Chapter One - Current sensor technologies for in situ and on-line measurement of soil nitrogen for variable rate fertilization: A review
Introduction
Nitrogen (N) is the most crucial nutrient for plant growth and crop yield in agriculture, and its availability to a crop is mainly determined by the soil mineral N (SMN) concentration at any given time during the growing season. However, SMN is traditionally viewed as the most difficult nutrient to manage because of the very large spatial and temporal variability. Because of this, N is uniformly applied across the entire field, based on an N recommendation that takes into account a number of factors, of which SMN is probably the most important one. Traditionally, SMN is measured on one composite sample taken from 0 to 30 cm (or sometimes 0–60 or 0–90 cm) depth, and the N recommendation may or may not take into account other factors related to specific soil and management factors. With uniform N fertilization, parts of the field with high fertility or N content will receive over-application of fertilizers leading to inefficient N use and environmental pollution risk including N leaching, whereas other parts will receive sub-optimal fertilizer rates, leading to reduced crop performance and yield (Koch et al., 2004). One way to modify traditional practices and contribute to the environmental remediation is to implement precision agriculture (PA) for within-field variability management (Delin and Stenberg, 2014). PA includes the use of technology for site-specific application of farm input resources that helps to manage production inputs (e.g., N fertilizers), site-specifically, in a manner and levels specific to the site, with optimal goals of maximizing yield and nutrient use efficiency (Bongiovanni and Lowenberg-Deboer, 2004). However, for the implementation of precision management of N fertilization, data on SMN is required at high sampling resolution to allow assessment of the spatial and temporal variability in available N to the crop. Traditional methods of laboratory analyses to determine SMN are not suitable, as they are costly, time-consuming, slow, require experience operator and dispose of chemicals to the environment. Therefore, a new paradigm in data collection using new sensing technologies is required to overcome the shortcomings of traditional laboratory methods. This can be achieved by means of remote sensing, proximal sensing or the fusion of data from both sensing technologies to acquire information not only on soil N but other affecting parameters and crop attributes.
Data collected by proximal sensors are valuable to farmers and land managers to build site-specific databases that show the association between soil nutrient concentration, plant reaction and crop development during the growing season with resulting yield (Viscarra Rossel and Bouma, 2016). Over the years, soil sensors have become smaller, more rugged, faster, more accurate, more energy-efficient, wireless and more intelligent (Viscarra Rossel and Bouma, 2016). Proximal soil sensors (PSS) enable the measurement of key soil properties at high sampling resolution, to allow the quantification of within-field spatial and temporal variability necessary for PA applications. They concern the measurement of soil attributes while the sensor is in direct contact with the soil or within 2–3 m distance from the soil surface (Viscarra Rossel et al., 2011). Proximal soil sensing can be divided into laboratory, in situ (non-mobile), or on-line (mobile) measurement modes. While the laboratory mode requires soil samples to be transferred from its natural position to the laboratory for pre-treatment and laboratory chemical analysis, in situ and on-line modes refer to the measurement of soil in their natural position. Although the laboratory measurement with PSS has advantages over the chemical analyses, they are still demanding extra sample transportation and preparation (e.g., drying, grinding and sieving). Advancements in the sensing technology have led to the development of in situ and on-line sensors. While the former group of sensors allows performing field measurements without the need to take soil samples to a laboratory, it is still time-consuming and does not allow to produce high sampling resolution data in a short period of time. The use of on-line sensors enables a greater amount of data to be obtained (> 500 samples per ha) and allows to generate more detailed maps for better and science-based within-field soil and crop management. Perhaps one of the main advantages of the on-line sensors in addition to the high-resolution maps, is that they can also be used for real-time control of inputs (Adamchuk and Jasa, 2002), such as the case of the sensor-based variable rate (VR) applications.
Kuang et al. (2012) categorized PSS into five main categories: (1) reflectance, (2) conductivity, resistivity and permittivity, (3) passive radiometric, (4) soil strength and (5) electrochemical. Each of these categories targets some specific soil parameters. Among these different PSS technologies, the reflectance and electrochemical-based sensors were proven to accurately quantify total nitrogen (TN) and SMN in soils, respectively (Kuang et al., 2012). For in situ and on-line applications, the visible and near-infrared (vis-NIR) spectroscopy (vis-NIRS) technique seems the best candidate to measure TN, because it is robust against vibration, less sensitive to moisture compared to mid-infrared (MIR) spectroscopy (MIRS), which has not been implemented in on-line measurement mode so far. However, they can only measure TN. On-line measurement of SMN with electrochemical sensing has got limited research attention because of difficulties associated with preparing a soil solution, while travelling with a speed of 3–5 km/h, in addition to the time required for electrodes exchange reaction. Although few studies have reported the development of on-line sensing platforms (Adamchuk et al., 2005; Adsett et al., 1999; Sibley et al., 2008, Sibley et al., 2009), problems associated with duration of extraction and analysis, noise and achievable level of accuracy are the most affecting factors hindering the development of robust electrochemical sensing system for on-line measurement of SMN.
