Current sensor technologies for in situ and on-line measurement of soil nitrogen for variable rate fertilization : a review

Recommendations for nitrogen (N) fertilization are based on the analysis of the soil mineral N (SMN) content of one composite sample that is assumed to represent the average concentration, but that almost by definition fails to properly represent the inherent spatial variability of the field. Improvement in N management can thus be achieved by high-resolution measurements of the soil N status. However, soil sampling, transportation, sample preparation and laboratory analysis are laborious, costly and time-consuming. Additionally, errors associated with all these steps (particularly sampling and sample preparation) are notorious sources of inaccuracy. In situ and on-line measurement modes could overcome these drawbacks. This paper presents a state-of-the-art review of proximal soil sensing technologies for in situ and on-line measurements of total nitrogen (TN) and mineral N in soils. Visible and near-infrared spectroscopy (vis-NIRS) and mid-infrared spectroscopy (MIRS) have been successfully used in in situ and on-line measurement platforms for determining TN. The review also focuses on data fusion (DF) approaches not only to improve the prediction performance of TN by vis-NIRS and MIRS but also to delineate management zones for site-specific N management. Electrochemical sensors such as ion-selective electrodes (ISE) or ion-selective field-effect transistors (ISFET) were found to be the most suitable to measure SMN. However, it was found that both sensor technologies are difficult to use for on-line measurements because they require the preparation of a soil solution and sufficient time delay is needed to reach a stable plateau. These limitations restrict the implementation of electrochemical sensing tools for site-specific N management, although they are the best methods to measure SMN, a needed parameter as a direct measure to determine N recommendations. Research on proximal soil sensing (PSS) for N measurement in soils should focus on the development of robust and accurate in situ and on-line sensors for SMN, whereas improvement in the performance of vis-NIRS and MIRS for TN prediction by using advanced modelling and DF techniques is recommended. This development is essential to promote the implementation of variable rate N fertilization (VRNF). The review suggested the fusion of on-line collected soil data with crop data, present and historical yield, topography and weather conditions as the optimal solution for map-based and sensor-based VRNF scenarios. Furthermore, a new approach for VRNF, designated as the map-sensor-based approach, is proposed. In this approach, management zones (MZ) maps for the soil is developed in advance at the beginning of the cropping season, and this information is combined with current crop growth status [e.g., normalized differential vegetation index (NDVI)], measured with a proximal crop sensor to calculate the recommendation and implement VRNF in real-time.

Publication year:2021

Accessibility:Closed