There is enormous scope and prospective of crop yield prediction at finer scale for both farm-level crop management as well as for crop insurance settlement at gram panchayat (GP) level in India. Now with the advent of satellite sensors like the MSI from Sentilnel-2 with fine spatial resolution, the possibility of generating frequent information on crop condition at field scale is increasing. This study demonstrated the combined use of high-resolution data from Sentinel-2 (10 m and 20 m); moderate-resolution data from MODIS (500 m) and coarser-resolution radiation data from INSAT-3D (4 km) for estimating yield of three major crops of India at GP and taluka level using a semi-physical model based on crop-specific radiation use efficiency. The novelty of this study lies in the data fusion approach using parameters from multiple spatial resolution of Geostationary and Lower Earth Orbiting satellites within the basic semi-physical model framework. The methodology has been demonstrated in Cuttack district of Odisha for rice; Rajkot district of Gujarat for cotton; and Indore district of MP and Fatehabad district of Haryana for wheat. We validated our result at GP, taluka and district level. At GP level, the root mean square error (RMSE) was found to be 16.5% for rice and 5.8% for wheat in Indore district. At taluka level, the RMSE was found to be 15%, 5.7%, 4.4% and 7.4% for rice, wheat in Indore district, wheat in Fatehabad district and cotton, respectively. The study concluded that high resolution remote sensing data would be of immense use for finer scale yield estimation, which can be aggregated at GP and taluka level with satisfactory accuracy (p = 95%).

在印度，对于农场一级的作物管理以及在gram panchayat（GP）级别的作物保险结算，在更精细的规模上对作物产量的预测具有巨大的范围和前景。现在，随着具有良好空间分辨率的卫星传感器（如Sentilnel-2的MSI）的出现，在田间尺度上生成有关作物状况的频繁信息的可能性正在增加。这项研究证明了结合使用Sentinel-2（10 m和20 m）的高分辨率数据。来自MODIS的中等分辨率数据（500 m）和来自INSAT-3D的较高分辨率数据（4 km），使用基于作物特定辐射的半物理模型估算了GP和taluka水平的印度三种主要农作物的产量使用效率。这项研究的新颖之处在于在基本半物理模型框架内使用对地静止和近地轨道卫星的多个空间分辨率中的参数进行数据融合的方法。该方法已经在奥里萨邦的Cuttack区进行了稻米的示范。古吉拉特邦拉杰科特地区的棉花；和MP的Indore区和哈里亚纳邦的Fatehabad区的小麦。我们在GP，taluka和地区一级验证了我们的结果。在GP级别，印多尔地区的水稻均方根误差（RMSE）为16.5％，小麦为5.8％。在塔卢卡水平上，稻米，印多尔地区的小麦，法特哈巴德地区的小麦和棉花的RMSE分别为15％，5.7％，4.4％和7.4％。研究得出的结论是，高分辨率遥感数据可用于更精细的产量估算，p  = 95％）。