Evaluating Spatial Management Zones Delineated by Historical Yield Data for the Purpose of Variable Rate Seeding of Corn

John A Hettinga, Purdue University

Abstract

Variable rate seeding is a farming practice that continues to generate interest among growers. Since yield is a product of the interaction between soil properties and production inputs, spatial yield data is a potential data source for the delineation of management zones to be used for variable rate seeding. The primary objective of this study was to do determine whether the use of multiple years of spatial yield data was effective for delineating management zones for variable rate seeding. The secondary objectives were to determine (i) whether or not high, medium, and low yielding areas of the field produced different mean yields and (ii) if different optimum populations were present within each designated zone. The 6 fields used for this study had at between 5 and 8 years of historical corn and soybean yield data available and represented different geographical areas and soil associations of Indiana. The historical yield data were processed and cleaned in order to remove any anomalous points prior to the zone delineation process. The cleaned yield data were normalized to the means of each crop + year combination and then interpolated by IDW. The interpolated grid surfaces were reclassified into “states” that represented areas directly above or below the mean (MM) or field areas ≤ 90% , 90 to 110%, or ≥ 110% of the mean (M10) yield in each individual year of historical yield data. For each field, all years of the reclassified yield data were analyzed using the Markov Chain frequency analysis tool extension for ArcView 3.2. The resulting frequency map was then classified by Equal Intervals into three yield zones that represented areas of historically low, medium, and high yielding areas of the field. Alternatives evaluated during the zone delineation procedure included a) the use of historical yield data from whole fields or field interiors (minus low yielding field edges), b) reclassification thresholds based simply on mean yield (MM) or ± 10% of mean yield (M10), and c) the use of historical yield data from corn only, soybean only, or corn plus soybean. The inclusion of end rows and field edges influenced the delineation of yield zones, but the visual differences were very minor. The M10 method was hypothesized to delineate zones of uniform higher and lower yield as well as reduced within zone variability. However, the M10 reclassification method was unable to consistently identify a lower yielding Zone 1, higher yielding Zone 3, and less variable Zones 1 and 3 in all but 1 field. There was also minimal effect of the M10 reclassification on the area and pattern of zones delineated in a field. Differences in patterns between crop source dataset was minor at all trial locations besides NC1. When at least 3 years of yield data for a particular crop were available, zone patterns and area between the two zone maps as well as the combine zone maps were subtle. However, when only 2 years of one crop were available spatial differences between maps became more evident. Seeding rate trials with target seeding rates ranging from 61,750 to 103,740 pph were conducted in 2015 in 3 of the fields and 2016 in the other 3 fields to determine whether yield response to plant population varied among zones. Stand counts were taken when > 50% of the field was at the V6 leaf stage to estimate plant population for each plot. The main effects of yield zone (YZ) and seeding rate (SR) on grain yield were significant in all 6 trials. The interactions of YZ x SR were significant at 3 of 4 trials and the interactions of Hybrid x YZ x SR were significant in the 2 trials where two hybrids were used. However, there were no significant regression responses at the four locations on a whole field basis or within zones. Consequently, economic optimum plant populations were the same for all 3 zones and variable rate seeding was not economically justified. Variable rate seeding was economically justified for one hybrid at the SE1 location as Zones 2 and 3 responded to plant population while Zone 1 did not. A profit increase of $86.01 ha–1 was estimated when the higher EOPP (89,765 pph) was used for Zones 2 and 3 as opposed to the lower EOPP (59,940 pph) estimated for Zone 1. At the SE2 location, a different hybrid was responsive to plant population in all 3 zones, however the estimated EOPP’s for all three zones were the same (65,847 pph). Because of this, variable rate seeding was still not economically justified at this location even though yield responded to plant population.

Degree

M.S.

Advisors

Camberato, Purdue University.

Subject Area

Agronomy

COinS