Abstract Rice growth monitoring using Synthetic Aperture Radar (SAR) is recognized as a promising approach for tracking the development of this important crop. Accurate spatio-temporal information of rice inventories is required for water resource management, production risk occurrence, and yield forecasting. This research investigates the potential of the proposed Generalized volume scattering model based Radar Vegetation Index (GRVI) for monitoring rice growth at different phenological stages. The GRVI is derived using the concept of a geodesic distance (GD) between Kennaugh matrices projected on a unit sphere. We utilized this concept of GD to quantify a similarity measure between the observed Kennaugh matrix (representation of observed Polarimetric SAR information) and the Kennaugh matrix of a generalized volume scattering model (a realization of scattering media). The similarity measure is then modulated with a factor estimated from the ratio of the minimum to the maximum GD between the observed Kennaugh matrix and the set of elementary targets: trihedral, cylinder, dihedral, and narrow dihedral. In this work, we utilize a time series of C-band quad-pol RADARSAT-2 observations over a semi-arid region in Vijayawada, India. Among the several rice cultivation practices adopted in this region, we analyze the growth stages of direct seeded rice (DSR) and conventional tansplanted rice (TR) with the GRVI and crop biophysical parameters viz., Plant Area Index – PAI. The GRVI is compared for both rice types against the Radar Vegetation Index (RVI) proposed by Kim and van Zyl. A temporal analysis of the GRVI with crop biophysical parameters at different phenological stages confirms its trend with the plant growth stages. Also, the linear regression analysis confirms that the GRVI outperforms RVI with significant correlations with PAI (r ≥ 0.83 for both DSR and TR). In addition, PAI estimations from GRVI show promising retrieval accuracy with Root Mean Square Error (RMSE)

中文翻译：

---

使用源自 RADARSAT-2 极化 SAR 数据的广义雷达植被指数评估印度半干旱地区的水稻生长条件

摘要 使用合成孔径雷达 (SAR) 进行水稻生长监测被认为是跟踪这一重要作物生长的一种很有前景的方法。水资源管理、生产风险发生和产量预测需要水稻库存的准确时空信息。本研究调查了所提出的基于雷达植被指数 (GRVI) 的广义体积散射模型在监测不同物候阶段水稻生长的潜力。GRVI 是使用投影到单位球体上的肯诺矩阵之间的测地距离 (GD) 的概念推导出来的。我们利用 GD 的这个概念来量化观察到的肯诺矩阵（观察到的极化 SAR 信息的表示）和广义体积散射模型（散射介质的实现）的肯诺矩阵之间的相似性度量。然后使用从观察到的肯诺矩阵和基本目标集：三面体、圆柱体、二面体和窄二面体之间的最小与最大 GD 之比估计的因子来调制相似性度量。在这项工作中，我们利用了印度 Vijayawada 半干旱地区的 C 波段四极 RADARSAT-2 观测的时间序列。在该地区采用的几种水稻种植实践中，我们使用 GRVI 和作物生物物理参数分析了直播水稻 (DSR) 和常规插秧水稻 (TR) 的生长阶段，即，植物面积指数——PAI。GRVI 针对两种水稻类型与 Kim 和 van Zyl 提出的雷达植被指数 (RVI) 进行了比较。GRVI 与不同物候阶段作物生物物理参数的时间分析证实了其随植物生长阶段的趋势。此外，线性回归分析证实 GRVI 优于 RVI，与 PAI 显着相关（DSR 和 TR 的 r ≥ 0.83）。此外，来自 GRVI 的 PAI 估计显示出有希望的均方根误差 (RMSE) 检索精度 线性回归分析证实 GRVI 优于 RVI，且与 PAI 显着相关（DSR 和 TR 的 r ≥ 0.83）。此外，来自 GRVI 的 PAI 估计显示出有希望的均方根误差 (RMSE) 检索精度 线性回归分析证实 GRVI 优于 RVI，且与 PAI 显着相关（DSR 和 TR 的 r ≥ 0.83）。此外，来自 GRVI 的 PAI 估计显示出有希望的均方根误差 (RMSE) 检索精度