Frontiers in On-Farm Experimentation
Statement of Issues and Justification
Identification and Significance of a Problem or Opportunity
The problem we propose to address is the worldwide, systemic inefficient application of crop inputs on farm fields. A principal focus will be on the chronic mismanagement of nitrogen fertilizer. Our society manipulates the nitrogen (N) cycle to great benefit, but chronic inefficient use of nitrogen fertilizer has led to the hypoxic “dead zone” in the Gulf of Mexico and the leaching of nitrates into groundwater. The National Academy of Engineering (2012) has declared managing the nitrogen cycle a “Grand Challenge,” and the N cycle has been labeled a “planetary boundary that has been transgressed” (Rockström, et al. 2009). But currently agricultural science is far from understanding the processes determining crop yields, and therefore from discovering how efficient input management practices vary over time and space. Reimer, et al. (2017, p. 6A) described the problem well:
The incredible complexities of the biophysical systems alone are still not well understood by researchers, advisors, or farmers. There are a myriad of variables involved in a decision about how much N to apply in a given growing season … This makes providing reliable recommendations inherently difficult, both for researchers trying to distill complex science into useable information and for advisors trying to craft recommendations for individual farms and fields. (Reimer, et al., 2017, p. 6A)
Unless scientists can gain a better understanding of these complexities, farmers will continue applying fertilizer in ways that send chemicals that everyone wants to go into crops actually go down rivers. The objective of the proposed project is to develop a research infrastructure and encourage the development of a commercial infrastructure that will generate the data needed for this better understanding. If successful, the research will increase farm income and enhance the quality of the nations’ waters.
Of late, great hope has been placed in using the components of digital agriculture, including precision agriculture technology, agricultural Big Data, remote and proximal sensing, and on-farm experimentation to increase farm management efficiency. Schlam (2019) offered a typical quote:
The use of analytics will be crucial to enabling smarter agriculture, with farmers leveraging data on soil, local diseases and pests, climate, and other environmental factors to optimize yields and seed selection. ... Soil sensors, aerial drones, GPS-enabled tractors, and more will generate the data upon which analytics solutions will rely … .
But, however impressive these technologies and techniques are individually, and despite public and commercial enthusiasm about them, academic agriculturists looking to the future are expressing frustration with the current state of their use, and have recognized a need to bring them together systematically:
Digital agriculture … has been trying to attract customers before the ecosystem has been properly constructed. What we believe is missing is a standardized way to gather and interpret data, and then translate actionable insights to commercial users—insights which then, in turn, can deliver value to growers. (Zuckerberg, 2017)
In the following, we propose an integrated project to drive the creation of the system Zuckerberg calls for above. We will argue that the key impediment to the development of digital agriculture is lack of information about crop responses to factors of production, but that components of digital agriculture can be used to generate that same information.
A Conceptual Framework and the Fundamental Question
We conceptualize the problem of crop input management in terms of four types of variables. The first variable, y, is crop yield. The second type we represent by a multi-element vector c, of unmanaged, spatially distributed “field characteristics.” The third is a multi-element vector z, of unmanaged and temporally stochastic variables (principally, weather). The fourth is a multi-element vector x of “managed input variables,” (e.g., fertilizer application rate, seeding rate). We conceptualize yield, y, as resulting from a natural process described by a function f, which depends on farmer choices, field characteristics, and weather: y = f(x, c, z).
The fundamental research question for crop input management is, “What is f?” Agricultural scientists have been attempting to generate data to estimate f for major row crops for almost two hundred years (Odell, et al. 1984). Indeed, estimating f was motivated the advance of modern statistical theory, as R.A. Fischer (1935) developed his pioneering statistical research on randomization in experimental design to address estimating f using data from agronomic experiments (Antle 2019). The extraordinary efforts to learn about f have been made because f reveals how crop yields respond to producer choices, and how yield responses change with growing conditions. Better estimates of f are key to better scientific support of farm management and all the social benefits that follow.
(Figure 1 - see attachments section)
Figure 1 provides the simplest of illustrations. Versions of figure 1 and explanations of the implications for producer management choices appear in most introductory microeconomics textbooks (e.g., Pindyck and Rubinfeld 2013, pp. 204-208), and are typically presented in the first few weeks of introductory microeconomics courses. It is assumed in figure 1 that there is only one managed input, x. The characteristics of a site A on which the crop is grown are cA. For simplicity, it is assumed that when a value of x is chosen, weather is assumed known and given as the vector of constant values z2019. The input price and output price are assumed constant at levels w and p. Elementary calculus shows that the profit-maximizing input application rate, x*A,19(p, w), is that value at which the slope of the yield response curve is equal to the price ratio, w/p. When multiple inputs are chosen, similar mathematical rules apply, optimizing over more than two dimensions. Space limitations prevent us from illustrating more complicated situations.
Technical Feasibility of the Work
The focus of the proposed project is on-farm precision experimentation (OFPE) and the analysis of OFPE data, which are illustrated in figure 2. The first panel of figure 2 shows the randomized design of a nitrogen rate OFPE on corn. In form, the design is like those of traditional small-plot trials, but covers a much wider area. The field in figure 2 is 37 ha in size. The second panel shows the trial being “put in the ground” using variable input rate application is accomplished by using GPS-based computer software to pre-program a variable application rate “plan” into a computer aboard farm machinery. That program “instructs” application equipment to apply inputs at the planned rates on the designated plots as the farmer just drives the farm equipment through the field in the usual manner. The final figure shows the resultant data.
(Figure 2 - see attachments section)
Led by PI David Bullock, the Data-Intensive Farm Management (DIFM) project has now worked with participating farmers to conduct over one hundred large-scale OFPEs in eight US states, Argentina, Brazil, and South Africa. Trials have been run on cotton, corn, soybeans, and wheat. DIFM has collected, processed, and analyzed the data, communicated the analytical results to the participating farmers, and published journal articles describing the techniques and results. The basic elements of OFPE research, then, are clearly feasible. Those techniques inexpensively provide excellent data, and one very legitimate aim of the proposed multistate project will simply be to pass the proven techniques to scientists currently unfamiliar with them, so that they might work with farmers to conduct OFPEs. But the project goals are much more ambitious than simply teaching other researchers OFPE techniques. The aim is to build a research infrastructure and encourage the development of a commercial infrastructure that will permit tens of thousands of OFPEs to be designed and conducted farmers and their crop consultants annually, as well as the handling, processing, and analysis of the resulting data. The system would annually generate vastly more field trial data than has been generated since the first agronomic field trials were conducted in the first half of the nineteenth century.
The main elements of the strategy to scale up data generation involve the creation of a cloud-based, “on-farm precision experiment design” software system. That system will allow crop consultants, who have received training from Extension personnel or others, to upload basic information about a farmer’s field, such as a geo-referenced file of the field’s border, and the sizes of the farmer’s machinery, and then, with some “pointing and clicking,” design a statistically legitimate agronomic experiment over the farmer’s entire field. All data would be transferred wireless to and from farm machinery. Writers of this proposal have already developed parts of this automated trial design system, and are confident that developing the rest will take hard work, but is feasible.
The principal element of scaling up OFPE data analysis will be an automated “analytical engine,” which can import OFPE data, and then with minimal “human-in-the-loop” effort, employ econometric analysis and a variety of machine learning methods to develop management recommendations. For example, methods based on reinforcement learning can be used to optimize prescriptions for seeding or applying fertilizers or herbicides. Learning algorithms such as those based on deep learning (e.g., convolutional networks or long short-term memories), locally (spatial or temporal) random forests, or Gaussian processes can be used to capture spatial and temporal properties in fields. Other techniques based on a variety of heuristics (local search) and metaheuristics (evolutionary and swarm-based search) in constraint satisfaction, Markov random fields, and even natural language models to describe agricultural processes can be applied to derive data-based farm management support.
The statistical and machine learning models will be accompanied by responsive, interactive statistical visualizations to explain which data have the most impact on the predictions (the “why” of the statistical models), helping build the trust of farmers and consultants in the “black box” models used for prediction. In addition, OFPE data will be used to improve the calibrations of existing crop growth models, which in turn can be used to model yield response and estimate optimal management strategies. Automating OFPE data analysis will be a major challenge. We envision a future in which significant parts of crop science and agricultural economics academia devote themselves to this very task. But it will be feasible, and fascinating, to make a convincing start to this research endeavor in the five years of the proposed project.
Automating the communication of the implications of a farm’s OFPE data will involve the creation of a cloud-based “decision tool” software system, which displays in user-friendly, visual ways, OFPE data results and implications. Several such “decision tools” are currently available on commercial markets. Several of the 135 interviews of farmers, professional crop consultants, and extension personnel conducted by two of the writers of this proposal (Montesdeoca, et al. 2018) made clear that many farmers and consultants have limited faith in their effectiveness. The crop science and agricultural economics literature impels skepticism of current commercial decision tool software packages. They are based on long debunked yield-based input application algorithms (Rodriguez, et al. 2019) and crop growth models with parameters that in general cannot be well enough calibrated to allow the models to provide adequate management advice (Antle 2018).
The DIFM project has proved that conducting trials and analyzing the data is feasible. We believe that the scaling-up of that work is also feasible within the five-year timeline. But, certainly, the project faces risks. Data analysis is difficult, and to develop automated techniques for it that work well enough to provide farmers with profitable management recommendations will be a challenge. But this is why a multistate project is needed--we plan to put dozens of very skilled, highly talented researchers to the tasks at hand, for five years. We are confident that we can develop a workable infrastructure by then, and that the next generation of agronomic science and farm management rely upon, and continue to advance that infrastructure.
The Importance of the Work, and Consequences of It not Being Done
“Site-specific” agriculture technology appeared on the market in the 1990s. But, as Reimer describes in the quote above, very little about the yield response process is known. As a result, little is known about how to use precision agriculture technology well. Farmers complain that they do not know what to do with their data. In truth, in many ways neither does anyone else. If the proposed work is not done, the ignorance that Reimer describes will continue for longer than necessary, and farmers will continue having to make input application decisions that depress their incomes and are bad for the environment.
