Published in Soil Sci. Soc. Am. J. 69:343-346 (2005).
© 2005 Soil Science Society of America
677 S. Segoe Rd., Madison, WI 53711 USA
Division S-1—Notes
HIGH-RESOLUTION MONITORING OF SURFACE-FLOW DEPTH WITH FREQUENCY-DOMAIN PROBES
Helmer Schack-Kirchner*,
Thomas Schmid and
Ernst Hildebrand
Albert-Ludwigs Univ., Freiburg Institut für Bodenkunde und Waldernährungslehre (Institute of Soil Science and Forest Nutrition), 79085 Freiburg im Breisgau, Germany
* Corresponding author (helmer.schack-kirchner{at}bodenkunde.uni-freiburg.de)
 |
ABSTRACT
|
|---|
The need exists for a measurement technique to accurately determine the depth of surface runoff on natural soil surfaces. The objective of this note was to determine the sensitivity to the immersion depth of commercially available frequency-domain probes. Within the range of 0 to 25 mm, measurement accuracies better than 0.5 mm were reached. Interestingly, soiling of the rods or droplets did not influence the results significantly. The influence of the electrical conductivity of the water or internal characteristics of different probes did not lead to a bias of more than 1 mm. A field test with grid-like mounted frequency-domain probes showed a stable output signal and a high sensitivity to hydrograph changes. We concluded that frequency-domain probes were well suited for tracking shallow depths of surface runoff in field studies and allow high spatial and time resolution. They provide higher accuracy than alternative systems in the water-level range <5 mm. However, when the measurement precision must be higher than 0.5 mm, an individual calibration of the probes is necessary as well as a correction for salinity.
|
INTRODUCTION
|
|---|
FLOW DEPTH OF RUNOFF is a key parameter in overland-flow hydraulics because it is closely related to discharge rate, flow velocity, surface roughness, and by this way to soil erosion (Hillel, 1998). In situ measurement of shallow water levels on natural soil surfaces is challenging. In most cases, when level sensing is used in runoff and erosion research, it is performed in collection tanks or in flumes using floats, pressure transducers, or bubblers. However, this approach requires berming the area to converge runoff to the flume. When measuring water levels on natural surfaces one must deal with different phenomena. Typical flow depth of interrill runoff ranges between 0 and 20 mm (Nearing et al., 1991; Parsons et al., 1994). Considering a roughness length in the same range it becomes evident, that the surface will often only be partially inundated and roughness elements protrude through the flow. In this case the runoff does not form a flat water level but follows depressions between the roughness elements (Lawrence, 1997). A sensor for the monitoring of flow depths on soil surfaces under natural conditions should therefore meet the following specifications:- small diameter to fit to the depressions of the soil microrelief and to minimize disturbance of the flow hydraulics
- a vertical resolution between 0 and 20 mm with high accuracy in the low range for detecting water films
- high time resolution to allow short storm tracking
- durability and low susceptibility to litter and soil splash
- automatic operation for unattended field use
To which extent do the common level-sensing techniques1 fulfill these specifications?
- Bubble pressure sensors (Dedrick and Clements, 1984) reach maximum accuracies between 1.7 and 3 mm (measuring range <100 cm) but it is necessary to install the bubbling tube outlet a few centimeters (e.g., the Campbell Inc. double bubbler: >15 cm) below the lowest water level of interest. Water levels in the millimeter range can therefore not be assessed.
- Some float sensors have high accuracy (e.g., the Celesco Inc., Chatsworth, CA, cable extension position transducer PT 420 with 0- to 50-mm measuring range: 0.15 mm). However, the mentioned float sensor requires a mass of >300 g to balance the spring-loaded cable tension. Float sensors need a minimum flat water level to work correctly, depending on the water displacement from the float. For their use in flumes, a stilling well is recommended. On rough and inclined soil surfaces it is unlikely that floats, even when miniaturized, yield accurate results when the water level does not reach at least a few millimeters of flow depth. Additionally any miniaturization of floats increases their susceptibility to sedimentation or other mechanical impact.
- Submersible pressure sensors are sensitive to the stress exerted by the water column above the sensor. They must then be installed below the lowest water level. This requires either the digging of a stilling well in the soil or to accept a measuring offset of the height of the sensor. The accuracy of commercially available pressure sensors for level sensing reach 0.6 mm (e.g., the Druck Ltd. Leicester, UK, GB PTX 1830 with a measuring range of 0–0.7 m).
