Wang et al. (2025) Remote Sensing Inversion and Spatiotemporal Dynamics of Multi-Depth Soil Salinity in a Typical Arid Wetland: A Case Study of Ebinur Wetland Reserve, Xinjiang

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

Identification

• Journal: Remote Sensing
• Year: 2025
• Date: 2025-12-07
• Authors: Jinjie Wang, Jinming Zhang, Zihan Zhang
• DOI: 10.3390/rs17243958

Research Groups

[Information not provided in the given paper text.]

Short Summary

This study developed a six-layer (0–100 cm) soil salinity inversion framework integrating multi-year field samples and Landsat imagery for the Ebinur Lake wetland. The framework, particularly using a Convolutional Neural Network with Random Frog Leaping Algorithm-optimized features, accurately retrieved multi-depth soil salinity and revealed distinct salinity migration patterns across different land types.

Objective

• To construct a six-layer (0–100 cm) soil salinity inversion framework by integrating multi-year field samples and Landsat imagery (1996–2024) to overcome the lack of soil depth information and limited spatiotemporal monitoring in arid regions.

Study Configuration

• Spatial Scale: Ebinur Lake wetland, Xinjiang, China.
• Temporal Scale: 1996–2024 (Landsat imagery).

Methodology and Data

• Models used: Convolutional Neural Network (CNN), Long Short-Term Memory Network (LSTM), Random Forest (RF). Feature optimization was performed using the Random Frog Leaping Algorithm (RFLA).
• Data sources: Multi-year field samples, Landsat imagery.

Main Results

• The Random Frog Leaping Algorithm (RFLA) effectively identified high-contribution spectral features, enhancing model efficiency and reducing data redundancy.
• The Convolutional Neural Network (CNN) outperformed LSTM and RF models in capturing spatial salinity, achieving R² values of 0.75, 0.59, 0.63, 0.69, 0.57, and 0.56 for the six soil layers (0–100 cm).
• Distinct salinity migration patterns were identified: surface enrichment, mid-layer buffering, and deep-layer accumulation.
• In oases, surface salinity declined while deep layers accumulated; conversely, in deserts, surface salinity increased.

Contributions

• Developed an accurate multi-depth (0–100 cm) soil salinity inversion framework by integrating multi-year field samples and multi-source remote sensing data.
• Demonstrated the effectiveness of RFLA for optimizing spectral features in soil salinity retrieval.
• Provided a robust technical framework for spatiotemporal monitoring, irrigation management, and ecological assessment of soil salinization in arid regions.
• Revealed nuanced multi-depth salinity migration patterns specific to different land types (oases vs. deserts).

Funding

[Information not provided in the given paper text.]

Citation

@article{Wang2025Remote,
  author = {Wang, Jinjie and Zhang, Jinming and Zhang, Zihan},
  title = {Remote Sensing Inversion and Spatiotemporal Dynamics of Multi-Depth Soil Salinity in a Typical Arid Wetland: A Case Study of Ebinur Wetland Reserve, Xinjiang},
  journal = {Remote Sensing},
  year = {2025},
  doi = {10.3390/rs17243958},
  url = {https://doi.org/10.3390/rs17243958}
}