Synergistic Retrieval of Soil Moisture in Arid Regions Using GF ‐3 SAR and Sentinel‐2 Optical Data
Bibliographic Data
As a vital component of the land–atmosphere interaction system, soil moisture plays an indispensable role. In arid and semi‐arid regions, soil moisture serves as a key indicator of ecosystem vulnerability and also functions as an essential role in drought monitoring, climate research, agricultural water resource management, and land management. This study integrates radar and optical remote sensing data combined with intelligent algorithms based on machine learning to construct and evaluate soil moisture estimation frameworks across diverse data combinations. The Yutian Oasis was selected as the study area, and four hybrid models (ACO‐RF, ACO‐SVM, SSA‐RF, and SSA‐SVM) were developed by optimizing standalone random forest (RF) and support vector machine (SVM) models using ant colony optimization (ACO) and sparrow search algorithm (SSA). Three different combinations of input data sources were constructed based on GF‐3 and Sentinel‐2 data. A total of six models were employed to assess soil moisture throughout the study area under three different data source scenarios. The results indicated that compared to the other two single‐source datasets, all models achieved the highest prediction accuracy when using the GF‐3 + Sentinel‐2 datasets. Specifically, the ACO‐RF model exhibited superior results, with R 2 values in the test set improved by 9.59% and 5.26% compared to models using GF‐3 or Sentinel‐2 data alone, respectively. Across all models, the four hybrid models outperformed the standalone RF and SVM models. Among the hybrid models, ACO‐RF demonstrated the best overall outcomes, achieving an R 2 of 0.80, RMSE of 3.07%, and RPD of 2.30. Therefore, integrating radar and optical data with intelligent algorithm‐optimized machine learning strategies improves soil moisture estimation precision, offering significant support for sustainable oasis agriculture and land management in arid regions
Arid · Data set · Mean squared error · Moisture · Radar · Random forest · Support vector machine · Water content · Remote Sensing in Agriculture · Soil Geostatistics and Mapping · Soil Moisture and Remote Sensing
The use of the Normalized Difference Water Index (NDWI) in the delineation of open water features
Modification of normalised difference water index (NDWI) to enhance open water features in remotely sensed imagery
Investigating soil moisture–climate interactions in a changing climate
Feature Selection with the Boruta Package
Support vector machines in remote sensing
Random Forests
| Highly cited | No |
|---|