Evidence for Stakeholder Identification
Evidence for stakeholder identification of the need for the proposed project’s outputs was provided in abundance during the 135 interviews of farmers, professional crop consultants, and extension personnel, mentioned above. That study concluded that many farmers are already experimenters. Certified Crop Advisors (CCAs) and farmers told us repeatedly that they were placing various types of informal “test plots” in their fields, to see how alternative management strategies fared against their actual strategies. But CCAs also told us that carrying out field trials was a great deal of work, and that they realized that the statistical designs of their trials were technically flawed.
Another principal conclusion from the interviews is that many farmers are frustrated that they now possess multitudes of data that they do not know how to analyze. A few representative quotes follow:
“Once you get all this data, you need to hire someone to analyze it. If you can trust them. … Lots of people are just trying to sell you something.” -western Illinois corn grower
“Our seed rates? Pretty much a shot in the dark.” -three south Texas cotton growers
“It would be wonderful if my decision-making were more analytical instead of just off-the-hip SWAG (‘Scientific Wild-A** Guesses’)!” –northern Illinois corn grower
The interviews also showed that neither farmers nor their CCAs want “black-box” management recommendations from decision tool. They want intuitive agronomic explanations of how field characteristics determine the responses of yields to managed variables.
The Advantages of Doing the Work as a Multistate Effort
The proposed project’s collaborators come from all over the US. Including multiple geographically dispersed universities allows trials in crops from different regions: maize, soybeans, white winter wheat, red spring wheat, barley, cotton, canola, sugarcane, and sorghum, under varied topography, soil characteristics, and climates. The multiple-university team of researchers and extension personnel is comprised of personnel with significant experience already working together, and its professionals are capable of providing the considerable needed expertise.
Likely Impacts of Successfully Completing the Work
The proposed project’s principal goal is ambitious: we want to revolutionize agronomic research and how the information from that research is transferred to those who actually make farming decisions. It will make farmers central to the research process. The data will allow crop consultants to provide farmers management advice derived from data from the same fields for which that advice needs to be applied, instead of data conducted by researchers somewhere else. Farmers will improve their input management strategies, basing them on data, not speculation. Farm income will be increased, and environmental damage from agricultural chemicals will decrease. In addition, the commercial crop consultancy industry will change, providing advantages to those who understand experimental design and statistics. It will attract highly skilled workers into rural areas, which will have multiple positive social and economic impacts in those regions.
Related, Current and Previous Work
Conducting Field Trials to Estimate Yield Response
A principal objective of agronomic and agricultural economic sciences is to provide farmers with science-based farm input management advice, which can lead to economic and environmental benefits (Scott et al., 2015; Wolfe et al., 2016). Knowledge of yield response functions is key to providing that advice. For almost two hundred years, agricultural scientists have relied on small-plot randomized agronomic field trials to generate the data needed to estimate yield response to inputs (e.g., Spillman, 1933; Heady and Pesek, 1954; Waugh et al., 1973; Stauber et al., 1975; Odell, 1984; Ackello-Ogutu et al., 1985; Cerrato and Blackmer, 1990; Paris, 1992; Bullock and Bullock, 1994; Makowski and Wallach, 2002; Tembo et al., 2008; Marenya and Barrett, 2009; and Tumusiine et al., 2011), and small-plot trials remain the principal method of providing input-yield response data. In a typical trial, researchers apply inputs and harvest the crop by hand or with specialized machinery, and analyze the resulting data to understand input-yield relationships. However, despite the concerted efforts, the generation of small-plot field trial data has always been limited because implementing the trials have been labor-intensive, involving numerous workers conducting hours of fieldwork, and thus expensive.
We believe that the much of the potential of precision technology and digital agriculture in general remain unfulfilled because we remain largely ignorant about crop yield response. Reimer, et al. (2017, p. 6A) states this point succinctly in the context of nitrogen fertilizer management:
The incredible complexities of the biophysical systems alone are still not well understood by researchers, advisors, or farmers. There are a myriad of variables involved in a decision about how much N to apply in a given growing season. … This makes providing reliable recommendations inherently difficult, both for researchers trying to distill complex science into useable information and for advisors trying to craft recommendations for individual farms and fields. (Reimer et al., 2017, p. 6A)
On-farm precision experimentation (OFPE) uses precision agriculture technology to produce data from full-field agronomic trials. OFPE offers great promise to decrease the costs and thus increase the generation of data for use in yield response function estimation. OFPEs were first performed in short-run studies about twenty years ago (e.g., Cook, et al. 1999; Rund 2000). But use of the technique has begun to grow rapidly over the past few years. The proposed multistate project aims to conduct the research necessary to underlay a much broader OFPE network, in which farmers and professional crop consultants work with each other and with Extension personnel to design and conduct tens of thousands of OFPEs annually, use sensors and robotics to measure field and plant characteristics, and use an “analytical engine” and Cloud-based “decision tool” software to process and analyze the data, to provide data-intensive farm management advice for individual fields, every year.
OFPE is an expanding, exciting technique, which we believe can revolutionize agronomic research and farm management, and upon which we believe that the future success of precision agriculture and digital agriculture depend. The proposed OFPE-focused multistate project will create a forum for focused discussion on the narrow, but exceedingly important area of using OFPE in yield response function research. The aim is to provide the research needed to underlay a worldwide infrastructure, to meet the challenges recognized by Zuckerberg (2017):
Digital agriculture … has been trying to attract customers before the ecosystem has been properly constructed. What we believe is missing is a standardized way to gather and interpret data, and then translate actionable insights to commercial users—insights which then, in turn, can deliver value to growers.
Related Multistate Projects
In the following, we present short comparisons between the proposed project and the active multistate projects that have key elements in common with those of the proposed project. Should the proposed project be funded, there will be no more than minimal duplication of the efforts of existing projects.
NCERA 180: Precision Agriculture for Technologies for Food, Fiber, and Energy Production
The NCERA 180 projects states that its principle expected outcome “is an improvement in the understanding of how precision and prescriptive agricultural brings value to the farm and agriculture overall.” The proposed multistate project will focus itself on a narrower topic than does form NCERA 180, examining not so much on the use of precision technology “overall,” but rather more narrowly focusing on the idea of using precision technology to generate information about yield response functions. We believe that the principal bottleneck to progress in precision agriculture is ignorance about yield response functions. The proposed project will focus not so much on what is currently working well in digital agriculture, but rather on fixing what is working poorly.
SERA17: Organization to Minimize Nutrient Loss from the Landscape
The proposed project and SERA17 both aim to encourage the development of more efficient fertilizer management strategies. But SERA17 puts little focus on on-farm research, and does not mention the narrower but crucial topic of yield response function research on its nimms.org website. SERA17 also focuses more on modeling and understanding the effects of alternative fertilizer management strategies on water quality, whereas the proposed project hopes to have a more indirect positive effect on water quality by focusing on yield response functions and thus helping farmers not apply fertilizers in excess of their economically optimal rates.
SERA46: Framework for Nutrient Reduction Strategy Collaboration: The Role of Land Grant Universities
As with SERA17, the proposed project shares certain goals with SERA46, which is to encourage the development of more efficient fertilizer management strategies. But, as discussed above, the proposed project focuses on the narrower topic of yield response function estimation, and will apply to a broader geographical region than the twelve-states in the Mississippi River Basin that are developing ERA-requested Nutrient Loss Reduction Strategies.
NC1195: Enhancing Nitrogen Utilization in Corn Based Cropping Systems to Increase Yield, Improve Profitability, and Minimize Environmental Impacts
While the proposed multistate will conduct research on corn yield response to nitrogen fertilizer, to increase farm profitability and reduce environmental impacts, its focus on-farm precision experimentation is narrower than NC1195’s focus, and its application will be wider, covering multiple crops and multiple regions in US agriculture. The NC1195 website provides broad descriptions of collecting nitrogen use data and developing nitrogen management decision tools. The proposed research will generate a particular, and particularly powerful type of nitrogen data, not simply recording how farmers currently use nitrogen, but also generating data of the results of N application strategies that farmers do not use. It is not possible to understand the effects of changes in N management strategies without physical experiments that apply N using strategies that farmers do not commonly use. Also, decision tools developed within the context of the proposed multistate project will be based narrowly on yield response functions, the estimations of which will be based uniquely on on-farm field trial experimentation.
S1069: Research and Extension for Unmanned Aircraft Systems (UAS) in U.S. Agriculture and Natural Resources
As does S1069, the proposed project will collect data with drones. That data will be used to complement the field trial data, and all of the other types of field characteristics data collected. The complementarity of x, c, z, and y data discussed in the Issues & Justification section is key; UAS data alone is far less valuable than UAS data used along with field trial and other field characteristics data. One cannot learn how UAS data relates to the effects on yield of changing input management strategies unless one possesses UAS data that has been collected in a systematic pursuit of all the pertinent data. That is, one cannot know how to improve management with UAS data unless one conducts experiments that provide data that varies input management.
W3009: Integrated Systems Research and Development in Automation and Sensors for Sustainability of Specialty Crops
As does W3009, the proposed project will examine the use of robots and proximal sensors to measure field characteristics and plant phenotypes. But W3009 is focused exclusively on specialty crops, such as fruits, nuts, and vegetables, the proposed project will focus on major field crops, including corn, wheat, and cotton.
Other Related Previous and Current Work
Several on-farm research networks are currently active in the US. Of these, only the Data-Intensive Farm Management project is conducting OFPEs. But other networks are currently running other kinds of field trials with participating farmers, with whom they have developed longer-term relationships. These relationships develop only over time, and therefore the other on-farm research networks are well situated to begin running OFPEs. Researchers and Extension personnel from DIFM and five other on-farm research networks will participate in the proposed multistate project, as described below. Having representatives from these different networks and projects at the same meetings on a regular basis will help them establish common trial design, data processing, and data analysis protocols, discuss challenges in OFPE research, and move OFPE research and outreach activities as a whole, and in common directions.
