Several remote sensing-based methods have been developed to apply site-specific nitrogen (N) fertilization in crops. They consider spatial and temporal variability in the soil-plant-atmosphere continuum to modulate N applications to the actual crop nutrient status and requirements. However, deriving fertilizer N recommendations exclusively from remote proximal and remote sensing data can lead to substantial inaccuracies and new, more complex approaches are needed.

Therefore, this study presents an improved approach that integrates crop modelling, proximal sensing and forecasts weather data to manage site-specific N fertilization in winter wheat. This improved approach is based on four successive steps: (1) optimal N supply is estimated through the DSSAT crop model informed with a combination of observed and forecast weather data; (2) actual crop N uptake is estimated using proximal sensing; (3) N prescription maps are created merging crop model and proximal sensing information, considering also the contribution of the soil N mineralisation; (4) N-Variable Rate Application (N-VRA) is implemented in the field. A VRA method based on DSSAT fed with historical weather data and a business-as- usual uniform fertilization were also compared.

The methods were implemented in a 23.4 ha field in Northern Italy, cropped to wheat and characterized by large soil variability in texture and organic matter content. Results indicated that the model-based approaches consistently led to higher yields, agronomic efficiencies and gross margins than the uniform N application rate. Furthermore, the proximal sensing-based approach allowed capturing of the spatial variability in crop N uptake and led to a substantial reduction of the spatial variability in yield and protein content. This study grounds the development of web-based software as a friendly tool to optimize the N variable rate application in winter cereals.

已经开发了几种基于遥感的方法来在作物中应用特定地点的氮 (N) 施肥。他们考虑了土壤-植物-大气连续体的空间和时间变异性，以根据实际作物养分状况和需求调节氮的应用。然而，仅从远程近端和遥感数据推导肥料 N 建议可能会导致严重的不准确性，因此需要新的、更复杂的方法。

因此，本研究提出了一种改进的方法，该方法整合了作物建模、近端传感和预测天气数据，以管理冬小麦中特定地点的氮肥施肥。这种改进的方法基于四个连续步骤：（1）通过结合观测和预报天气数据的 DSSAT 作物模型估计最佳氮供应；(2)使用近端传感估计实际作物氮吸收量；(3) 结合作物模型和近端传感信息创建 N 处方图，同时考虑土壤 N 矿化的贡献；(4)在现场实施N-Variable Rate Application (N-VRA)。还比较了基于 DSSAT 输入历史天气数据的 VRA 方法和照常进行的均匀施肥。

这些方法在意大利北部 23.4 公顷的土地上实施，种植小麦，土壤质地和有机质含量差异很大。结果表明，与统一施氮量相比，基于模型的方法始终能带来更高的产量、农艺效率和毛利率。此外，基于近端传感的方法允许捕获作物氮吸收的空间变异性，并导致产量和蛋白质含量的空间变异性显着降低。这项研究将基于网络的软件的开发作为一种友好的工具来优化冬季谷物中的 N 可变比率应用。