ABSTRACT

We require spatio-temporal information about rice for executing and planning diverse management practices. In this regard, data obtained from Synthetic Aperture Radar (SAR) sensors are well suited for tracking morphological developments of rice across its phenology stages. This study proposes different target characterization parameters from polarimetric SAR data for rice phenology mapping. Six C-band Radarsat-2 images acquired over Vijayawada, India, are used for complete analysis. It is known that polarimetric information provides excellent sensitivity for identifying crop phenology stages. Hence, in this study, we assessed phenology classification results using a scattering-type parameter and scattering powers for full-polarimetric (FP) and extracted dual-polarimetric (DP) SAR data. Here, we utilized the real 4$\times$4 Kennaugh matrix elements to derive these parameters equivalently for the two polarimetric modes (i.e. FP and DP). We obtained better overall classification accuracy for each phenology stages using the proposed parameters than the existing ones from FP and DP SAR data. We noted that the overall classification accuracy using the DP SAR data was only marginally lower than the FP SAR data. This marginal difference in the accuracies could be due to the absence of the cross-polarized component in the DP SAR data. We also demonstrate the usefulness of the scattering powers from DP SAR data for rice phenology monitoring.

摘要

我们需要有关稻米的时空信息，以执行和规划各种管理实践。在这方面，从合成孔径雷达（SAR）传感器获得的数据非常适合跟踪整个水稻物候阶段的形态发育。这项研究从极化SAR数据中提出了用于水稻物候映射的不同目标表征参数。在印度Vijayawada上采集的六张C波段Radarsat-2图像用于完整分析。众所周知，极化信息为识别作物物候阶段提供了极好的敏感性。因此，在这项研究中，我们使用散射类型参数和全极化（FP）散射功率和提取的双极化（DP）SAR数据评估了物候分类结果。在这里，我们利用了真正的4$\times$4个Kennaugh矩阵元素，可以等效地导出两个极化模式（即FP和DP）的这些参数。与从FP和DP SAR数据中获得的现有参数相比，我们使用建议的参数获得了每个物候阶段更好的总体分类精度。我们注意到，使用DP SAR数据的总体分类准确性仅略低于FP SAR数据。精度的这种边际差异可能是由于DP SAR数据中不存在交叉极化分量。我们还证明了DP SAR数据的散射功率对水稻物候监测的有用性。