Crop growth models are powerful tools for predicting crop growth and yield. Gross primary production (GPP) is a major photosynthetic flux that is directly linked to crop grain yield. To better understand the potential of GPP for regional crop yield estimation, in this study, a novel crop data-model assimilation (CDMA) framework was proposed that assimilates accumulative GPP estimates from the satellite-based vegetation photosynthesis model (VPM) into the WOrld FOod STudies (WOFOST) model using the ensemble Kalman filter (EnKF) algorithm to estimate winter wheat GPP and grain yield. Results showed that the WOFOST simulated GPP agreed with the GPP_EC derived from eddy flux tower (R² = 0.74 and 0.47 in 2015 and 2016, respectively). Assimilating GPP_VPM into the WOFOST model improved site-scale GPP estimation (R² = 0.87 and 0.67 in 2015 and 2016, respectively), and also improved regional-scale winter wheat yield estimates (R² = 0.36 and 0.29; RMSE= 479 and 572 kg/ha in 2015 and 2016, respectively) compared with the open loop simulations (R² = 0.14 and 0.10; RMSE= 801 and 788 kg/ha in 2015 and 2016, respectively). Our study demonstrated that assimilation of remotely sensed GPP optimized the results of carbon simulation in the WOFOST model and highlighted the potential of GPP for regional winter wheat yield estimation using a data assimilation framework.

作物生长模型是预测作物生长和产量的有力工具。初级生产总值 (GPP) 是与作物谷物产量直接相关的主要光合通量。为了更好地了解 GPP 在区域作物产量估算中的潜力，在本研究中，提出了一种新的作物数据模型同化 (CDMA) 框架，该框架将来自基于卫星的植被光合作用模型 (VPM) 的累积 GPP 估算同化到 WOrld FOod STudies (WOFOST) 模型使用集成卡尔曼滤波器 (EnKF) 算法来估计冬小麦 GPP 和谷物产量。结果表明，WOFOST模拟的GPP与涡通量塔衍生的GPP _{EC一致（2015年和2016年R} ² = 0.74和0.47，分别）。同化 GPP _VPMWOFOST 模型改进了场地尺度的 GPP 估计（2015 年和 2016 年分别为 R ² = 0.87 和 0.67），并且还改进了区域尺度的冬小麦产量估计（R ² = 0.36 和 0.29；RMSE = 479 和 572 kg/分别在 2015 年和 2016 年公顷）与开环模拟（R ² = 0.14 和 0.10；RMSE = 801 和 788 公斤/公顷，分别在 2015 年和 2016 年）进行比较。我们的研究表明，遥感 GPP 的同化优化了 WOFOST 模型中的碳模拟结果，并突出了 GPP 使用数据同化框架估计区域冬小麦产量的潜力。