State spaces for agriculture: a meta-systematic design automation framework

Authors: RUNCK, BRYAN - University Of Minnesota; STREED, ADAM - Non ARS Employee; WANG, DIANE - Purdue University; Neupane, Dhurba; KANTAR, MICHAEL - University Of Hawaii; RAGHAVAN, BARATH - University Of Southern California

Submitted to: Proceedings of the National Academy of Sciences-Nexus
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 3/6/2023
Publication Date: 4/26/2023
Citation: Runck, B.C., Streed, A., Wang, D., Ewing, P.M., Kantar, M.B., Raghavan, B. 2023. State spaces for agriculture: a meta-systematic design automation framework. Proceedings of the National Academy of Sciences-Nexus. 2(4):1-8. https://doi.org/10.1093/pnasnexus/pgad084.
DOI: https://doi.org/10.1093/pnasnexus/pgad084

Interpretive Summary: Addressing emerging and acute challenges of agricultural systems requires imagining and testing innumerable combinations of crops, varieties, and management practices across multiple growing seasons in numerous locations. However, time, space, and financial resources preclude testing all combinations. Computer science has long-established modeling frameworks to explore configurations and infer outcomes in similarly poorly-defined problem. We describe how to use these frameworks to model snapshots of agricultural systems as “states.” Transitions between these states, for example from corn to soybeans, comes with benefits including crop yield and costs including inputs that vary given weather and management. Modeling these and additional transitions will allow predicting costs and benefits of moving to new, possibly unexplored agricultural systems. These represent candidates to be validated in the field. The overall framework can be applied to breeders, agronomists, and soil scientists working from the plant to the landscape scale. As a result, we expect next-generation agricultural models built with this framework will allow more rapid innovation in systems that stakeholders are willing to adopt. We expect that these systems will increase food security, improve farmer income, and rejuvenate national soil and water resources.

Technical Abstract: Agriculture is a designed system with the largest areal footprint of any human activity. In some cases, the designs within agriculture emerged over thousands of years, such as the use of rows for spatial organization of crops. In others, designs were deliberately chosen and implemented over decades as happened with the Green Revolution. Currently, a substantial amount of work in the agricultural sciences is about developing and evaluating designs that could improve agriculture’s sustainability. However, approaches to agricultural system design are diverse and fragmented, relying on individual intuition and disciplinary frameworks for how to meet the needs of stakeholders with semi-incompatible goals. This presents a risk that agricultural science will overlook non-obvious designs. Here, we introduce a state space framework for agriculture, a common approach from computer science for addressing the problem of proposing and evaluating designs computationally. This approach addresses current limitations of agricultural system design by enabling a general set of computational abstractions to explore, and then select from, a much larger agricultural design set, which can then be empirically tested.