Reading Assignment:

1. Variable Rate Technologies. In: The Precision Farming Guide for Agriculturists by Deere & Company, pages 79 to 93.

2. Sudduth, K.A., Hummel, J.W., and Birrell, S.J. Sensors for Site-Specific Management. In: Pierce, F.S.; Sadler, E.J. The State of Site-Specific Management for Agriculture. ASA/CSSA/SSSA, 1997. pages 183-207.

1.0. INTRODUCTION:

The ultimate goal of precision farming or site specific management is to manage the farm on a site-by-site basis. Knowledge of the soil and crop characteristics on a fine grid basis is therefore needed. Traditional soil and plant sampling and analysis methods are very expensive, tedious, and time consuming for obtaining soil and crop parameters on a fine grid and at a short time scale. Sensors capable of gathering information on-the-go are needed. They will be particularly useful to measure parameters that vary faster in time, such as nitrogen and soil water content.

Sensors are being developed and/or used for the following applications:

• Soil Properties Sensing: Soil Texture, Structure, and Physical Condition ; Soil Moisture; Soil Nutrients.
• Crop Sensing: Plant Population; Crop Stress and Nutrient Status.
• Yield Monitoring Systems: Crop Yield; Harvest Swath Width; Crop Moisture:
• Variable Rate Technology Systems: Fertilizer flow; Weed detection, pressure sensors.

Contact sensors, ultrasonic sensors, force, pressure, linear displacement, and optoelectronic sensors are usually applied to obtain indirectly the flow of produces in the combine and to evaluate the yield on-the-go.

2.0. SENSING SOIL PROPERTIES:

Electrical Conductivity:

The electrical conductivity of the soil depends on factors such as amount and type of clay, salinity, soil cation exchange capacity, soil moisture content, depth to the claypan, presence of specific ions, among others. Even though all the above factors affect crop yield, it is difficult to infer based on the soil EC which one could be responsible for decrement in yield in a certain field. The fact that some of these factors have temporal variation add to the difficulty. However, the EC can be mapped to identify contrasting soil proprieties and used as a tool for precision farming management.

Two methods are available nowadays to measure bulk EC on-the-go: the electromagnetic induction and the direct EC measurement method.

Tillage Force as an Indicator of Soil Strength:

A group headed by Upadhyaya in the Bio. & Agr. Engr. Department at the UCDAVIS is working on the development of a multiple sensor system to measure soil physical properties. The idea is to relate soil physical properties with an index, the texture compaction index – TCI .

The instrument described by Lui et al.(1996) uses two load cells to measure draft, another load cell to measure vertical load, a radar gun to measure ground speed, a displacement sensor, a moisture sensor, linear potentiometer to measure depth, and a DGPS receiver with sub meter accuracy (Fig. 05 and Fig. 06) . The moisture sensor measures changes in dielectric constant using a resonance frequency and phase lock technique. A data acquisition system (Fig 05) collects and stores the data from the sensors.

If the geometry of the instrument is kept unchanged, the draft of a tillage tool depends basically on soil physical properties, operating depth, and speed. If one operates the device at a constant ground speed (monitored by the radar) and depth (kept constant by the depth control wheel), the draft will be a function of soil physical properties only and can be expressed as:

D = f(BD, MC, TC)

Where D = draft of the tillage tool (measured by the load cells); BD = soil bulk density (compaction component); MC = moisture content; and, TC= texture.

If soil moisture is independently measured, than f(.) can be expressed as: D = g(BD, TC)* h(MC)

Where g(.) incorporates the effect of compaction and texture and is called TCI, that is: TCI = g(BD, TC)= D/h(MC)

Lui et at (1996) defined h(MC) as an exponential function of the form: h(MC) = exp(-k*q )
where q is the soil moisture content expressed on volume basis. The value of k was 0.11 for an untilled Yolo loam soil.

The system operated in a tomato field furrow irrigated. Measurements were taken at the bottom and top of the bed. The resulting map of the draft is presented in Fig. 7.

Soil Water Infiltration

Josiah et al. (1999) applied the ultrasonic displacement sensor technology to measure the water level in a flow measurement flume (Fig. 11). The ultimate goal was to come up with a way to quantify infiltration variability in furrow irrigated tomato field. The sensor performed very well in the field. A timer, which was trigged by the approaching wetting front, was used to measure the advance time of the wetting front.
 

