D.W. Franzen, V.L. Hofman and L.J. Cihacek


The basis of grid sampling is the premise that soil fertility patterns are developed in a random manner and that a systematic method of sampling is required to reduce sampling bias. Early references to grid sampling were developed to reveal patterns for a variable-rate fertilizer application (Linsley and Bauer, 1929), while later systematic sampling was intended to allow the determination of a fields' central tendency (Peck and Melsted, 1973) without the bias of sampler interpretation. There still is the tendency for researchers to believe that sampling bias is undesirable, choosing instead to recommend dense soil sample grids to reveal patterns (Wollenhaupt et al., 1994; Franzen and Peck, 1995). However, there is sufficient literature to suggest that there are some logical reasons why certain areas of a field may test at one fertility level and other areas test at a different level. The effect of topography on soil fertility levels is one logical reason to question the need to grid sample all fields for certain nutrients.

Topography, or landscape, is based on elevation measurements, but the result of elevation mapping is depiction of a surface, classified into landscape features that were named upland, pediment backslope, pediment footslope and alluvial toeslope by Ruhe (1960). Ruhe described that there are developmental differences in soil profile A horizon depth and clay content at each landscape zone. Troeh (1964) found that the curvature of slope (convex vs. concave) influenced soil drainage, which influences the flow of water at the surface and within the soil, and the subsequent growth of crops. Zaslavsky and Rogowski (1969) showed that soil water infiltration streamlines which govern water movement diverge on convex hillslopes and converge in concave areas. The higher water content in the concave areas lead to more pronounced profile development.

Sinai et al. (1981) recorded a linear correlation between soil surface curvature and soil water content, which influenced wheat yield. Stone et al. (1985) showed that corn grain yields were more consistently related to landscape position than to erosion class. Jones et al. (1989) related yield to landscape position, redefining Ruhe's landscape postions as upper interfluve, lower interfluve, shoulder, upper linear, lower linear and the footslope. Yield was related to position in corn, soybeans and sorghum in the following manner-

Lower interfluve>Foot>Upper interfluve>Shoulder>Upper linear>Lower linear

Additional studies have shown the relationship of crop yields and slope position (Aspinall and Hayes, 1995; Fiez et al., 1994; Halvorson and Doll, 1991; Miller et al., 1988; Miller et al., 1992; Schroeder, 1995; Simmons et al., 1989) . Many of these studies have noted the influence of soil water content as a probable reason for crop yield relationship to topography. Pennock et al. (1987) illustrated probable water movement and concentration associated with topography.

Several studies have recorded the relationship of soil physical and fertility factors with topography (Brubaker et al., 1993; Fiez et al., 1994; Jones et al., 1989; Miller et al., 1988; Simmons et al., 1989). Additional studies have described the relationship of soil nitrogen specifically with topography (Bruulsema et al., 1996; Cassel et al., 1996).

Residual soil nitrogen may be especially influenced by landscape due to the number of kinds of possible transformations and its movement in the soil. Landscape and associated soil water influences mineralization (Stevenson et al., 1995), nitrification, and denitrification (Stevenson, 1982). Soil water movement also directs the movement of soil nitrate to certain areas of the landscape. In North Dakota, saline areas, which develop due to a locally high water table as a consequence of subsurface water movement, often have a high level of nitrate (Seelig and Richardson, 1991). Differences in crop removal of nutrients due to landscape induced crop yield variability may also influence residual nitrogen and other nutrient levels.

The concept of directed sampling by landscape position is not without precedent. Carr et al. (1991) reported improvements in economic return to wheat by fertilizing based on soil type sampling compared to uniform application. Penney et al. (1996) describes the value of using a combination of crop yield maps with topography. Pocknee (1996) discusses methods of directed soil sampling. In North Dakota, a method of composite soil sampling for uniform fertilizer application used a method of reverse-bias soil sampling by not sampling unusual areas such as eroded areas, saline areas and other areas not representative of the greatest part of the field (Swenson et al. 1984). Since these unusual areas are important for site-specific management, could the same reasons why parts of a field were not sampled in the past now be reasons to define certain management zones?

Landscape may be a logical reason to define initial management zones. However, measuring elevations and determining where to draw zone boundaries from landscape may be difficult to conduct. Even if elevation is recorded and landscape maps produced, the question for some may be where to draw the lines for boundaries defining management zones. North Dakota studies are currently investigating several methods for directing zone sampling, including satellite imagery and aerial photography (Moraghan, 1996). This paper will address initial experiences with satellite imagery, yield monitor information and electrical conductivity as methods that may provide patterns similar to landscape patterns. Some or all of these techniques may be useful in defining management zones, with the knowledge that the patterns expressed have a logical basis, in part due to landscape.


