14 Application of NIR in Agriculture
333
14.2.1 Soil Analysis by NIR—A Technique in Development
There is great interest for farmers having a good knowledge of the quality of their soil.
Specifically, it is strategic and crucial to be able to adequately determine the physical
and chemical characteristics and properties of the soils proposed and used for agricultural purposes. Based on different quality parameters, farmers will decide which crop
to plant and which agricultural practices to follow in order to maximise the intrinsic
value of their plots. Adequate management of inputs is of prime importance in order
to maximise the benefits and to maintain the soil fertility. Soil analysis by classical techniques is tedious, time consuming and requires standardised methods using
significant amounts of reagents, making such analyses particularly unfriendly for
the environment. Several reviews have been published dealing with the perspectives
offered by NIRS for soil analysis [5–7]. Chabrillat et al. reviewed the achievements
and perspectives in soil mapping and monitoring based on imaging spectroscopy
from air and space-borne sensors [8]. This review underlines that the next generation
of hyperspectral satellite sensors could greatly help to meet the demand for global
soil mapping and monitoring. Recently, Hutengs and co-authors have compared the
performance of portable NIR and mid-infrared (MIR) devices for assessment of
organic carbon in soils. Even though a comparison has not been made on the same
sample batches, they conclude with the fact that handheld MIR gives significantly
more accurate results for on-field analyses. This work clearly demonstrates that the
prediction of soil properties (whatever the technique) is improved when the samples
are dried and ground (RMSE = 0.155 for NIR) instead of being analysed in situ
(RMSE = 0.243 for NIR). Ongoing projects also aim to develop the application
of NIR to tackle the current challenges faced by soil analysis. For instance, the
INDIGGES project (http://www.cra.wallonie.be/fr/indigges) aims to develop direct
and indirect indicators to evaluate greenhouse gas emissions and carbon storage at the
farm. In this project, the NIRS technique is tested for characterising both fresh and
dried soils. NIR benchtop, portable, handheld and hyperspectral imaging devices are
tested (Fig. 14.1). Another challenging issue where NIR approaches are interesting
is the detection of foreign material in soil. An example is the detection, identification
and quantification of macro- and micro-plastics in cultivated lands. Micro-plastics
are emerging as persistent contaminants of increasing concern. They come from
mulching film, sludge, wastewater irrigation and atmospheric deposition. It seems
that the micro-plastics influence soil physio-chemistry and biota [9].
A key issue regarding soil analysis is the heterogeneity of samples proposed for
classical or NIR analyses. It is crucial to take care and devote the required resources
to be sure that the sample specimen analysed is representative of the soil for which
we want to determine the physical and chemical properties.
Table 14.1 presents the performance of equations used in the REQUASUD
network for soil analysis (2018 status). Only the parameters for which a RPD sec
is higher than 2 are presented, i.e. organic carbon content, nitrogen, CEC and clay
content. The performance for soil from grasslands and lands under cultivation is
given [6, 10]. The NIR analysis protocol for soil and the spectral database has been
333
14.2.1 Soil Analysis by NIR—A Technique in Development
There is great interest for farmers having a good knowledge of the quality of their soil.
Specifically, it is strategic and crucial to be able to adequately determine the physical
and chemical characteristics and properties of the soils proposed and used for agricultural purposes. Based on different quality parameters, farmers will decide which crop
to plant and which agricultural practices to follow in order to maximise the intrinsic
value of their plots. Adequate management of inputs is of prime importance in order
to maximise the benefits and to maintain the soil fertility. Soil analysis by classical techniques is tedious, time consuming and requires standardised methods using
significant amounts of reagents, making such analyses particularly unfriendly for
the environment. Several reviews have been published dealing with the perspectives
offered by NIRS for soil analysis [5–7]. Chabrillat et al. reviewed the achievements
and perspectives in soil mapping and monitoring based on imaging spectroscopy
from air and space-borne sensors [8]. This review underlines that the next generation
of hyperspectral satellite sensors could greatly help to meet the demand for global
soil mapping and monitoring. Recently, Hutengs and co-authors have compared the
performance of portable NIR and mid-infrared (MIR) devices for assessment of
organic carbon in soils. Even though a comparison has not been made on the same
sample batches, they conclude with the fact that handheld MIR gives significantly
more accurate results for on-field analyses. This work clearly demonstrates that the
prediction of soil properties (whatever the technique) is improved when the samples
are dried and ground (RMSE = 0.155 for NIR) instead of being analysed in situ
(RMSE = 0.243 for NIR). Ongoing projects also aim to develop the application
of NIR to tackle the current challenges faced by soil analysis. For instance, the
INDIGGES project (http://www.cra.wallonie.be/fr/indigges) aims to develop direct
and indirect indicators to evaluate greenhouse gas emissions and carbon storage at the
farm. In this project, the NIRS technique is tested for characterising both fresh and
dried soils. NIR benchtop, portable, handheld and hyperspectral imaging devices are
tested (Fig. 14.1). Another challenging issue where NIR approaches are interesting
is the detection of foreign material in soil. An example is the detection, identification
and quantification of macro- and micro-plastics in cultivated lands. Micro-plastics
are emerging as persistent contaminants of increasing concern. They come from
mulching film, sludge, wastewater irrigation and atmospheric deposition. It seems
that the micro-plastics influence soil physio-chemistry and biota [9].
A key issue regarding soil analysis is the heterogeneity of samples proposed for
classical or NIR analyses. It is crucial to take care and devote the required resources
to be sure that the sample specimen analysed is representative of the soil for which
we want to determine the physical and chemical properties.
Table 14.1 presents the performance of equations used in the REQUASUD
network for soil analysis (2018 status). Only the parameters for which a RPD sec
is higher than 2 are presented, i.e. organic carbon content, nitrogen, CEC and clay
content. The performance for soil from grasslands and lands under cultivation is
given [6, 10]. The NIR analysis protocol for soil and the spectral database has been