Characterisation of Soils by the near Infrared Technique

Abstract

There are physical, physico-chemical and chemical parameters which can be used as a basis for the characterisations of soils. Some of these characterisation methods are slow, laborious and show very empiric character. Near infrared (NIR) reflectance spectroscopic methods were used for substitution of conventional soil analysis procedures. Seven different soil characteristics (humus, calcium carbonate, total nitrogen, dead water content, yarn test by Arany, mechanical composition, hygroscopic water content) were measured by NIR using Hungarian and ISO standards as reference methods. It was concluded that the humus content, which measures the total organic content of soil can be predicted from NIR spectra with good accuracy (SEP = 0.3%). There are two parameters (mechanical composition and hygroscopicity) which showed significant correlation with the changes in the status of water in soil and they were detectable with acceptable accuracy based on their NIR spectra. The other four parameters were not predictable either due to the lack of spectroscopic information or to the poor quality of reference methods. The developed method was very sensitive to the physical status of samples and an accurate standard sample preparation was needed. Spectral libraries were developed and validated for a quick and non-destructive qualification and selection of soil samples.

Determination of carbon and nitrogen contents in Alfisols, Oxisols and Ultisols from Africa and Brazil using NIRS analysis: Effects of sample grinding and set heterogeneity

“…However, data from Coûteaux et al (2003) for a fairly homogeneous set indicated that prediction accuracy was greater for Ct than for Nt (validation R 2 and CV were 0.99 vs. 0.98 and 6% vs. 10%, respectively). Similarly, data from Salgó et al (1998), Fystro (2002 and Russell (2003), for heterogeneous sample sets, showed more accurate prediction of Ct (or Corg) than Nt. Considering literature data as well as results from this study, it was thus difficult to find a clear trend when comparing prediction accuracy for Ct (or Corg) and Nt using sieved samples.…”

“…Moreover, some other authors have reported that normalisation did not improve the calibration (Shepherd and Walsh, 2002). First derivatives of the spectra have frequently been used for Ct, Corg and Nt prediction (Chang et al, 2001;Reeves et al, 2002;Coûteaux et al, 2003;Russell, 2003;Móron and Cozzolino, 2004) but second derivatives could sometimes result in more accurate results (Salgó et al, 1998;Fystro, 2002). Nevertheless, studying 2 mm sieved topsoil n 1 is the number of samples after the elimination of calibration outliers.…”

“…There also seems to be some confusion about the data preprocessing, such as derivation and scatter correction in order to optimise spectral features. For example, for NIRS prediction of soil carbon content, many authors used the first derivatives of the spectra (Chang et al, 2001;Reeves et al, 2002;Shepherd and Walsh, 2002;Coûteaux et al, 2003;Russell, 2003;Móron and Cozzolino, 2004) while others preferred the second derivatives (Salgó et al, 1998;Fystro, 2002). This paper assesses the effect of sample preparation (0.2 mm grinding vs. 2 mm sieving) and set heterogeneity on the accuracy of NIRS determination of soil total carbon (Ct) and soil total nitrogen (Nt) contents for a range of Alfisols, Oxisols and Ultisols from Africa and Brazil.…”

Soil carbon measurement in clods and sieved samples in a Mediterranean Vertisol by Visible and Near-Infrared Reflectance Spectroscopy

“…The first application of NIRS technology for soil analysis was reported in the 1960s (Bowers and Hanks, 1965), with accurate predictions of soil total C and total N contents using NIRS published in the 1970s (AlAbbas et al, 1972), and followed by a greater number of publications in the 1990s (Morra et al, 1991;Sudduth and Hummel, 1993). Numerous papers have addressed soil total C (Dalal and Henry, 1986;Reeves et al, 2002;Russell, 2003;Barthès et al, 2006), organic C (Confalonieri et al, 2001;Shepherd and Walsh, 2002) and inorganic C content (Ben-Dor and Banin, 1990;Salgó et al, 1997;McCarty et al, 2002;Brown et al, 2006) using NIRS in different types of soil and in different environments.…”

Accounting for surface roughness effects in the near-infrared reflectance sensing of soils

“…Some researchers see in this spectroscopic technique a fast, relatively inexpensive alternative to traditional laboratory methods for the determination of a wide range of soil parameters, like clay, silt, sand, nitrogen, organic matter, moisture, and heavy metal content, to name only a few (e.g., Dalal and Henry, 1986;Ben-Dor and Banin, 1995;Salgó et al, 1998;Hummel et al, 2001;Reeves and McCarty, 2001;Cozzolino and Morón, 2003;Brunet et al, 2007).…”