Soil-Sampling Issues for Precision Management of Crop Production

2

Figure 1. (A) Aerial photograph of 27 ha (67 ac) field six weeks after planting cotton.(B) management zones of field; and (C) 0.8 ha (2 ac) field grids.

A B C

take a GPS receiver (with differential correction) andlaptop computer, with GPS software, to the field andcollect field coordinates around the edge of the field. Theboundary file along with yield maps and aerial photos arethen used to develop management zones or field grids.Another method is to use aerial photos from an Internetwebsite:

http://www.terraserver.comMost fields in the United States have aerial photo-

graphs on this website that are easily found and can bedown-loaded for free, although they may be several yearsold (many photos over Georgia currently have a 1993date). These images have the geo-referenced coordinates(latitude and longitude) of the corners and can be used bysoftware packages to create a boundary around a field.Satellite images may also be purchased through Internetproviders such as Earthscan Network Inc.

www.earthscan.comThe images provided are usually taken every one to

two weeks and are available in a variety of formats forcreating field boundaries, analyzing vegetation growth,determining accurate field acreages, etc. This option iseasiest and does not require fieldwork but may create aless accurate boundary. To determine the best optionavailable, contact the software vendors.

Several software packages are available to create themanagement zones or field grids inside the field boun-dary. An aerial photograph is one of the most usefulpieces of information needed to create managementzones. An aerial photograph of our 27 ha (67 ac) examplefield approximately six weeks after the emergence ofcotton illustrates some of the history of the field, such asthe location of old fence lines and former pastureland,forest or buildings (Figure 1a). Using the aerial

photograph, the previous year’s yield map and historicaldata from the farmer, the field was divided into sevenmanagement zones of relatively homogeneous yields,similar history and/or soil types (Figure 1b). In this case,zone 3 was previously pasture and was recently (within 4years) converted to cropland. Zone 4 is a clay knoll thattradi-tionally has very low yields. Zone 1 is outside theend-gun on the center-pivot irrigation system and zone 7is a depressed wet-land. These zones take a little effort todraw out and may need “tweeking” the first couple ofyears but are representative of field variability based onsoils data, crop yields and field history.

Grids are usually easier to create and are generallymore widely supported by soil sampling and/or fertilizersuppliers and applicators. The number of grids dependson the size of the field and the size of each grid square,rectangle or other shape. However, grids are createdusing arbitrary boundaries that may have variable soilproperties within a grid. Consequently, a variable rateapplication of fertilizer or lime may still not account forfield variability with any improved precision (Rains andThomas, 2000). A software package was used to dividethe example field into 0.4-0.8 ha (1-2 ac) rectangles(Figure 1c). Soil samples can be collected at the center ofeach grid or at the corners where the grid areas meet.Some software packages that are capable of creating afield grid and management zones are:

Agris’s Aglink/Fieldlink, - 800-795-7995 -www.agris.com

RedHens FarmHMS/FarmGPS - 800-237-4182 -www.farmsoft.com

Farmworks FarmSite and FarmSite Mate - 219-488-3492 www.farmworks.com

Overall, it is recommended to use management zones

3

Figure 2. Lime application map for example field.Rates are based on soil test results in each zone.

to determine fertilizer and lime application rates. It ismore difficult to develop the zones, but they usuallyrequire less soil sampling than field grids and can moreaccurately represent the variability in the field. Figure 2illustrates the example field with variable rates of limeprescribed for each zone based on the soil sample testresults and the producer’s yield goal. In this case, oneyield goal was used for the entire field.

Collecting Soil Sample CoresAccurate representation of the soil characteristics is

extremely important to determining the recommendedapplication rates. Recommendations for soil samplingdepth are provided in the Soil Test Handbook for Georgia(Plank, 1989) Recommended practices are also availablein UGA Leaflet No. 99;

www.ces.uga.edu/pubcd/L99.htmFollow these practices and guidelines for each of the

soil samples collected for precision soil management. Inour study, it was shown that an inaccurate soil depthsignificantly changed the measured soil properties (Rainset al., 2001). The exact inaccuracies were dependent onwhich soil properties were analyzed. Phosphorous, phos-phate and lime recommendations were the most affectedby the depth of sample (test results of a 15 cm [6 in] soilsample core were statistically different from a 7.5 cm [3in] sample).

Although this study was a unique case for one field, itis expected that sample depth will affect the soil-testresults and recommendations for fertilizer and lime in allfields. The level it affects a field will depend on the soiltypes, topography, tillage practices, crops planted andwater availability. It is difficult to determine how accu-rate the sampling must be to avoid altering the soil testresults. In some cases, the soils may be homogeneousover a fairly large depth, while in other cases, soilstratification could largely affect results. A recommen-dation to sample within ±1 inch of the desired sampledepth is recommended here to reduce inaccuracies indesired application rate recommendations.

A GPS receiver and portable computer (laptop orhandheld) loaded with one of the software packageslisted earlier can provide a method for navigating to soilsample sites. Take samples in management zones fromtwo to three locations within that zone and mixed intoone sample.

Soil Test Laboratory Choice Lab-to-lab variability can affect soil-test sample

results as well as the recommended application rates.

When comparing soil samples that were split and sent totwo separate laboratories, results between the labs weresignificantly different for pH, phosphorous and potas-sium levels in the soil, even though both laboratoriesused the same soil test procedures (mehlich-1 test for Pand K, and pH meter) and analytical equipment (Rains etal., 2001). Recommended application rates from the twolaboratories were also different. One laboratory used theUniversity of Georgia crop-response functions and theother used “in-house” developed functions to makeapplication recommendations. Consequently, it is best toremain with one lab for soil analyses and application raterecommendations. This is especially true when theproducer is monitoring year-to-year changes in soilcharacteristics and yield responses. If two laboratoriesare used, find out from each laboratory what basis wasused to make application rate recommendations beforemaking any management decisions. If they do not use thesame crop response functions and soil testing procedures,the two lab results and amendment recommendations arenot comparable.

Some laboratories will create application maps if theyare given a computer file with the field boundary,management zones or grids and the location of the soilsample points. The laboratory will create application raterecommendations in each zone based on the soil test

The University of Georgia and Ft. Valley State University, the U.S. Department of Agriculture and counties of the state cooperating. Cooperative Extension, the University of Georgia College of Agricultural and Environmental Sciences, offers educational programs, assistance and materials to all people without regard to race, color, national origin, age, gender or disability.

An Equal Opportunity Employer/Affirmative Action Organization Committed to a Diverse Work Force

Bulletin 1208 Reviewed May, 2009