- Ultrasonic level sensors may be advantageous due to the avoidance of contact with the soil. Depending on the beam angle (3–8°) and the minimum measuring distance (>30 cm) a flat water surface with a minimum 3-cm diam. is necessary to produce accurate signals. Accuracy of commercially available ultrasonic level sensors is between 2 mm (e.g., BadgerMeter Inc., Milwaukee, WI, Ultrasonic 2500) and 10 mm (Campbell Scientific, Logan, UT, Ultrasonic).
- A modified TDR-System for level sensing in tanks was tested by Thomsen et al. (2000). The accuracy of this system was about 2 mm.
We were unable to find any publications where one of these common methods of water level sensing had been used to measure shallow water levels on natural soil surfaces. Direct measurements of the depth of runoff on soil surfaces are scarcely reported. Parsons et al. (1994) measured flow depth in field trials with a "thin millimeter scale", which was inserted into the flow and Dunkerley (2001) used a stepping motor controlled needle gauge. In most studies, the flow depth was estimated indirectly from the discharge rate and the flow velocity (Katz et al., 1995). However, these measuring techniques do not fulfill the above mentioned specifications, especially the possibility of unattended field use. Hoover (1990) used a water activated printed-circuit switch to automatically detect surface runoff, but this device did not provide any information on the depth of the water level.
On testing frequency-domain probes (ThetaProbe, Delta-T Devices Ltd. Cambridge, UK) to measure soil moisture, sensitivity of the output signal to the depth of immersion was observed, that lead to testing these probes for high-resolution measurement of flow depth.
In this paper theoretical considerations were made about the ability of frequency-domain probes to measure surface-water depth and report laboratory and preliminary field experiments with their results.
|
Measurement Theory
|
|---|
The measurement theory of the ThetaProbe was extensively documented by Gaskin and Miller (1996). A schematic cross cut through a ThetaProbe is shown in Fig. 1. An oscillator at the top of the transmission line produces a 100-MHz sinoidal electromagnetic wave propagating along the waveguide. When the impedance along the waveguide changes, a portion of the signal is reflected. The squared impedance is proportional to the dielectric constant of the surrounding media and depends mainly on the water content. For soil moisture measurement the ThetaProbe is completely inserted into the soil so that the main impedance change and signal reflection occur at the junction between the internal and external transmission line (P1 in Fig. 1). Within the internal transmission line, the reflected signal is added to the outgoing wave. The travel time of the outgoing and incoming signal at Point P0 differs by half a wavelength (two times the distance between P1 and P0) and therefore the reflected wave interferes destructively with the emitted signal. At P1, located at the reflection point, the peak value is the amplitude added to the reflected signal. The difference in tension peak values measured at P1 and P0 is therefore proportional to the reflection coefficient and generates the output signal of the ThetaProbe. However, when the ThetaProbe is used to measure water level height the reflection coefficient remains constant, but the position of the main reflection point moves from P1 to the water level. If the air section increases, the distance between P0 and the reflection point is slightly greater than one quarter the wavelength. As a consequence, the effect of destructive interference decreases and the tension peak value at P0 increases. The additive effect of interference hereby decreases the peak tension at P1. Any shift of the main reflection point causes a change in the output signal of the Theta-probe.
|
Materials and Methods
|
|---|
In our laboratory tests we used a slide equipped with a digital position control by which the ThetaProbes could be lowered into a water-filled jar. The first test was performed with tap water (electrical conductivity of 24.8 mS m–1). Immersion depths between complete extraction of the rods out of the water and 30 mm were selected in advance by a random generator. In this way, hysteretic effects due to remaining droplets were included in the data. The effect of splashed soil materials on the measuring rods was tested by wetting and powdering the rods with silt dominated soil material. To test the influence of the fluid properties, five different solutions with increasing electrical conductivity were tested at immersion depths of 5 and 15 mm. Electrical conductivity was adjusted with NaCl between 0.13 (distilled water) and 45.6 mS m–1. We tested with soil suspension and soiled rods as well. The variability between different ThetaProbes was tested by comparing the output signal of 35 probes at tap-water depths of 5 and 15 mm.
