The Data‐Intensive Farm Management (DIFM) project works with participating farmers, using precision technology to inexpensively design and run randomized agronomic field trials on whole commercial farm fields, to provide data‐based, site‐specific farm input management guidance, thus providing economic and environmental benefits. This article lays out a conceptual framework used by the multidisciplinary DIFM research team to facilitate collaboration and then presents details of DIFM's procedures for what it calls on‐farm precision experimentation (OFPE), which includes field trial design and implementation, data generation, processing, and management, and analysis. It is argued that DIFM's data and the agricultural “Big Data” currently being collected with remote and proximal sensors are complementary; that is, more of either increases the value of the other. In 2019, DIFM and affiliates conducted over 120 trials, ranging from 10 to 100 ha in size, on maize, wheat, soybeans, cotton, and barley in eight US states, Argentina, Brazil, and South Africa. The DIFM is developing cyberinfrastructure to “scale up” its activities, 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 agronomists and farmers about implementing data‐driven input management decisions. The proposed framework allows researchers, agronomists, and farmers to carry out on‐farm precision experimentation using novel digital tools.Core Ideas The Data‐Intensive Farm Management project's on‐farm trials can generate massive amounts varied managed input data. The Data‐Intensive Farm Management project's data fill a gap in agricultural “Big Data,” to enable data‐intensive crop management. The Data‐Intensive Farm Management project's protocols support trial design, data processing and analysis. The Data‐Intensive Farm Management project can be implemented by researchers, consultants, and farmers in diverse agronomic scenarios.

Within-field variability of crop yield levels has been extensively investigated, but the spatial variability of crop yield responses to agronomic treatments is less understood. On-farm precision experimentation (OFPE) can be a valuable tool for the estimation of in-field variation of optimal input rates and thus improve agronomic decisions. Therefore, the objectives of this study were to investigate the spatial variability of optimal input rates in OFPE and the potential economic benefit of site-specific input management. Mixed geographically weighted regression (GWR) models were used to estimate local yield response functions. The methodology was applied to investigate the spatial variability in corn response to nitrogen and seed rates in four cornfields in Illinois, USA. The results showed that spatial heterogeneity of model parameters was significant in all four fields evaluated. On average, the RMSE of the fitted yield decreased from 1.2 Mg ha⁻¹ in the non-spatial global model to 0.7 Mg ha⁻¹ in the GWR model, and the r-squared increased from 10 to 68%. The average potential gain of using optimized uniform rates of seed and nitrogen was US$ 65.00 ha⁻¹, while the added potential gain of the site-specific application was US$ 58.00 ha⁻¹. The combination of OFPE and GWR proved to be an effective tool for testing precision agriculture’s central hypothesis of whether optimal input application rates display adequate spatial variability to justify the costs of the variable rate technology itself. The reported results encourage more research on response-based input management recommendations instead of the still widespread focus on yield-based algorithms.

Soybean maturity is a trait of critical importance for the development of new soybean cultivars, nevertheless, its characterization based on visual ratings has many challenges. Unmanned aerial vehicles (UAVs) imagery-based high-throughput phenotyping methodologies have been proposed as an alternative to the traditional visual ratings of pod senescence. However, the lack of scalable and accurate methods to extract the desired information from the images remains a significant bottleneck in breeding programs. The objective of this study was to develop an image-based high-throughput phenotyping system for evaluating soybean maturity in breeding programs. Images were acquired twice a week, starting when the earlier lines began maturation until the latest ones were mature. Two complementary convolutional neural networks (CNN) were developed to predict the maturity date. The first using a single date and the second using the five best image dates identified by the first model. The proposed CNN architecture was validated using more than 15,000 ground truth observations from five trials, including data from three growing seasons and two countries. The trained model showed good generalization capability with a root mean squared error lower than two days in four out of five trials. Four methods of estimating prediction uncertainty showed potential at identifying different sources of errors in the maturity date predictions. The architecture developed solves limitations of previous research and can be used at scale in commercial breeding programs.

Nitrogen (N) management is critical to sustain soil fertility and crop production while minimizing the negative environmental impact, but is challenging to optimize. This paper proposes an intelligent N management system using deep reinforcement learning (RL) and crop simulations with Decision Support System for Agrotechnology Transfer (DSSAT). We first formulate the N management problem as an RL problem. We then train management policies with deep Q-network and soft actor-critic algorithms, and the Gym-DSSAT interface that allows for daily interactions between the simulated crop environment and RL agents. According to the experiments on the maize crop in both Iowa and Florida in the US, our RL-trained policies outperform previous empirical methods by achieving higher or similar yield while using less fertilizers.

Crop breeding is as ancient as the invention of cultivation. In essence, the objective of crop breeding is to improve plant fitness under human cultivation conditions, making crops more productive while maintaining consistency in life cycle and quality. Predictive breeding has been demonstrated in the agricultural industry and in public breeding programs for over a decade. The massive stores of data that have been generated by industry, farmers, and scholars through several decades have finally been recognized as a potential asset that can be brought to bear on specific breeding decisions. A wide range of analytical methods that were initially developed for various other quantitative disciplines, such as machine learning, deep learning, and artificial intelligence, are now being adapted for application in crop breeding to support analytics and decision making processes. This convergence between data science and crop breeding analytics is expected to address long‐standing gaps in crop breeding analytics, and realize the potential of applying advanced analytics to multidimensional data such as geospatial variables, a multitude of phenotypic responses, and genetic information. Here, we summarize the few existing examples followed by perspectives on where else these technologies would have applications to accelerate operational aspects of crop breeding and agricultural product development efforts.