In the global context of population growth and climate change, monitoring crops is necessary to sustain agriculture and conserve natural resources. While many studies have demonstrated the ability of optical and SAR remotely sensed data to estimate crop parameters, these data have not been compared or combined to predict crop phenological stages. Despite the high sensitivity of SAR polarimetric data to crop phenological stages, no study has used high temporal resolution data. The freely available SAR Sentinel-1 (S-1) and optical Sentinel-2 (S-2) time series provide a unique opportunity to monitor crop phenology at a high spatial resolution on a weekly basis. The objective of this study was to evaluate the potential of S-1 data alone, S-2 data alone, and their combined use to predict wheat and rapeseed phenological stages. We first analyzed temporal profiles of spectral bands, vegetation indices and leaf area index (LAI) derived from S-2 data, and backscattering coefficients and polarimetric indicators derived from S-1 data. Then, an incremental procedure was used to estimate the contribution of S-1 and S-2 features to the classification of principal and secondary phenological stages of wheat and rapeseed. Results for both crops showed that the classification obtained with combined S-1 & 2 data (mean kappa = 0.53–0.82 and 0.74–0.92 for wheat and rapeseed, respectively) was more accurate than those obtained with S-2 data alone (mean kappa = 0.54–0.75 and 0.67–0.86 for wheat and rapeseed, respectively) or S-1 data alone (mean kappa = 0.48–0.61 and 0.61–0.64 for wheat and rapeseed, respectively). Combining S-1 & 2 data allowed better identification of the beginning and end of tillering for wheat and the beginning and end of ripening for rapeseed. Among S-2 features, the most important were LAI for wheat and the NDVI for rapeseed. For both crops, the S2REP index was one of the most important vegetation indices, while MCARI was less important. For S-1 features, results highlighted the large contribution of the backscatter ratio (σ◦VH:σ◦VV) and the value of using polarimetric indicators (Shannon entropy and span) to monitor rapeseed and wheat phenology. The main novelties of this work are the use of S-1 polarimetric indicators to identify phenological stages of wheat and rapeseed and the mapping of wheat and rapeseed secondary phenological stages using remotely sensed data.

在全球人口增长和气候变化的背景下，监测作物对于维持农业和保护自然资源是必要的。尽管许多研究表明光学和SAR遥感数据可以估算作物参数，但尚未对这些数据进行比较或组合以预测作物物候阶段。尽管SAR极化数据对作物物候阶段具有很高的敏感性，但尚无研究使用高时间分辨率数据。可免费获得的SAR Sentinel-1（S-1）和光学Sentinel-2（S-2）时间序列提供了独特的机会，可以每周以高空间分辨率监视作物物候。这项研究的目的是评估单独使用S-1数据，单独使用S-2数据以及将其组合用于预测小麦和油菜籽物候期的潜力。我们首先分析了根据S-2数据得出的光谱带，植被指数和叶面积指数（LAI）的时间剖面，以及根据S-1数据得出的后向散射系数和极化指标。然后，使用增量程序来估计S-1和S-2特征对小麦和油菜籽的主要和次要物候期分类的贡献。两种作物的结果均表明，结合S-1和2数据获得的分类（小麦和油菜籽的平均Kappa分别为0.53–0.82和0.74–0.92）比单独利用S-2数据获得的分类（平均Kappa）更准确。单独的数据分别为小麦和油菜籽= 0.54-0.75和0.67-0.86）或单独的S-1数据（小麦和油菜籽的平均kappa分别为0.48-0.61和0.61-0.64）。结合S-1和 2个数据可以更好地识别小麦分till的开始和结束以及油菜籽的成熟开始和结束。在S-2特征中，最重要的是小麦的LAI和油菜籽的NDVI。对于这两种作物，S2REP指数都是最重要的植被指数之一，而MCARI则不那么重要。对于S-1特征，结果强调了背向散射比（σ◦VH：σ◦VV）的巨大贡献以及使用极化指示剂（香农熵和跨度）监测油菜籽和小麦物候的价值。这项工作的主要新颖之处是使用S-1极化指示剂来识别小麦和油菜籽的物候阶段，并使用遥感数据绘制小麦和油菜籽的第二物候阶段。