The Data-Intensive Farm Management Project
The NIFA-funded Data-Intensive Farm Management project has lain much of the foundation for the proposed project. Bullock, et al. (2019), describe that project in detail, summarized here. DIFM works with farmers, using precision technology to inexpensively design and conduct OFPEs that provide data-based, site-specific management guidance for farm inputs. DIFM’s data and the agricultural “Big Data” currently being collected with remote and proximal sensors are complementary; that is, more of each increases the value of the other. In 2019, DIFM and affiliates are conducting over one hundred OFPEs, ranging from 10 to 100 ha in size, on maize, wheat, soybeans, cotton, and barley in eight US states, Argentina, Brazil, and South Africa. DIFM has developed the beginnings of an OFPE cyberinfrastructure, which must be “scaled up” to permit researchers and crop consultants worldwide to work with farmers to conduct trials, then process and manage the data. In addition, DIFM is in the early stages of developing a software system for semi-automatic data analytics, and a cloud-based farm management aid, the purpose of which is to facilitate conversations between crop consultants and their farmer-clients about implementing data-driven input management decisions.
On-farm Experimentation Networks and Projects
On-farm experimentation (though not necessarily on-farm precision experimentation) is currently being organized and conducted by at several U.S. universities. The Nebraska On-Farm Research Network (NOFRN) is a statewide program that addresses critical farmer production, profitability and natural resource questions. Growers take an active role in the project, which is sponsored by Nebraska Extension in partnership with the Nebraska Corn Growers Association, Nebraska Corn Board, Nebraska Soybean Checkoff and Nebraska Dry Bean Commission. (Nebraska On-Farm Research Network 2019). Farmers have conducted research as part of the NOFRN since 1990; annually 60-100 on-farm research projects are conducted and meetings are held to share the local, reliable research-based information. Among its many on-farm research pursuits, NOFRN has cooperated with DIFM project, working with participating farmers to run “checkboard trials” since 2016. Laura Thompson is a Co-coordinator of NOFRN, and will play an important role in the proposed multistate project.
Ohio State University’s eFields On-Farm Research program (E-Fields On-Farm Research 2019) utilizes modern technologies and information to conduct on-farm studies with an educational and demonstration component used to help farmers and advisors understand how new practices and techniques can improve farm efficiency and profitability. The program is also dedicated to delivering timely and relevant, data-driven, actionable information. Current projects are focused on precision nutrient management strategies and technologies to improve efficiency of fertilizer placement, enable on-farm evaluation, automate machine functionality, enhance placement of pesticides and seed, and to develop analytical tools for digital agriculture. Though eFields does not currently conduct OFPE research, it has the institutional backing along with the experienced research and extension personnel to do so, having run ninety-five on-farm, full-field strip trials in 2019. Ag Technologies Extension Field Specialist Dr. Elizabeth Hawkins is the Coordinator of eFields, and will play an important role in the proposed multistate project.
Washington State University’s Farmers Network (WSU Farmers Network 2019) advances soil and nutrient management, crop productivity, farm sustainability and profit through collaborative research, extension and on-farm participatory learning. By directly participating in the research on their own farms, farmers learn first-hand which practices are better choices and how to fine-tune their own nutrient management systems. Along with other research activities, in 2019 WSFN began cooperating with the DIFM project to conduct OFPE research. Dr. Haiying Tao is the Director of WSFN, and will play a key role in the proposed multistate project.
At the University of Wisconsin, Dr. Paul Mitchell is the lead on the NIFA-funded project “Applications of Reinforcement Learning Algorithms to Improve Crop Input Use” (grant: number 2019-67023-29418). He has assembled a multi-disciplinary research team and a small group of cooperating grain and commercial vegetable farmers to field test algorithms developed by project research assistant Yuji Saikai (who will graduate in May 2020). An important focus of their research is to use Bayesian optimization and other machine learning techniques to improve plot placement and other farm management in on-farm field trials (Sakai et al. 2018, Saikai and Mitchell 2019, Saikai et al. 2019).
The Iowa Soybean Association’s On-Farm Network® (Iowa Soybean Association 2019) runs field-length strip trials, including soybean and corn seed rate trials and corn N rate trials. Working with participating farmers, ISA completed over 200 trials in 2018. The ISA strip trials are simpler than typical OFPEs, to help farmers successfully implement trials on their own. Because ISA runs hundreds of trials annually, farmers need to be able to implement them with minimal guidance and less disturbance to their common operations. An important objective of the proposed multistate project is to design software that will simplify and automate the design and implementation of OFPEs, enabling ISA and other organizations to run them on hundreds of farms annually without increasing their costs for trial implementation and technical support.
OFPE Data Analysis Activities
Anselin, et al. (2004) used spatial econometrics to analyze yield monitor data from a full-field N strip trial in Argentina. Their principal result that the estimates of the profitability of site-specific N management depended on model specification, and that spatial econometric methods predicted profitability of site-specific management, whereas non-spatial econometric analysis did not. Ruffo, et al. (2006) estimated yield response functions using data from early OFPEs on eight Illinois cornfields, and analyzed the data using geospatial techniques. They concluded that terrain attributes could be used as surrogates for soil water content, and that results of the Illinois Soil Nitrogen Test could be used to proxy soil mineralizable nitrogen in estimated yield response to nitrogen fertilizer. Bullock, et al. (2009) also used OFPE data from Illinois fields to measure the value of information in nitrogen management. They concluded that, excluding consideration of the costs of precision technology equipment, variable rate nitrogen application would have been profitable on six of the eight fields on which the OFPEs were conducted.
Trevisan, Bullock, and Martín (2019) demonstrated the intriguing potential of geographically weighted regression in the analysis of on-farm precision experiment data. The method proposed revealed the spatial variability in corn response to nitrogen and seed rate in four corn fields in Illinois, USA. The within-field variation in the economically optimum seed and nitrogen rates varied from 10 to 20 kseeds ha-1 and from 20 to 60 kg ha-1, respectively. Applying the knowledge obtained with the OFPE results showed potential guiding profitable input reduction and yield improvement. Whether the greatest benefits would come from adjusting the field’s uniform rate or from accounting for the spatial variability in crop response and implementing variable rate application strategies was field-specific.
Piepho, et al. (2017) review statistical issues arising in the design and analysis of on-farm experiments. They outline basic design principles, and consider both the classical set-up with either a single datum per randomization unit (plot) or subsampled data that are not geo-referenced, as well as the case of within-plot, geo-referenced data as arising in precision farming. They emphasize the importance of replicating trials in several environments, focusing on the implications for both design and analysis. Two OFPE case studies illustrate the concepts they review.
Kindred, et al. (2017) conduct OFPE experiments to examine variation in N requirements, and to develop and test systems for its prediction and predictability. Results demonstrated substantial within-field variability in fertilizer N requirements, typically greater than150 kg per ha. But only increases in profitability were modest (compared to the uniform average rate).
Activities Involving Development of the Analytical Engine
Considerable progress has been made in the fields of machine learning (especially deep learning) and real-time control (especially in the context of autonomous vehicles). From a computational standpoint, issues related to high-dimensional, noisy, dynamic optimization remain (Duan et al, 2016; Ho & Ermon, 2016). Within the context of precision agriculture in general, and data intensive agriculture in particular, fundamental problems to be addressed in this project range from crop and weed identification to field characterization to yield prediction and beyond. The work proposed here approaches these problems computationally by expanding upon new methods in multi-task reinforcement learning under uncertainty, transfer learning capitalizing on models trained from different fields, and high-dimensional, multi-/many-objective optimization (Li et al, 2009; van Moffaert & Nowe, 2014; Liu et al, 2015; Omishafiei et al., 2017; Devin et al. 2017).
To date, the vast majority of prior work done has either focused on limited experiments over limited time frames and fields, or has attempted to create models aggregating over a variety of field and crop types (Kaul et al., 2005; Panda et al., 2010; You et al., 2017). At the heart of these problems is scalability – how well can we develop methods and algorithms to scale to large feature spaces and large state spaces, especially in the presence of uncertain and dynamically changing environments? Furthermore, how well can we adapt our methods to exploit rather than ignore the unique characteristics of the specific farms and even plots on the farms being managed? Ultimately, the advances in the algorithms developed or expanded here will contribute to the larger goal by being integrated together to enable closed-loop operations in support of effective farm management.
Field and Plant Characterization
Over the past two decades, tremendous have been put into research on measuring and interpreting how field and plant characteristics affect yield response to managed inputs. But because this kind of c data is highly complemented by the x and y data that OFPEs generate, using these data together is far more valuable than using them apart. The proposed multistate project will indeed work with the latest technology for measurement and mapping of field characteristics. But the main part of the research that makes it differ from other projects measuring and mapping field characteristics is that it will combine the use of c data with its x and y data in the analyses.
A major barrier to the adoption of more sustainable agricultural techniques is the shortage of labor needed to conduct intricate and skilled tasks for maintaining and researching the systems. Automating large equipment only solves a part of the problem because of issues with soil-compaction and challenges in ensuring safety of large unmonitored equipment. As such, small robot teams are a promising technology for these complex systems, with future developments leading to fundamental theoretical advances directly impacting dexterity of manipulation through soft, compliant, and continuum robots. These technological advances have the potential to make sustainable agricultural practices more feasible for individual growers, while also enabling practicing engineers in the industry to get involved.
Unmanned Aircraft Systems
There has been a pressing need to advance field-based plant phenotyping capabilities to match currently available genotyping tools. Recent advances in UAS and sensor technology have made it possible to accurately assess overall crop growth and health status with fine spatial and high temporal resolutions previously unobtainable. These platforms enable fast and accurate data collection throughout the growing season. UAS-based remote sensing technologies are becoming an important tool for agriculture scientists (Awika et al., 2019; Jung et al., 2018) and growers (Anderson et al., 2019; Pugh et al., 2018; Enciso et al., 2018) as they allow more affordable, efficient evaluation of the performance of many experimental treatments under varyinf field conditions (Akash et al., 2019; Yeom et al., 2018; Pugh et al., 2017; Chang et al., 2017).