Soil Moisture Content

Pelletier et al. (1996) used near infrared reflectance to determine soil moisture content. They tested the methodology in laboratory using the NIRs System, 2500 Spectra Photometer for six soil types representing a diverse range of soil texture. Seven levels of moisture content were tested. The PLS equation which used the wavelengths from 820-880 nm and 920-960 nm only resulted in a very good linear fit (r²=.91) with a standard error of 2.4% of m.c.. Laboratory validation of this calibration was performed by using the calibration equation to predict an independently prepared and scanned set of spectra from the Yolo loam soil, resulting in r2=.85 (standard error=1.9%.) The results were not acceptable according to the authors.
A better calibration was obtained when absorption bands for water corresponding to 1450nm and 1910nm were added to the model (r²=0.98, standard error=1.21% m.c.).

Results in the 700-960nm range are not as good as in the 1400 –2000nm, however, the smaller range can be detected using inexpensive silicon photo-diodes, while the higher range requires lead sulfide (PbS) detectors, which are more expensive.

The methodology applied to field did not produce as good a result unless new calibration to correct difference in soil particle size variation was applied (r²=0.97, standard error=1.25 % m.c.)

Rossel and McBratney (1998) conducted a laboratory experiment and found the 1600, 1800, 2000, and 2100 nm to be the most suitable wavelengths for simultaneous measurements of clay and soil water content.
 

Soil Organic Matter
 

Hummel et al (1996) developed an optoelectronic based soil organic matter (SOM) sensor. They presented a single wavelength sensor and a multiple wavelength sensor. The single-wavelength sensor although inexpensive needs to be recalibrated for the soil and moisture conditions that prevail at the time of use. The multiple-wavelength sensor can utilize a single calibration to predict SOM over a range of soil moistures and soil types that occur within a geographical area of several hundreds of kilometers. In addition, it can be used to sense soil moisture and cation exchange capacity (CEC).

Excellent correlation (r²=0.92, standard error=0.34% for SOM) was obtained with laboratory data when the NIR data were smoothed to a 60 nm data point spacing for wavelengths in the 1720-2380 nm range, for a total of only 12 data points used. (Sudduth and Hummel, 1991). Field application of the sensor did not yield acceptable results (standard error =0.91% for SOM).

Soil Nitrogen

Nitrogen fertilizer is usually applied uniformly throughout a field, although it is known that soil fertility varies considerably within a field. Mineralization (conversion of soil organic matter to mineral-N by microbes), denitrification (breakdown of NO^-₃ to N₂O by microbes), and leaching are some of the major reasons for the variability of nitrogen levels in soil. Since increasing nitrate concentration in the ground water is a concern, it is wise to use techniques that avoid excessive use of fertilizer (Ehsani et al. 1999).

Sensors and instrumentation to determine soil texture and mineral-N rapidly in the field are essential to prescribe the correct amounts of fertilizer on a site-specific basis. Development of these sensors, in particular a nitrate sensor, is a major bottleneck in implementing precision agriculture.(Ehsani et al. 1999).

Upadhyaya et al. 1994 concluded that NIR absorbance data in the 1800 to 2300 nm range could be used to determine soil mineral-N content reasonably well. Ehsani et al. (1999) applied Partial Least Squares (PLS) and Principal Component Regression (PCR) to relate soil mineral-N to absorbance data obtained using NIR spectroscopy. They tested two soils (Yolo loam and Copay clay) and three sources of nitrogen (ammonium sulfate, calcium nitrate, and ammonium nitrate.) in both laboratory and field conditions. Spectral response over many wavelengths (1800 nm to 2500 nm in increments of 2 nm) were obtained.

They concluded that the NIR response of soil in the wavelength range of 1800 to 2300 nm can be used to determine the nitrate content of soil successfully, provided that the calibration equation is derived from soil samples obtained from the same location. The practical implication of this result is that multi-wavelength NIR absorbance spectroscopy has the potential to map the nitrate variation over a large area of several hundred acres provided a few soil samples about 50 were analyzed using analytical laboratory techniques to provide a calibration curve for that site.

Adsett and Zoerb (1991) developed a real-time nitrate sensor using ion selective electrodes. The automatic field monitoring system consisting of a soil sampler, a nitrate extraction unit, a flow cell, and a controller were tested under field and laboratory conditions. The major disadvantage of the system was the time required to extract nitrate.

A hand-held model which utilizes similar concept is commercially available from Spectrum Technologies, Inc (www.specmeters.com ).

