A low-cost microprocessor and infrared sensor system ...

Computers and Electronics in Agriculture 53 (2006) 122–129

A low-cost microprocessor and infrared sensor system for automating water infiltration measurements Katherine Milla a,∗ , Stephen Kish b a

College of Engineering Sciences, Technology and Agriculture, Florida A&M University, Tallahassee, FL 32307, USA b Department of Geological Sciences, Florida State University, Tallahassee, FL 32306, USA

Received 1 August 2005; received in revised form 1 February 2006; accepted 11 May 2006

Abstract Understanding the nature of soil infiltration properties is important for a number of applications, including planning efficient irrigation and erosion control practices, evaluating effectiveness of linings for sanitary landfills and waste lagoons, and landscape design and management. A common method for measuring infiltration is with the use of cylinder infiltrometers. This method can be very labor intensive and time-consuming, thus limiting its practicality. We describe the construction of an inexpensive system that can be readily adapted to single- and dual-ring infiltrometers to provide automated measurements on the scale of seconds to days. The system is comprised of an infrared distance-measuring sensor and microcontroller that can be programmed to collect water level measurements at various time intervals. Data are logged by the system in the field and can be downloaded for processing and analysis. This system is not only inexpensive, but also small, lightweight and versatile, and can be readily adapted to collect additional parameters. © 2006 Elsevier B.V. All rights reserved. Keywords: Infiltration; Infiltrometer; Instrumentation; Microprocessor; Infrared sensor

1. Introduction Infiltration describes the process by which water, applied to a soil surface, moves into the soil. Infiltration properties of a soil have important implications for plant production, soil and water conservation, and irrigation management. Infiltration is a major factor in controlling crop yields and delivering water and agricultural chemicals to the soil profile (Ersahin and Karaman, 2000). Low infiltration rates can result in loss of irrigation water to runoff and loss of soil to erosion. Knowledge of soil infiltration characteristics of a soil can be used in farm management to predict and manage the performance of irrigation systems (Cahoon et al., 1993), and is important in evaluating the effectiveness of clay linings or other treatments to reduce seepage from sanitary landfills (Bouwer, 1986). Infiltration rates are of particular importance to turfgrass specialists for proper golf course design and maintenance (Ok and Anderson, 2003; Gregory et al., 2005).

∗

Corresponding author. E-mail address: [email protected] (K. Milla).

0168-1699/$ – see front matter © 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.compag.2006.05.001

K. Milla, S. Kish / Computers and Electronics in Agriculture 53 (2006) 122–129

123

Local soil infiltration properties may vary significantly (Ersahin and Karaman, 2000; Josiah et al., 2001; Ersahin, 2003; Cichota et al., 2003) due to a number of factors, including tillage practices, biological activity in the soil, crop or plant influences (Vervoort et al., 2001), natural variations in soil physical properties, and compaction. Depending on soil characteristics, it can take several hours to several days to make one reliable infiltration rate measurement (ASTM, 1994; Diamond and Shanley, 2003). Characterizing infiltration properties of a site can be very tedious and time consuming, especially when many measurements must be made. Although several methods exist for measuring vertical soil infiltration rates, use of a cylinder infiltrometer is one of the most common. This method is simple and equipment costs are relatively inexpensive. The infiltrometer consists of either a single metal ring or two open concentric metal rings that are inserted into the ground and filled with water. Measurements are made on water volume over time as it moves into the soil. In the double-ring infiltrometer, water infiltrating from the outer ring is considered to promote one-dimensional vertical flow beneath the inner ring (ASTM, 1994). Bouwer (1986) cites evidence that this notion is erroneous, and that increasing the size of the cylinder to at least 1 m diameter is the only way to obtain truly accurate measurements of infiltration rates. Even though smaller infiltrometers generally overestimate infiltration rates, the double-ring test using a 15–30 cm inner ring diameter is a common, well documented and relatively simple field test that can yield valuable information on general magnitude of infiltration rates and variability of soil infiltration properties. There are two techniques for determining infiltration rate with a double ring infiltrometer. For both methods water is maintained at a constant level in the outer ring. In the falling head test, the inner ring is filled with water and the decrease in water level is measured at regular intervals until the water disappears from the ring. In this method the head decreases as water level drops. In the constant head technique, the water in the inner ring is maintained at a constant level and the volume of water used to maintain this level is measured over time. A common method for maintaining water level is with the use of a Mariotte siphon. This consists of a sealed container filled with water, into which are inserted two tubes. One tube supplies water to the infiltrometer. The second tube is an air intake tube that regulates the water level within the inner ring of the infiltrometer (Bouwer, 1986). Numerical modeling suggests that both falling and constant head methods yield similar results for fine-textured soils, but that the falling head technique overestimates infiltration rate in coarse textured soils (Wu et al., 1997). We were only able to identify a few other methods in the literature for automating infiltration rate determinations using cylinder infiltrometers. Constantz and Murphy (1989) devised a method of determining infiltration rate by logging the gas pressure changes within a Mariotte reservoir during water outflow using a single pressure transducer. Gas pressure readings were converted by the data logger to produce a record of infiltration rate versus time. Prieksat et al. (1992) developed an automated, self-regulating system for use with a single-ring infiltrometer. The system employs two pressure transducers to measure water height changes in a Mariotte reservoir. Data are recorded by a datalogger. Maheshwari (1996) described a design for an automated double ring infiltrometer that employs two electrical contact water level sensors (for inner and outer rings), a capacitance type depth sensor and solenoid valves. A 12 V wet-cell car battery is used to power the solenoid valve and a laptop computer is required to control the valves and collect data from the sensor. This paper discusses the construction of a simple, inexpensive infrared sensor system for making automated water level measurements. The system has a number of potential applications. In this paper we demonstrate the automation of soil infiltration measurements using a double-ring cylinder infiltrometer. Although various types of liquid level sensors are commercially available, we were unable to locate any commercial source for a completely automated system that can collect and store measurements unattended over a period of time. The system described can be constructed for less than US$ 200 at the time of publication. 2. IR sensor/microprocessor infiltrometer system 2.1. Description of system The system described here has several advantages over those discussed above. The sensor is small, inexpensive, lightweight and versatile. It can be mounted on either single- or double-ring infiltrometers of any size or on the Mariotte reservoir. Additional sensors can be connected to the system, thus enabling collection of parameters such as water temperature and soil moisture. The sensor measures distance to water level, and can thus be used for other applications requiring water level measurements.