- Our first supporting objective is to lay the research groundwork needed to support an on-farm precision experimentation (OFPE) infrastructure, sufficiently automated to be scaled up to enable
Comments: 1) the running of tens of thousands of OFPEs per year, worldwide; 2) the collection of field characteristics and weather data on the OFPE fields; 3) the processing and management of the resultant data, which will make it possible to apply advanced statistical and artificial intelligence methodologies to data analysis, thereby 4) providing farmers with management recommendations based on data from their own fields, as well as on data from other fields, worldwide. The resultant increase in input use efficiency will not only 5) increase farm income, but also 6) enhance the nation’s water quality, as more fertilizers enter plants as building blocks to growth and production instead of waterways (Martinez-Feria et al., 2018; Puntel et al., 2016).
- The second supporting objective is to lay the research groundwork necessary to begin the co-revolutionization of the U.S. land grant university extension system and the private crop consulting industry.
Comments: In the new system, extension educators will receive concentrated quantitative and computational training, to enable them to work with crop consultants, providing advice and guidance as those consultants working with their farmer-clients to conduct OFPEs, and together use the cloud-based “decision tool” software developed by the research of the proposed project to conduct truly data-based farm management.
In this section, we describe the methods to be followed to satisfy the principle objective of the proposed multistate project, which is to lay the research groundwork needed to support an on-farm precision experimentation (OFPE) infrastructure, sufficiently automated to be scaled up to enable
1) the running of tens of thousands of OFPEs per year, worldwide;
2) the efficient collection of field characteristics and weather data on the OFPE fields;
3) the processing and management of the resultant data, which will make it possible to apply advanced statistical and artificial intelligence methodologies to data analysis, thereby
4) providing farmers with management recommendations based on data from their own fields, as well as on data from other fields, worldwide.
How On-Farm Precision Field Trials Are Run, and the Data They Provide
Figure 1 illustrates how OFPEs are designed and implemented, and why the data are useful. The left-hand panel depicts the design of an actual field trial recently conducted with a participating farmer on a 37-ha Illinois cornfield. The trial was designed using the DIFMR R software package, following rigorous spatial-statistical principles. The software provided spatially-specific input application “instructions” to GPS-linked variable rate application equipment to apply an input at different randomized rates on the different plots (grid cells). The trial design shown in figure 1 was comprised of 240 plots, each of size approximately 0.115 ha, with dimensions 64 m x 18.5 m (the width of the applicator and combine head). Most importantly, because the machinery’s actual application rates were changed by an on-board computer and GPS-linked variable rate equipment as the machinery moved through the field, the implementation of the trial was simple and of little bother to the farmer. For example, the center panel of figure 1 shows a farmer implementing an N-rate trial while simply driving his equipment through the field in the usual manner. Yield data is gathered at harvest, using a GPS-linked yield monitor. As is shown in the right-hand panel of figure 1, this new experimental design and implementation technique allows very labor-extensive, inexpensive generation of large amounts of varied x data, and accompanying y data.
Methods Used for Scaling up OFPE
In 2019, approximately one hundred OFPEs were conducted around the world, mostly by the Data-Intensive Farm Management Project. While the current methodology provides excellent data at low costs, significant work from professional researchers is required for trial design and implementation, data processing and analysis, and communicating with farmers about the management implications of the analytical results. While current methods serve well if the objective is simply to publish academic journal articles, running thousands of trials annually will require increased automation of current methods. The proposed multistate project aims to lay research groundwork needed to push forward this automation. While the scaling-up of OFPE will require the development of new software programs that decrease the labor and expertise needed for trial design, data gathering, data processing, data analysis, and communication with farmers, it is not the expressed purpose of the proposed project to design that software; rather, the purpose is to conduct the research need to develop the analytical engine to which the software gives practical access. The goal is that said engine will allow consultants to provide their clients with more profitable input management advice, which will give private markets incentive go develop the needed software. To provide the intellectual foundation for a new OFPE infrastructure, the following activities will be necessary:
Participating On-Farm Research Networks Conduct OFPEs
The more x, y, c, and z data available, the more researchers will be able to learn about yield response, and therefore the better will be the input management advice that is ultimately given to farmers. Many members of the proposed project will not initially have experience in running OFPEs, though they have significant experience conducting other kinds of on-farm research. Therefore, some of the initial activities of the project will involve discussions and training members in OFPE methods.
During January and February of 2020, representatives from the following “farmer networks” will hold remote meetings to begin planning OFPEs to be run in 2020, with hopes and in anticipation of approval of the multistate project in the spring. In those remote meetings, preliminary plans will be made about the which research networks will run how many OFPEs. The project will start small, running approximately forty OFPEs in its first year. Provided time and resources are adequate, the various centers hope to accomplish the following:
- The Data-Intensive Farm Management Project at will use funding from various sources to run approximately twenty corn and soybean trials in Illinois in 2020.
- The e-Fields program at The Ohio State University will switch approximately four of its planned stripped trials to OFPEs.
- The Iowa Soybean Association will conduct approximately two soybean seed rate trials in 2020.
- The Washington State University Farmers Network will conduct approximately four OFPEs in 2020, studying wheat yield and protein response to N.
- The University of Nebraska’s On-Farm Research Network will conduct approximately four OFPEs in 2020, studying corn and/or soy yield response.
- Louisiana State University’s Ag Center will conduct approximately four OFPEs in 2020, studying corn, soy, and cotton response
Other registered members of the proposed project are researchers and Extension personnel from Cornell, Kansas State, Oklahoma State, Michigan State, Minnesota, Wisconsin, Mississippi State, Purdue, and two USDA-ARS research stations in Arkansas. They will attend the Spring 2020 meeting to gather information, with the aim of finding resources to begin conducting OFPEs in 2021. Because of the inherent multidisciplinary nature of OFPE, a wide array of research specialties will be represented in the project, including crop scientists, soil scientists, agricultural engineers, agricultural economists, computer scientists, and statisticians.
Increase Efficiency in Generating and Processing Data on Field and Plant Characteristics
Great efforts are currently being made by researchers in academic and commercial settings to generate characteristics data: drones, robots, satellites, and many kinds of proximal sensors have been or are being created and used. The current situation has a major shortcoming: while there exists a huge academic literature on how to use characteristics data to establish field “management zones” (e.g., De Caires, et al. 2015), there is very little discussion on how to manage said management zones. The multistate project will focus on not just gathering characteristics data, but learning if and how that data is most useful. Members will also work together to developing a common database structure, creating methods of cleaning, processing, and then “stacking and packing” the spatial characteristics data into common resolutions. The multistate project will organize the collection of the following types of field and plant characteristics data:
Efficient and accurate methods of measuring spatial variations in soil properties are important for precision agriculture and digital soil mapping. Mobile sensor systems that can collect dense datasets in situ provide several advantages over traditional measurement methods that involve soil sample collection, removal from the field, and lab analysis. Apparent soil electrical conductivity (EC), a widely used proximal soil sensing technology, is related to several important soil properties, including salinity, clay content, and bulk density (Suddoth, et al., 2013). Soil EC can serve as a proxy for soil physical properties such as organic matter (Jaynes, et al., 1994), clay content (Williams and Hoey, 1987), and cation exchange capacity (McBride, et al., 1990). These properties have a significant effect on water and nutrient-holding capacity, which are major drivers of yield (Jaynes, et al., 1995). The relationship between soil EC and yield has been reported and quantified by others (Kitchen, et al., 1996; Fleming, et al., 1998). The proposed multistate project will encourage the collection of EC data by all members conducting OFPEs. Discussions will be held about how to process and store that data under common protocols. How EC data can be used to guide management decisions will be a frequent topic of research and conversation.
Data from Unmanned Aircraft Systems.
There has been a pressing need to advance field-based plant phenotyping capabilities to match currently available genotyping tools. Recent advances in UAS (Unmanned Aircraft Systems) and sensor technology have made it possible to accurately assess overall crop growth and health status with fine spatial and high temporal resolutions previously unobtainable from traditional remote sensing platforms. These advanced platforms enable fast and accurate data collection throughout the growing season. UAS-based remote sensing technologies are becoming an important tool for agriculture scientists (Awika et al., 2019; Jung et al., 2018) and growers (Anderson et al., 2019; Pugh et al., 2018; Enciso et al., 2018) as they provide the capability to efficiently evaluate the performance of many experimental treatments under field conditions (Akash et al., 2019; Yeom et al., 2018; Pugh et al., 2017; Chang et al., 2017) at a relatively low cost. UAS based remote sensing platforms are expected to revolutionize on-farm research by providing means to collect fine spatial and high temporal resolution data. The data will be used to decipher complex interactions between genotypes and environments at field scales, which will eventually result in developing best farm management practices to increase input management efficiency. The proposed multistate project will encourage the collection of UAS data by all members conducting OFPEs. Discussions will be held about how to process and store that data under common protocols. How UAS data can be used to guide management decisions will be a frequent topic of research and conversation
Data from Robots
Collectively measuring different aspects of the plant, soil, and the surrounding microclimate is the key challenge facing scientists working to improve the yield and sustainability of productive agroforestry systems. Field experiments are necessary to advance both plant breeding and agricultural production systems research because it is well known that plants grow very differently inside greenhouses. However, scientists must account for the variability in the field environment to ensure that the results from their field experiments are statistically significant. The inability to collectively measure data that can capture this variability has led to the so-called phenotyping bottleneck: The inability to accurately understand the relationship between crop performance and crop genetics as a function of the plant’s local environment. This critical gap is significantly hampering agroforestry productivity by limiting yield potential, delaying detection of stressors, and precluding accurate predictions. But robotics is beginning to bridge said gap enabling plant phenotyping, including stem counts using side facing RGB cameras, plant height using vertically oriented LiDAR, plant width using RGB and side-facing LiDAR, and Leaf-Area index using vertically facing fish-eye cameras. A machine learning pipeline that could correlate visual data with data from other sensors could be developed. The proposed multistate project will encourage the collection of “agbot” data by all members conducting OFPEs. Discussions will be held about how to process and store that data under common protocols. How agbot data can be used to guide management decisions will be a frequent topic of research and conversation.
