Han et al. (2025) Optimization of ecological restoration efficiency in Qinghai-Tibet Plateau using the Cubist regression tree model: A study of environmental adaptability models

Identification

• Journal: PLoS ONE
• Year: 2025
• Date: 2025-11-12
• Authors: Yuan Han, Wu Jin, Heng Liu, Wei Wang, Jie Ma, Wei Zhao
• DOI: 10.1371/journal.pone.0335056

Research Groups

Xining Natural Resources Comprehensive Survey Center, China Geological Survey, Xining, China

Short Summary

This study develops an integrated Cubist-BiGRU-SA regression tree model to enhance the prediction accuracy and environmental adaptability of ecological restoration efficiency on the Qinghai-Tibet Plateau (QTP). The model achieves high accuracy (over 96%) in predicting vegetation restoration rates and soil quality improvements, providing quantitative guidelines for manual intervention measures.

Objective

• To address inefficiencies in ecological restoration on the Qinghai-Tibet Plateau, specifically prolonged vegetation restoration cycles, slow soil quality improvement, and difficulties in quantifying manual intervention measures.
• To develop an integrated Cubist regression tree model combining a lightweight self-attention mechanism (SA) with bidirectional gated recurrent units (BiGRU) to enhance the accuracy and adaptability of restoration efficiency prediction.
• To achieve a quantitative evaluation of manual intervention measures and provide optimized decision support for ecological restoration on the QTP.

Study Configuration

• Spatial Scale: Qinghai-Tibet Plateau (QTP), focusing on multiple typical ecological restoration demonstration zones on its eastern edge, covering low-altitude (< 3500 meters), middle-altitude (3500–4500 meters), and high-altitude (> 4500 meters) regions.
• Temporal Scale: 2019 to 2023.

Methodology and Data

• Models used:
  • Integrated Cubist-BiGRU-SA model (Cubist regression tree, Bidirectional Gated Recurrent Units, Self-Attention mechanism).
  • Comparative models: Cubist-GRU, RF-LSTM, BiGRU, and a model by Nguyen et al. (2025).
• Data sources:
  • Meteorological data: Daily temperature and precipitation from meteorological stations within the QTP.
  • Remote sensing (RS) data: Satellite-derived soil moisture and fractional vegetation cover (FVC).
  • Manual restoration data: Field surveys providing irrigation volume, planting density, vegetation survival rate, and soil physicochemical properties.
  • Dataset size: 3,217 independent time-series samples.

Main Results

• The Cubist-BiGRU-SA model achieved less than 5% error in vegetation restoration rate prediction, with correlation coefficients exceeding 0.90.
• The model demonstrated 96% accuracy in soil improvement prediction.
• Temperature was the most influential factor with a contribution rate of 32%, followed by precipitation at 25%.
• Soil moisture and fractional vegetation cover (FVC) jointly contributed 25% to the prediction.
• Prediction accuracy remained above 90% across different altitude zones (low, middle, and high-altitude regions), indicating strong regional adaptability.
• In areas with annual precipitation below 200 millimeters, every 10% increase in irrigation volume led to approximately 15% improvement in vegetation survival rate.

Contributions

• Proposes an innovative hybrid Cubist-BiGRU-SA model that effectively integrates multi-source spatiotemporal data for ecological restoration efficiency prediction on the QTP.
• Significantly enhances prediction accuracy and environmental adaptability by dynamically adjusting environmental factor weights (Self-Attention) and optimizing temporal feature representation (BiGRU).
• Provides quantitative and operational intervention guidelines for plateau ecological restoration, enabling a shift from experience-driven to data-driven management.
• Quantifies the synergistic effects between natural conditions and manual intervention measures, offering scientific decision support for localized restoration strategies.

Funding

• Xining Natural Resources Comprehensive Survey Center, China Geological Survey (Earth Science Data Integration and Knowledge Services, NO. DD20230616).

Citation

@article{Han2025Optimization,
  author = {Han, Yuan and Jin, Wu and Liu, Heng and Wang, Wei and Ma, Jie and Zhao, Wei},
  title = {Optimization of ecological restoration efficiency in Qinghai-Tibet Plateau using the Cubist regression tree model: A study of environmental adaptability models},
  journal = {PLoS ONE},
  year = {2025},
  doi = {10.1371/journal.pone.0335056},
  url = {https://doi.org/10.1371/journal.pone.0335056}
}