For the field tests, we used a frame (length 2 m, width 1 m) with up to five crossbars each with six ThetaProbes mounted on pipe clamps (Fig. 2). The probes were adjusted vertically allowing the distance of the middle rod to the soil surface to be as small as possible without contacting the soil surface. Fresh water was supplied upslope by trickling it uniformly on the soil over the width of the frame. The supply rate could be regulated to a maximum of 0.2 m3 h–1. On the downhill side of the frame, the surface runoff was captured and quantified by a combination of a tipping-bucket sensor and turbine flowmeter. All sensors were connected to a multi-channel data logger.
View larger version (37K):
[in this window]
[in a new window]
|
Fig. 2. Measuring frame used for field tests. Lateral water loss was prevented by aluminum sheets mounted to the sides of the frame and sealed to the soil with polyurethane foam.
|
|
We tested this equipment on a landfill surface prepared with steel production residues. The slope was 25% and the surface of the landfill was relatively smooth with a maximum difference between depressions and elevations of about 3 cm that could mainly be attributed to surface gravel. The only vegetation was shallow mosses, covering one third of the area.
|
Results and Discussion
|
|---|
There was a clear relationship between the immersion depth and the output signal of the probe (Fig. 3). When the rod tips touched the water surface, the signal increased abruptly from 0 to above 50 mV. At an immersion depth of 25 mm, the maximum signal of 1100 mV was reached. Because of the undulated form of the curve, the best fit was obtained with a sixth-order polynomial. Ninety percent of the residuals were between –0.37 and 0.39 mm, and only eight values exceeded ±0.5 mm with a maximum of 1 mm.
View larger version (33K):
[in this window]
[in a new window]
|
Fig. 3. Relationship between immersion depth and output signal of 255 points. Line of best fit was calculated using 6th order polynomial. The dashed line symbolizes the 95% confidence limit of individual points.
|
|
The influence of the electrical conductivity (Fig. 4) of the liquid depended on the depth of immersion. The deviations of the adjusted immersion depths from the calculated results of the model in Fig. 3 were used as a test index. At 5-mm immersion depth, no significant effect was visible, but at the 15-mm a systematic influence of salinity was observed. However, when conductivity was between 20 (e.g., tap water) and 50 mS m–1 (soil suspension of a non-saline soil) the model bias did not exceed 0.5 mm.
When the reference model was applied to 35 individual ThetaProbes at immersion depths of 15 and 5 mm, the maximum bias was <1 mm and the 25th and the 75th percentiles were <0.5 mm. The output signal was relatively insensitive to water droplets on the rods with and without additional soil material. The signal did not change more than 10 mV as a consequence of these contaminations. Therefore water splashes carrying soil particles did not affect the measurements.
The synchronous progression of the runoff-rate and water-level measuring curves from the field tests (Fig. 5) revealed a close relationship between runoff rate and flow depth. For reasons of clarity, the intermediate results at a distance of 94 and 175 cm from the supply are not displayed. The delay in response of the water level sensors to the increasing upslope supply increased with their distance from the supply. The level sensor signals were stable with fluctuations <0.5 mm when the upslope supply remained constant. Mean water level decreased with increasing distance from the upslope supply. However, due to the very high lateral variability of the water level on the slope, these differences could not be found significant (Non-parametric analysis of variance, Kruskal Wallis Test performed at four points in time with different mean water levels). The maximum recorded flow depth for a single probe was 4.5 mm while one probe remained dry throughout the experiment.
View larger version (31K):
[in this window]
[in a new window]
|
Fig. 5. Results of a field test. Top: Time series of the upslope water supply and surface discharge rate 2-m downhill. Bottom: Time series of water level with time resolution of 30 s. Each line represents the mean value of six probes mounted on one crossbar with a defined distance to the upslope supply (Fig. 2). The vertical bars represent the standard deviation of the water level at the respective points. The arrow points to an abrupt increase in water supply to reveal the lag in water level changes.
|
|
|
Conclusions
|
|---|
Results show that the ThetaProbe is well suited for unattended monitoring of surface runoff in the field with high accuracy and high resolution in space and time. The system showed a high sensitivity to the occurrence of shallow water levels even in the submillimeter range but was insensitive to droplets and splashes. The accuracy of water level measurement, even without individual calibration of the probes, is higher or at least equal to the usual level sensing techniques in flumes and to the methods where measurement of shallow water films on soil surfaces have been reported (e.g., Parsons et al., 1994; Dunkerley, 2001). One main advantage in using ThetaProbes for level sensing is the particularly high sensitivity to very low water levels in the range <5 mm. However, to obtain an accuracy better than 0.5 mm, an individual calibration of the probes, as well as a correction term for salinity, are required. Because the ThetaProbes can be directly connected to standard data loggers, their use does not require special technical skills. The disturbance to flow hydraulics may be neglected due to the small diameter of the measuring rods. If many probes are mounted gridlike, like in our frame construction, they may alter the kinetic energy of the rain reaching the soil. For monitoring of surface runoff in sprinkler experiments or with natural rain we recommend increasing the distance between the probes.