124

K. Milla, S. Kish / Computers and Electronics in Agriculture 53 (2006) 122–129

Fig. 1. Schematic diagram of the wiring configuration for the BX-24 computer IR sensor system.

2.1.1. Microcontroller A BX-24 microcontroller assembly produced by NetMedia was selected for interfacing with sensors used for infiltration measurements. The BX-24 uses an Atmel AT90S8535 microprocessor, a 32 kbyte EEPROM, and a 5 V, low-voltage dropout regulator/monitor, which can work from external voltages ranging from 5.5 to 12 V. The BX-24 was mounted on a Parallax Basic Stamp II carrier board. The board has serial and battery connectors and plug-pin I/O pin connectors. The carrier board eliminates the need for extensive hand wiring of connections. A wiring schematic for the circuit is shown in Fig. 1. Two LEDs with internal current-limiting resistors are used to communicate the status of programming operations and sensor measurements. A momentary-on, push button switch is used to initiate and suspend sensor measurements. Sensor measurements and a time stamp are stored to EEPROM and transferred to a desk or laptop computer following fieldwork. The Atmel microprocessor has several features that make it desirable for incorporation into a field based data acquisition system. It has 16 general purpose I/O lines that are TTL and CMOS compatible. Eight of the 16 lines can be used with an internal, 8 channel, 10-bit analog to digital converter (ADC). No additional hardware is necessary for working with analog voltage signals from the sensor system. The ADC inputs must be within a range of 0–5 V and sensors normally have a common ground with the BX-24. The 32 kbyte EEPROM (Atmel AT35656) offers sufficient memory to store the operating program and several thousands sets of measurements (e.g., time, distance to water, temperature). An external LT1129 voltage regulator provides power to sensors. The external voltage regulator is necessary because the onboard BX-24 can only support a maximum current of 20 mA per individual output and a combined maximum current load of 80 mA, which may not be sufficient to support all necessary hardware and sensors. The LT1129 offers an additional advantage in that it can be shut down at times when sensors do not need to be active. This can provide a significant power saving capability to the system. 2.1.2. Infrared distance sensor The drop in water height during an infiltration measurement is determined by a Sharp GP2D12 infrared distancemeasuring sensor. The sensor contains an IR-LED transmitter and a large surface area photodiode position sensitive detector (PSD). The transmitter and detector are housed in a single module, which contains a lens for both the transmitter and receiver. The IR beam illuminates a 1.5 cm spot on a target surface at a distance of 10–30 cm. The

