Broad vision for OFPE

Precision Management and Experimentation: Wheat Yield and Protein as a Proof-of-Concept For Increasing the Value of Northern Great Plains Crops

Wheat is a unique crop because farmers can get paid a premium price for a crop quality trait (protein). Protein is the bellwether for many Human Health Quality (HHQ) traits that are likely to add future value to crops and economy. Increasing nutritional density in crops can contribute to more efficient food production on limited global arable land. Therefore, we propose that we demonstrate how protein can be manipulated in wheat with management practices informed by environmental data to maximize profit to producers and minimize pollution from crop production inputs. Our precision agriculture research has previously demonstrated how to put the massive data streaming from combine harvesters and the internet to work through complex algorithms to provide farmers field-specific nitrogen fertilizer management recommendations to maximize profits (Lawrence et al. 2015) and minimize nitrogen fertilizer load (Hegedus et al. 2022). We demonstrated how to use plot experiments, automatically laid out in the field with GIS, to apply different rates of fertilizer from standard rate controllers on fertilizer application equipment. When the crop matured, the yield monitor and protein sensor on the combine provided georeferenced yield and protein concentrations every few feet while traveling through the field allowing us to use this data accompanied by internet available satellite imagery to parametrize response functions and make predictions of profit maximizing nitrogen fertilizer strategies. Our optimizations suggest that precision application of nitrogen fertilizer will increase net returns by 18% over the standard uniform application. This could conservatively mean an infusion of $26 million per year to the Montana farm economy alone if half the acres in wheat employed our site-specific experimentation and subsequent nitrogen fertilizer recommendations. We are testing and thereby demonstrating this approach on small grain farms across Montana.

When farmers can see the performance of site-specific crop and pest management strategies on their field their adaptation to profit maximizing and pollution minimizing strategies are more likely than from results of small plot experiments miles away at Research Centers. However, the response functions have a high degree of uncertainty arising from a range of soil, crop sequence, crop pest, weeds, climate/weather, input cost, crop price and protein premium variables that cannot be controlled. Thus a major focus of our research has been aimed at refining the response functions and automating the gathering, centralizing, and cleaning the “big data” stream coming from the combine sensors, climate and weather data, soil samples, remote sensing, etc. That is, the proof-of-concept is highly reliant on automating the data management process and therefore demonstrating the value in site-specific crop and input management. Response functions are being tested on a dispersed set of fields with different climates, environmental conditions and management histories. In addition, we will determine how continuous wheat, crop-fallow wheat and pulse-wheat crop rotations in conventional and organic systems influence the response functions. Placing our adaptive management on-field precision experiment (OFPE) framework in the context of new agronomic developments, economics and rural communities will require identifying the best mechanisms for wide adoption in Montana.

General study objectives: 1) Develop software that automates OFPE from experiments to site- specific input (seeding rate for cover crops and cash crops and nitrogen fertilizer) recommendations that will optimize inputs based on maximizing net returns and minimizing pollution. 2) Demonstrate OFPE experiments on 20 fields on 8 farms growing wheat each year, with fields representing at least 2 different rotations and a regime of climate variability, and on conventional input intensive farms as well as certified organic farms. In addition, we have established collaboration with Washington State University, University of Illinois, University of Nebraska, Argentina and Western Australian scientists. This will lead to 36 wheat fields to monitor for increased net returns from the use of OFPE (Table 1). 3) Evaluate based on economics and adoption. MREDI has allowed us to initiate this process with the promise that we will continue to seek funding through competitive grant programs.

How the experimentation proceeds.

Table 1. Collaborator farms, fields, previous 3 years of crops plus 2017 crop, crop used for nitrogen treatment stratification in OFPE, OFRD or OFE year.

1WW = winter wheat, CF = chemical fallow, P = peas, MB = malt barley, B = barley, SW = spring wheat, AL = alfalfa, WT = organic winter triticale wheat, C = canola, SF = safflower; M = maize, S = soybean, S2 = late soybeans, bold = 2016 OFPE experiments, red = 2017 OFRD experiments and blue = 2017 OFE experiments.