Publicly Available Remote Sensing Data
A great deal of remote sensing data is currently available, free of charge, for the purposes of academic research: Sentinel 2 satellite data (European Space Agency 2019) (in particular, vegetative index data that is now frequently used by agricultural scientists attempting to identify precision agricultural “management zones” (Mulla 2013)), LiDAR digital elevation model data (United States Geographic Survey 2019), USGS-POLARIS probabilistic soil data (Chaney et al. 2016), and WSS SSURGO soil-type data (United States Department of Agriculture 2019). Combining this publicly available agricultural “Big Data” with the (x, y) data from OFPEs increases the value of both types of data, leading to greater understanding of why yield responds differently to managed inputs at different locations. Without (x, y), c, and zdata, the yield response y = f(x,c,z) cannot be estimated, and this estimation is the key to data-intensive farm management. Discussions will be held about how to process and store this publicly available remote sensing data under common protocols. How said data can be used to guide management decisions will be a frequent topic of research and conversation.
Establish Consistent Protocols
Especially in the early years of the proposed multistate project, it will also be important to discuss and use common OFPE protocols, such as establishing consistent data variable names and units, and consistency (though not necessarily complete uniformity) in trials design, data cleaning and processing, and spatial data resolutions. The proposed multistate project will not aim to make all the resultant research data available to all participants for analysis. But it will aim to build a highly secure, password-protected common dataset, with which researchers who generate data can grant access to that data to others, if desired. Creating such a dataset, with agreed on units, variable names, and protocols will greatly increase research efficiency over the long-run by allowing new OFPE projects to go from raw data collection to cleaned data analysis in relatively simple, well-established steps, instead of having to independently solve the same data cleaning and management problems that earlier OFPE projects have investigated and dealt with.
Enhance Soil-Crop Model Performance and Utility.
Process-based models offer a way to understand the underlying crop and soil processes driving yield and environmental outcomes, but this comes at the cost of extensive input data requirements (Basso et al., 2012; Puntel et al., 2018, refs). Successful yield forecasting approaches using crop models (Morell et al., 2016; Togliatti et al., 2017: Carberry et al., 2009) have shown potential to couple explanatory with predictive power to evaluate and deliver adaptive management strategies (Jones et al., 2017). We will enhance soil-crop model performance and utility within a cyber-infrastructure by using field-specific information, derived from the OFPEs, that generate high-resolution spatio-temporal data on crop response to field conditions and management. Although the algorithms behind crop models continue to be refined to better represent biophysical processes, knowledge gaps still limit their use (Rötter et al., 2018; Li et al., 2019). Continuous development and testing of field-scale agronomic crop models with OFPE data can therefore improve our understanding of fundamental science, identify gaps in model function, and improve current predictive methods to better account for extreme events (Peng et al., 2018). This research will use the Adapt-N and APSIM models, which are deterministic-stochastic models that simulate nitrogen in the soil-crop continuum. Adapt-N (Melkonian et al. 2008) was launched in 2008 and is already used by producers and researchers in commercial settings in the US and Canada, and APSIM (the Agricultural Production Systems sIMulator; Holzworth et al., 2014)) has been widely used by researchers in numerous countries and commercial applications (Yield Prophet). Both models have been extensively validated in controlled trials, including long-term studies, and on-farm strip trials (Sogbedji et al., 2006; Marjerison et al., 2016; Sela et al., 2017; Puntel et al., 2016).
Start up and Push forward a Literature on Statistical Analysis of OFPE Data.
There are many methods of regression analysis that might be applied to OFPE data to learn about yield response. Without doubt, an active literature will come about over coming decades examining and debating about which statistical methods work best in which situations. One important activity or the proposed product will be to start up that research, and that research discussion. We envision project meetings to feature presentations of statistical research, and discussions about that research. That research will eventually go into journals articles, and create and press forward a new lite4rature on statistical analysis and OFPEs. As more is learned about which analytical methods work best in which situations, it will be easier to design more automated statistical analysis techniques, which will be the primary inputs used for private companies and/or university research to begin to develop automated research analysis software.
Start up and Push forward a Literature on Machine Learning and OFPE Design
Machine learning is a broad term covering many types of algorithms (Coble et al. 2018), but at this time, their application to agricultural production contexts for decision making still remains rudimentary (Mishra et al. 2016; Shine et al. 2018; Sun et al. 2017; Ip et al. 2018). One impediment to the conduct of OFPEs is that is that they may involve short-run profit losses from either yield losses from substantially low input use for some treatments, and high costs for substantially high input use for others. Machine learning algorithms offer the opportunity to increase the efficiency of on-farm experimentation by targeting placement of “plots” within fields based on grid/zone sampling data. The traditional randomization and replication used for placement of small plot experiments is essentially a random “brute force” approach to identify the production function, while a more targeted approach can acquire the same information with fewer plots (Saikai and Mitchell 2018). Fewer plots would further reduce the cost of on-farm experimentation. Automated, easy-to-use, and low-cost systems are key to achieving wider adoption of on-farm experimentation to locally optimize crop management (Griffin 2018). In the proposed multistate project, researchers specializing in machine learning will work together to improve the efficiency of OFPE trial designs, through discussion, collaboration, and critique of each other’s reported research
Measurement of Progress and Results
- Data Comments: Data. The proposed project will generate field trial data from hundreds of OFPEs. As OFPEs catch on in the private sector over time, thousands of annual field trials are foreseeable. The raw data from a typical OFPE will include tens of thousands of “as-applied”, and/or “as-planted” observations, and tens of thousands of yield observations. Those OFPEs will be designed to estimate yield response to inputs and other factors of production, including one or more of the following: randomized or stratified seed rate, N fertilizer rate, and P fertilizer rate data, for several crops, including corn, soy, cotton, barley, and various kinds of wheat. The proposed project will also help create the cyber-infrastructure needed to gather, process, and “stack and pack” many kinds of field characteristics data. The success of the data generation will be measured in numbers of OFPEs conducted and analyzed. The project goal is to generate data from 400 OFPEs to be run over five years.
- The Intellectual Foundation for a Looped OFPE Cyber-infrastructure Comments: The proposed project will provide the intellectual foundation needed for the development of the looped OFPE infrastructure shown in figure 2. This foundation will facilitate the development of the infrastructure itself both in academia and in the private sector. The infrastructure will be made up of the following components, which will be either direct outputs of the project, or outputs developed in the private sector, but dependent on the proposed project’s research. Success in generating this infrastructure will be measured by numbers of consultants and farmers that use it to design OFPEs and develop data-intensive management strategies. The project goal after five years is for 300 farmers to be working with their consultants to manage their farmers data-intensively, using OFPEs. Component A: “Front End” Portal. Panel A in Figure 2 depicts a farmer and a consultant working at a laptop computer. Farmers and their consultants will have access, through a user-friendly “front end” portal, to a cloud-based software system that bookends the infrastructure’s loop. Component B: An Automated OFPE System. The portal will provide access to OFPE design software, allowing consultants/farmers to design and implement statistically sound field trials. For each field in an OFPE, the system will request basic information from the farmer/consultant, and use it to build a trial design, in the form of a polygonized geospatial file, assigning randomized input application rates to the “plots” of the trial design. Component C: Weather Data System. The proposed project will provide the intellectual foundation for a data system to store and manipulate daily weather data using the DAYMET platform. Component D: Characteristics Data System The cyber-infrastructure will provide a secure, centralized data system, in which researchers, consultants, and clients can store, access, and manipulate their OFPE characteristics data in proven, efficient ways. That data will be encrypted, and only accessible to those who own it. Component E: Automated Data Processing, Analysis and Management Recommendation System OFPE researchers currently process data one field at a time. But to move from a system of dozens of annual trials to one facilitating thousands of annual trials, the proposed project will lay in intellectual foundation for a (mostly) automated data processing system. Currently, OFPE data is analyzed trial-by-trial, with great care. But to scale up the infrastructure, the project will lay the intellectual foundation for the development of a (mostly) automated analytical engine, to provide practical, profitable management advice and data interpretation to farmers and their consultants through the front-end decision tool. Component F: Decision Tool. Panel F of figure 2 shows the end of the cyber-infrastructure’s loop, where the user interface gives the farmer/consultant access to input management “decision tool” software. Many commercial agricultural enterprises currently offer decision-tool software, but their effectiveness could be greatly enhanced by the development of the OFPE infrastructure. Component G: OFPE Community Discussion Forum. Area G of figure 2 shows how the web-based OFPE forum joins all OFPE participants in discussions about trial design and implementation, and the practical management implications of the data and resulting analyses. The forum will involve a global cyber-environment including all stakeholders, from university professors working in areas of complex mathematical observation to the farmers dealing with malfunctioning variable rate application machinery.
- The Beginnings of a Revolutionized System of Public Outreach Comments: The current model of Extension education has Extension educators working directly with farmers, with little participation of the private sector. The proposed project will provide the intellectual capital needed to involve the private crop-consulting sector. Extension educators will focus more on training private consultants to use a cloud-based software system to work with their farmer-clients to both design OFPEs, put them “in the ground,” gather the field trial data, and analyze that data with cloud-based “decision tool” software, the “guts” of which are the analytical algorithms and methods developed by the proposed project.
Outcomes or Projected Impacts
- Increased Farm Income from More Efficient Crop Input Management The proposed project will increase the efficiency of crop input management, enabling increased profitable use of site-specific crop input management. The economic benefits from precision farming come because a ‘‘compromise’’ in different areas’ input application rates need not then be made (Bullock and Bullock 2000). This increased efficiency, whether coming from lower costs of input application, higher yields, or both, increase farm income from their activities in private markets, but decrease the political incentives for government direct and indirect subsidization of agriculture, lowering taxpayer costs, and government debt.