K. Milla, S. Kish / Computers and Electronics in Agriculture 53 (2006) 122–129

125

Fig. 2. Digital ADC values for 300 sensor-target measurements made at a distance of approximately 12 cm. Each increase in digital value represents a voltage difference of 4.89 mV. Vertical axis represents counts, horizontal axis is digital value of sensor response.

image of the illuminated spot is focused through the lens located in front of the PSD. The position of the focused spot on the PSD surface is controlled by the distance between the two lenses of the sensor (approximately 20 mm) and the distance (L) between the sensor and the illuminated surface. The position of the focused beam on the PSD produces a differential current between the two ends of the PSD. The difference or ratio of the current will be proportional to 1/L. For the GP2D12 the working range of the unit is 10–80 cm, however, the maximum sensitivity is restricted to a range of 10–30 cm. In this range a change of 1 mm in distance produces a 20 mV change in the output signal. This is approximately four times greater than the lower resolution limit of the BX-24 ADC (4.89 mV). 2.2. System construction and calibration The Sharp GP2D12 (http://document.sharpsma.com/files/Inf-Use of Optical Sensors.pdf) is connected to the BX24 computer through a 4-wire telephone cable and wall bracket module. Individual sensor measurements are prone to non-Gaussian errors produced by voltage spikes during the firing of the IR LED. Signal conditioning and data processing are necessary to produce acceptable accuracy and precision (±1 mm). The manufacturer recommends placement of a 10 ␮F capacitor between the Vcc and ground connection of the sensor housing. Even with signal conditioning, some voltage spikes are present (Fig. 2). A rudimentary means of removing outlying measurements is to use ADC digital values that have the greatest frequency of occurrence. This procedure is a form of “consensus” averaging. The GP2D12 sensor was calibrated with the sensor mounted on a movable clamp attached to a ring stand. A rule with 1 mm divisions was used to measure the distance between the sensor and a white plastic target. A calibration curve for measurements in the 10–25 cm range is presented in Fig. 3. In this range, a change of sensor-target distance of 1 mm produces a change of four digital units on the ADC. Repeated calibration of the GP2D12 in the 10–30 cm range demonstrates a very high degree of both precision and accuracy (±1 mm) and long-term (1–3 h) stability for measurements, if signal noise is reduce by appropriate signal conditioning and data processing. Tests of the sensor with different housing configurations and using the sensor with the IR beam passing through a glass window of a Mariotte siphon produce minor offsets in the voltage–distance response. Relative distance response was very similar. During testing is was found that direct sunlight swamps the sensor’s IR signal, therefore all measurements must be made with a protective shade such as a beach umbrella or a small “tent” constructed from aluminized Mylar film. IR measurements in the Mariotte siphon use the container cap as a sun shield. Enclosures for the BX-24 computer and GP2D12 IR sensor can be constructed of inexpensive PVC plumbing components (Fig. 4). For normal use, the housing only needs to be moisture resistant. The different portions of the housing can be assembled in the field and sealed with electrical tape. We found this construction to be rugged enough for prolonged use under normal conditions. An inexpensive Mariotte siphon was constructed from a new, 15–20 L, hermetically sealed plastic paint container. The container cap has a circular hole in which a female PVC coupling is attached and the edges are sealed with silicon caulking. The female coupling serves as a base for a square-end PVC plug (Fig. 5). The end of the plug is cut off and

126

K. Milla, S. Kish / Computers and Electronics in Agriculture 53 (2006) 122–129

Fig. 3. Reciprocal distance vs. voltage calibration curve for the GP2D12 IR sensor. Calibration curve formula is a second order polynomial curve.

a Pyrex glass disc is sealed in the base of the plug. This serves as a window, allowing the Mariotte siphon to remain under partial atmospheric pressure, while the IR sensor can range water height via a floating Styrofoam target disc. A vent tube controls the height of water at the infiltration ring. Water is transferred from the bottle to the ring by a drain located at the base of the bottle. If the IR sensor is used to take falling head measurements, the sensor is placed directly over the infiltration ring. A Styrofoam float covered with white plastic laminate is placed in the ring as a reflecting surface (Fig. 6). A sunshade must be used to prevent direct sunlight from interfering with the IR signal from the sensor. For continuous head measurements using a Mariotte reservoir, the sensor is placed in a housing attached to the top of the reservoir. The reflective float is placed inside the reservoir. The sensor makes measurements through a sealed, 7.6 cm (3 in.) glass window. This allows

Fig. 4. Diagrammatic representation of BX-24 computer and IR sensor housing modules. The computer hosing is constructed from 7 cm (3 in.) PVC pipe and components. The IR sensor housing is constructed from 5 cm (2 in.) PVC components.

