Section 4: Remote Sensing in Precision Agriculture

Definition of Precision Agriculture

"Precision Farming is the title given to a method of crop management by which areas of land/crop within a field may be managed with different levels of input depending upon the yield potential of the crop in that particular area of land. The benefits of so doing are two fold:

    1. the cost of producing the crop in that area can be reduced and,
    2. the risk of environmental pollution from agrochemicals applied at levels greater than those required by the crop can be reduced" (Earl et al, 1996).

Precision farming is an integrated agricultural management system incorporating several technologies. The technological tools often include the global positioning system, geographical information system, yield monitor, variable rate technology, and remote sensing.

The global positioning system ("GPS") is a network of satellites developed for and managed by the U.S. Defense Department. The GPS constellation of 24 satellites orbiting the earth, transmit precise satellite time and location information to ground receivers. The ground receiving units are able to receive this location information from several satellites at a time for use in calculating a triangulation fix thus determining the exact location of the receiver.

planetani4.gif (178825 bytes)

Figure 4.1 Global Positioning System


A geographical information system ("GIS") consists of a computer software data base system used to input, store, retrieve, analyze, and display, in map like form, spatially referenced geographical information.

gis-9.gif (106624 bytes)

Figure 4.2  Data Integrated Through a Geographical Information System

Yield monitors are crop yield measuring devices installed on harvesting equipment. The yield data from the monitor is recorded and stored at regular intervals along with positional data received from the GPS unit. GIS software takes the yield data and produces yield maps.

pf-sm2.jpg (19401 bytes) flow-jd.gif (7709 bytes)
Figure 4.3  Combine Yield Monitor Figure 4.4  Combine Grain Tank Flow Sensor

Variable rate technology ("VRT") consists of farm field equipment with the ability to precisely control the rate of application of crop inputs and tillage operations.

systems.jpg (39598 bytes)

Figure 4.5  VRT Spreader

Remote sensing image data from the soil and crops is processed and then added to the GIS database.

00033a7n.jpg (13987 bytes)

Figure 4.6 Normalized Vegetation Index Image of Farm Field


Goal of Precision Farming

The goal of precision farming is to gather and analyze information about the variability of soil and crop conditions in order to maximize the efficiency of crop inputs within small areas of the farm field. To meet this efficiency goal the variability within the field must be controllable.

Efficiency in the use of crop inputs means that fewer crop inputs such as fertilizer and chemicals will be used and placed where needed. The benefits from this efficiency will be both economical and environmental. Environmental costs are difficult to quantify in monetary terms. The reduction of soil and groundwater pollution from farming activities has a desirable benefit to the farmer and to society.



Precision farming is an integration of several technologies. U.S. and foreign governments originally paid for the development and support of technologies such as GPS, Remote Sensing, and GIS, for military or other civilian purposes long before the emergence of precision farming.

gpsslide.jpg (25057 bytes)

Figure 4.7 GPS Survey Equipment

It is the use of these advanced technologies that has generated enormous amounts of data to process with computers. A very basic question that still needs to be answered by researchers is what does all of this data mean in order for farmers to make profitable management decisions? In other words we have the technology to gather the data from the field but we don't yet have the knowledge to transform the data into answers for agricultural management decisions.

Universities around the world, and foreign and U.S. government entities such as the Agricultural Research Stations of the USDA are conducting extensive research on precision farming.  Research projects will apply technologies such as remote sensing, GPS, GIS, and VRT to create management decision support systems. A goal of many publicly funded research institutions is to promote technology transfer from government agencies to the private sector.


Application of Remote Sensing in Precision Agriculture

Soil and Drainage Maps

Management Zones and Soil Maps

Soil maps are also sometimes used to determine management zones. Soil maps are becoming part of the GIS database.

The grid sampling technique takes separate soil samples from uniform sized grids laid out over the field. A problem with this type of sampling is the variability that can exist in soil types with in each grid. This variability makes it much tougher to determine soil characteristics within the grid for crop input management purposes. To minimize this problem smaller grids are required which then requires many more soil samples to be take for a larger number of grids. Soil samples can become a major cost of precision farming.