- Improved Water Quality from More Efficient Crop Input Management A major purpose of agricultural research, education, and outreach is to increase the efficiency of input application management (“to grow two blades of grass, where only one grew before”). The most dramatic example of the need for increased input use efficiency is presented by the current environmental damage from inefficient nitrogen (N) fertilizer application. Society manipulates the N cycle to great benefit, particularly for food and agricultural systems, but chronic inefficient use of N fertilizer has led to the hypoxic “dead zone” in the Gulf of Mexico and the leaching of nitrates into groundwater (Khanna, et al. 2019). The National Academy of Engineering (2019) has declared managing the nitrogen cycle a “Grand Challenge.” The proposed project will enable farmers to fertilize more where needed, and less where not needed; increasingly, N fertilizer will go into building plant structure, rather than the Gulf of Mexico, Puget Sound, Chesapeake Bay, the Great Lakes, and drinking water.
- High-skill Job Opportunities in Rural Economies The improvement in farm income and water quality discussed above needs to be achieved without neglecting the health of rural economies, which are already under considerable stress. Ideally, the benefits of better N management would be achieved in ways that encourage high-skilled job growth in rural areas. The system described here will provide high-skill job opportunities in rural communities, especially in the crop consulting industry, strengthening local economies. Numerous workshops, seminars, and symposia will communicate findings to a broad audience of academics and industry professionals, and will provide abundant training to further the advances in digital agriculture, including initiatives focusing on underrepresented farmer groups.
- High Social Returns from Increased Public Support for Agricultural Research It is well established by empirical research that there are high social returns to public agricultural research (Hurley, et al., 2014; Jin and Huffman 2016). The proposed project will involve farmers directly in research, changing the dominate research paradigm from one in which professional researchers run small-plot trials to gather data to one in which farmers and their consultants are the driving force behind agronomic research, which professional researchers “supervise” the process by providing the intellectual tools that underlay the on-farm research. This new system of public participation will not only increase the amount of agronomic research conducted, but also increase public support for government policies that encourage and support agronomic research.
- Improved Public Policy Analysis due to Improved Crop Growth Models Citing a long line of recent research, Antle (2019) concluded that crop growth models “are in need of significant improvement before they are capable of being sufficiently reliable for use to predict management practices on a field.” The principal problem is that crop growth models do not “move well.” To perform well, crop growth model coefficients must be calibrated to consider local conditions. But, in general, sufficient data do not exist for proper coefficient calibration. The proposed project will lead to the generation of exactly the kind of data needed. Since the OFPEs will be run in many locations over many years, they will generate the data needed to “move” crop growth models so that they accurately predict how input management, field characteristics, and weather affect crop production, improving scientists’ ability to model the effects of many events on agricultural supply, increasing the accuracy of the many kinds of public models of agricultural policy, climate change, and other events. This will enable better public decision making about how to deal with these large-scale public challenges.
(2020):Timeline During entire project duration: Conduct research to expand and improve automated data processing system, database management system, automated analysis engine. Work with agribusiness to incorporate algorithms into decision tool software. Very early versions of the software are planned to be ready by early 2021. Every year following, improvements are made. By 2024, the software and the system are capable of handling thousands of OFPEs per year. Every March and every November-December: Extension educators develop training programs, and train consultants to work with their farmer-clients to run OFPEs, and to use available software to draw inferences about the management implications of the data analyses. January-March 2020. Hold remote meetings to begin organization of 2020 activities, in anticipation of project approval. Participating farmers for the 2020 growing season will be recruited during this time. April 2020. First multistate project meeting to plan field trials. Two days of field trial design training, working together to complete field trial design shapefiles for the fast-approaching crop season. Growing Season, 2020. OFPEs conducted, including planting, fertilization, and harvests. Plant phenotype data collected using UAS and robots. Publicly available vegetative index data from satellites is collected. November-December 2020. Electroconductivity data generated. Extension educators train crop February 2021-2024. Second through Fifth All-Project meetings. Present, discuss research on analyses of earlier OFPEs. Discuss successes and challenges of project in previous year. Growing Seasons, 2021-2024. OFPEs conducted, including planting, fertilization, and harvests. Plant phenotype data collected using UAS and robots. Publicly available vegetative index data from satellites is collected.
View Appendix E: Participation
We envision a world in which farmers run OFPEs on parts of every field every few years, similar to today’s soil sampling and chemical analyses. This will depend on a public education and outreach infrastructure to assist crop consultants and professional agronomists in running OFPEs, handling the data, and conducting statistical analyses. Of course, it will not be feasible to provide every crop consultant with advanced courses in data management and analysis. We envision public outreach that contains one more layer than does the traditional land grand university Extension framework, in which Extension Educators in many ways “compete” with commercial crop consultants and crop input suppliers in providing input management advice directly to farmers. Rather, Extension Educators will rely more on the private sector to reach out directly to farmers, and focus instead on educating commercial crop consultants, agronomists, and privately employed agronomists about how to use the OFPE infrastructure to the greatest advantage of their farmer-clients.
The public outreach objective of the proposed multistate will be to develop a program to teach professional crop consultants and agronomists to work with farmers at the beginning and end of the looped infrastructure illustrated by figure 2 in the Methods section. Panel A of that figure shows consultants using cloud-based software to work with farmers to design and conduct OFPEs. The DIFM project has designed early versions of the field trial design and implementation software. Extension Educators participating in the proposed multistate project will work together to develop common training session materials and presentations, to use the software in its current and later versions. Panel G also shows crop consultants using “decision tool” software with their farmer-clients, using “decision tool” software. A good deal of commercial “decision tool” software is currently available and used by crop consultants. But existing decision tools are based on data-extensive and intellectually dated analytical algorithms. The proposed project will develop data-intensive analytical algorithms, to serve as the “guts” of new commercial software packages, which will depend heavily on OFPE data.
Multistate NC_temp1210 “Frontiers in On-Farm Experimentation”
Organization and Governance
Project Management Committee.
The Project Management Committee will be made up of the Chair, Vice-Chair, and Secretary. Each will serve a 5-year term. The Chair will be in charge of the general progress of the project, delegating responsibilities for meeting organization, logistics, and communication. The Vice Chair will assist the Chair, and fill the Chair position in the events of retirement or absence of the Chair. The Secretary will fill the Vice-Chair if needed, and is responsible for keeping minutes and organizing the writing of reports. Should the Secretary retire from office, the Chair will appoint a new Secretary from project membership, in consultation with the Vice Chair. The Project Management Committee will meet in person immediately before or after the annual project meeting, and will stay in regular contact via email and remote meetings. The Project Management Committee will be responsible for organizing the various research efforts of group members, working to make members’ research efforts complementary, where feasible. The Project Management Committee will also be responsible for organizing group efforts in obtaining outside research funding.
The project will be organized into five groups: the Field Trial Design Group, the Field Trial Implementation Group, the Characteristics Mapping Group, and the Empirical Analysis Group and the OFPE Systems Group. A project member will be welcome to participate in one or more groups. Each group will have a Group Leader, who will serve a two-year term. Group leaders will be appointed by the Project Management Committee.
The Field Trial Design Group will be responsible for writing and using computer programs that produce OFPE trial designs, and into investigating the effectiveness of different design types.
The Field Trial Implementation and Farmer Communications Group will be made up of research and extension personnel who work with participating farmers to actually put the OFPEs “in the ground.” Representatives from several existing on-farm research groups will come together discuss and debate challenges in implementing OFPEs, receiving data from farm machinery, and working with participating farmers. The project’s outreach activities will be conducted under the purview of this group, with members collaborating in the formation and organization of common outreach methods and software.
The Characteristics Mapping Group will consist of researchers who develop and/or work with proximal sensors, remote sensors, and robots to collect field characteristics and plant phenotype data. This group will also be in charge of bringing weather data into the system.
The Empirical Analysis Group will consist of those whose research involves applying statistical and machine-learning techniques to estimate yield response with OFPE, field characteristics, and weather data. Members of the group will support each other, and no doubt sometimes critique each other, in both applied yield response estimation, and in theoretical analytical research.
The OFPE Systems Group will be responsible for making the different components of the system work together efficiently. One part of their job will be to develop, or at least find from other providers, and manage a central OFPE database. This group will facilitate inter-group communication and data transfer.
Ackello-Ogutu, C., Q. Paris, and W.A. Williams. 1985. “Testing a von Liebig Crop Response Function against Polynomial Specifications.” American Journal of Agricultural Economics 67 (4): 873–80. https://doi.org/10.2307/1241829.
Anderson, S.L., S.C. Murray, L. Malambo, C. Ratcliff, S. Popescu, D. Cope, A. Chang, J. Jung, and J.A. Thomasson. 2019. “Prediction of Maize Grain Yield before Maturity Using Improved Temporal Height Estimates of Unmanned Aerial Systems.” The Plant Phenome 2 (1). https://doi.org/10.2135/tppj2019.02.0004.
Anselin, L., R. Bongiovanni, and J. Lowenberg-DeBoer. 2004. “A Spatial Econometric Approach to the Economics of Site-Specific Nitrogen Management in Corn Production.” American Journal of Agricultural Economics 86 (3): 675–87.https://doi.org/10.1111/j.0002-9092.2004.00610.x.
Antle, J.M. “Data, Economics and Computational Agricultural Science.” 2018 Fellows AAEA Fellows Address. Washington, DC, August 7, 2018.
Ashapure, A., J. Jung, J. Yeom, A. Chang, M. Maeda, A. Maeda, and J. Landivar. 2019. “A Novel Framework to Detect Conventional Tillage and No-Tillage Cropping System Effect on Cotton Growth and Development Using Multi-Temporal UAS Data.” ISPRS Journal of Photogrammetry and Remote Sensing 152 (June): 49–64. https://doi.org/10.1016/j.isprsjprs.2019.04.003.
Awika, H.O., R. Bedre, J. Yeom, T.G. Marconi, J. Enciso, K.K. Mandadi, J. Jung, and C.A. Avila. 2019. “Developing Growth-Associated Molecular Markers Via High-Throughput Phenotyping in Spinach.” The Plant Genome 12 (3). https://doi.org/10.3835/plantgenome2019.03.0027.