K. Milla, S. Kish / Computers and Electronics in Agriculture 53 (2006) 122–129

127

Fig. 5. Diagrammatic representation of Mariotte siphon/IR sensor system. The Mariotte siphon is a 15 L, hermetically sealable plastic paint container. The IR window assembly is constructed from 7 cm (3 in.) PVC pipe components.

the internal pressure within the Mariotte siphon to be controlled by a single vent tube. The Mariotte siphon housing acts as an effective sunscreen and no additional sunlight protection is necessary. Water temperature in the infiltration ring is measured by a LM35 precision analog centigrade temperature sensor enclosed in a heat shrink covered copper tube that is attached to the inside rim of the infiltrometer cylinder. Temperature measurements are taken after approximately 5 min after the addition of water to the system. This allows time for temperature equilibration between the sensor and water. During normal operation the BX-24 and IR ranging system will require an average supply current of 75 mA. If the GP2D12 IR sensor runs continuously, a 9 V alkaline battery will operate the system for approximately 8 h. If the LT1129 voltage regulator is used to shut the sensor down when measurements are not being made, the system may operate for up to 24 h on a 9 V cell. This amount of time should be sufficient for most infiltration measurements. 3. System testing and results An evaluation of the performance of the IR infiltrometer system was made at an experimental agricultural plot on the campus of Florida A&M University. The test site had a gently sloping surface (5–7◦ ), part of which was cultivated

Fig. 6. Photograph of IR sensor system set up to make falling head measurements in the infiltrometer inner ring. Divisions on ruler are 5 cm.

128

K. Milla, S. Kish / Computers and Electronics in Agriculture 53 (2006) 122–129

Fig. 7. Plots of water level measurements vs. time for experimental infiltration tests. (A) Falling head measurements. Plus symbols represent manual measurements for comparison in an uncultivated portion of the test site. The trend line shown through these points is a linear regression line with an R-squared of 0.99. Diamond symbols represent measurements for two IR sensor runs at this same site. R-squared for open diamonds is 0.99. R-squared for closed diamonds is 0.98. Circles represent IR sensor measurements for two runs in a cultivated portion of the test site. R-squared for closed circles is 0.98. R-squared for open circles is 0.98. The inflection points in the open and closed circle trends may represent leakage from the infiltrometer. (B) Constant head measurements using Mariotte siphon and IR sensor. Water height on Y-axis represents equivalent water levels in the inner ring as calculated from Mariotte reservoir readings. For clarity, points on both (A) and (B) are plotted at greater time intervals than collected by the sensor.

and planted with corn. Soil at the site is described as Orangeburg fine sandy loam (Sanders, 1981). The relatively high permeability of this soil produced nearly constant infiltration rates over the 10–60 min test runs. For each trial, a Turf-Tec IN5-W double ring infiltrometer (15 cm inside diameter, 30 cm outside diameter) was set in the soil to a depth of approximately one half of its height (10 cm). Several falling head trials were made in both cultivated and undisturbed portions of the plot. Falling head measurements made at a smooth, uncultivated, partially shaded site produced results that were uniform throughout the trials. A manual falling head measurement at this site had a value of 16.9 cm/h (plus symbols on Fig. 7A). Two IR runs at the same site made under different soil moisture conditions yielded values of 13.7 and 19.8 cm/h (open and closed diamond symbols on Fig. 7A). Preliminary trails using manual measurements in a cultivated area yielded relatively high infiltration rates (20–35 cm/h) (for clarity, manual measurements at the cultivated site are not shown on Fig. 7A). Two falling head IR-sensor runs at the cultivated site produced infiltration rates similar to those obtained by the manual method (open and closed circles on Fig. 7A). The high infiltration rates observed at this location may have been affected by leakage problems produced by incomplete sealing of the infiltration rings. Leakage is suggested by increases in infiltration rates that appear as inflections on the IR plot. For all IR runs, precision is comparable to the manual method (see caption for Fig. 7). Constant head measurements were made at the second site using the IR sensor system in conjunction with a Mariotte siphon. The IR sensor collected water elevation readings from a Mariotte siphon supplying the inner ring of the infiltrometer. A second Mariotte siphon was used to maintain a constant head in the outer ring. A 1.1 h run (Fig. 7B) yielded a nearly constant rate of 13.7 cm/h, very similar to manual and IR-sensor falling head measurements at this location (Fig. 7A).

