Published: January 2027
Title: From patterns to prediction: detecting crop sequence and interpreting crop selection with explainable AI at field level in Denmark
João G. Serra, Edwin Haas, Diego Abalos, Søren Kolind Hvid, Tommy Dalgaard, Lars Uldall-Jessen, Franca Giannini-Kurina, Meshach Ojo Aderele, Bo Thiesson, Jørgen Eivind Olesen, Klaus Butterbach-Bahl, Jaber Rahimi
The paper is published by Elsevier and in it you can read:
- Field-level crop sequences across Denmark were characterised using a 10-year dataset.
- A heuristic algorithm detected recurring sequence patterns from longitudinal data.
- ML and DL models predicted annual crop choice using lagged structural and management predictors.
- Crop selection patterns were strongly associated with short-term crop sequences and farm typology.
- Pedoclimatic and management variables contributed minimally to model predictions.