Applying the correct dose of nitrogen (N) fertilizer to crops is extremely important. The current predictive models of yield and soil–crop dynamics during the crop growing season currently combine information about soil, climate, crops, and agricultural practices to predict the N needs of plants and optimize its application. Recent advances in remote sensing technology have also contributed to digital modelling of crop N requirements. These sensors provide detailed data, allowing for real-time adjustments in order to increase nutrient application accuracy. Combining these with other tools such as geographic information systems, data analysis, and their integration in modelling with experimental approaches in techniques such as machine learning (ML) and artificial intelligence, it is possible to develop digital twins for complex agricultural systems. Creating digital twins from the physical field can simulate the impact of different events and actions. In this article, we review the state-of-the-art of modelling N needs by crops, starting by exploring N dynamics in the soil−plant system; we demonstrate different classical approaches to modelling these dynamics so as to predict the needs and to define the optimal fertilization doses of this nutrient. Therefore, this article reviews the currently available information from Google Scholar and ScienceDirect, using relevant studies on N dynamics in agricultural systems, different modelling approaches used to simulate crop growth and N dynamics, and the application of digital tools and technologies for modelling proposed crops. The cited articles were selected following the exclusion criteria, resulting in a total of 66 articles. Finally, we present digital tools and technologies that increase the accuracy of model estimates and improve the simulation and presentation of estimated results to the manager in order to facilitate decision-making processes.

Nitrogen use efficiency (NUE) is a central issue to address regarding the nitrogen (N) uptake by crops, and can be improved by applying the correct dose of fertilizers at specific points in the fields according to the plants status. The N nutrition index (NNI) was developed to diagnose plant N status. However, its determination requires destructive, time-consuming measurements of plant N content (PNC) and plant dry matter (PDM). To overcome logistical and economic problems, it is necessary to assesses crop NNI rapidly and non-destructively. According to the literature which we reviewed, it, as well as PNC and PDM, can be estimated using vegetation indices obtained from remote sensing. While sensory techniques are useful for measuring PNC, crop growth models estimate crop N requirements. Research has indicated that the accuracy of the estimate is increased through the integration of remote sensing data to periodically update the model, considering the spatial variability in the plot. However, this combination of data presents some difficulties. On one hand, at the level of remote sensing is the identification of the most appropriate sensor for each situation, and on the other hand, at the level of crop growth models is the estimation of the needs of crops in the interest stages of growth. The methods used to couple remote sensing data with the needs of crops estimated by crop growth models must be very well calibrated, especially for the crop parameters and for the environment around this crop. Therefore, this paper reviews currently available information from Google Scholar and ScienceDirect to identify studies relevant to crops N nutrition status, to assess crop NNI through non-destructive methods, and to integrate the remote sensing data on crop models from which the cited articles were selected. Finally, we discuss further research on PNC determination via remote sensing and algorithms to help farmers with field application. Although some knowledge about this determination is still necessary, we can define three guidelines to aid in choosing a correct platform.