K. Milla, S. Kish / Computers and Electronics in Agriculture 53 (2006) 122–129

129

Table 1 Cost of Materials for Infiltrometer System Basic 24X Microprocessor Serial cable Parallax Basic Stamp Carrier Board Sharp GP2D12 IR ranger (2) Miscellaneous Hardwarea Total

$50 $5 $25 $35 $55 $170

a Plastic buckets and tubing, PVC fittings for IR ranger, electrical parts (batteries, LED’s, voltage regulator and push button switches). Other miscellaneous items (wire, solder, glue) not included in cost tabulation. Cost does not include the double ring infiltrometer. We used a Turf Tech IN5-W ($150).

4. Cost effectiveness A tabulation of material costs for the system is presented in Table 1. A complete system, including sensor, controller and Mariotte siphon can be constructed for less than US$ 200. No commercial source for automated infiltrometer systems could be located for comparison. Construction of an automated system from commercially available “off-the-shelf” components would cost considerably more. For example a Campbell Scientific CR10X datalogger, a measurement and control module commonly used with various types of soil and meteorological sensors, was recently quoted at US$ 1250. 5. Conclusion Preliminary testing of the IR/microprocessor sensor system for automation of infiltration measurements demonstrates that it is capable of reproducing results obtained from manual measurements. When used with a Mariotte siphon to maintain constant head in the infiltration ring, the IR system can automate the measurement process, thus significantly reducing the amount of labor involved and the potential for human error. Because the materials are inexpensive, a network of systems could be deployed in a relatively cost-effective manner. References ASTM, 1994. D3385-94, Standard test method for infiltration rate of soils in field using double-ring infiltrometer. In: Annual Book of ASTM Standards 04.08, pp. 356–361. Bouwer, H., 1986. Intake rate: cylinder infiltrometer. In: Methods of Soil Analysis, Part 1. Physical and Mineralogical Methods, Agronomy Monograph no. 9, second ed. American Society of Agronomy-Soil Science Society of America, pp. 825–844. Cahoon, J., Klocke, N., Kranz, W., 1993. Crop residue and irrigation water management. University of Nebraska Cooperative Extension Publication G93-1154-A (http://ianrpubs.unl.edu/irrigation/g1154.htm). Cichota, R., van Lier, J., Rojas, L., 2003. Spatial variability of infiltration in an Alfisol. Revista Brasileira de Ciˆencia do Solo 27, 789–798. Constantz, J., Murphy, F., 1989. An automated technique for flow measurements from Mariotte reservoirs. Soil Sci. Soc. Am. J. 51, 252–254. Diamond, J., Shanley, T., 2003. Infiltration rate assessment of some major soils. Irish Geogr. 36, 32–46. Ersahin, S., Karaman, M.R., 2000. Assessment of infiltration rate parameters for site-specific water management. Paper presented at International Symposium on Desertification, June 13–17. Konya, Soil Science Society of Turkey (http://www.toprak.org.tr/isd/isd 48.htm). Ersahin, S., 2003. Comparing ordinary kriging and cokriging to estimate infiltration rate. Soil Sci. Soc. Am. J. 67, 1848–1855. Gregory, J.H., Dukes, M.D., Miller, G.L., Jones, H., 2005. Analysis of double-ring infiltration techniques and development of a simple automatic water delivery system. Appl. Turfgrass Sci. (May) (http://www.plantmanagementnetwork.org/pub/ats/guide/2005/ring/). Josiah, M.N., Upadhyaya, S.K., Rosa, U., Andrade, P., Mattson, M., 2001. Mapping field variability in infiltration rate and evapotranspiration in a tomato production system. In: American Society of Agricultural Engineers International Meeting, Sacramento, California, July 29–August 1. Maheshwari, B.L., 1996. Development of an automated double ring infiltrometer. Aust. J. Soil Res. 34, 709–714. Ok, C., Anderson, S., 2003. Top Soil. Grounds Maintenance, June 1 (online at http://www.grounds-mag.com). Prieksat, M.A., Ankeny, M.D., Kaspar, T.C., 1992. Design for an automated, self-regulating, single-ring infiltrometer. Soil Sci. Soc. Am. J. 56, 1409–1411. Sanders, T.E., 1981. Soil Survey of Leon County, Florida. United States Department of Agriculture, Soil Conservation Service. Vervoort, R.W., Dabney, S.M., R¨omkens, M.J.M., 2001. Tillage and row position effects on water and solute infiltration characteristics. Soil Sci. Soc. Am. J. 65, 1227–1234. Wu, L., Pan, L., Robertson, M., Shouse, P., 1997. Numerical evaluation of ring infiltrometers under various soil conditions. Soil Sci. 162, 771–777.