Liang et al. (2025) GNSS-IR retrieval of soil moisture at hourly resolution taking into account corrections for inter-orbit phase bias of satellites

Identification

• Journal: Advances in Space Research
• Year: 2025
• Date: 2025-12-02
• Authors: Yueji Liang, Xingyu Zhao, Binglin Zhu, Guo Xi, Chao Ren, Xianjian Lu, Z. L. Feng, Jinlong Pan
• DOI: 10.1016/j.asr.2025.11.101

Research Groups

• College of Geomatics and Geoinformation, Guilin University of Technology, Guilin, China
• Bureau of Natural Resources of Hezhou City, Hezhou, China

Short Summary

This study proposes an hourly-resolution soil moisture retrieval method by fusing multi-system GNSS observations and correcting for inter-orbit phase biases using a low-order polynomial fitting approach. The method, validated with machine learning models, achieved a correlation coefficient of 0.95 and a 40.1 % reduction in root mean square error for hourly soil moisture retrieval.

Objective

• To propose and validate an hourly-resolution soil moisture retrieval method using multi-system Global Navigation Satellite System Interferometric Reflectometry (GNSS-IR) observations, specifically addressing and correcting inter-orbit phase biases among different satellites and systems.

Study Configuration

• Spatial Scale: Single point location (P041 station, Colorado, USA).
• Temporal Scale: Hourly resolution over a period of approximately 213 days (5112 theoretical hourly slots).

Methodology and Data

• Models used: Low-order polynomial fitting (for phase bias correction), Backpropagation Neural Network, Random Forest, CatBoost, Gradient Boosted Trees (for soil moisture retrieval).
• Data sources: Multi-system Global Navigation Satellite System (GNSS) observations (GPS and GLONASS) from the P041 station.

Main Results

• The proposed method achieved near-complete temporal coverage for hourly soil moisture retrieval, with 5084 valid hourly slots out of 5112 theoretical ones by fusing GPS and GLONASS data.
• The CatBoost machine learning model demonstrated optimal performance, achieving a correlation coefficient of 0.95 on the test set.
• The root mean square error (RMSE) for soil moisture retrieval decreased by approximately 40.1 % compared to the uncorrected case, highlighting the effectiveness of the phase bias correction.
• Low-order polynomial fitting successfully rectified linear phase biases across different satellite orbits and systems.

Contributions

• Proposes a novel method for hourly-resolution soil moisture retrieval using multi-system GNSS-IR, addressing a critical need for high-temporal-resolution data.
• Introduces an effective approach using low-order polynomial fitting to correct significant inter-orbit phase biases in GNSS-IR data, improving retrieval accuracy.
• Demonstrates the superior performance of machine learning models, particularly CatBoost, for high-resolution soil moisture estimation from corrected GNSS-IR data.
• Provides a robust technical framework for high-resolution environmental monitoring applications.

Funding

• Not specified in the provided text.

Citation

@article{Liang2025GNSSIR,
  author = {Liang, Yueji and Zhao, Xingyu and Zhu, Binglin and Xi, Guo and Ren, Chao and Lu, Xianjian and Feng, Z. L. and Pan, Jinlong},
  title = {GNSS-IR retrieval of soil moisture at hourly resolution taking into account corrections for inter-orbit phase bias of satellites},
  journal = {Advances in Space Research},
  year = {2025},
  doi = {10.1016/j.asr.2025.11.101},
  url = {https://doi.org/10.1016/j.asr.2025.11.101}
}