Kobayashi et al. (2026) Correlation between field-measured grain moisture content and C-band SAR backscatter: Potential use of Sentinel-1 SAR data to estimate optimal harvest timing in rice crop cultivation

Identification

• Journal: Remote Sensing Applications Society and Environment
• Year: 2026
• Date: 2026-01-01
• Authors: Shoko Kobayashi, J. P. Matthews
• DOI: 10.1016/j.rsase.2026.101935

Research Groups

• College of Agriculture, Tamagawa University, Japan
• Environmental Satellite Applications (ENSA), UK
• Institute of Liberal Arts and Sciences, Kyoto University, Japan

Short Summary

This study investigates the relationship between C-band SAR backscatter and field-measured rice grain moisture content (GMC) to improve the prediction of optimal harvest timing. It found that co-polarized C-band SAR backscatter (VV) at shallow incidence angles (around 45°) exhibits a significant correlation with GMC, demonstrating the potential of Sentinel-1 data for this application.

Objective

• To examine in depth the relationship between rice crop biophysical parameters and C-band SAR backscatter, specifically focusing on the link between field-measured grain moisture content and satellite SAR backscatter using data collected over a 2-year period.
• To test the hypothesis that C-band backscatter is predominantly affected by moisture content in the grain during the ripening stage, where plant physical structure changes minimally.

Study Configuration

• Spatial Scale: Rice-crop fields in Saku City, Nagano Prefecture, Japan. Average field size was 2288 square meters in 2020 and 2685 square meters in 2022, with a minimum of 1115 square meters and a maximum of 4073 square meters.
• Temporal Scale: Field observations were conducted 9 times in 2020 and 10 times in 2022, spanning from June to September. Satellite data were acquired for the years 2020 and 2022, with a focus on the ripening stage (after August 15).

Methodology and Data

• Models used: Partial Least Squares Regression (PLSR) with 5-fold cross-validation, Savitzky-Golay (S-G) filter for time-series smoothing, Median Absolute Deviation (MAD) and Grubbs’ test for outlier detection.
• Data sources:
  • Satellite: Sentinel-1 C-band (wavelength: 5.55 cm) Synthetic Aperture Radar (SAR) data in Interferometric Wide-swath (IW) acquisition mode, dual-polarization (VV and VH). Data from relative orbit 119 (incidence angle 35°-36°) and orbit 46 (incidence angle 45°-46°).
  • Observation (Field-measured): Plant length (centimeters), number of tillers per hill, tiller volume (cubic meters per square meter), leaf and tiller moisture content (%), and grain moisture content (GMC) (%).
  • Ancillary: Automated Meteorological Data Acquisition System (AMeDAS) weather data, land parcel polygons from the Ministry of Agriculture, Forestry and Fisheries (MAFF) of Japan.

Main Results

• Co-polarized C-band VV backscatter obtained at shallower incidence angles (45°-46°) showed a significant negative correlation with grain moisture content (GMC) during the ripening stage (R² = 0.669 ± 0.184), with GMC having the highest variable importance in projection (VIP = 1.517). This indicates that VV backscatter increases as rice grains dry.
• Cross-polarized VH backscatter and the VV/VH ratio at shallower incidence angles (45°-46°) also showed correlations with leaf and tiller moisture content and GMC, but with lower R² values (VH: R² = 0.297 ± 0.328; VV/VH: R² = 0.492 ± 0.122).
• SAR backscatter signals acquired at steeper incidence angles (35°-36°) did not exhibit statistically significant correlations with rice crop biophysical parameters during the ripening stage.
• Seasonal trends in backscatter were observed, with VV and VH increasing after transplanting, and VV (at 45°-46°) decreasing during panicle formation due to signal attenuation, then increasing during ripening as GMC declined.

Contributions

• This study provides the first robust, generalized correlation between field-measured grain moisture content (GMC) and C-band SAR backscatter using a comprehensive 2-year dataset, distinguishing GMC from general vegetation moisture content (VMC) which was the focus of previous studies.
• It demonstrates the critical role of radar incidence angle, identifying that shallower incidence angles (greater than 40°) are more effective for monitoring GMC in rice crops during the ripening stage.
• The research offers a practical remote sensing methodology for estimating optimal harvest timing at the individual field level, which is crucial for mitigating climate change impacts on rice quality and ensuring stable food supplies.
• It addresses limitations of prior research by integrating extensive field measurements with satellite data over multiple years, providing a more robust understanding of SAR signal interaction with rice crops.

Funding

• Japan Society for the Promotion of Science (JSPS) KAKENHI Grant Number 18K11626.

Citation

@article{Kobayashi2026Correlation,
  author = {Kobayashi, Shoko and Matthews, J. P.},
  title = {Correlation between field-measured grain moisture content and C-band SAR backscatter: Potential use of Sentinel-1 SAR data to estimate optimal harvest timing in rice crop cultivation},
  journal = {Remote Sensing Applications Society and Environment},
  year = {2026},
  doi = {10.1016/j.rsase.2026.101935},
  url = {https://doi.org/10.1016/j.rsase.2026.101935}
}