EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

A transparent crop model-driven and AI-assisted framework for agricultural decision support

Bangyou Zheng, Yusuke Kikuchi, Enli Wang, Kirsten Verburg

Climate Smart Agriculture · 2026-11 · 3 (4) · 100143

DOI: 10.1016/j.csag.2026.100143 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Climate Smart Agriculture, 2026-11. Verfasst von Bangyou Zheng und 3 weiteren Autor:innen. Die Arbeit nennt 40 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: CSIRO.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Agricultural decision-making requires evaluation of complex interactions under variable environment and management conditions. Process-based crop models are widely used to support these decisions by providing scientifically grounded analysis of management alternatives. However, their application remains constrained by technical complexity and the difficulty of translating informal decision contexts into complete and reproducible model analyses without expert knowledge. Recent advances in artificial intelligence (AI) offer opportunities to improve accessibility, but questions remain regarding transparency, reproducibility, trustworthiness, and the appropriate role of AI in scientific decision-support systems. We present Agricultural Large Language Model (AgriLLM), a transparent crop model-driven and AI-assisted framework for agricultural decision support that explicitly separates AI-assisted interpretation, deterministic decision processing, and scientific inference. A central feature of the framework is its target-driven design, in which decision objectives rather than user phrasing determine the information requirements, simulation design, and evaluation procedures needed to support decision-making. Using Agricultural Production Systems Simulator (APSIM) Next Generation as a proof-of-concept platform, we demonstrate the framework across cultivar selection, time of sowing, and nitrogen management scenarios. The results show how diverse and incomplete user inputs can be transformed into transparent and reproducible decision workflows while maintaining explicit representation of assumptions, evidence, and decision logic. The framework demonstrates how AI can be integrated with process-based crop modelling to improve accessibility and usability while preserving transparency, reproducibility, and scientific integrity in agricultural decision support.

Förderung

  • CSIRO

Zitieren

APA

Zheng, B., Kikuchi, Y., Wang, E., & Verburg, K. (2026). A transparent crop model-driven and AI-assisted framework for agricultural decision support. Climate Smart Agriculture, 3(4), 100143. https://doi.org/10.1016/j.csag.2026.100143

Vancouver

Zheng B, Kikuchi Y, Wang E, Verburg K. A transparent crop model-driven and AI-assisted framework for agricultural decision support. Climate Smart Agriculture. 2026;3(4):100143. doi:10.1016/j.csag.2026.100143

BibTeX
@article{Zheng2026,
  author = {Zheng, Bangyou and Kikuchi, Yusuke and Wang, Enli and Verburg, Kirsten},
  title = {A transparent crop model-driven and AI-assisted framework for agricultural decision support},
  journal = {Climate Smart Agriculture},
  year = {2026},
  volume = {3},
  number = {4},
  pages = {100143},
  doi = {10.1016/j.csag.2026.100143}
}
Kostenlos starten →