Dr. Dave Franzen
Extension Soil Specialist
North Dakota State University

INTRODUCTION

Precision farming can be defined as a farming method that uses on-farm data to improve the accuracy of inputs to each field. Under this definition, starting to soil test for the first time and improving monitoring of planting rates or combine settings during the field operations may be included as precision farming. But given that many producers are already using these and other techniques to manage each field, there are now new methods to improve information gathering and input management within fields. These new methods are based on being able to give a reproducible and recordable position, or address, to field measurements and associated management decisions.

Positions of individual field measurements are possible by means of a system of United States Department of Defense satellites known as the Global Positioning System (GPS) . These twenty-four GPS satellites emit signals, which when received by a GPS receiver, can be triangulated automatically within the receiver, giving the user the longitude and latitude of the location within about 100 meters (300 ft.). The 100 meter unreliability of the signal is a result of the random error built into the signal by the Department of Defense. However, using a correction signal from a stationary tower, the US Coast Guard, commercial satellite or FM radio, correction for randomness can be conducted within the GPS receiver. The correction signal is called the Differential GPS signal (DGPS) . If the user is near a navigable waterway within the range of a Coast Guard differential signal, this DGPS signal is free. Other options may be purchased from a commercial DGPS differential signal provider for $75 to $600/yr (US) depending on the accuracy of differential needed. A stationary tower may also be used to placed a differential signal by the user, but this option may be expensive. Coarse resolution of about 3 meters (10 ft.) is usually chosen by users in agriculture, except for planter markers, and other equipment guidance systems, which need less than 0.3 meter (1 ft.) accuracy.

Position is important, because it can be linked to a computer which automatically records each measurement. Since GPS information is in a global gridding system used for hundreds of years, the latitude and longitude system, the measurements can be linked to a Geographic Information System (015) program. 015 lets the user compare different kinds of information, called layers, related by common position. Literally hundreds of layers of information can be compared using a GIS package, although the comparison of three to four layers are usually enough to keep anyone busy presently.

Researchers have demonstrated what producers have sensed for years; that nearly every field growth factor is variable across the field. Soil series, soil texture, soil water, landscape position, nitrogen, organic matter, phosphate, potassium, soil pH, yield, grain quality, micronutrients, weed populations, insect levels, and disease incidence, are all variable. The major questions raised by this knowledge include:

1. What kinds of information can be gathered and how might that information be used?

2. How might the field be sampled practically to gather information?

3. How reliable are estimates from sampling patterns?

4. What are the economic returns from gathering information and changing management within the field?

GATHERING AND USING INFORMATION

Yield monitoring

A yield monitor is often the first step that producers take into precision farming. Costing from $6,000 to $10,000 (US), combine monitors are relatively simple to install and can provide yield and moisture measurements every second if desired when the combine is operating. Beginning producers sometimes think that yield monitors without GPS and a recording computer would be useful, but their use is mostly to give a running display of yield across the field. This information can seldom be used to manage within field variability. If a producer needs to address within field issues, then DOPS and a recording computer with data card are essential.

Yield monitor packages should include a mapping system, so that the user can see and analyze the field data after harvest is completed. It would be a good idea to work with a consultant to help interpret data, especially if a producer is not very proficient with a computer. Files produced from one data card are huge, so do not ask for a data printout. Just ask for the map.

Once a yield map is produced, many producers will be surprised at first by the wide variability in most fields. One producer cooperator in the North Dakota precision farming project found that barley yields varied from 0 bushels to 140 bu/Acre in one field. Questions will arise regarding the origin of variability. If field visits have not been made during the season, it may not be possible to begin to answer these questions in one year of work.

Some producers believe that a yield map might be a guide to soil testing. In some situations this may be helpful. In the Mandan, ND map in Figure 1, sunflower yields in the north-central part of the map were useful in displaying a drainage problem not evident in the topography map in Figure 2. By including this area in the topographic sampling data set, a better picture of nitrate-N levels in the field was possible. However, producers may also find that because of wide differences in soil moisture holding capacity, presence of salts, weeds, disease and insect levels not necessarily associated with soil nutrient levels, yield monitor data should not stand alone as a soil sampling directive, but should be a supplement to sample results.