Four forty-acre sites were sampled in a 110 foot grid. One of the sites is located near Valley City, North Dakota in the glacial till plain region. The Valley City site was sampled each fall from 1994 to 1997. The field was in spring wheat in 1994, sunflower in 1995, spring wheat in 1996 and barley in 1997. A uniform rate of N and P was applied each year based on a composite soil test, except in 1997 when a variable-rate of N as anhydrous ammonia was applied prior to the barley crop. The Valley City site was included in this sugarbeet oriented paper even though it is not in the Valley because it may be useful to note that at other sites the relationship of nutrients follow similar patterns as those in the Valley studies, giving more confidence to our conclusions to date.

The second field is located southwest of Gardner, ND. This field was divided into a north 15 acre that was in third year alfalfa in 1994 when the study began, experienced winter kill during 1994-95, but remained in declining alfalfa until after the 1996 sampling. The south 25 acres was in spring wheat in 1994, barley in 1995 and seeded to alfalfa in the spring of 1996. The Gardner site sampling study was terminated after the 1996 sampling.

The third site is located southwest of Colfax, North Dakota in the Red River Valley. The Colfax site was sampled each fall during 1995 and 1996. The Colfax site was in corn in 1995 and spring wheat in 1996, as part of a sugarbeet rotation.

The fourth site is located about four miles west of the Gardner site, east of Hunter, ND. The Hunter site consists of Bearden and associated soil types, and the field is in a sugarbeet rotation with spring wheat grown in 1997.

At each site, five to eight soil cores were taken at each sample location at two depths, 0-6 inches and 6-24 inches. Nitrate-N and P were analyzed on the 0-6 inch depth. Nitrate-N was also analyzed on the 6-24 inch depth. Nitrate-N was reported as the total of two depths. Elevation was measured using a laser-surveying device, except at the Hunter site because of rough surface conditions in the fall of 1997. Elevation measurements were taken in a 110 foot grid. At Hunter, elevation will be measured in the spring of 1998. In this paper, relationships with topography are estimated based on the assumption that organic matter is related to topography.

The SPOT image of the Colfax site was taken late in the 1995 corn growing season prior to the first soil sampling. Pixel size is approximately 30 feet square. The electrical conductivity mapping at Valley City was conducted using a Veris Corporation soil conductivity instrument and an EM-38 (Geonics, Ltd., Missisauga, ON). The Veris apparatus consists of conductivity sensors attached to straight-set tillage discs that run in the soil at a depth of about 1 inch while being pulled through the field with a pickup truck. The sensor is coupled to a DGPS receiver and the data recorded in readings taken about 1 second apart. The EM-38 measurements were made in a 50 foot grid. Mapping of elevation, nutrient levels and conductivity was conducted using Surfer for Windows (Golden Software, Inc., Golden CO). Inverse distance squared was the interpolation method utilized with eight nearest neighbors and a simple search. Correlation was conducted using SYSTAT for Windows, Evanston, IL.


Relationship of nutrient levels with topography at Valley City

The nitrate-N maps from 1994, 1995 and 1996 at Valley City form complex but similar patterns between years (Figure 1). The stability of these nitrate-N patterns over years and rotational crops formed the reason for the researchers to explore topography as the source of pattern stability. Elevation measurement produced the map shown in Figure 2. The field varied from one side of the field to the other by about 40 feet. The simple contour map in Figure 3 was not adequate to define landscape, compared to the 3-dimensional map in Figure 2. Elevation itself is not related to nutrient levels in this field. The structure of the landscape-hilltop, sideslope, concave areas- appear to be the defining elements in the nutrient/landscape relationships. The correlation of topography-based sampling compared to a 220 foot grid is shown in Table 1. The topography based map for 1995 compared to the 110 foot and 220 foot grids is shown in Figure 4. In 1994 and 1995, comparisons between topography were made with a topography map based on five sampling zones. These zones were chosen based on the zones of similar topography patterns from the 3-dimensional mapping.