Bullock, D.G., and D.S. Bullock. 1994. “Quadratic and Quadratic-Plus-Plateau Models for Predicting Optimal Nitrogen Rate of Corn: A Comparison.” Agronomy Journal 86 (1): 191–95. https://doi.org/10.2134/agronj1994.00021962008600010033x.
Bullock, D.S., Matías L.R, D.G. Bullock, and G.A. Bollero. 2009. “The Value of Variable Rate Technology: An Information-Theoretic Approach.” American Journal of Agricultural Economics 91 (1): 209–23. https://doi.org/10.1111/j.1467-8276.2008.01157.x.
Cerrato, M. E., and A. M. Blackmer. 1990. “Comparison of Models for Describing; Corn Yield Response to Nitrogen Fertilizer.” Agronomy Journal 82 (1): 138–43. https://doi.org/10.2134/agronj1990.00021962008200010030x.
Chang, A., J. Jung, M.M. Maeda, and J. Landivar. 2017. “Crop Height Monitoring with Digital Imagery from Unmanned Aerial System (UAS).” Computers and Electronics in Agriculture 141 (September): 232–37. https://doi.org/10.1016/j.compag.2017.07.008.
Coble, K.H., A.K. Mishra, S. Ferrell, and T. Griffin. 2018. “Big Data in Agriculture: A Challenge for the Future.” Applied Economic Perspectives and Policy 40 (1): 79–96. https://doi.org/10.1093/aepp/ppx056.
Cook, S.E., M.L. Adams and R.J. Corner. 1999. “On-farm experimentation to determine site-specific responses to variable inputs.” Precision Agriculture: Proceedings of the 4th International Conference (Part A), edited by P. C. Robert, R. H. Rust and W. E. Larson (ASA, CSSA, SSSA, Madison, Wisconsin, USA), 611–621.
Corwin, D.L., and S.M. Lesch. 2003. “Application of Soil Electrical Conductivity to Precision Agriculture.” Agronomy Journal 95 (3): 455. https://doi.org/10.2134/agronj2003.0455.
De Caires, S.A., M.N. Wuddivira, and I. Bekele. “Spatial Analysis for Management Zone Dilineation in a Huid Tropic Cocoa Plantation. Precision Agriculture 16(2015): 129-147.
“E-Fields On-Farm Research | Digital Ag.” Accessed 18 Nov. 2019. https://digitalag.osu.edu/efields.
Enciso, J., J. Jung, A. Chang and G. Cavazos. 2018. “Assessing Land Leveling Needs and Performance with Unmanned Aerial System.” Journal of Applied Remote Sensing 12 (01): 1–8. https://doi.org/10.1117/1.jrs.12.016001.
Fischer, R.A. The Design of Experiments. Edinburgh: Oliver and Boyd, 1935.
Fleming, K.L., D.G. Westphall, D.W. Wiens, L.E. Rothe, J.E. Cipra, and D.F. Heerman. 1998. Proceedings of 4th International Conference on Precision Agriculture.
Griffin, T. 2018. “Collating and Analysing Small Data to Make Big Decisions — Can It Improve Farm Productivity and Profitability?” Grains Research and Development Corporation. 2018. https://grdc.com.au/resources-and-publications/grdc-update-papers/tab-content/grdc-update-papers/2018/02/collating-and-analysing-small-data-to-make-big-decisions.
Heady, E.O., and J. Pesek. 1954. “A Fertilizer Production Surface with Specification of Economic Optima for Corn Grown on Calcareous IDA Silt Loam.” Journal of Farm Economics 36 (3): 466–82. https://doi.org/10.2307/1233014.
Heiniger, R.W., R.G. Mcbride, and D.E. Clay. 2003. “Using Soil Electrical Conductivity to Improve Nutrient Management.” Agronomy Journal 95: 508–19.
Hoselton, G.S.W., M.A. Boerngen, and D.S. Bullock. In Preparation. “Farmers’ Awareness of and Concerns about Nutrient Loss.” Journal of Soil and Water Conservation.
Hoselton, G.S.W., M.A. Boerngen, and D.S. Bullock. In Preparation. “Illinois Corn Farmers’ Adoption of Best Management Practices in Response to Nutrient Loss Concerns.” Journal of Environmental Quality.
Hurley, T.M., X. Rao, and P.G. Pardey. 2014. “Re-Examining the Reported Rates of Return to Food and Agricultural Research and Development.” American Journal of Agricultural Economics 96 (5): 1492–1504. https://doi.org/10.1093/ajae/aau047.
IEPA, IDOA, and University of Illinois Extension. 2015. Illinois Nutrient Loss Reduction Strategy. n.d. Https://Www2.Illinois.Gov/Epa/Documents/Iepa/Water-Quality/Watershed-Management/Nlrs/Nlrs-Final-Revised-083115.Pdf.
“Iowa Soybean Association.” 2016. Iasoybeans.Com. Accessed 18 Nov. 2019. https://www.iasoybeans.com/programs/isa-research/what-we-do/.
Ip, R.H.L., L. Ang, K.P. Seng, J.C. Broster, and J.E. Pratley. 2018. “Big Data and Machine Learning for Crop Protection.” Computers and Electronics in Agriculture 151 (August): 376–83. https://doi.org/10.1016/j.compag.2018.06.008.
Jaynes, D. B., J. M. Novak, T. B. Moorman, and C. A. Cambardella. 1994. “Estimating Herbicide Partition Coefficients from Electromagnetic Induction Measurements.” Journal of Environment Quality 24 (1): 36–41.
Jaynes, D. B., J. M. Novak, T. B. Moorman, and C. A. Cambardella. 1995. “Estimating Herbicide Partition Coefficients from Electromagnetic Induction Measurements.” Journal of Environment Quality 24 (1): 36–41. https://doi.org/10.2134/jeq1995.00472425002400010005x.
Jin, Y., and W.E. Huffman. 2015. “Measuring Public Agricultural Research and Extension and Estimating Their Impacts on Agricultural Productivity: New Insights from U.S. Evidence.” Agricultural Economics 47 (1): 15–31. https://doi.org/10.1111/agec.12206.
Jones, R.W., S.J. Rathke, D.A. Laird, and J.F. McClelland. 2013. “Real-Time Sensing of Soil Nitrate Concentration in the Parts per Million Range While the Soil Is in Motion.” Applied Spectroscopy 67 (9): 1106–10. https://doi.org/10.1366/13-07064.
Jung, J., M. Maeda, A. Chang, J. Landivar, J. Yeom, and J. McGinty. 2018. “Unmanned Aerial System Assisted Framework for the Selection of High Yielding Cotton Genotypes.” Computers and Electronics in Agriculture 152 (September): 74–81. https://doi.org/10.1016/j.compag.2018.06.051.
Kindred, D.R., R. Sylvester-Bradley, A.E. Milne, B. Marchant, D. Hatley, S.L. Kendall, S. Clarke, K. Storer, and P.M. Berry. 2017. “Spatial Variation in Nitrogen Requirements of Cereals, and Their Interpretation.” Advances in Animal Biosciences8 (2): 303–7. https://doi.org/10.1017/S2040470017001327.
Kitchen, N.R., K.A. Sudduth, and S.T. Drummond. 1996. “Mapping of Sand Deposition from 1993 Midwest Floods with Electromagnetic Induction Measurements.” J. Soil Water Conserv. 51 (4): 336–40.
Kitchen, N.R., K.A. Sudduth, and S.T. Drummond. 1996. “Mapping of Sand Deposition from 1996 Midwest Floods with Electromagnetic Induction Measurements.” J. Soil Water Conserv. 51 (4): 336–40.
Kitchen, N.R., S.T. Drummond, E.D. Lund, K.A. Sudduth, and G.W. Buchleiter. 2003. “Soil Electrical Conductivity and Topography Related to Yield for Three Contrasting Soil–Crop Systems.” Agronomy Journal 95 (3): 483–95. https://doi.org/10.2134/agronj2003.0483.
Kweon, G. 2012. “Toward the Ultimate Soil Survey: Sensing Multiple Soil and Landscape Properties in One Pass.” Agronomy Journal 104 (6): 1547–57. https://doi.org/10.2134/agronj2012.0156.
Lund E.D., C.S. Christy, and P.E. Drummond. 2000. “Using Yield and Soil Electrical Conductivity (EC) Maps to Derive Crop Production Performance Information.” Retrieved on January 5, 2004 from Https://Pdfs.Semanticscholar.Org/3dd6/F8241781d07ca40593ec193c16d61b05c6a5.Pdf?_ga=2.41031266.1997047317.1574390094-134168074.1574390094.
Makowski, D., and D. Wallach. 2002. “It Pays to Base Parameter Estimation on a Realistic Description of Model Errors.” Agronomie 22 (2): 179–89. https://doi.org/10.1051/agro:2002002.
Marenya, P.P., and C.B. Barrett. 2009. “State-Conditional Fertilizer Yield Response on Western Kenyan Farms.” American Journal of Agricultural Economics 91 (4): 991–1006. https://doi.org/10.1111/j.1467-8276.2009.01313.x.
Marks, B., and M.A. Boerngen. 2019. “A Farming Community’s Perspective on Nutrient Loss Reduction.” Agricultural & Environmental Letters 4 (1): 0. https://doi.org/10.2134/ael2019.02.0004.
Martinez-Feria, Rafael A., Michael J. Castellano, Ranae N. Dietzel, Matt J. Helmers, Matt Liebman, Isaiah Huber, and Sotirios V. Archontoulis. 2018. “Linking Crop- and Soil-Based Approaches to Evaluate System Nitrogen-Use Efficiency and Tradeoffs.” Agriculture, Ecosystems & Environment 256: 131–43. https://doi.org/10.1016/j.agee.2018.01.002.
McBride, R.A., S.C. Shrive, and A.M. Gordon. 1990. “Estimating Forest Soil Quality from Terrain Measurements of Apparent Electrical Conductivity.” Soil Science Society of America Journal 54 (1): 290–93. https://doi.org/10.2136/sssaj1990.03615995005400010047x.
