Wang et al. (2026) A novel two-step method for ratoon rice mapping using Sentinel-1/2 time series

Identification

• Journal: International Journal of Applied Earth Observation and Geoinformation
• Year: 2026
• Date: 2026-01-08
• Authors: Yue Wang, Yuechen Li, Xiaolin Zhu, J CHEN, Ruyin Cao, Xiong Yao, Wujun Zhang
• DOI: 10.1016/j.jag.2025.105083

Research Groups

• Chongqing Engineering Research Center for Remote Sensing Big Data Application, Chongqing Jinfo Mountain National Field Scientific Observation and Research Station for Karst Ecosystem, School of Geographical Sciences, Southwest University, Chongqing, China
• Key Laboratory of Monitoring, Evaluation and Early Warning of Territorial Spatial Planning Implementation, Ministry of Natural Resources, Chongqing, China
• Key Laboratory of Remote Sensing Application and Innovation, Chongqing, China
• Department of Land Surveying and Geo-Informatics, The Hong Kong Polytechnic University, Hong Kong, China
• State Key Laboratory of Remote Sensing Science, Faculty of Geographical Science, Beijing Normal University, Beijing, China
• School of Resources and Environment, University of Electronic Science and Technology of China, Chengdu, Sichuan, China
• Chongqing Rice Ratooning Research Center, Chongqing Academy of Agricultural Sciences, Chongqing, China

Short Summary

This study proposes a novel two-step ratoon rice (TSRR) mapping method integrating Sentinel-1A synthetic aperture radar (SAR) and Sentinel-2 optical imagery at the parcel scale. The method achieved an average overall accuracy of 0.87 in distinguishing ratoon rice from other paddy rice types across diverse regions without relying on detailed phenological information.

Objective

• To develop a robust and automated two-step ratoon rice (TSRR) mapping method using multi-source remote sensing data (Sentinel-1A and Sentinel-2) that can accurately distinguish ratoon rice from other rice types across diverse climatic and surface conditions, without relying on detailed phenological information, and mitigating cloud cover issues and "salt-and-pepper" noise.

Study Configuration

• Spatial Scale: Parcel scale, applied to three heterogeneous study sites in China (Fushun County, Sichuan Province; border of Anxiang County, Hunan Province and Shishou City, Hubei Province; Xuanzhou District, Anhui Province). Spatial resolution of 10 meters.
• Temporal Scale: Sentinel-1A time series data from 2019 to 2021 (specific periods for each site, e.g., January 11 to December 24, 2020 for site 1) with a 12-day revisit capability. Sentinel-2 optical imagery for corresponding study periods. Field data collected from August 19–26, 2023.

Methodology and Data

• Models used:
  • Two-Step Ratoon Rice (TSRR) mapping method.
  • Object-based segmentation (Canny edge detection, Watershed segmentation).
  • SAR-based Paddy Rice Index (SPRI).
  • SAR-based Ratoon Rice Index (SRRI).
  • Modified Neighborhood Similar Pixel Interpolator (MNSPI) for cloud removal.
  • Refined Lee filter for speckle noise reduction in SAR data.
  • Savitzky-Golay filter for temporal smoothing of SAR data.
• Data sources:
  • Sentinel-1A Synthetic Aperture Radar (SAR) Ground Range Detected (GRD) data (VH polarization, 10 m spatial resolution, 12-day temporal resolution) from Google Earth Engine (GEE).
  • Sentinel-2 Level-2A optical imagery (10 m resolution bands: red, green, blue, near-infrared) from GEE, with cloud cover threshold less than 10%.
  • Field collection data (633 paddy rice samples, 318 ratoon rice samples, 462 non-paddy rice samples) with GPS coordinates (RTK surveying instrument) and photographs.
  • Google Earth historical imagery for sample validation.
  • Derived indices: Normalized Difference Vegetation Index (NDVI), Normalized Difference Water Index (NDWI).

Main Results

• The TSRR method effectively distinguished ratoon rice from other paddy rice types, achieving an average overall accuracy (OA) of 0.87, an F1 score of 0.84, a Kappa coefficient of 0.71, a user's accuracy (UA) of 0.91, and a producer's accuracy (PA) of 0.77 across the three study sites.
• The method demonstrated strong robustness and transferability across different regions with varying climatic and surface conditions, without requiring prior phenological information.
• The SAR-based Paddy Rice Index (SPRI) effectively distinguished paddy rice from non-rice features with a unified threshold of 0.25, achieving OA, UA, PA, and F1 scores exceeding 0.89, and Kappa above 0.81.
• TSRR values for ratoon rice were generally distributed between 0.3 and 0.8, significantly higher than double- and single-season rice (mostly near zero), indicating strong discriminative ability.
• The two-step TSRR method outperformed a one-step SRRI method, which resulted in increased misclassification of other land cover types (e.g., construction land, forest, water bodies) as ratoon rice, with overclassification areas expanding by 43.81 km² to 162.25 km².
• While pixel-based classification generally achieved slightly higher accuracy (differences mostly within 0.1), the object-based method provided better visual consistency and alignment with agricultural management practices by reducing "salt-and-pepper" noise.

Contributions

• Proposes a novel two-step ratoon rice (TSRR) mapping method that integrates Sentinel-1A SAR and Sentinel-2 optical imagery, overcoming limitations of existing methods.
• Eliminates the reliance on detailed phenological information, enhancing cross-regional adaptability and transferability, especially in areas with complex or poorly documented crop calendars.
• Effectively mitigates challenges posed by frequent cloud cover and rainfall by leveraging cloud-penetrating SAR data, making it suitable for cloud-prone regions.
• Introduces object-based segmentation to reduce "salt-and-pepper" noise and provide spatially consistent classification results aligned with agricultural parcel management.
• Develops two novel indices, SAR-based Paddy Rice Index (SPRI) and SAR-based Ratoon Rice Index (SRRI), for robust identification of paddy rice and subsequent ratoon rice.
• Demonstrates high accuracy in distinguishing ratoon rice from other morphologically similar rice types (e.g., double-season rice), which is a significant challenge for existing methods.
• Provides a reliable solution for large-scale paddy rice and ratoon rice mapping, supporting crop monitoring, yield estimation, and national agricultural inventory initiatives.

Funding

• Key Laboratory of Remote Sensing Application and Innovation [grant numbers LRSAI-2025009]
• Sichuan Provincial International Science and Technology Innovation Cooperation/Hong Kong, Macao and Taiwan Science and Technology Innovation Cooperation Project [grant numbers 2025YFHZ0228]
• Key Project of Chongqing Technology Innovation and Application Development Special [grant numbers CSTB2024TIAD-KPX0107]
• Natural Science Foundation of Chongqing [grant numbers CSTB2025NSCQ-GPX1065, CSTB2022NSCQ-MSX1535]

Citation

@article{Wang2026novel,
  author = {Wang, Yue and Li, Yuechen and Zhu, Xiaolin and CHEN, J and Cao, Ruyin and Yao, Xiong and Zhang, Wujun},
  title = {A novel two-step method for ratoon rice mapping using Sentinel-1/2 time series},
  journal = {International Journal of Applied Earth Observation and Geoinformation},
  year = {2026},
  doi = {10.1016/j.jag.2025.105083},
  url = {https://doi.org/10.1016/j.jag.2025.105083}
}