The ThetaProbes allow unattended measurements of the surface-runoff depths as long as the soil level does not change (e.g., due to erosion or bioturbation). When erosion at the measurement points is expected, the ThetaProbes can be mounted vertically staggered allowing even soil loss or sedimentation to be monitored.
According to Govers et al. (2000) and Parsons et al. (1994), direct measurements of flow depth and discharge appear to give more reliable results when used to parametrize models of surface-runoff hydraulics than the use of dye-tracing (e.g., described at Katz et al., 1995) or falling limb regression modeling on the hydrograph (e.g., described at Weltz et al., 1992). By using ThetaProbes flow depth measurements are little laborious and can be performed with high precision in the laboratory as well as in field experiments. The technique possibly allows an easier parametrization of runoff equations but also the monitoring of emergence and extent of shallow surface runoff under field conditions.
|
ACKNOWLEDGMENTS
|
|---|
The work has been funded by the Bundesministerium für Bildung und Forschung (Project No. 0339767) and by the AG der Dillinger Hüttenwerke (Dillingen, Saar).
|
NOTES
|
|---|
1 A comprehensive overview about level sensors is given in the ITRC Report No. R99-002 available online at the Irrigation Training and Research Center, Cal Poly State Univ., San Luis Obispo, CA 93407 (http://www.itrc.org/reports/WaterLevelSensor/WaterLevelSensor.html, verified 19 Nov. 2004) 
Received for publication March 29, 2004.
|
REFERENCES
|
|---|
- Delta-T Devices Ltd. 1999. ThetaProbe soil moisture sensor. Type ML2 user manual ML2x-UM-1.21. Delta-T Devices Ltd. Burwell Cambridge U.K.
- Dedrick, A.R., and A.J. Clemmens. 1984. Double-Bubblers coupled with pressure transducers for water level sensing. Trans. ASAE 27:797–783.
- Dunkerley, D. 2001. Estimating the mean spead of laminar overland flow using dye injection– uncertainty on rough surfaces. Earth Surf. Processes Landforms 26:363–374.
- Gaskin, G.J., and J.D. Miller. 1996. Measurement of soil water content using a simplified impedance measuring technique. J. Agric. Eng. Res. 63:153–160.
- Govers, G., I. Takken, and K. Helming. 2000. Soil roughness and overland flow. Agronomie 20:131–146.[ISI]
- Hillel, D. 1998. Environmental soil physics. Academic Press San Diego, CA
- Hoover, J.R. 1990. Seep and runoff detector design and performance to determine extent and duration of seep/runoff zone from precipitation on hillsides. Trans. ASAE 33:1843–1850.
- Katz, D.M., F.J. Watts, and E.R. Burroughs. 1995. Effects of surface roughness and rainfall impact on overland flow. J. Hydraul. Eng. 121:546–553.
- Lawrence, D.S. 1997. Macroscale surface roughness and frictional resistance in overland flow. Earth Surf. Processes Landforms 22:365–382.
- Nearing, M.A., J.M. Bradford, and S.C. Parker. 1991. Soil detachment by shallow flow at low slopes. Soil Sci. Soc. Am. J. 55:339–344.[Abstract/Free Full Text]
- Parsons, A.J., A.D. Abrahams, and J. Wainwright. 1994. On determining resistance to interrill overland flow. Water Resour. Res. 30:3515–3521.
- Thomsen, A., B. Hansen, and K. Schelde. 2000. Application of TDR to water level measurement. J. Hydrol 236:252–258.
- Weltz, A.M., A.B. Arslan, and L.J. Lane. 2000. Hydraulic roughness coefficients for native rangelands. J. Irrig. Drain. Eng. 118:776–790.
TOP
ABSTRACT
INTRODUCTION
