Computational design in agriculture is the use of data‐driven systems and tools to propose and evaluate alternative configurations of agricultural systems. It is unique from digital agriculture in that it integrates computational and crop science approaches to formulate problems rather than mitigating problems by applying digital technologies. In this special issue, we highlight how computational design could be used to adapt agricultural systems to better meet societal goals more rapidly and at lower cost. Many disciplines within crop sciences are represented, from breeding to cropping systems agronomy. Using a symposium at a major scientific conference as a case study, we also demonstrate how this framing of computational design can facilitate transdisciplinary research. Critically, all participants highlighted the potential of computational design to facilitate stakeholder engagement through eliciting, formalizing, and evaluating their values and experiences. This is especially important within the grand challenge contexts of changing climates and market demands, where intuition developed in the past may break down. By leveraging the power of computational design, we can make informed decisions to create agricultural systems that maximize productivity while minimizing environmental impact under current and future environments.

Corn (Zea mays L.) yield productivity is driven by a multitude of factors, specifically genetics, environment, and management practices, along with their corresponding interactions. Despite continuous monitoring through proximal or remote sensors and advanced predictive models, understanding these complex interactions remains challenging. While predictive models are improving with regard to accurate predictions, they often fail to explain causal relationships, rendering them less interpretable than desired. Process-based or biophysical models such as the Agricultural Production Systems sIMulator (APSIM) incorporate these causalities, but the multitude of interactions are difficult to tease apart and are largely sensitive to external drivers, which often include stochastic variations. To address this limitation, we developed a novel methodology that reveals these hidden causal structures. We simulated corn production under varied conditions, including different planting dates, nitrogen fertilizer amounts, irrigation rules, soil and environmental conditions, and climate change scenarios. We then used the simulation results to rank features having the largest impact on corn yield through Random Forest modeling. The Random Forest model identified nitrogen uptake and annual transpiration as the most influential variables on corn yield, similar to the existing research. However, this analysis alone provided limited insight into how or why these features ranked highest and how the features interact with each other. Building on these results, we deployed a Causal Bayesian model, using a hybrid approach of score-based (hill climb) and constraint-based (injecting domain knowledge) models. The causal analysis provides a deeper understanding by revealing that genetics, environment, and management factors had causal impacts on nitrogen uptake and annual transpiration, which ultimately affected yield. Our methodology allows researchers and practitioners to unpack the “black box” of crop production systems, enabling more targeted and effective model development and management recommendations for optimizing corn production.

Weed management practices strive to reduce weeds, which compete with crops for nutrients, sunlight, and water and are thus important for maintaining yield. In most weed management practices, the first step is to identify or classify weeds. However, efficient identification and classification of weeds are challenging using conventional manual methods such as field visits. Therefore, in this study, we propose to classify four common weeds in corn field of North Dakota (common lambsquarters, common purslane, horseweed, and redroot pigweed), also applicable to other regions, using computer vision methods. Weeds were grown in plastic trays under natural conditions from the field soil, and images were collected using an RGB camera. For each weed species, 21 shape features were extracted through a developed ImageJ plugin. The handcrafted simple image processing approach was highly successful in distinguishing common lambsquarters and redroot pigweed from horseweed, while did not perform well with other species combinations and classifying them collectively owing to the similarity of the shapes. However, the three advanced non-parametric machine learning (ML) models, namely, k-nearest neighbor (KNN), random forest (RF), and support vector machine (SVM), resulted in high accuracies with RF outperforming the others. It is recommended that the handcrafted simple image processing algorithm should be tried first, due to its simpler nature, for the identification and classification of weeds before resorting to advanced and complex versatile ML modeling approaches. Such tools and methodologies using open-source platforms will be helpful to farmers, producers, and other stakeholders related to crop production.