Birrell and Hummel (1993) investigated the possible use of ion selective field effect transistor- ISFET to measure soil nitrate levels. They suggested ISFET has several advantages over ion selective electrodes such as small dimension, low output impedance, high signal/noise ratio, fast response and the ability to integrate several sensors in a single chip, the ability to use small sample volumes and sense multiple species simultaneously. Disadvantage of ISFET include greater long-term drift and hysterisis than ion selective electrodes. However, these disadvantages are eliminated in dynamic measurements. They tested ISFET with four integrated sensors in a Flow Injection System using four different flow rates (0.04, 0.09, 0.14, and 0.19 mL/s), five sample injection times (0.25, 0.50, 0.75, 1.2 sec), and three washout times (0.75, 1.0, 2.0 sec). The correlation coefficients of linear regression of the peak height against the logarithm of the nitrate concentration of standard solutions were within the range of 0.89 - 0.99, except for the lowest flow rate (0.75 to 0.99).

A hand-held pH measurements model is commercially available from Spectrum Technologies, Inc (www.specmeters.com ).

IQ Scientific Instruments www.phmeters.com (San Diego, CA) has a hand-held product for measuring pH that suits several applications.
 
 

3. CROP SENSING:

Plant Population:

Plattner et al. (1996) developed and tested an optoelectronic sensor for determining maize population. The sensor consisted of a photoelectric emitter and receiver pair and measures in-row distance between plants. The sensor was mounted on a maize-head on a combine. Average plant spacing was estimated with ± 3.1% error an early growth stage, and ± 6.2 % error at harvest.

Photosynthetically Active Radiation(PAR):

Kebabian, et al. (1998) designed and built an optical sensor for real-time in situ sensing of photosynthetic activity in plant. The sensor operates on the principle that as light collected from the fluorescing plants is passed through a cell containing oxygen at low pressure, the oxygen will absorb the energy and subsequently re-emit photons which can be detected by a photomultiplier tube.

Spectrum Technologies, Inc produces an optoelectronic sensor to measure chlorophyll in plants (SPAD 502 Chlorophyll Meter). Possibility of relating chlorophyll measurements with nitrogen deficiencies in plants can be a major use for this device.

4. YIELD MONITORING SYSTEMS:

Wallace (1999) evaluated an optoelectronic cotton yield monitoring system and found a significant linear relationship (r²=0.99) between values provided by the system and measured weight.

Zycom Corporation, Bedford, MA(USA), and Ag Leader Technology, Inc, Ames, IA (USA). ( www.agleader.com/pf3k-cotton-sensor.htm ) have both developed cotton yield monitoring system based on optoelectronic sensors.

The Ag Leader flow sensor (Fig. 9) is comprised of an emitter unit and a detector unit. The emitter unit emits five beams of infrared light to the detector unit mounted on the opposite side of a picker chute. The beams of light are interrupted by the cotton that is conveyed to the basket. This information is computed into seed cotton weight. Then, with the area calculation, the monitor computes and displays pounds/acre or kilograms/hectare (metric mode).

In a yield monitoring system, a significant source of error is introduced when one assumes the actual width of crop cut is the same as that of the header width. Stafford et al. (1997) used an ultrasonic displacement sensor to continuously monitor cut width in a wheat crop. The sensors could eliminate nearly 10% error in yield that results when a constant, full width is assumed.

5. VARIABLE RATE TECHNOLOGY SYSTEMS:

Mass Flow of Fertilizer

Swisher et al. (1999) designed, built, and laboratory tested an optical sensor to measure granular fertilizer flow in an air system. The sensor consists of a laser line generator that continuously transmits light across a trapezoidal chamber to two 16-element photodiode arrays. Every time an air-born grain of fertilizer crosses the light an interruption is detected by the diode and an electrical count is generated. These counts are then converted to mass flow. The sensor performed well for both static and dynamic measurements, however, calibration have to be performed for every product being sensed.

Weed Detection:

Biller (1998) used optoelectronic sensor for detection weeds against a plant or soil background. The system distinguishes green plants from soil by their different light reflection properties. In a maize field he succeeded in identifying the weeds between crop rows. The system saved 30-70% herbicide compared to normal application and eliminated 100% of weeds.

Feyaerts et al. (1999) went one step farther and used a spectrograph to collect the spectral signature of the plants. Under controlled conditions, the system could distinguish weeds from corn and sugarbeet with a success rate of up to 90%.

