KIT | KIT-Bibliothek | Impressum | Datenschutz

Unraveling the impact of environmental factors on wheat yield across the European Union via explainable machine learning

Zhou, Wei; Zhou, Wang ; Cammarano, Davide; Butterbach-Bahl, Klaus 1; Olesen, Jørgen E.; Lin, Zhixian; Huang, Tianjin; Cai, Gaochao; Zhang, Jingwen; Qiu, Jianxiu; Wang, Sheng
1 Institut für Meteorologie und Klimaforschung Atmosphärische Umweltforschung (IMKIFU), Karlsruher Institut für Technologie (KIT)

Abstract (englisch):

Accurately predicting crop yield and identifying its response to environmental drivers are critical for ensuring food security and optimizing commodities trading financial operations under climate change. Soft wheat is the most widely cultivated crop across the European Union (EU). While studies have assessed the impacts of environmental factors locally, a comprehensive continental-scale analysis of their impacts on soft wheat yield across the EU remains unexplored. To address this research gap, we evaluated the performance of six machine learning models in predicting the Nomenclature of territorial units for statistics (NUTS)-level wheat yield of the EU from 2000 to 2020 using a suite of comprehensive environmental factors, including air temperature, precipitation, solar radiation, soil moisture, and air humidity. Using explainable machine learning, we quantified the importance of these variables in explaining the spatial and temporal variations of wheat yield across the EU throughout the growing seasons. The results demonstrated that the bidirectional long short-term memory (BiLSTM) model outperformed other models with R$^2$, RMSE, NRMSE, and Bias of 0.75, 1.17 t/ha, 20.86% and −0.01 t/ha, respectively, using the Leave-One-Year-Out (LOYO) cross-validation approach. ... mehr


Originalveröffentlichung
DOI: 10.1016/j.compag.2025.111268
Zugehörige Institution(en) am KIT Institut für Meteorologie und Klimaforschung Atmosphärische Umweltforschung (IMKIFU)
Publikationstyp Zeitschriftenaufsatz
Publikationsmonat/-jahr 02.2026
Sprache Englisch
Identifikator ISSN: 0168-1699, 1872-7107
KITopen-ID: 1000189249
Erschienen in Computers and Electronics in Agriculture
Verlag Elsevier
Band 241
Seiten 111268
Vorab online veröffentlicht am 10.12.2025
Schlagwörter Crop yield, Environmental drivers, Machine learning, European Union wheat, BiLSTM, SHAP, Deep learning
Nachgewiesen in Scopus
OpenAlex
KIT – Die Universität in der Helmholtz-Gemeinschaft
KITopen Landing Page