With the rising cost of inputs and the shrinking profit margins in agriculture, farmers are looking to manage at the plant level to increase crop yields. As the physical size of agricultural field machinery continues to grow, many agriculture professionals recognize the negative effects of increasing gross vehicle weights on soil structure, health and productivity. The persistent trend of increased machinery size and gross weights thus exacerbating soil compaction which reduces crop yields and impacts profitability. This manuscript focuses on assessing the adverse impact of high axle loads on field productivity for corn production.
Historically, many studies were performed using axle loads ranging from 10 T to 20 T. Few, if any, studies were conducted at axle loads in excess of 20 T. A better understanding of higher axle loads is needed in view of the trend of increasing equipment size where axle loads now approach 50 T. Development of a compaction model combined with data tools will allow users to process remote sensed imagery and CANbus data to better visualize and estimate the yield-related effects of compaction. The overarching goal of the envisioned tool is to provide farm managers and decision makers with actionable information as they assess the ever-expanding number of equipment options available in the marketplace. By coupling remote sensed imagery, yield monitor data, CANbus data and field trial results, the envisioned tool aids producers in making informed decisions specific to their equipment complements and soils via information extraction and synthesis from the ever-expanding quantity of data being collected on their farms.
This manuscript details a series on investigations undertaken to better understand the potential effects of each pass of machinery over a field. These investigations were designed to: 1) develop an empirical model framework to predict the magnitude of compaction events and the resulting yield penalty based on axle load from previous investigations; 2) conduct fields trials that enhance the models ability to estimate yield penalty at higher axles loads; 3) expand model to include model parameters to account for number of passes, use of tracks vs. tires, and higher axle loads; and 4) determine if remote sensed imagery and yield monitor data can be used to identify and assess yield loss in compaction zones.
An empirical model was created using previously completed research on crop response to axle loads. A graphical user interface (GUI) was created to support implementation of the model at the grower/consultant level. Model results were compared and contrasted with results from two existing investigations. The first study (Ahlers, 2012) was designed to quantify the “pinch-row” effects of large, central fill planters. The model successfully bracketed the study results. The second comparison was to a 2014 compaction study conducted at the Farm Science Review site near London, OH in which the model predictions were within 2.0% of the actual yield loss.
Compaction plots were created at the Beck’s Hybrid field site near London, OH using two grain carts with three undercarriages; Equalizer tracks, regular tracks, and flotation tires and axle loads in excess of 48 T. Performance of three hybrids were compared during this study along with multiple grain cart passes (one, two, and three passes) for a total of 540 plot observations. Additionally, soil measurements were taken to quantify soil strength (soil cone penetrometer), volumetric soil moisture content, and soil surface deformation from the grain cart passes. The compaction events occurred in the early spring after deep ripping to reset the soil conditions. Throughout the growing season numerous over flights and image collects were completed using UAS and manned aircraft. These data facilitated development of a second model for redistributing yield monitor data based on vegetative indices from the remote sensed imagery. The effective resolution of the resulting yield map improved significantly.
Wheeled grain carts generated yield reductions of the 23.5% across all events regardless of the number of passes. Grain loss from Equalizer tracks averaged 19.1 % across all number of pass treatments compared to 19.6% for the regular tracks. This allowed for the creation of a tracked undercarriage correction factor in the empirical model. No significant yield difference was found between the number of passes. The range on mean yields or one to three passes were: Equalizer tracks 129 bu/ac to 125 bu/ac, regular tracks 130 bu/ac to 108 bu/ac, and wheels 117 bu/ac to 110 bu/ac.
When looking at the 0 to 6 in. soil depth, soil strength (soil cone penetrometer readings) for the Equalizer tracks average 93 psi higher than the control; regular tracks, 84 psi greater; and wheels, 15 psi greater. All differences were significant. When looking at volumetric soil moisture contents for the 0 to 8 in. depth, Equalizer track soils were 6% wetter, regular tracks were 5% wetter, and wheels were 9% wetter. For the 0 to 3.0 in. depth, soil moisture contents for the Equalizer tracks were 10% wetter, regular tracks, 7%; and wheels, 11%. All differences (between the trafficked and control areas) were significant. The adjusted compaction yield reduction model, with new correction factors, predicted yield reductions ranging from 15% to 21% for wheels, 19% to 28% for regular tracks, and 20% to 28% for Equalizer tracks.
The yield map post-processing model used linear regression to establish the relationship between corn grain yield and organic carbon, digital elevation and vegetative indices derived from RGB remote sensing iamgery. To avoid multicollinearity between explanatory variables in the regression model, a variance inflation factors (VIF) was used. The final post-processing model used plant pigment ratio, digital elevation and soil organic carbon to produce a yield maps at a resolution consistent with remote sensed image. The R2 values for the post-processing model was 0.50. Yield predictions from this model were within 1.0 to 7.0 bu/ac for the control regions. The post-processing model accurately predicted yield losses for the wheeled grain cart while it over-predicted yield losses by 17.0 bu/ac for both of the tracked undercarriage configurations. Yield reductions for all compaction field plots ranged from 19% to 25% (all undercarriages) while prediction from the post-processing model ranged from 23% to 30% (all undercarriages).