Topography‐Based Sampling Compared with Grid Sampling in the Northern Great Plains

Abstract

A practical variable‐rate fertilizer application should be based on information gathered at low cost that represents field fertility levels. The number of soil samples gathered and analyzed may limit the effectiveness of some variable‐rate fertilizer applications. Topography‐based soil sampling is attractive because it suggests a lower number of samples needed to characterize fertility levels and patterns in a field than some current grid sampling recommendations. A 40‐acre North Dakota field consisting of Barnes loam (fine loamy, mixed, Udic Haploborolls), Swenoda loam (mixed, Pachic Udic Haploborolls), and Wyard clay loam (fine loamy, mixed, frigid Typic Haplaquolls) soil types was sampled in a 110 ft grid each fall from 1994 to 1996. Nitrate N, P, sulfate S, and chloride level patterns were similar in all 3 yr. Correlation of both area‐ (sample cores obtained from a wide area within a topography zone) and point‐based (a topography zone represented by a composite taken from a point location) topography sampling with the 110 ft grid values was compared with correlation from a 220 ft, 330 ft, and 5 acre grid. Area‐based topography sampling for nitrate N was superior in correlation to the 220 ft grid values in 2 of 3 yr. Area‐based topography sampling with P was superior to the 220 ft grid in only 1 of 3 yr, but was superior to the 5 acre grid in all years. The 220 ft grid was superior to topography sampling for sulfate S in all years, but area‐based topography sampling was better than the 330 ft grid in 2 of 3 yr and superior to the 5 acre grid in all years. Area‐based topography sampling for chloride was superior to the 220 ft grid in 1 of 3 yr, but was more highly correlated with the 110 ft grid than the 330 ft and 5 acre grid in all years. Area‐based topography sampling was more highly correlated than point‐based topography sampling in seven of 12 comparisons. Topography‐based sampling for nitrate N better represented fertility patterns than did grid sampling. Research Question In order for variable‐rate fertilizer application to be practical, soil testing information must be gathered at low cost, but at the same time represent the variation in field soil fertility levels. Since collecting and analyzing soil samples is expensive, minimizing the number of samples collected while maintaining a high level of soil fertility information is important. This study of one North Dakota field compared the use of topography‐based sampling with selected grid methods for nitrate N, P, sulfate S, and chloride to determine whether topography‐based sampling might compare favorably with information gathered from grid sampling methods, while decreasing the number of samples needed to provide similar or superior soil fertility information. Literature Summary Previous work in Wisconsin and Illinois has suggested that a one sample per acre grid might be required to gather soil fertility information needed for a variable‐rate fertilizer application. Other studies have demonstrated that some soil fertility factors may be related to landscape. Topography‐based sampling is attractive because the number of samples needed to characterize a field may be reduced compared with a dense grid. Study Description A 40‐acre field in North Dakota was sampled in a 110ft grid to 2 ft in depth, dividing the core into a 0 to 6 in. depth and a 6 to 24 in. depth. Phosphate was analyzed on the 0 to 6 in. depth only, while nitrate N, sulfate S, and chloride were analyzed on both depths and reported for the 0 to 2 ft depth. Maps of 220 ft grid, 330 ft grid, and 5 acre grid nutrient estimates were developed by deleting points from the 110 ft data set between grid sample values and then using an inverse‐distance squared estimator to calculate values between grid spacings. Elevation was measured using a laser‐survey device at 110 ft intervals. Five to six topographic zones were determined from a three dimensional elevation map by drawing lines around areas of similar landscape. Topography estimates were determined by both point sampled and area sampled methods. The point sampling based topography was developed from the 110 ft grid by choosing a central point value from each topography zone and representing the zone with that value. The area‐based topography sampling was developed by choosing six close points surrounding the point value and giving the average value to each point within the topographic zone. Correlation was made of grid sampling value estimates and topography estimates with the values from the 110 ft grid. Like points within grids and topography were deleted to avoid autocorrelation between data sets. Applied Questions Were soil fertility patterns similar between years? Soil fertility patterns for all nutrients were similar between years, suggesting that landscape relationships may be important in controlling residual nutrient levels within the field. How did correlation of grid estimates compare with topography‐based sampling? When compared with 110 ft grid patterns, area‐based topography sampling for nitrate N was superior to the 220 ft grid values in 2 of 3 yr. Area‐based topography sampling with P was superior to the 220 ft grid in only 1 of 3 yr, but was superior to the 5 acre grid in all years. The 220 ft grid was superior to topography sampling for sulfate S in all years, but area‐based topography sampling was better than the 330 ft in 2 of 3 yr and superior to the 5 acre grid in all years. Area‐based topography sampling for chloride was superior to the 220 ft grid in 1 of 3 yr, but was more highly correlated with the 110 ft grid than the 330 ft and 5 acre grid in all years. How did area‐based topography sampling compare with point‐based sampling methods? Area‐based topography was superior to point based nitrate N and sulfate S in all years. Point‐based topography sampling was slightly better than area‐based topography for P in all years. Area‐based topography sampling was most consistent for chloride correlation, however, point‐based topography for chloride was more highly correlated in 2 of 3 yr. Recommendation Grid sampling might be recommended for a field when: The field history is unknown. The field has had a history of manure application. Nonmobile nutrient levels are important, such as P, K, or Zn. High levels of fertilizer have been previously applied. Small fields have been merged to form larger fields. Topography sampling might be recommended for a field when: Remote‐sensing or yield monitor information reveal a relationship to landscape. The field has no history of manure application. Relatively low (less than maintenance) rates of fertilizer have been previously applied. Mobile nutrients, especially N, are important. It is conceivable that both methods, a dense grid and topography sampling, might be used within a single field based on a previous sampling or information from field history.