简报

On March 13, 2026, Dr. Steffen Fritz, principal research scholar of the International Institute for Applied Systems Analysis (IIASA) in Austria and an international collaborator of the RESIL FARM project, visited the Institute of Agricultural Resources and Regional Planning (IARRP) of the Chinese Academy of Agricultural Sciences (CAAS) and attended the second Expert Academic Seminar of the project team. Project leader Dr. Yu Qiangyi, project task leader Professor Shi Wenjiao from the Institute of Geographic Sciences and Natural Resources Research of the Chinese Academy of Sciences (CAS), associate researcher Dr. Zhang Miao from the Aerospace Information Research Institute (AIRI) of CAS, and other project members were present at the seminar.

Centering on the prominent contradiction between farmland infrastructure development and biodiversity conservation in the research of this project, the seminar conducted in-depth exchanges on practical issues including the selection of globally representative regions, similarity matching analysis of Chinese and international farmland research sites, and the development of a sustainable resilience index for farmland systems. Three key research priorities were identified through the discussion: First, farmland infrastructure. The discussion highlighted regional issues such as high post-harvest grain loss caused by inadequate storage capacity in Africa and the game between agricultural product storage and pricing in South America, clarifying that grain storage facilities, in addition to conventional facilities such as irrigation and road systems, should be incorporated into the research scope. Second, diversified crop cultivation. Based on the integrated analysis of remote sensing and field data, the research will monitor the rotation patterns of major crops such as wheat and rice, and develop monitoring technologies for small-scale planting diversity. Third, remote sensing identification of farmland boundaries. By adopting image comparison combined with time-series remote sensing data to assist in the identification of physical and functional farmland boundaries, the project will explore the technical application of fusing empirical knowledge with machine learning, and enrich the dataset through crowdsourcing of farmland boundary delineation and open-source algorithms.

The seminar also identified three critical technical challenges: (1) Remote sensing technologies face great difficulties in identifying small farm plots and complex planting patterns; (2) Machine learning models have the "black box" problem, where irrelevant variables are prone to being misjudged as key factors and thus require verification with professional domain knowledge; (3) Conflicts between agricultural policies and actual production practices exist in both China and other countries—for instance, restrictions on certain types of cultivation in the construction of well-facilitated farmland in China, and the reluctance of some European farmers to switch to conservation agriculture.

Meanwhile, the project team discussed and formulated the follow-up work plan: launch a crowdsourcing campaign for farmland boundary delineation, improve data digitalization by combining open-source algorithms, and optimize technologies for farmland identification and plot size estimation; further strengthen global cross-regional comparative analysis to explore synergistic pathways between green infrastructure and agricultural production; refine model development, compare the contribution of variables across different algorithms, and enhance the interpretability and accuracy of the models; intensify collaboration with the international research team, share research data and methodologies, quantify the global regional differences in farmland infrastructure development through interdisciplinary approaches, and address the core challenge of balancing efficiency improvement and ecological conservation.