An alternative to grid sampling is targeted or zone sampling. The soil samples are located in homogeneous management zones instead of uniformly spaced grids (Searcy, 1997). The zones are laid out using a process similar to computer based unsupervised image classification. Images obtained from multispectral remote sensors are taken of the vegetated areas of the field. The pixel digital numbers for each band are separated into statistically separable clusters that are classified into homogeneous zones. This cuts down on the soil, terrain, plant growth, and other variability within each area to be managed; thus fewer soil samples are needed for each area (Anderson et al, 1996).

Except for county soil surveys remote sensing has not gained wide acceptance as a mapping tool for soil characteristics.  This is because "the reflectance characteristics of the desired soil properties (e.g., organic matter, texture, iron content) are often confused by variability in soil moisture content, surface roughness, climate factors, solar zenith angle, and view angle"( Moran et al, 1997).

Drainage Maps

Subsurface drainage tile lines that have been installed, as long as 50 or more years ago are still partially or totally functional today. Often the existence or location of older tile lines has been lost as landowners die or sell their property. Some states, including Iowa (Iowa Code, 1997), are now starting to require landowners to prepare and file drainage plats with county recorders when new tile lines are installed. It is desirable to have accurate drain tile maps for maintenance purposes or for the installation of new additional tile line systems. Installation of new tile lines may cut through old tile lines at unknown locations. Building livestock manure lagoons, which cut through old unknown and uncharted tile lines may cause environmental damage from manure leaking through the old tile lines.

wpe1.gif (52256 bytes) wpe2.jpg (9688 bytes)
Figure 4.8  Normal Color Photo of Dry Soil Figure 4.9  Normal Color Photo of Soil After Rain

Color infrared ("CIR") aerial photographs have been shown to be an effective tool in locating unknown subsurface tile lines. The image data is digitized for preprocessing and then geo-referenced using ground control points. The CIR photographs show different tones of gray depending on soil type and moisture. By filtering out spectral reflectance differences due to soil type, soil moisture content in dry soils that have a higher reflectance can be identified from lower reflectance wet soils. The resulting image shows were the tile lines are located and whether they are working properly (Verma et al, 1997).

Normal color aerial photographs can also be used to locate tile lines.  Simple color photographs offer tile line images similar to CIR but at a lower cost.  If the soil is too dry such as that shown in Figure 4.8 the tile lines will not be visible in the image.  The images similar to Figure 4.9 must be acquired when the soil is bare and within a few days after an adequate rain.  High resolution and on demand temporal availability  make images acquired from aircraft platforms ideal for acquiring this kind of image data.

Variable Rate Technology

One method of controlling variability within the field is VRT. VRT allows the grower to apply the quantity of crop inputs needed at a precise location in the field based on the individual characteristics of that location. Crop inputs that can be varied in their application commonly include tillage, fertilizer, weed control, insect control, plant variety, plant population, and irrigation.

Typical VRT system components include a computer controller, GPS receiver, and GIS map database. The computer controller adjusts the equipment application rate of the crop input applied. The computer controller is integrated with the GIS database, which contains the flow rate instructions for the application equipment. A GPS receiver is linked to the computer. The computer controller uses the location coordinates from the GPS unit to find the equipment location on the map provided by the GIS unit. The computer controller reads the instructions from the GIS system and varies the rate of the crop input being applied as the equipment crosses the field. The computer controller will record the actual rates applied at each location in the field and store the information in the GIS system, thus maintaining precise field maps of materials applied.

Although VRT can control inputs applied to crops, it cannot control factors such as soil type, weather climate, and topography that are fixed.

Monitor Crop Health

Remote sensing data and images provide farmers with the ability to monitor the health and condition of crops. Multispectral remote sensing can detect reflected light that is not visible to the naked eye.  The chlorophyll in the plant leaf reflects green light while absorbing most of the blue and red lightwaves emitted from the sun.  Stressed plants reflect various wavelengths of light that are different from healthy plants. Healthy plants reflect more infrared energy from the spongy mesophyll plant leaf tissue than stressed plants.  By being able to detect areas of plant stress before its becomes visible, farmers will have additional time to analyze the problem area and apply a treatment.                                   

Water Stress

The use of remote sensors to directly measure soil moisture has had very limited success. Synthetic Aperture Radar ("SAR") sensors are sensitive to soil moisture and they have been used to directly measure soil moisture. SAR data requires extensive use of processing to remove surface induced noise such as soil surface roughness, vegetation, and topography.

A crop evapotranspiration rate decrease is an indicator of crop water stress or other crop problems such as plant disease or insect infestation. Remote sensing images have been combined with a crop water stress index ("CWSI") model to measure field variations (Moran et al, 1997).