P levels at Valley City were not as correlated with topography-based sampling as nitrate-N (Table 1). The farm cooperator stated that forty years ago, their was a small feedlot in the northwest corner of the area currently being examined. The old feedlot area may be masking landscape affects on P levels in this portion of the field. The map of Valley City P levels is given in Figure 5. Outside of the old feedlot area, the patterns of high and low P follow closely the patterns of nitrate-N.

Alternative methods of management zone definition at Valley City

In 1996, the first yield map at Valley City was produced using a John Deere yield monitor. The yield map (Figure 6) showed a very low yielding area in the northwest corner of the field. By separating out this area as a separate management zone, the 1996 correlation of topography N with the 110 foot grid was better correlated than in 1995 (Table 1). Because of the many factors producing yield differences, using a yield map as the primary basis for directed sampling may not be prudent. However, by using the landscape zones as the main basis for establishing sampling zones, and then dividing a zone into two zones based on an area of low yield in the northwest corner, correlation with zone sampling with the 110 foot nitrate-N map is increased.

In the fall of 1997, an on-the-go soil conductivity sensor was used to map the Valley City field for salinity. The resulting map of the 0-1 foot soil conductivity is shown in Figure 7. The stated purpose of the conductivity sensor by the manufacturer is not to determine nitrate levels, other nutrient levels, depth to clay, depth to sand, clay content or moisture. The purpose is to detect zones of similar conductivity. The user is then encouraged to sample within those areas of similar conductivity and determine what is the source of the differences between zones.

The patterns of soil conductivity are very similar to the patterns of nitrate-N. However, correlation of soil conductivity with soil nitrate-N levels may not be appropriate. The high conductivity in the northwest is similar to the high nitrate-N levels in the same area. However, in the "horsehead-shaped" pattern of nitrate-N in the center of the field, nitrate-N levels are relatively higher than its surroundings. Soil conductivity values in the same area has the "horsehead-shaped" pattern, but the values are relatively low compared to the surroundings. A simple correlation of conductivity with nitrate-N levels would result in an overestimation of nitrate-N levels in that area. By using the patterns of conductivity differences as a way to define management zones, then sampling within those zones, conductivity zone sampling would be expected to give similar boundaries and similar correlation as topography sampling. A similar conductivity map was produced using a EM-38 conductivity detector (Geonics, Ltd., Missisauga, ON, Canada), with the assistance of NRCS personnel Norm Procnow and Hal Weiser. The map produced with the EM-38 showed similar patterns with those of the Veris instrument.

Influence of previous crop on spatial variability and topography relationships at Gardner

In 1994, the south field after spring wheat was spatially variable in soil nitrate-N, but the north field of vigorous alfalfa was not. After the winter kill of alfalfa in 1994-95, the nitrate-N levels in 1995 of both the north and south fields were spatially variable, and the relationship of nitrate-N levels with topography is shown in Table 2. In 1996, after further decline of the north alfalfa stand, continued wet soil conditions and after spring seeding and growth of a vigorous alfalfa stand in the south field, spatial variability of nitrate-N was not present and the relationship with topography was not evident. The Gardner study showed that previous crop was important, and suggests that soil nitrate-N levels immediately following a vigorous stand of alfalfa would be effectively represented by a composite soil test rather than gridding or directed sampling. Patterns of P were stable during the study, although the mean values were influenced by cropping.

Relationship of topography and nutrient levels at Colfax

Nitrate-N levels Colfax in 1995 and 1996 are shown in Figure 8. The maps are similar between years, suggesting again that the nitrate-N levels are related to landscape. The elevation mapping, shown in the contour map in Figure 9, shows that the patterns of nitrate-N follow landscape patterns. In this field, where the elevation differences are only about 2 feet, a contour map is sufficient to define fertility boundaries. Correlation of nitrate-N levels with the 110 foot and 220 foot grids are shown in Table 2.

The SPOT image of the Colfax site in 1994 is shown in Figure 10. Using the map to define management zones based on patterns in the image result in patterns similar to those defined by topography. In 1994, the poor corn growth areas were the result of excessive rainfall during the season, which stunted the corn and probably lowered the level of nitrate in those areas through denitrification. The areas of better corn growth were at higher elevations where there was no standing water. By identifying areas of poor corn growth due to water stress, the image also identified areas of low and high N.