Heisel et al. (1999) developed a sensor for weed detection that uses a digital camera. Preliminary experiments have shown that the sensor system can detect small weed seedlings when the camera travels at a speed of 45 km/h. The system was useful to estimate the percentage leaf area as well.

Slaughter et al. (1999) developed an color machine vision based automatic guidance system for precision guidance of an agricultural cultivator. The guidance system was designed to operate in weedy row crop fields at the time of first cultivation. The system consisted of two solid-state color video cameras with CCD-Iris, a computer, two manual iris video camera lenses (8.5 mm focal length, f 1:1.5), two high-speed 24 bit color video frame acquisition boards, and an electronically controlled, proportional hydraulic control valve. When tested on tomato, cotton and lettuce fields, the system’s performance varied from a root mean square (RMS) guidance error of 7 mm under low weed loads to 12 mm under high weed loads, and was capable of operating at travel speeds up to 16 km/h.

A1. ELECTRICAL CONDUCTIVITY

A1.1. Electromagnetic induction (EM):

This technology uses electromagnetic energy to measure the apparent conductivity of the soil. The device is composed of a transmitter and a receiver coil installed usually 1.0 m (3.3 ft) apart in a non-conductive (wooden) bar in the opposite ends of the instrument (Fig. 01). The transmitter coil is energized with an alternating current, generating a time-varying magnetic field in the earth. This magnetic field causes current to flow in the soil, and a secondary magnetic field is generated. The ratio of the secondary to the primary magnetic field is proportional to the ground conductivity of the soil (McNeill, 1980; Sudduth et al., 1993).

Variations in electromagnetic response are related to changes in the ionic concentration of the soil. Soil parameters such as moisture content, amount and type of ions in the soil water, amount and type of clay in the soil matrix are correlated to the response of the system (Doolittle et. al. 1994).

Applications :

1. Estimation the depths to claypans : Sudduth and Kitchen (1993), Doolittle, et al. (1994) correlated depth to clay pan to electrical conductivity of the soil. The technology works better if the instrument is locally calibrated.

2. Mapping of sand deposition from floods : Kitchen et al (1996) coupled EM reading with GPS data to map the post flood sand deposition depth at farm land in four sites along the Missouri River. The methodology worked well when a well defined pre flooded layer of soil was present. Local and temporal calibration improved the results as well.

Advantages: Low cost per hectare, non-destructive technique, maps very fast.

Disadvantages: Measurement will reflect conjunctive effect of physical and chemical properties of the soil. Need to do the mapping during dry season to isolate the effect of water content. Need to stay away from any metallic structure or electricity source.

Commercially available systems: EM38 by Geonics Limited, Mississauga, Ontario, Canada at www.geonics.com .
 Back to Soil Properties
A1.2. Direct Measurement:

A pair of coulter-electrode is inserted into the soil. Current is applied at one electrode and the voltage drop across the two electrodes is measured (Fig. 02). The conductivity is then computed as the ratio of current to voltage difference. The system can be stationary or mobile, and paired electrodes can be installed at different depths.

Commercially available systems: SOIL EC MAPPING SYSTEMS from VERIS technologies (www.veristech.com ) (Fig. 03). Maps of  EC and corn yield are presented in Fig. 04.

 Back to Soil Properties
 

A2. OPTOELECTRONIC SENSORS:

Optoelectronics deals with the interaction of electronic processes with light and optical processes. Devices where such interaction takes place are called optoelectronic devices.(Bhattacharya, 1997).

Optoelectronic sensors in general use the characteristics of different materials to have a defined spectral signature. The system is composed of a light source (emitter) that emits light in a wavelength range, and a sensor (detector) for specific ranges of wavelength.
Examples of emitter/detector are the infrared emitting diodes devices built by LITE ON, Inc(www.liteon.com  ) presented in Fig. 08. There one can see the spectral signature of a pair of infrared emitter/detector diodes in the range of 940 nm. An application of such technology is presented in the yield monitoring sensor of Ag. Leader  (Fig. 09).
The optoelectronics technology has been successfully applied for weed detection(Biller et al., 1997; Biller, 1998),  measurement of granular fertilizer flow in air stream (Swisher et al., 1999), as component of yield monitoring system for cotton (Wallace, 1999), for soil organic matter detection (Hummel et al., 1996), soil texture identification and soil water content measurement (Rossel and McBratney, 1998).