Simple panchromatic aerial photographs have been used to spot irrigation equipment problems. Strips in the vegetation images point to problems with water application rates from defective water nozzles (Univ. of Georgia, 1995).

Weed Management

One goal of precision farming is to cut crop production inputs, which result in cost and environmental savings. Conventional farming methods apply herbicides to the entire field. Site-specific variable-rate application puts the herbicide where the weeds are.

Aerial remote sensing has not yet proved to be very useful in monitoring and locating dispersed weed populations. Some difficulties encountered are that weeds often will be dispersed throughout a crop that is spectrally similar, and very large-scale high resolution images will be needed for detection and identification (Ryerson, Curran, P. and Stephens 1997).

The use of machine vision technology systems to detect and identify weeds places remote sensors directly on the sprayer equipment. Being close to the crop allows for very high spatial resolutions. Machine vision systems have the ability to be used in the field with the real-time capabilities that are necessary to control sprayer equipment (Steward and Tian, 1998).     

Insect Detection

Aerial or satellite remote sensing has not been successfully used to identify and locate insects directly. Indirect detection of insects through the detection of plant stress has generally not been used in annual crops. The economic injury level for treatment is usually exceeded by the time plant stress is detected by remote sensing. Entomologists prefer to do direct in field scouting in order to detect insects in time for chemical treatments to be effective and economical.

Nutrient Stress

Plant nitrogen stress areas can be located in the field using high-resolution color infrared aerial images. The reflectance of near infrared, visible red and visible green wavelengths have a high correlation to the amount of applied nitrogen in the field. Canopy reflectance of red provides a good estimate of actual crop yields (GeopalaPillai, Tian, and Beal 1998).

Yield Forecasting

Plant tissue absorbs much of the red light band and is very reflective of energy in near infrared ("NIR") wavebands. The ratio of these two bands is referred to as the vegetation index ("VI"). The difference of red and NIR measurements divided by their sum is normalized difference VI ("NDVI").

For crops such as grain sorghum, production yields, leaf area index ("LAI"), crop height and biomass have been correlated with NDVI data obtained from multispectral images (Anderson et al, 1996). In order to get reasonably accurate yield predictions this data must be combined with input from weather models during the growing season (Moran et al, 1997).

Management Decision Support Systems

Just having information about variability within the field doesn't solve any problems unless there is some kind of decision support system ("DSS") in order to make VRT recommendations. Russo and Dantinne (Russo et al, 1997) have suggested the following steps for a DSS:

    1. Identify environmental and biological states and processes in the field that can be monitored and manipulated for the betterment of crop production.
    2. Choose sensors and supporting equipment to record data on these states and processes.
    3. Collect, store and communicate the field-recorded data.
    4. Process and manipulate the data into useful information and knowledge.
    5. Present the information and knowledge in a form that can be interpreted to make decisions.

Choose an action associated with a decision to change the identified state or process in a way that makes it more favorable to profitable crop production.

mis2.gif (80734 bytes)

Figure 4.11  Decision Support System Software


Future Prospects and Developments

Future satellite systems to be launched within the next year such as Ball Corporation's Quickbird will have a four-band multispectral pushbroom sensor with a resolution of .8 m panchromatic and 4.5 m multispectral. EarthWatch Incorporated of Longmont, Colorado will distribute Quickbird images.

Future satellites will have better spatial and spectral resolutions. Launching more satellites will also improve temporal resolution.

The delivery time of remote sensing data to the customer will improve. We will someday have real-time satellite remote sensing systems.

University research will concentrate more on the cause of soil and crop variability verses just being able to measure that variability.  A greater emphasis will be placed on technology transfer from universities to commercial agribusiness industry.

Decision support systems will become the main link to convert the spatial data collected into detailed management recommendations at the farmer level.  Decision support systems is what will add the most value to remote sensing data for the farmer.

The future of remote sensing in precision agriculture will depend upon meeting the needs of the end user, the farmer. Right now remote sensing  for agricultural use is still in an early stage of commercial development with unproven economic benefits to the farm producer.

The cost of remote sensing data and the other systems associated with precision agriculture will come down to be in line with the benefits received. This is likely to happen in the future as more agricultural information technology companies enter the marketplace.   

Top             Contents             Previous Page            Next Page