Most yield monitors are equipped with a computer that allows records of field observations to be entered into the computer memory along with the automatic position record of the observation. Some producers are currently manually entering weed sightings through their yield monitors as they operate the combine through the field. These entrees can be reproduced in separate maps, which may then be modified and used to direct the application of soil applied herbicides later that fall or next spring.

Remote sensing

The area representing the finest definition by a satellite image is called a pixel. The area represented by satellite data ranges from one pixel per 160 acres to one pixel representing about a 100 square ft. The greater the definition, the higher the expense. There are companies that can commercially make satellite images available to producers. Satellite images will improve in definition and frequency of images during the next few years. Satellite images may be helpful in defining differences between good growth and poor growth during the growing season. Research at NDSU has shown that sugarbeet leaf color is associated with soil nitrate-N levels. Sugarbeet leaf color patterns in the field near harvest may help define boundaries of nitrate-N levels, which may then help modify fertilizer N needs for the following crop.

Aerial photography may also be useful, especially from new digital cameras. Digital cameras may be developed directly into a computer software program. Digital pictures would be expected to have more resolution than computer scanned photographs from ordinary cameras. Canopy cover and organic matter levels (if no salt or carbonate is at the surface) may be estimated using the camera. Aerial photographs can also be taken when satellite images are obscured by clouds.

Satellite images and aerial photography must be ground truthed to have any real use. However, these tools might make it easier to scout a field if the problem areas were already identified.

Soil Sampling

Soil sampling was one of the first uses of position technology in agriculture. This was because variable-rate fertilizer application was one of the first control technologies available for producers to begin managing variability. There have been successful variable-rate fertilizer programs operated by retail fertilizer companies since 19B9 Other companies before 1989 had tried the system, including one in North Dakota, but were not successful because of computer hardware, software and equipment problems. Currently, there are very few problems with variable rate equipment. The cost of setting up a system ranges from expensive systems designed to vary the rate of multiple products, to less expensive systems customized to vary the rate of one input.

Soil samples are currently collected using grids (equally spaced soil sample locations) , soil type, topography, and remote sensing. It is likely that in individually fields, there are situations when any of these would be appropriate. It is also likely that within a single field only one would be most useful. Soil sample collection and analysis costs are the greatest expense in variable-rate nutrient management.

Sensors

Yield monitors are an example of a sensor. Yields are not really weighed in the combine, but usually a force against a plate is calibrated with weight over distance traveled. Organic matter sensors are being developed based on light reflectance from different wavelengths. Organic matter sensors may be useful in providing detailed organic matter levels for use in modifying soil applied herbicides or directing soil sampling sites. However, soils with salt or carbonate levels are much less likely to give favorable results, since in an organic matter sensor, colors are assumed to be related only to the modifying effects of organic matter, and not the white to cream color of salts and soil minerals.

For many years, sensors for use in analyzing soil tests on-the-go have been tried. One sensor is currently commercially available for use in recognizing soil nitrate-N levels, however, it also responds to soil chloride and sulfate levels. If chloride and sulfate levels were uniform across the field, this sensor might have some use. However, North Dakota studies show that sulfate and chloride are as variable as nitrate, and may not vary in the same places. Also, fertilizer-N levels are based on two foot cores, while this nitrate sensor only analyzes a small segment of that core near the surface. Other sensors with multiple nutrients are being developed, but the one most promising only uses a small segment of the surface as well. Unfortunately, the soil is variable not only horizontally, but vertically. Another sensor uses a membrane that selects for nitrate with the exclusion of chloride and sulfate. However, it currently works only in soilless solutions. It is not known if soil particles would plug the membrane, nor is it known how the top 2 ft. of soil would be extracted on-the-go.

One sensor with a wide range of uses is the EM-38. Demonstrated years ago by staff at North Dakota State University and others, the EM-38 is a device that looks like a carpenters level and measures electrical conductivity from one end of the device to the other across the soil surface and down into the soil profile It can be calibrated in North Dakota soils to measure salt content In areas without salt, such as in Missouri, the EM-3B has been used to measure depth to a root limiting layer. The measurements can be coupled to GPS and measurements can be recorded on-the-go with a computer.