Soil color and reflectance are function of factors such as soil organic matter, moisture, texture, mineralogy, and parent material. Sensors which use only one or a few pieces of spectral information will not be able to explain the isolated effect of one of these factors. The use of multiple range of wavelengths has been verified by Pelletier et al. (1996) and Hummel et. al. (1996) as an effective way to generalize results and eliminate the need for calibration on a local basis.

Other important component in the optoelectronic sensor system is the algorithm applied to classify the response of the sensor according to the material being sensed. Common methodologies include maximum likelihood classification (Price and Gaultney, 1993), partial least square regression (PLSR) (Hummel et al.,1996), stepwise multilinear regression-SMLR (Sudduth and Hummel, 1993c), Partial Least Squares (PLS) and Principal Component Regression (PCR) (Ehsani et al., 1999) , and artificial neural networks-ANN (Yang, 1997).

Advantages: High speed of analysis, simple sample preparation, nondestructive analysis of the sample and omission of chemicals (pollution free) (Updhyaya, 1996), low cost, easy to implement.

Disadvantages: results may be sensitive to light changes in exposure; results are largely empirical, therefore, calibrations need to be performed for each medium being sensed.

Applications:

1. Organic matter, moisture content, and clay content of soil;
2. Mineral nitrogen in soil;
3. Weed detection;
4. Granular fertilizer flow;
5. Yield monitoring system for cotton;
6. Nitrogen in soil.

A3. ION SELECTIVE ELECTRODES:

An Ion Selective Electrode measures the potential of a specific ion in solution. (The pH electrode is an ISE for the Hydrogen ion.) This potential is measured against a stable reference electrode of constant potential. The potential difference between the two electrodes will depend upon the activity of the specific ion in solution. This activity is related to the concentration of that specific ion, therefore allowing the end-user to make an analytical measurement of that specific ion. Several ISE's have been developed for a variety of different ions.(Omega, 2000). A complete description on how to operate Ion Selective Electrodes devices is presented in the Omega Inc. home page at www.omega.com/techref/ph-5.html  .

 Back to Soil Properties

A4. ULTRASONIC DISPLACEMENT SENSORS:

Ultrasonic wave sensors are composed of a transmitter device, a receiver device, and a circuit control (Fig. 10) . When the waves from the emitter reach an object, part of the energy is reflected. In many practical cases, this energy is reflected in a diffuse way, i.e., regardless of the direction where the energy comes from, it is reflected almost uniformly within a wide solid angle, which may approach 180^o (Fraden, 1997).

The distance L to the object can be calculated by:

where: v = speed of the ultrasonic wave in the media, t = time for the wave to propagate from the transmitter to the object and back to the receiver.

Typically the transmitter and receiver components are piezoelectric devices that convert mechanical energy into electrical energy (and vice-versa for the transmitter).

Josiah et al. (1999) applied a linear model to relate output voltage(Vout) to the distance to the target and temperature.

Vout = a + b * L + c * T

Where L is the distance between the target object and the sensor, T the temperature, and a, b, and c empirical constants.

  Back to Soil Properties - Infiltration

Figure 01: Electromagnetic induction (adapted from McNeill, 1980)
Back to Text

Figure 02: Direct measurement of EC (Veris, Tech 1999)
 Back to Text

Figure 03: Field equipment for Direct EC measurement (Veris, Tech 1999)
 Back to Text

Figure 04: EC map and corn yield map (Veris, Tech 1999)
 Back to Text

Figure 05 - TCI : Data acquisition system and sensors
Back to Text

Figure 06 : TCI system
Back to Text

Figure 7: Texture / Soil Compaction Index (TCI) Map
Back to Text

Fig 08. PHOTODIODES
Back to Optoelectronic sensors
 

Figure 9: Optoelectronic cotton yield monitoring sensor from Ag. Leader.
  Back to Yield Monitoring Systems or Back to Optoelectronic sensors
 

 

  Back to Text

Figure 11: Flume with ultrasonic sensor.

 Back to Soil Properties - Infiltration

BIBLIOGRAPHY:
 

This site is funded by grants from USDA and ARI, and developed by California Polytechnic State University, San Luis Obispo, California State University, Fresno, and the University of California, Davis
Web site development by ATI-Net
Distribution of this material is under the GNU General Public License
http://www.gnu.org/copyleft/gpl.html