The development of multivariate and machine learning calibration models to assess key soil fertility parameters including TN using sensor data (e.g., spectra, images, output voltage) have greatly improved measuring the complexity of the N-cycle in agricultural processes. Moreover, data fusion (DF) analysis has increased the accuracy of prediction and opened new horizons for nutrient management. DF involves complex processes related to data collection, storage and analysis. Different sources of information are merged using mathematical, statistical and geo-statistical techniques to create relationships between parameters that yield robust prediction models. In addition, DF can be used to combine soil and crop data to obtain a better view of the N dynamics in soil and plants (De Benedetto et al., 2013; Pantazi et al., 2015). It has been used for creating fertility and recommendation maps for variable-rate nitrogen fertilization (VRNF), e.g., by means of management zones (MZ) maps (Nawar et al., 2017). This solution is optimal when no data on SMN is available; hence, a field should be divided into zones having different fertility classes, necessary for VRNF.
This paper provides a critical review of the current state-of-the-art of technologies for the measurement of TN and SMN in soils. The focus will be on field applications, including the in situ and on-line sensing modes. The review will present spectra modeling approaches for the prediction of key soil fertility parameters associated with N fertilization and the potential of DF approaches for the delineation of MZs in deriving VRNF recommendations based-on decision support systems. Finally, the review introduces for the first time a new approach for VRNF by combining in advance developed soil fertility maps obtained from PSS with real-time collected crop growth data using proximal crop sensors to calculate recommendations and implement VRNF in real-time.
Section snippets
Traditional methods
Traditional laboratory methods for TN measurement based-on wet chemistry include wet-oxidation or dry-combustion techniques. Both Kjeldahl-digestion and Dumas-combustion are the most common methods, respectively. The Kjeldahl digestion method relays on the conversion of organic and inorganic N into NH4+ followed by the measurement of its concentration. The dry-combustion method involves initial oxidation followed by a reduction of NOx, where N2 gases are measured (McGill and Figueiredo, 1993).
Spectral data analysis
In order to obtain quantitative or qualitative information about the soil TN status from the vis-NIR and MIR spectral data, spectra pre-processing and modelling are required. Researchers observed from vis-NIR spectral analysis that higher reflectance in the entire spectral range (350–2500 nm) is associated with lower TN content and vice versa (Fystro, 2002; Gillon et al., 1999; Udelhoven et al., 2003). Calibration techniques of spectra can be divided into linear and nonlinear approaches. The
Integration and future prospects
One of the most important challenges for VRNF is the on-line measurement of SMN, necessary to create high-resolution recommendation maps. As discussed above that the electrochemical sensors are proven technologies for SMN measurement under in situ (non-mobile) and laboratory conditions (Table 2). However, less success was reported for the on-line measurement mode. From the literature review above, it can be concluded that ISEs can be used for successful measurement of SMN. However, the majority
Conclusions
This review shows that optical reflectance and electrochemical sensors have been widely used for soil nitrogen (N) assessment. These sensor technologies can measure soil total nitrogen (TN) and soil mineral nitrogen (SMN), respectively, with high accuracy under laboratory conditions, despite less accuracy is to be expected under both the in situ and on-line measurements modes. When electrochemical sensors are adopted for on-line measurement of SMN, a challenging task of preparing a soil
Acknowledgment
Authors acknowledge the financial support received from the Research Foundation—Flanders (FWO) for Odysseus I SiTeMan Project (Nr. G0F9216N).
References (124)
Direct measurement of soil chemical properties on-the-go using ion-selective electrodes
Comput. Electron. Agric.
(2005)- et al.
Combining frequency domain reflectometry and visible and near infrared spectroscopy for assessment of soil bulk density
Soil Tillage Res.
(2014) Eliminating the interference of soil moisture and particle size on predicting soil total nitrogen content using a NIRS-based portable detector
Comput. Electron. Agric.
(2015)- et al.
Application of ion sensitive field effect transistor based sensors to soil analysis
Comput. Electron. Agric.
(2001) - et al.
Near-infrared (NIR) and mid-infrared (MIR) spectroscopic techniques for assessing the amount of carbon stock in soils - critical review and research perspectives
Soil Biol. Biochem.
(2011) - et al.
Real-time multi ISFET/FIA soil analysis system with automatic sample extraction
Comput. Electron. Agric.
(2001) Use of EMI, gamma-ray emission and GPS height as multi-sensor data for soil characterisation
Geoderma
(2012)Real-time measurement of soil attributes using on-the-go near infrared reflectance spectroscopy
Comput. Electron. Agric.
(2008)- et al.
Near infrared reflectance spectroscopy for determination of organic matter fractions including microbial biomass in coniferous forest soils
Soil Biol. Biochem.
(2003) An approach for delineating homogeneous zones by using multi-sensor data
Geoderma
(2013)