Optical sensors aid agriculture

Transcription

1 Optical sensors aid agriculture Farmers turn to information from optical sensors to make environmentally sound use of fertilizers, herbicides, pesticides, and water. Geofiey J. Shropshire and James A. DeShazer 1 Ihe efficiency of US agricultural production has increased through the use of machinery, biotechnology, and chemicals so that one farmer is now able to feed 100 Americans. During this past decade, however, many people, including those in the agricultural industry, have felt that more attention must be directed toward the preservation of the environment. Because agricultural practices such as overuse of fertilizers create nonpointsource pollution of waterways and airways, we are developing new technologies, both physical and biological, to reduce the use of chemicals and conserve natural resources. Optical-sensor development, control systems, and machinery designs are addressing these issues to provide site-specific or spatially variable farming techniques. Opticalbased sensors used in agriculture represent one area in which stides are being made to analyze and improve our environment. Soil analysis Soil in which crops are grown is complex and variable material and is often difficult to characterize precisely. Some properties important to cropland management have been selected for automated field measurement. For GEOFFREY J. SHROPSHIRE is assistant professor and JAMES A. DESHAZER is professor and department head of the Agricultural Engineering Department at the University of Idaho, Moscow, ID FIGURE 1. Data from sensor on bottom of plow shank measuring soil organic matter and moisture will help determine how much fertilizer to deposit at this point in the field. example, the amount of organic matter in a soil affects its texture and water-holding capacity and is often an indicator of fertility. The effectiveness of many herbicides is strongly dependent on the organic-matter content of the soil: an increase from 1% to 2% organic matter may increase the required amount of herbicide by 100%. (Typical levels are less than 5% organic matter by weight.) Current chemical-application equipment is designed to apply a chemical at a constant rate over the treated area. Organic matter, though, can vary greatly across a single field. The application rate is often set to a level that will be effective at all expected levels of soil organic matter. This leads to areas with herbicide rates that are higher or lower than needed. In the past, constant-rate application was not viewed as a problem, because there was usually no crop damage and the excess herbicide was not recognized as an environmental problem. Farmers and others are now trying to reduce the total amount of herbicides used, for environmental and economic reasons. One way to do this is to measure the soil organic matter and continuously adjust the herbicide application rate accordingly. Soils are now characterized by collecting samples and sending them to a laboratory for analysis. Because collecting and analyzing numerous samples would be prohibitively expensive, a device is needed to measure soil organic matter automatically and rapidly in the field. The sensor could be mounted on the applicator machine directly, or it could be used to map the organic matter for later use. Highorganic-matter soils are visually darker, but many other factors, such as moisture content and base material, also affect soil color. Researchers have taken two approaches to the soil organic-matter measurement problem. A group at Purdue University (West Lafayette, IN) developed a sensor using a single spectral band around 660 nm.' A tillage tool propelled by a vehicle scrapes a groove in the soil about 100 mm deep. Light-emitting diodes, mounted on the bottom of the tool and shielded LASER FOCUS WORLD MAY

2 from external light, illuminate a small area of soil over a range of approximately 25 mm. A photodetector then measures the reflected light in the 660- nm band. One version of this single-band sensor is available commercially from Tyler Company (Benson, MN). The device is simule. robust, and relativelv inexpensive, but it is sensitive to variations in soil moisture and soil base material. Therefore its practical use requires calibration for each landscape, and the device must be used at a time when soil moisture will be relatively uniform. Kenneth Sudduth at the University of Missouri (Columbia, MO) and John Hummel at the University of Illinois (Urbana, IL) have built a spectrophotometer that measures soil reflectance at numerous wavelengths over the range from 1.6 to 3.0 pm. The device uses a rotating circular variable filter to modify radiation from a broadband near-infrared (near-ir) source, and a lead sulfide (PbS) photodetector measures the light reflected from the soil sample. Readings are collected and analyzed with a computer. The researchers found that accurate results could be obtained from as few as 12 spectral bands. Because this spectrophotometer, with its matching data-analysis algorithm, works with more data than a single-band instrument, it can separate information on soil organic matter, soil moisture, and, to some extent, soil clay content (see Fig. 1). It does not require recalibration for the range of soils tested, and the data-analysis FIGURE 2. Thermal IR sensors measure temperature of potato leaves. Rising temperatures indicate a site-specific need for crop irrigation. program takes only about 10 s to process one soil sample. So far, this device has been used primarily in the laboratory, although a field unit has been constructed. The near-ir spectrophotometer.would be more expensive than the single-band sensor but could be used on a wider range of soils without calibration. Ammonia sensing Nitrogen is a major plant nutrient and is often the most expensive chemical input to crop production. The application of high levels of nitrogen fertilizer to cropland is believed to be responsible for nitrates detected in groundwater. For economic and environmental reasons, agricultural scientists and engineers are often interested in the fate of soil nitrogen in its various forms. Nitrogen can be applied to cropland as ammonia (NH3), and, under some 80 CIRCLE NO. 72

