Frequency Domain Analysis for Extending Time Domain Reflectometry Water Content Measurement in Highly Saline Soils

TTA especially with the filtering of higher frequencies in cables longer than 3.2 m resulting in increased signal Water content and electrical conductivity of soils are routinely rise time (Heimovaara, 1993; Noborio, 2001). Timing or determined using time-domain reflectometry (TDR) based on analysis length measurement errors associated with finite pixel of signal travel time and attenuation along embedded probes. In soils with appreciable salinity, time domain analysis becomes progressively resolution can also be problematic. Shorter probes are inaccurate due to signal attenuation to the point of failure. Our objecappealing because losses due to electrical conductivity tives were to test whether dielectric permittivity, which is inextricable decrease proportionally with probe length reduction. using TDR travel time analysis (TTA) in saline soils could be extracted Using TTA, Mojid (2002) compared water content deusing frequency domain techniques applied to waveforms from shorter terminations using three-rod TDR probes varying in TDR probes that reduce signal attenuation. The methodology was length from 2 to 10 cm. Although waveforms deteriotested using coaxial cells and three-wire TDR probes in sand and silt rated and became more trough-shaped and rounded loam soil under a wide range of saturated solution soil electrical with reduced probe length and water content, reasonconductivities. Three different interpretation techniques were comable correlation to gravimetrically determined water pared; conventional TTA, scatter function fitting (SFF), and resonant content was found for all probes from 10 cm down to frequency analysis (RFA). Using a range of probe lengths (2, 3, 6, 10, and 15 cm) soil bulk dielectric permittivity estimates were obtained 2.5 cm in length. In contrast, Lin (2003b) showed a using all three techniques in solution electrical conductivities up to significant enhancement in the modeled bulk dielectric 48 dS m 1. Both SFF and RFA produced similar permittivity estimates, for probe lengths shorter than 10 cm in length while which generally increased with increasing solution electrical conduclengths between 10 and 50 cm showed little influence tivity. Network analyzer permittivity measurements (0.5–1.5 GHz) on permittivity. Persson and Haridy (2003) compared were greater than estimates using TTA, which were both greater than two-rod 2-cm long TDR probe measurements of b and values from SFF and RFA in a saturated silt loam soil. Although εb for estimating volumetric water content under condependent upon dielectric permittivity and electrical conductivity, stant soil solution electrical conductivity. Their thirdfrequency domain analysis results indicate a 3-cm probe is optimal order polynomial fit suggested improved accuracy using for maintaining a interpretable scatter function in the TDR frequency b rather than dielectric measurements for water content band while providing maximum extension of dielectric determination under lossy conditions in saturated soils. determination with these short probes. Time-domain methods employ a step voltage that propagates down a low-loss coaxial line, transitioning through fittings, multiplexers, and other lossy connecT ime domain reflectometry is gradually becoming tions resulting in cumulative system losses acting as a one of the most accurate and reliable measurement low-pass filter that tends to spread the frequency conmethods for concurrent determination of bulk pertent of the return signal seen by the TDR. The signal mittivity (εb) for water content and bulk electrical contransition from the cable to the soil in the probe head ductivity ( b) in soils and other porous media (Jones is where the major mismatch in impedance occurs. The et al., 2002; Robinson et al., 2003b). However, even at reflection from the end of the probe forms multiple moderate soil b levels (typically at b 2 dS m 1) reflections that result in a build-up of the waveform attenuation of the TDR signal degrades the ability to amplitude illustrated in Fig. 1. These undesirable losses determine water content from TTA of the waveform in the system combine with electrical and dielectric (Fig. 1). The attenuation in soil is complex and proporlosses we are interested in measuring in the soil. The tional to b, probe length, clay content and mineralogy, total attenuation measured in the waveform at low freand water content. An alternative to travel time deterquencies (long distances) yields information on the soil’s mination of the bulk dielectric permittivity (εb) is offered b, but separation of the undesirable losses is difficult. by transformation of the TDR waveform from the time Castiglione and Shouse (2003) recently demonstrated a to frequency domain, which is not based on waveform difference reflection calibration method scaling b meaTTA. The key for implementation of such an approach surements between an open and short-circuit of the is the potential use of very short TDR probes (e.g., TDR probe that removes unwanted system losses from 10 cm, Jones and Or, 2001). Reduced probe length the measurement of b. leads to increased measurement error using standard The extent to which electrical and even dielectric losses influence the measured bulk permittivity is not S.B. Jones, Dep. of Plants, Soils, and Biometeorology, Utah State well understood and there seem to be three different Univ., Logan, UT 84322-4820; D. Or, Dep. of Civil and Environmental views as to the effect of b on ε. Engineering, Univ. of Connecticut, Storrs, CT 06269-2037. Received 29 Jan. 2004. *Corresponding author ([email protected]). Abbreviations: DFFT, discrete fast Fourier transform; EM, electromagnetic; NAM, network analyzer measurement; RFA, resonant frePublished in Soil Sci. Soc. Am. J. 68:1568–1577 (2004).  Soil Science Society of America quency analysis; S11, scatter function; SFF, scatter function fitting; TDR, time domain reflectometry; TTA, travel time analysis. 677 S. Segoe Rd., Madison, WI 53711 USA