Evaluation of SAR C-band radar vegetation indices for rice crop monitoring in Tamil Nadu, India

Abstract

Rice (Oryza sativa) cultivation plays a critical role in food security across Asia, where smallholder farmers depend heavily on timely information about crop development and field conditions. Monitoring these changes using optical remote sensing is constrained by persistent cloud cover during monsoon-driven growing seasons, underscoring the necessity of Synthetic Aperture Radar (SAR) for continuous observation. This study evaluates the capability of Sentinel-1 C-band SAR for tracking rice phenology in two smallholder fields in Mayiladuthurai District, Tamil Nadu, during the Late Samba season (September to January). Field-scale analysis of VV, VH, and NDVI time series for 2023–2024 captured key phenological transitions, with polarization showing a strong correlation with NDVI (Farm 1: r = 0.75; Farm 2: r = 0.73). Radar Vegetation Indices (RVI, mRVI, and RVI4S1) were computed from multi-year Sentinel-1 data (2018–2023) and compared with MODIS NDVI. Although the radar indices showed high inter-correlation (r > 0.90), their relationship with NDVI remained weak (0.15–0.30). Machine learning experiments over a 1.5 × 1.5 km region (2018–2022) demonstrated that a Linear Regression model (6.92 × 10−5) outperformed Random Forest Regression (0.000258) in predicting RVI4S1 from VV and VH, indicating linear relationship between radar channels and the index. The study highlights the suitability of Sentinel-1 SAR-particularly VH polarization – for phenology tracking in smallholder contexts, especially where optical data are limited by cloud cover.

Plant phenotyping relevance

SARセンシングとレーダー植生指数を用いてイネの生育・フェノロジーを推定し、光学指標との比較および機械学習による技術評価を行っており、表現型取得法が中心である。

Code and data availability

公開論文であることは確認できましたが、現在の公式API・許可済み取得経路では本文を自動取得できませんでした。

No evidence-backed public reproduction asset is currently recorded.