3 LIDAR Solutions from the Compact Laser Innovator Don't settle for an off-the-shelf lab laser when a compact, rugged system designed to vour specifications may be available at a comparable price. I Environmentafly Hardened I Lightweight I Flight Qualified I Rep Rates to 200 HZ I Wavelengths from UVtoIR Call Us For Details SKY LASER, TECHNOLOGIES. INC. Your Source for Compact Laser Systems '.O. BOX 8100 SOZEMAN. MT CIRCLE NO. 74 PHONE FAX I l l WCH SO2 FT Q KlHWE'I' HE!, FIGURE 3. Aerial thermal IR view of a center-pivot-irrigated field shows stressed plants in the red area, indicating either plant disease or a sandy area that does not hold water well. conditions, existing soil nitrogen can be converted to gaseous ammonia. While ammonia released into the atmosphere in small amounts is not usually an environmental problem, it does represent a monetary loss. J. L. Walsh and researchers from the Georgia Institute of Technology (Atlanta, GA) and the University of Georgia (Griffin, GA) are developing a sensor to detect small amounts of ammonia lost from the soil into the air.3 The device uses a thin-film waveguide with an organic coating that is sensitive to the ammonia vapors. Absorption of ammonia molecules changes the refractive index of the coating. These changes are then measured either by attenuation of the guided light beam or by using interferometry to detect phase shifts in two modes of transmitted light. Detecting plant stress Increasing constraints on water usage are prompting the search for better ways to evaluate the amount and timing of irrigation water. Several vendors sell hand-held IR temperaturemeasuring devices for detecting water stress in crops: as the soil dries out, the rate of water uptake and transpiration by the plants decreases. Evaporation from the leaves cools the plants so their daytime temperature is higher when water is restricted. This temperature change can be detected by ground-based or aerial devices and can be used to determine the most efficient timing of the next irrigation (see Fig. 2). In a similar manner, IR temperature measurement can detect insect activity, which reduces plant vigor and, thus, the amount of water used, again increasing the daytime leaf temperature. Although IR temperature measurement cannot determine causes of plant stress, aerial thermal IR imagery can be useful to map the location and extent of the stress. Ground-based inspection can then identify the problem. Private firms in several parts of the USA offer combinations of aerial video imagery in the IR, near-ir, and visible wavelength bands for agricultural use. These data enable farmers to detect areas of disease, insect, or nutrient problems in cropland, or even locate a faulty nozzle in an irrigation system (see Fig. 3). Aerial imaging can be expensive due to the cost of the aircraft, imaging hardware, and the software and expertise to interpret data. It can become cost-effective, however, if a large area of cropland is involved. Finding weeds In some areas where there is not enough rainfall for annual crop production, the ground is left fallow for one year. During the fallow year, the soil builds up moisture so that two years of moisture are used to produce one crop. Because weeds will con-

4 FIGURE 4. Growing conditions monitored by optical sensors in wheat field are found to be different from one end to the other. Spatially variable farming requires that chemicai input and irrigation be varied to fit the needs of each part of the field. sume water, they must be prevented from growing on the fallow ground. Typically, weeds are killed by occasional tilling or herbicide spraying. This process can be inefficient because the entire field must be treated even though weeds grow nonuniformly. Several devices have been proposed to detect individual weeds or weedy areas and activate herbicide spraying only for those areas. This reduces the amount of chemicals used, with obvious environmental and economic benefits. In their original device, D. E. Haggar and others at the Agricultural Research Council (Weed Research Organization, Yarnton, England) used two photoresistors and optical filters to measure the ratio of reflected red to near-ir light from the ground at a range of about 0.5 m.'. Living plants absorb red light but strongly reflect near-ir light. Other objects and substances found on typical cropland tend to reflect these bands more uniformly. One of us (Shropshire) used this instrument to detect plants rowing between rows of soybeans! The performance was characterized and determined to be susceptible to noise because of variation in the ambient CIRCLE NO

5 ~ YES-369 V lighting level. K. Von Bargen and his colleagues at the University of Nebraska (Omaha, NE) improved the device by adding a third detector to measure and compensate for the ambient light level6 A similar commercial device, Detectspray, is being marketed in Australia, Canada, and the USA by Concord Inc. (Fargo, ND).' The development of sensors is a high priority for agriculture. Economic, environmental, and regulatory issues are putting pressure on farmers to manage their croplands more carefully by supplying fertilizers, herbicides, pesticides, and water at carefully controlled and variable rates (see Fig. 4). To accomplish this, sensors will need to provide input to partially automated machines. Thus, we expect increased use of optical sensors in future agricultural activities as we learn more precisely how to harmonize farming activities with the world economy and environment. 0 CIRCLE NO. 76 REFERENCES 1.1. L. Shonk et al., Trans. Am. SOC. of Agricultural Eng. 43(5), 1978 (1991). 2.K. A. Sudduth and J. W. Hummei, to be included in Proc. SPlE lnt'l Conf. 7836, OEl Technology '92: Optics in Agriculture and Forestry, Boston, MA (Nov. 1992). 3. J. L. Walsh, F. C. Boswell, and D. P. Campbell, Am. SOC. Agricultural Eng. 90, 1632 (1990). 4.D. E. Haggar, C. J. Stent, and S. Issac, 1. Agricultural Eng. Res. 28, 349 (1983). 5.G. J. Shropshire, Weed detection in row crops using the red near-ir reflectance ratio and frequency transforms of video images," doctoral dissertation, Univ. of Nebraska, Lincoln, NE (1989). 6. K. Von Bargen et al., to be included in Proc. SflE lnt'l Conf. 7836, OElTechnology '92: Optics in Agriculture and Forestry, Boston, MA (Nov. 1992). 7. W. L. Felton and K. R. McCloy, Agricultural Eng., 9 (Nov. 1992). 84 Odics for Research PRECISION OPTICAL PRODUCTS Bulk reprints of all LFWorld articles can be ordered from June Bozartk, reprints manager, Pennwell Publishing Co., at (800) or (918) YOUR RATING, PLEASE... Is this article of value to you? Please circle appropriate number on the Reader Service Card. NO-370