XGBoost modeling of climate impact on corn yield under varying fertilizer rates
Soil Science Society of America Journal · 2026-11 · 90 (6)
DOI: 10.1002/saj2.70410 ↗Kurz gefasst
Fachartikel in der Zeitschrift Soil Science Society of America Journal, online erschienen am 18.11.2026. Verfasst von Umesh Acharya und 2 weiteren Autor:innen. Die Arbeit nennt 47 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
Crop yield stability faces major challenges from increasing climate variability and extreme weather events. Combining manure with mineral fertilizers offers a promising strategy to enhance corn ( Zea mays L .) yield and improve climate‐based yield predictability. In a 114‐year, long‐term experiment (1912–2013), we evaluated the effects of manure (none vs. applied) and six nitrogen (N) fertilizer rates (0, 45, 90, 135, 180 kg N ha −1 , plus a 135 kg N ha −1 + 56 kg P ha −1 ) on corn yield under changing climatic conditions. This study employed XGBoost (extreme gradient boosting) models to predict corn yield using yearly and monthly climate variables, examine the relative importance of weather predictors across treatments, and compare climate‐yield relationships among fertility treatments during 1991–2013. Our results show a positive response of corn yield to increasing N rates, both with and without manure; however, manure application significantly increased overall yields and stability over the 23‐year period. Model performance varied among the treatments, with lower predictive performance in the no‐manure, low‐N treatments ( R 2 ≤ 0.76; root mean squared error ≥ 1.02 Mg ha −1 ), indicating that yield in these treatments was less accurately predicted from the selected climate variables. Model interpretation, utilizing feature importance gain, beeswarm, and waterfall plots, indicated relatively high predictive importance for soil and air temperature, reference evapotranspiration, and solar radiation as the key climate variables influencing corn yield prediction. Monthly analyses indicated that several climate variables had relatively high predictive importance during June, July, and August. Overall, the results demonstrate that climate‐yield relationships varied among fertility treatments, and that key climatic predictors were concentrated in critical growing‐season months.
Zitieren
APA
Acharya, U., Daigh, A. L. M., & Maharjan, B. (2026). XGBoost modeling of climate impact on corn yield under varying fertilizer rates. Soil Science Society of America Journal, 90(6). https://doi.org/10.1002/saj2.70410
Vancouver
Acharya U, Daigh ALM, Maharjan B. XGBoost modeling of climate impact on corn yield under varying fertilizer rates. Soil Science Society of America Journal. 2026;90(6). doi:10.1002/saj2.70410
BibTeX
@article{Acharya2026,
author = {Acharya, Umesh and Daigh, Aaron Lee M. and Maharjan, Bijesh},
title = {XGBoost modeling of climate impact on corn yield under varying fertilizer rates},
journal = {Soil Science Society of America Journal},
year = {2026},
volume = {90},
number = {6},
doi = {10.1002/saj2.70410}
}
EUVIMED