil specific heat using a heat-pulse method. Soil Sci. Soc. Am. J. 55:291–293.[Abstract/Free Full Text]
- Ferre, T.P.A., and G.C. Topp. 2002. Time domain reflectometry. p. 434–446. In J.H. Dane and G.C. Topp (ed.) Methods of soil analysis. Part 4. Physical methods. SSSA Book Ser. 5. SSSA, Madison, WI.
- Heimovaara, T.J. 1993. Design of triple-wire time domain reflectometry probes in practice and theory. Soil Sci. Soc. Am. J. 57:1410–1417.[Abstract/Free Full Text]
- Heitman, J.L., J.M. Basinger, G.J. Kluitenberg, J.M. Ham, J.M. Frank, and P.L. Barnes. 2003. Field evaluation of the dual-probe heat-pulse method for measuring soil water content. Available at www.vadosezonejournal.org. Vadose Zone J. 2:552–560.[Abstract/Free Full Text]
- Kay, B.D., and J.B. Goit. 1975. Temperature-dependent specific heats of dry soil materials. Can. Geotech. J. 12:209–212.
- Kluitenberg, G.J. 2002. Heat capacity and specific heat. p. 1201–1208. In J.H. Dane and G.C. Topp (ed.) Methods of soil analysis. Part 4. Physical methods. SSSA Book Ser. 5. SSSA, Madison, Wisconsin.
- Knight, J.H., and G.J. Kluitenberg. 2004. Simplified computational approach for dual-probe heat-pulse method. Soil Sci. Soc. Am. J. 68:447–449.[Abstract/Free Full Text]
- Logsdon, S.D. 2000. Effect of cable length on time domain reflectometry calibration for high surface area soils. Soil Sci. Soc. Am. J. 64:54–61.[Abstract/Free Full Text]
- Or, D., and J.M. Wraith. 1999. Temperature effects on soil bulk dielectric permittivity measured by time domain reflectometry; a physical model. Water Resour. Res. 35:371–383.[CrossRef]
- Pennell, K.D. 2002. Specific surface area. p. 308–312. In J.H. Dane and G.C. Topp (ed.) Methods of soil analysis. Part 4. Physical methods. SSSA Book Ser. 5. SSSA, Madison, WI.
- Pepin, S., N.J. Livingston, and W.R. Hook. 1995. Temperature-dependent measurement errors in time domain reflectometry determinations of soil water. Soil Sci. Soc. Am. J. 59:38–43.[Abstract/Free Full Text]
- Ren, T., T.E. Ochsner, and R. Horton. 2003a. Development of thermo-time domain reflectometry for vadose zone measurements. Available at www.vadosezonejournal.org. Vadose Zone J. 2:544–551.[Abstract/Free Full Text]
- Ren, T., T.E. Ochsner, R. Horton, and Z. Ju. 2003b. Heat-pulse method for soil water content measurement: Influence of the specific heat of the soil solids. Soil Sci. Soc. Am. J. 67:1631–1634.[Abstract/Free Full Text]
- Song, Y., J.M. Ham, M.B. Kirkham, and G.J. Kluitenberg. 1998. Measuring soil water content under turfgrass using the dual-probe heat-pulse technique. J. Am. Soc. Hortic. Sci. 123:937–941.[Abstract/Free Full Text]
- Tarara, J.M., and J.M. Ham. 1997. Measuring soil water content in the laboratory and field with dual-probe heat-capacity sensors. Agron. J. 89:535–542.[Abstract/Free Full Text]
- Topp, G.C., J.L. Davis, and A.P. Annan. 1980. Electomagnetic determination of soil water content: Measurements in coaxial transmission lines. Water Resour. Res. 16:574–582.[CrossRef]
- Weast, R.C. 1972. CRC handbook of chemistry and physics. 53 ed. The Chemical Rubber Co., Cleveland, OH.
- Wraith, J.M., and D. Or. 1999. Temperature effects on soil bulk dielectric permittivity measured by time domain reflectometry: Experimental evidence and hypothesis development. Water Resour. Res. 35:361–369.[CrossRef]
This article has been cited by other articles:
|
|
|
|
 
M. H. Young, G. S. Campbell, and J. Yin
Correcting Dual-Probe Heat-Pulse Readings for Changes in Ambient Temperature
Vadose Zone J.,
January 23, 2008;
7(1):
22 - 30.
[Abstract]
[Full Text]
[PDF]
|
|
|
TOP
ABSTRACT
INTRODUCTION