Mishra, S., D. Mishra, and G.H. Santra. 2016. “Applications of Machine Learning Techniques in Agricultural Crop Production: A Review Paper.” Indian Journal of Science and Technology 9 (38). https://doi.org/10.17485/ijst/2016/v9i38/95032.
Montesdeoca, K., D.S. Bullock, K. Johansen, L. Puntel, and D. Schreiber. “Data-Intensive Farm Management.” National Science Foundation Innovation Corps Program. October 2016 - April 2017.
Morris, T.F., T. Scott Murrell, D.B. Beegle, J.J. Camberato, R.B. Ferguson, J. Grove, Q. Ketterings, et al. 2018. “Strengths and Limitations of Nitrogen Rate Recommendations for Corn and Opportunities for Improvement.” Agronomy Journal 110 (1): 1. https://doi.org/10.2134/agronj2017.02.0112.
“Nebraska On-Farm Research Network.” 2015. CropWatch, Accessed 18 Nov. 2019.
Paris, Q. 1992. “The von Liebig Hypothesis.” American Journal of Agricultural Economics 74 (4): 1019. https://doi.org/10.2307/1243200.
Paris, Q., and K. Knapp. 1989. “Estimation of von Liebig Response Functions.” American Journal of Agricultural Economics71 (1): 178. https://doi.org/10.2307/1241786.
Piepho, H.P, C. Richter, J. Spilke, K. Hartung, A. Kunick, and H. Thöle. 2011. “Statistical Aspects of On-Farm Experimentation.” Crop and Pasture Science 62 (9): 721–35. https://doi.org/10.1071/CP11175.
Pindyck, R., and D. Rubinfeld. Microeconomics. 8th edition. London: Pearson, 2013.
Proximal Soil Sensing. 2010. “Proximal Soil Sensing | Raphael A. Viscarra Rossel | Springer.” https://www.springer.com/gp/book/9789048188581.
Pugh, N.A., D.W. Horne, S.C. Murray, G. Carvalho, L. Malambo, J. Jung, A. Chang, et al. 2018. “Temporal Estimates of Crop Growth in Sorghum and Maize Breeding Enabled by Unmanned Aerial Systems.” The Plant Phenome 1 (1): 0. https://doi.org/10.2135/tppj2017.08.0006.
Pugh, N.A., X. Han, S.D. Collins, J.A. Thomasson, D. Cope, A. Chang, J. Jung, et al. 2018. “Estimation of Plant Health in a Sorghum Field Infected with Anthracnose Using a Fixed-Wing Unmanned Aerial System.” Journal of Crop Improvement 32 (6): 861–77. https://doi.org/10.1080/15427528.2018.1535462.
Puntel, L.A., J.E. Sawyer, D.W. Barker, R. Dietzel, H. Poffenbarger, M.J. Castellano, K.J. Moore, P. Thorburn, and S.V. Archontoulis. “Modeling Long-Term Corn Yield Response to Nitrogen Rate and Crop Rotation.” Frontiers in Plant Science 7 (November 2016). https://doi.org/10.3389/fpls.2016.01630.
Reimer, A., J.E. Doll, B. Basso, S.T. Marquart-Pyatt, G.P. Robertson, D. Stuart, and J. Zhao. 2016. “Moving toward Sustainable Farming Systems: Insights from Private and Public Sector Dialogues on Nitrogen Management.” Journal of Soil and Water Conservation 72 (1): 5A-9A. https://doi.org/10.2489/jswc.72.1.5a.
Robert, P.C., R.H. Rust, W.E. Larson, S. E. Cook, M. L. Adams, and R. J. Corner. 1999. “On-Farm Experimentation to Determine Site-Specific Responses to Variable Inputs.” ACSESS Publications. https://doi.org/10.2134/1999.precisionagproc4.c60.
Rodriguez, D.G.P., D.S. Bullock, and M.A. Boerngen. 2019. “The Origins, Implications, and Consequences of Yield-Based Nitrogen Fertilizer Management.” Agronomy Journal 111 (2): 725. https://doi.org/10.2134/agronj2018.07.0479.
Ruffo, M.L., G.A. Bollero, D.S. Bullock, and D.G. Bullock. 2006. “Site-Specific Production Functions for Variable Rate Corn Nitrogen Fertilization.” Precision Agriculture 7 (5): 327–42. https://doi.org/10.1007/s11119-006-9016-7.
Rund, Q. 2000. “Enhanced Farm Research Analyst: Tools for on-Farm Crop Production Research.” In ESRI Conference. Accessed November 18, 2019. Available at http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.138.3728&rep=rep1&type=pdf
Sadi S.G., Q. Paris and W.A. Williams. 1987. “A von Liebig Model for Water and Nitrogen Crop Response.” Western Journal of Agricultural Economics 12 (2): 182–92.
Saikai, Y., and P.D. Mitchell. 2019. “Bayesian Optimization for Precision Agriculture.” Presented at the 2019 Agricultural and Applied Economics Association Annual Meeting, Atlanta, GA. https://doi.org/10.22004/ag.econ.291207
Saikai, Y., P.D. Mitchell, and J. Colquhoun. 2018. “Multi-Armed Bandit for Experimental Plot Selection.” Presented at the 2018 Agricultural and Applied Economics Association Annual Meeting, Washington, D.C. https://doi.org/10.22004/ag.econ.274353
Saikai, Y., V. Patel, S.P. Conley, and P.D. Mitchell. 2019. “Adaptive Experimental Design Using Bayesian Optimization to Improve the Cost Efficiency of Small-Plot Field Trials.” Under Review at PLOS ONE.
Shine, P., M.D. Murphy, J. Upton, and T. Scully. 2018. “Machine-Learning Algorithms for Predicting on-Farm Direct Water and Electricity Consumption on Pasture Based Dairy Farms.” Computers and Electronics in Agriculture 150 (July): 74–87. https://doi.org/10.1016/j.compag.2018.03.023.
Stanford, G. 1966. “Nitrogen requirements of crops for maximum yield.” W.H. McVickar et al., editors, Agricultural anhydrous ammonia technology and use. Agricultural Ammonia Institute, Memphis, TN, ASA and SSSA, Madison, WI. p. 237-257.
Stanford, G. 1973. “Rationale for Optimum Nitrogen Fertilization in Corn Production1.” Journal of Environment Quality 2 (2): 159. https://doi.org/10.2134/jeq1973.00472425000200020001x.
Stauber, M.S., O.R. Burt, and F. Linse. 1975. “An Economic Evaluation of Nitrogen Fertilization of Grasses When Carry-over Is Significant.” American Journal of Agricultural Economics 57 (3): 463–71. https://doi.org/10.2307/1238409.
Sudduth, K.A., D.B. Myers, N.R. Kitchen, and S.T. Drummond. 2013. “Modeling Soil Electrical Conductivity–Depth Relationships with Data from Proximal and Penetrating ECa Sensors.” Geoderma 199 (May): 12–21. https://doi.org/10.1016/j.geoderma.2012.10.006.
Sun, L., Y. Yang, J. Hu, D. Porter, T. Marek, and C. Hillyer. 2017. “Reinforcement Learning Control for Water-Efficient Agricultural Irrigation.” 2017 IEEE International Symposium on Parallel and Distributed Processing with Applications and 2017 IEEE International Conference on Ubiquitous Computing and Communications (ISPA/IUCC), December. https://doi.org/10.1109/ispa/iucc.2017.00203.
Tembo, G., B.W. Brorsen, F.M. Epplin, and E. Tostão. 2008. “Crop Input Response Functions with Stochastic Plateaus.” American Journal of Agricultural Economics 90 (2): 424–34. https://doi.org/10.1111/j.1467-8276.2007.01123.x.
Trevisan, R.G., D.S. Bullock, and N.F. Martin. n.d. “Site-Specific Treatment Responses in On-Farm Precision Experimentation.” Preprints 2019, 2019020007.
Tumusiime, E., B. Wade, J. Mosali, J. Johnson, J. Locke, and J.T. Biermacher. 2011. “Determining Optimal Levels of Nitrogen Fertilizer Using Random Parameter Models.” Journal of Agricultural and Applied Economics 43 (4): 541–52. https://doi.org/10.1017/s1074070800000067.
Viscarra, R.A., V.I. Adamchuk, K.A. Sudduth, N.J. McKenzie, and C. Lobsey. 2011. “Proximal Soil Sensing: An Effective Approach for Soil Measurements in Space and Time.” Advances in Agronomy, iii. https://doi.org/10.1016/b978-0-12-386473-4.00010-5.
Waugh, D.L, R.B. Cate, L. Nelson, and International Soil Fertility Evaluation And Improvement Program. 1973. Discontinuous Models for Rapid Correlation, Interpretation, and Utilization of Soil Analysis and Fertilizer Response Data. Raleigh, N.C.: International Soil Fertility Evaluation And Improvement Program.
Williams, BG, and D. Hoey. 1987. “The Use of Electromagnetic Induction to Detect the Spatial Variability of the Salt and Clay Contents of Soils.” Soil Research 25 (1): 21–27. https://doi.org/10.1071/SR9870021.
“WSU Farmers Network.” Accessed 18 Nov. 2019. https://armersnetwork.wsu.edu/
Yeom, J., J. Jung, A. Chang, M. Maeda, and J. Landivar. 2018. “Automated Open Cotton Boll Detection for Yield Estimation Using Unmanned Aircraft Vehicle (UAV) Data.” Remote Sensing 10 (12): 1895. https://doi.org/10.3390/rs10121895.
Zuckerberg, K. (2017, May). Cracking the Code on Precision Farming and Digital Agriculture. [Blog Post] Retrieved from https://research.rabobank.com/far/en/sectors/farm-inputs/cracking-the-code-on-precision-farming-and-digital-ag.html
Land Grant Participating States/Institutions
CA, CT, IA, IL, IN, KS, LA, MI, MN, MS, MT, ND, NE, NY, OH, OK, TX, WI
Non Land Grant Participating States/Institutions
Illinois State University, Iowa Soybean Association, Purdue University, USDA-ARS