Si et al. (2026) Differential Effects of Soil Moisture and Air Temperature on Vegetation Dynamics in Northwest China’s Warming and Wetting Region: An LSTM Modeling Approach

⚠️ Warning: This summary was generated from the abstract only, as the full text was not available.

Identification

• Journal: Plants
• Year: 2026
• Date: 2026-05-19
• Authors: Yajun Si, Junpo Yu, Geng Li, Jesus Carrera, Jiming Jin, Haihua Bai
• DOI: 10.3390/plants15101542

Research Groups

Not specified in the provided text.

Short Summary

The study employs a bivariate LSTM model to simulate Leaf Area Index (LAI) in Northwest China, revealing that while hydrothermal drivers explain seasonal phenology and grassland variability, they are insufficient to capture long-term forest greening.

Objective

• To evaluate the capacity of a bivariate Long Short-Term Memory (LSTM) model to simulate LAI dynamics across four vegetation types using air temperature and soil moisture as predictors.

Study Configuration

• Spatial Scale: Northwest China.
• Temporal Scale: Seasonal and interannual scales.

Methodology and Data

• Models used: Bivariate Long Short-Term Memory (LSTM) model.
• Data sources: Air temperature and soil moisture (predictors); Leaf Area Index (LAI) (target variable).

Main Results

• Seasonal Scale: The model effectively reproduced seasonal vegetation phenology across all types with $R^2 > 0.9$.
• Interannual Scale:
  • Grasslands in water-limited environments were reasonably represented ($R^2 = 0.31$).
  • The model failed to reproduce long-term greening trends in forests.
• Mechanistic Insight: Grassland dynamics are driven by high-frequency hydroclimatic variability, whereas forest growth is governed by low-frequency drivers such as $\text{CO}_2$ fertilization, nitrogen deposition, and ecological inertia.

Contributions

• Highlights the transition from water-limited to energy- and process-limited controls across different vegetation types.
• Demonstrates the limitations of purely climate-driven machine learning models in predicting long-term vegetation changes, suggesting the need to integrate biogeochemical processes into such frameworks.

Funding

Not specified in the provided text.

Citation

@article{Si2026Differential,
  author = {Si, Yajun and Yu, Junpo and Li, Geng and Carrera, Jesus and Jin, Jiming and Bai, Haihua},
  title = {Differential Effects of Soil Moisture and Air Temperature on Vegetation Dynamics in Northwest China’s Warming and Wetting Region: An LSTM Modeling Approach},
  journal = {Plants},
  year = {2026},
  doi = {10.3390/plants15101542},
  url = {https://doi.org/10.3390/plants15101542}
}