Generic crop rotation pattern-matching algorithm revealed dominant rotational systems for France

Crop rotations remain poorly documented at large spatial scale, despite their central role in agroecosystem sustainability. This work introduces a generic pattern-matching algorithm for the systematic identification and classification of crop rotations from field-level crop sequence datasets. By formalizing rotation recognition as a tunable matching problem, proposed approach achieves high recognition rates with minimal crop aggregation, enabling spatially explicit extraction of representative rotations across multiple spatial scales. Applied to mainland France, the method identified crop rotations on 90% of arable land. Four-year rotations, typically including two to three years of flexible crops, were most common. Nationally, the top 20 rotations covered 30% of arable land, and the top 52 covered 50%. A spatial typology of French agricultural districts revealed that cropping systems structure around 25 dominant rotational systems grouped into eight categories: (i) maize grain monocropping, (ii) maize grain – winter wheat rotations, (iii) sunflower – winter wheat rotations, (iv) grass-based systems, (v) maize silage – winter wheat rotations, (vi) winter wheat – barley – oilseed rape rotations, (vii) root crop-based rotations, and (viii) specialized production. Organic rotations were longer and more flexible than conventional ones, and rotations of larger farms were longer and temporally more diverse, but exhibited lower spatial diversity. This scalable, data-driven approach is readily applicable to other regions, periods and data sources, from administrative datasets to earth observation products. While the method does not assess agronomic performance or environmental impacts, it supports large-scale agroecosystem assessment and modelling, offering quantitative evidences to guide the design and evaluation of cropping systems transition strategies.