Study concerns about the transition from Agriculture to Agriautomation:

All studies have underlying concerns and we want to be absolutely transparent about our position on the work we are doing and how it is contributing to a future of agriculture that might not be acceptable by society. We believe that we have an ethical responsibility to provide our assessment and vision of where the technology we are developing is taking agriculture and society.

The automation of determining optimum agricultural outcomes through on-field experimentation and data analytics can remove farmers from the decision process and through its objectivity provide safer, healthier food and non-polluting production practices. This endpoint to industrialized agriculture may be no different than what robotics did for automobile manufacturing. Safer, reliable and more efficient cars were the result. However, it eliminated many assembly line jobs and a portion of the population angry about the technology. Do we want to remove people (farmers) from agriculture? Agriculture could see a similar result to the automobile industry. Machine automation similar to that employed now in the mining industry could see operation of large equipment from remote locations only requiring mechanics on site to conduct repairs. Generations of handed-down knowledge could be removed representing a loss of "culture" from agriculture. Culture infers art and thus agriculture has been largely an art-form requiring generations of observations to recognize and respond to the complex ecological interactions in crop and animal production. Should we leave that knowledge base behind now that we have the ability to empirically understand the complex ecology well enough to derive best management practices? The conundrum is that agriautomation results in a more ecologically based agriculture that can simultaneously focus on food nutritional density and minimization of pollution, but ultimately removes human and historic mechanisms of knowledge. Agriautomation, the end point of industrialized agriculture, will likely increase farm size and consolidation leading to a fully automated food system controlled by modern data analytics (primarily machine learning).

We ask ourselves if there is a way to preserve the "culture" in agriculture and at the same time draw on the new and rapidly evolving data stream and analytics to envision a future that maintains an agrarian component to society. In part, we addressed this question with the concept of precision agroecology (Duff et al. 2022). However, there is much more to do as we consider the conundrum of agriculture and agriautomation.


Duff, H., P. Hegedus, S. Loewen, T. Bass and B.D. Maxwell. 2022. Precision Agroecology. Sustainability

14(1), 106.

Hegedus, P.B., B.D. Maxwell, J. Sheppard, S. Loewen, H. Duff, G. Morales-Luna and A. Peerlinck. 202_. On-Farm Precision Experiments (OFPE) framework: tapping local data to optimize crop sub-field scale decisions. Agriculture. In review Submitted 12/2/2022

Hegedus, P.B. and B.D. Maxwell. 202_. Best use of modern data for field-specific decision support. Precision Agriculture. In review Submitted 6/2022

Hegedus, P.B., S. Ewing, C. Jones and B.D. Maxwell. 202_. Using spatially variable nitrogen application and crop responses to evaluate crop nitrogen use efficiency. Nutrient Cycling in Agroecosystems. In review. Submitted 9/30/2022

Morales, G., J. Sheppard, P. Hegedus, B. Maxwell. 202_. Improved Yield Prediction of Winter Wheat Using a Novel Two-Dimensional Deep Regression Neural Network Trained via Remote Sensing. Sensors In review submitted 11/13/2022

Hegedus, P.B., T. Mieno and B.D. Maxwell. 2022. Assessing performance of empirical models for forecasting crop responses to variable fertilizer rates using on-farm precision experimentation. Precision Agriculture. 23: Accepted 10/5/2022

Hegedus, P.B. and B.D. Maxwell. 2022. Field-specific precision agriculture based experimentation demonstrates field-to-field and year-to-year variation. Agriculture Ecosystems and Environment. Accepted 7/2/22.

Morales, G., Sheppard, J., Peerlinck, A., Hegedus, P., & Maxwell, B. (2022). Generation of Site-specific Nitrogen Response Curves for Winter Wheat using Deep Learning. Proceedings of the 15th International Conference on Precision Agriculture. Minneapolis, MN, USA.