Initial variability of soil nutrients at Hunter and the relationship of soil nutrients with topography estimated by organic matter levels

Organic matter levels at Hunter are shown in Figure 11. The field is represented by a low area along the road in the east, with patterns of soil that suggest areas of high salt due to a road ditch affect. The northeast corner of the field is the field entry culvert, which is the area where sugarbeet tare piles are typically dumped. From the northeast corner southwest to the south-center of the field is a relatively low area. West of this line are two higher elevation areas which are a little coarser in texture than the soils to the east. Higher nitrate-N levels tended to follow the path of the lower elevation from northeast to southwest. Some localized depressions tended to be low in N, perhaps due to greater crop production or denitrification losses (Figure 12). The relationship of nitrate-N, P, and sulfate with topography (estimated by organic matter levels) and gridding is shown in Table 4.

Table 1 Figure 1 Figure 5 Figure 9
Table 2 Figure 2 Figure 6 Figure 10
Table 3 Figure 3 Figure 7 Figure 11
Table 4 Figure 4 Figure 8 Figure 12


Soil nitrate-N levels were related to topography at four sites in North Dakota, depending on the previous crop. Soil P levels were not as related to topography perhaps because of the past existence of a feedlot in a portion of the Valley City, and the use of high P buildup rates in the Valley, generally. Using a yield map to identify low yielding areas within a zone may be a reason to divide one zone into two zones and strengthen a management zone strategy in a field. Soil conductivity measurements also defined zones similar to topography. Satellite imagery was related to topography patterns at Colfax. Landscape may serve as a logical basis for expecting the presence of nutrient patterns, especially nitrate-N, to appear in fields. However, other methods, such as satellite imagery, aerial photography and satellite imagery may help to determine where to draw sampling zone boundaries. Use of a zone sampling strategy should not rely exclusively on elevation mapping as its only method of zone definition. Zone sampling should be considered an iterative process, where additional sources of information about the field are used to further define areas of importance to crop growth, including more intensive grid sampling when appropriate. Zone sampling may be most useful in regions where plant or soil testing for nitrogen requirement is justified and the cost of annual intensive grid sampling is prohibitive.


Aspinall, D. and A. Hayes, 1995. The relationship of corn and soybean yields, soil properties to slope positions and landform. pp. 1-5 In:Poster Paper Abstracts, Information Agriculture Conference. June 27-30, 1995. Champaign, IL. Potash & Phosphate Institute, Norcross, GA.

Brubaker, S.C., A.J. Johnes, D.T. Lewis, and K. Frank. 1993. Soil properties associated with landscape positions. Soil Sci. Soc. Am. J. 57:235-239.

Bruulsema, T.W., G.L. Malzer, P.C. Robert, J.G. Davis and P.J. Copeland. 1996. Spatial relationships of soil nitrogen with corn yield response to applied nitrogen. pp. 505-512. In:Precision Agriculture. Proceedings of the 3rd International Conference, June 23-26, 1996, Minneapolis, MN. ASA-CSSA-SSSA, Madison, WI.

Carr, P.M., G.R. Carlson, J.S. Jacobsen, G.A. Nielsen, and E.O. Skogley. 1991. Farming soils, not fields:A strategy for increasing fertilizer profitability. J. Prod. Agric. 4:57-61.

Cassel, D.K., E.J. Kamprath, and F.W. Simmons. 1996. Nitrogen-sulfur relationships in corn as affected by landscape attributes and tillage. Agron. J. 88:133-140.

Fiez, T.E., B.C. Miller, and W.L. Pan. 1994. Winter wheat yield and grain protein across varied landscape positions. Agron. J. 86:1026-1032.

Franzen, D.W. and T.R. Peck. 1995. Field soil sampling density for variable rate fertilization. J. Prod. Agric. 8:568-574.

Halvorson, G.A. and E.C. Doll. 1991. Topographic effects on spring wheat yields and water use. Soil Sci. Soc. Am. J. 55:1680-1685.

Jones, A.J., L.N. Mielke, C.A. Bartles, and C.A. Miller. 1989. Relationship of landscape position and properties to crop production. J. Soil Water Conserv. 44:328-332.

Linsley, C.M. and F.C. Bauer. 1929. Test your soil for acidity. Univ. of Il. Col. of Ag. and Ag. Exp. Sta. Circ. 346.

Miller, M.P., M.J. Singer, and D.R. Nielsen. 1988. Spatial variability of wheat yield and soil properties on complex hills. Soil Sci. Soc. Am. J. 52:1133-1141.

