Previous work by Franzen and Peck (1995) in Illinois concluded that a one sample per acre grid is needed to begin a variable-rate program for P, K and pH. This work supported a similar finding in Wisconsin (Wollenhaupt, et al., 1994). However, soil sampling expenses for nitrate-N are much higher than for P, K and pH because of sampling depth, the need for more expensive sampling equipment, and greater time needed to gather each sample. A sampling study was initiated to look at the variability of soil nutrients and what sampling method would represent field nutrient levels at the least density.

Four sites have been sampled for two or three years in a dense grid. The Gardner site is a square, 40 acre, heavy textured soil in the Red River Valley. The Colfax site is a square, 40 acre, sandy loam to clay loam textured Red River Valley site. The Valley City site is a square, 40 acre, sandy loam to clay loam textured site in the central drift prairie. The Mandan site is a rectangular, 80 acre, loam to silt loam textured field representative of the western part of the state in climate and soils. The Mandan site is on a farm operated by the USDA-ARS Northern Great Plains Research Laboratory in Mandan, ND., with Dr. Ardell Halvorson, director. The Mandan site is a cooperative effort between NDSU extension and USDA-ARS.

The Gardner, Colfax, Valley City sites and the east 30 acres of the Mandan site were sampled on a 110 ft. grid. The west 50 acres of Mandan was sampled in a 150 ft. grid. All samples at all sites were sampled to two feet, separating the samples into a 0-6 inch and 6-24 inch core. Three to eight sample cores represented each sample location. At the Gardner and Colfax locations, an additional core depth from 24-48 inches was also taken. Grid size comparison was conducted by mapping each field based on selected sample locations from the dense sampling, throwing out identical locations and then running a correlation analysis on the remaining locations within the field. A 220 ft. (one sample per acre), 330 ft. (one sample per 2.5 acres) and 5 acre grid were compared to the 110 ft. or 150 ft. grids.

Topography of all sites was measured using a laser-leveling survey device and DGPS to record the position of each elevation, except at Colfax, where organic matter levels have been used as an estimate of elevation differences. Elevation information was taken on a 110 ft. grid.

At Gardner (Table 1) in 1994, the 5 acre grid performed relatively well compared to the 220 ft. and 330 ft. grids for nitrate-N. However, in 1995, the 5 acre grid was much less correlated compared to the 220 ft. or 330 ft. grid. Correlation of the 5 acre grid at Valley City, Mandan and Colfax was generally low compared to other grid sizes.

When topography was compared to grid size, topography was similar or superior to the 220 ft. grid nitrate-N levels in two of four locations, and better than the 5 acre grid at all four locations. The similarity of topography based nitrate-N levels is compared to the original sample grid in Figure 3 (Mandan) and Figure 4 (Valley City). At Valley City, the topography was mapped using 5 sample locations, while at Mandan only 14 locations were used.

Based on this research and another related survey in the Valley (Hollands, 1996), some producers are using topography to guide their soil sampling for nitrate-N. However, nitrate-N was most often related to topography, while sulfate-S and chloride were related, but tended to be more clustered. Phosphate was least related to topography, although at Mandan and Valley City, the relationship was relatively high.

ECONOMICS OF PRECISION FARMING

In Red River Valley sugarbeet production, three studies during the 1994 and 1995 growing seasons at University of Minnesota, Crookston and at North Dakota State University have shown that returns from variable-rate N applications based on a 4-5 acre grid could produce from $50-$70 (US) per acre net returns. The 1996 returns are being calculated presently and should be available for the conference. At both Mandan and Colfax, wheat was treated with variable-rate N treatments. Phosphate at planting was also varied at Mandan, and sulfur was varied at Colfax. Nitrogen and P was also varied for sunflower at Mandan. Yields were taken using a yield monitor at both Mandan and Colfax. An analysis of the data should be ready by the Manitoba-North Dakota No-till Association meeting.

Economics for use of precision farming will be site-specific. Responses to variable-rate fertilizers will be based on sampling costs, variability of the field, the probability of response to the nutrient and the scale of the nutrient response. Just like the application of any fertilizer, there is never a guarantee of profitability in a given year.

PUTTING TOGETHER INFORMATION

There is no standard way of conducting precision farming. For some, yield monitoring will be useful in managing decisions regarding field profitability, drainage, weed locations, comparing the effects of fertilizers, varieties and herbicides. For others, grid sampling, because of a history of heavy fertilizer or manure application, will be important. Topography sampling will help other producers reduce sampling cost on landscapes with historically modest fertilizer application and a lack of manure application. Remote sensing will help producers identify problems before yield is completely determined, and may help draw fertility boundary lines and outline other needs to help make future management decisions. Soil sensors such as the EM-38 for electrical conductivity may help map soil factors without costly analysis. These methods are all to be found in the Precision Farming Toolbox. Just like any tool, a producer should select the one right for the job and if it doesn't work well, select another.

REFERENCES

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

Hollands, K. 1996. Relationship of nitrogen and topography. pp. 123-128.IN:1995 Sugarbeet Research and Extension Reports. N.Dak. St. Univ. Ext. Serv., Fargo, ND.

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.Table 1. Correlation of nitrate-N from sampling grids and topography with original grid values.

Table 1. Correlation of nitrate-N from sampling grids and topography with original grid values.

Site	Sampling grid or method
	220 ft.	330 ft.	5 acre	topography
	correlation (r)
Gardner, 1994	0.513	0.351	0.158	-
Gardner, 1995	0.391	0.226	0.043	0.312
Valley City, 1994	0.175	0.065	0.073	-
Valley City, 1995	0.501	0.211	0.207	0.352
Mandan, 1995	0.290	0.442	0.225	0.755
Colfax, 1995	0.616	0.448	0.061	0.320

ACKNOWLEDGEMENTS

Special thanks go to the following people, companies and institutions for their financial and/or technical support for this on-going project. Agrium, Denver, CO; Agsco, Grand Forks, ND; Agvise Labs, Northwood,ND; Noel Anderson, Case-IH, Fargo, ND; Centrol, Mooreton, ND; P&PI, Norcross, GA; Sugarbeet Research and Education Board, MN and ND; the Soil Conservation Districts of Wild Rice and Stutsman county in North Dakota; USDA-ARS, Northern Great Plains Research Laboratory, Mandan, ND, Ardell Halvorson, director; and USEPA.