EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

A Machine Learning Coupled Model Reveals Spatial Patterns of Background Ammonia Emissions in Yangtze River Delta Rice Paddies

Jiabei Sun, Shihang Wang, Yujuan Liu, Siyuan Cai, Xu Zhao

Agronomy · 08.06.2026 · 16 (12) · 1128

DOI: 10.3390/agronomy16121128 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Agronomy, online erschienen am 08.06.2026. Verfasst von Jiabei Sun und 4 weiteren Autor:innen. Die Arbeit nennt 66 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Institute of Soil Science, Chinese Academy of Sciences, National Key R&D Program of China, National Natural Science Foundation of China.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Background ammonia (NH3) volatilisation—defined as inherent NH3 loss without nitrogen (N) fertilisation—is essential for accurate emission factor (EF) establishment and regional mitigation strategies. The Yangtze River Delta (YRD), accounting for ~20% of China’s rice production with the highest fertilisation intensity nationwide, represents a critical hotspot where current inventories predominantly focus on fertiliser-induced emissions while neglecting spatial heterogeneity in background volatilisation, causing substantial uncertainty in net emission estimates. To elucidate the spatial differentiation of NH3 volatilisation EFs and background emissions across the YRD and to overcome the spatiotemporal precision limitations of current regional-scale assessments, we compiled a comprehensive rice-growing season NH3 database through a systematic literature review. Regression models were used to decompose NH3 volatilisation into EFs and background volatilisation. Subsequently, Random Forest was applied to quantify environmental drivers and generate county-level background emission maps. The linear model effectively distinguished fertiliser-induced emissions from background volatilisation. The Random Forest model exhibited robust predictive performance for background emissions (R2 = 0.66) but limited power for EFs (R2 = 0.12). TN and pH were the dominant predictors of background volatilisation. YRD background emissions ranged across 3.97–9.03 kg N ha−1 with a north–south gradient: Anhui had the highest (6.87 kg N ha−1), followed by Jiangsu (6.48 kg N ha−1), with Shanghai (6.29 kg N ha−1) and Zhejiang (6.11 kg N ha−1) with lower emissions. The pronounced spatial heterogeneity underscores the necessity of geographically specific mitigation strategies that account for regional edaphic conditions. This coupled approach provides critical data support for precision emission reduction in intensively cultivated rice systems.

Förderung

  • Institute of Soil Science, Chinese Academy of Sciences
  • National Key R&D Program of China
  • National Natural Science Foundation of China
  • Natural Science Foundation of Jiangsu Province

Zitieren

APA

Sun, J., Wang, S., Liu, Y., Cai, S., & Zhao, X. (2026). A Machine Learning Coupled Model Reveals Spatial Patterns of Background Ammonia Emissions in Yangtze River Delta Rice Paddies. Agronomy, 16(12), 1128. https://doi.org/10.3390/agronomy16121128

Vancouver

Sun J, Wang S, Liu Y, Cai S, Zhao X. A Machine Learning Coupled Model Reveals Spatial Patterns of Background Ammonia Emissions in Yangtze River Delta Rice Paddies. Agronomy. 2026;16(12):1128. doi:10.3390/agronomy16121128

BibTeX
@article{Sun2026,
  author = {Sun, Jiabei and Wang, Shihang and Liu, Yujuan and Cai, Siyuan and Zhao, Xu},
  title = {A Machine Learning Coupled Model Reveals Spatial Patterns of Background Ammonia Emissions in Yangtze River Delta Rice Paddies},
  journal = {Agronomy},
  year = {2026},
  volume = {16},
  number = {12},
  pages = {1128},
  doi = {10.3390/agronomy16121128}
}
Kostenlos starten →