Miller, B.C., T. Fiez, and W.L. Pan. 1992. Impact of landscape variability on grain yield and quality. pp. 3-6. In:Precision farming variable cropland:An introduction to variable management within whole fields, divided slopes and field strips. Proceedings 10th Northwest Conserv. Farming Conf., Pullman, WA. R.J. Veseth and B.C. Miller, ed. Wash. St. Univ., Pullman, WA.

Moraghan, J., K. Horsager, L.Smith, and A. Sims. 1997. Canopy reflectance and GPS technology in relation to sugar production. pp. 121-133. In:Sugarbeet Research and Extension Reports. Vol. 27. N. Dak. St. Univ. Ext. Serv., Fargo, ND.

Peck, T.R. and S.W. Melsted. 1973. Field sampling for soil testing. pp. 67-75. In:Soil Testing and Plant Analysis. Second edition. SSSA, Madison, WI.

Penney, D.C., R.C. McKenzie, S.C. Nolan, and T.W. Goddard. 1996. Use of crop yield and soil landscape attribute maps for variable rate fertilization. pp. 126-140. In:1996 Great Plains Soil Fertility Conference Proceedings, Denver, CO, March 5-6, 1996. J. Havlin, ed. K.St. Univ., Manhattan, KS.

Pennock, D.J., B.J. Zebarth, and E. DeJong. 1987. Landform classification and soil distribution in hummocky terrain, Saskatchawan, Canada. Geoderma 40:297-315.

Pocknee, S., B.C. Boydell, H.M. Green, D.J. Waters, and C.K. Kvien. 1996. Directed soil sampling. pp. 159-168. In:Precision Agriculture. Proceedings of the 3rd International Conference. June 23-26, 1996. Minneapolis, MN. ASA-CSSA-SSSA, Madison, WI.

Ruhe, R.V. 1960. Elements of the soil landscape. pp. 165-170. In:Transactions of the 7th International Congress of Soil Science. Vol. 4. Int. Soc. of Soil Science. Madison, WI.

Schroeder, S.A. 1995. Topographic influences on soil water and spring wheat yields on reclaimed mineland. Journ. Env. Qual. 467-471.

Seelig, B.D. and J.L. Richardson. 1991. Salinity and sodicity in North Dakota soils. N.Dak. Ext. Serv. Bull. EB-57. 16 pp.

Simmons, F.W., D.K. Cassel, and R.B. Daniels. 1989. Landscape and soil property effects in corn grain yield response to tillage. Soil Sci. Soc. Am. J. 53:534-539.

Sinai, G., D.Zaslavsky, and P. Golany. 1981. The effect of soil surface curvature on moisture and yield-Beer Sheba observations. Soil Sci. 132:367-375.

Stevenson, F.J. 1982. Origin and distribution of nitrogen in soil. pp. 1-42. In:Nitrogen in Agricultural Soils. Agonomy Monograph No. 22. F.J. Stevenson, ed. ASA-CSSA-SSSA, Madison, WI.

Stevenson, F.C., J.D. McKnight, and C. van Kessel. 1995. Dinitrogen fixation in pea:controls at the landscape- and micro-scale. Soil Sci. Soc. Am. J. 59:1603-1611.

Swenson, L.J., W.C. Dahnke, and D.D. Patterson. 1984. Sampling for soil testing. N. Dak. Dept of Soil Sci. Res. Report No. 8. April 1984.

Stone, J.R., J.W. Gilliam, D.K. Cassel, R.B. Daniels, L.A. Nelson, and H.J. Kleiss. 1985. Effect of erosion and landscape position on the productivity of Piedmont soils. Soil Sci. Soc. Am. J. 49:987-991.

Troeh, F.R. 1964. Landform parameters correlated to soil drainage. Soil Sci. Soc. Am. Proc. 28:808-812.

Wollenhaupt, N.C., R.P. Wolkowski, and M.K. Clayton. 1994. Mapping soil test phosphorus and potassium for variable-rate fertilizer application. J. Prod. Agric. 7:441- 448.

Zaslavsky, D. and A.S. Rogowski. 1969. Hydrologic and morphologic implications of anisotropy and infiltration in soil profile development. Soil Sci. Soc. Am. Proc. 33:594- 599.



Funding for this project was provided through the US-EPA 319 Water Quality grant program, the Minnesota-North Dakota Sugarbeet Research and Education Board, Agrium, Inc., the Soil Conservation Districts of Stutsman Co., Cass Co., and Wild Rice, and the Phosphate & Potash Institute.

1997 Sugarbeet Research and Extension Reports. Volume 28, pages 127-142.