Xie et al. (2026) Decadal changes and influencing factors of global industrial heat sources

Identification

• Journal: International Journal of Applied Earth Observation and Geoinformation
• Year: 2026
• Authors: Yakun Xie, Ruifeng Xia, Jianbo Lai, Youcai Zhao, Chaoda Song, Wen Song, Dejun Feng, Jun Zhu, HU Ya
• DOI: 10.1016/j.jag.2025.105058

Research Groups

• Faculty of Geosciences and Engineering, Southwest Jiaotong University, Chengdu, China.
• Emergency Management, Chengdu University, Chengdu, China.
• Department of Computer and Data Sciences, Case Western Reserve University, Cleveland, USA.
• State Key Laboratory of Bridge Intelligent and Green Construction, Southwest Jiaotong University, Chengdu, China.

Short Summary

This study develops a deep learning framework to automatically classify six major types of global industrial heat sources using high-resolution satellite imagery and thermal anomaly data. The researchers produced the first annual global dataset of industrial heat sources from 2013 to 2023, revealing decadal shifts in industrial activity and the impacts of the COVID-19 pandemic.

Objective

• To address the lack of consistent, cross-industry global monitoring of industrial heat sources (IHS) by developing an automated, high-resolution classification and tracking system.
• To analyze the spatiotemporal dynamics of six industrial categories: cement plants, steel plants, chemical plants, oil and gas platforms, coal chemical plants, and open-pit mines.

Study Configuration

• Spatial Scale: Global (excluding Antarctica), with detailed analysis across six continents (Asia, Europe, North America, South America, Africa, and Oceania).
• Temporal Scale: Annual monitoring over an 11-year period from 2013 to 2023.

Methodology and Data

• Models used: DBSCAN (Density-Based Spatial Clustering of Applications with Noise) for initial fire point clustering; EfficientNetV2 (deep learning) for multi-class image classification; Grad-CAM for model interpretability.
• Data sources: VIIRS (Visible Infrared Imaging Radiometer Suite) active fire data (VNP14IMG) at 375 m resolution; high-resolution optical satellite imagery from Google Earth (0.54 m to 2.15 m resolution); socio-economic data from the United Nations and World Bank.

Main Results

• Model Performance: The EfficientNetV2 model achieved an overall classification accuracy of 95.38% and an average precision of 94.38%, performing 98% faster than manual interpretation.
• Industrial Trends: Oil and gas platforms remained the dominant heat source (38–40% share); coal chemical plants showed a marked decline; steel and chemical plants exhibited moderate growth.
• Facility Dynamics: The number of newly built facilities declined from 2,643 in 2014 to 209 in 2023, while decommissioned facilities peaked in 2021 (2,594), indicating a global shift toward industrial consolidation and structural adjustment.
• Socio-economic Correlation: Industrial distribution is strongly correlated with population density and development stage; traditional heavy industries (steel/cement) are more prevalent in regions with lower per capita GDP and higher population counts.
• Pandemic Impact: A significant short-term decline in industrial heat signatures was recorded in 2020 due to COVID-19 lockdowns, with a gradual recovery beginning in 2022.

Contributions

• First Global Multi-class Dataset: Provides a high-resolution, annual dataset of six distinct industrial heat source types, filling a gap in fragmented, sector-specific records.
• Methodological Innovation: Integrates spatiotemporal-annual constraints with deep learning to distinguish persistent industrial activity from ephemeral biomass burning.
• Policy Support: Offers a scalable foundation for global carbon accounting, energy transition monitoring, and environmental policy assessment.

Funding

• National Natural Science Foundation of China (42301473).
• Sichuan Science and Technology Program (2024YFFK0421, 2024NSFSC0074, 2025ZNSFSC0322).
• National Earth Observation Data Center (NODAOP2024003).
• Chengdu Science and Technology Program (2025-YF05-00009-SN).

Citation

@article{Xie2026Decadal,
  author = {Xie, Yakun and Xia, Ruifeng and Lai, Jianbo and Zhao, Youcai and Song, Chaoda and Song, Wen and Feng, Dejun and Zhu, Jun and Ya, HU},
  title = {Decadal changes and influencing factors of global industrial heat sources},
  journal = {International Journal of Applied Earth Observation and Geoinformation},
  year = {2026},
  doi = {10.1016/j.jag.2025.105058},
  url = {https://doi.org/10.1016/j.jag.2025.105058}
}

Generated by BiblioAssistant using gemini-3-flash-preview (Google API)