Application of visible near-infrared absorbance spectroscopy for the determination of Soil pH and liming requirements for broad-acre agriculture

• Published in: Precision agriculture 2022-02, Vol.23 (1), p.194-218
• Main Authors: Sleep, Bethany, Mason, Sean, Janik, Les, Mosley, Luke
• Format: Article
• Language: English
• Subjects: Acidic soils; Acidification; Agriculture; Agronomy; Atmospheric Sciences; Biomedical and Life Sciences; Chemistry and Earth Sciences; Computer Science; Cores; Fertility; Infrared reflection; Infrared spectroscopy; Least squares method; Life Sciences; Lime; Liming; Near infrared radiation; pH effects; Physics; Regression models; Remote Sensing/Photogrammetry; Sampling; Sampling methods; Soil acidification; Soil chemistry; Soil fertility; Soil investigations; Soil lime; Soil management; Soil pH; Soil Science & Conservation; Soil surfaces; Soils; Spatial discrimination; Spatial resolution; Spectra; Spectroscopy; Statistics for Engineering
• ISSN: 1385-2256; 1573-1618
• DOI: 10.1007/s11119-021-09834-7

Soil acidification is a major and growing concern in many cropping regions globally. Whilst spatial variability in acidification is a common consideration in the management of soil health and fertility at sub-paddock scale, insufficient focus has been directed toward the identification of this variability. Suitability of portable visible near infrared reflectance (vis–NIR) spectroscopy was assessed in this study as a potential technique to achieve rapid, precise, inexpensive and spatially specific quantification of key soil parameters to inform lime requirements. Spectral fingerprints were taken using a 1 ha grid sampling approach, with four sampling protocols investigated as follows, scans: (i) directly on cleared soil surfaces; (ii) on 0–100 mm undisturbed cores; (iii) on dried 0–100 mm cores; and finally (iv) on dried, ground, sieved and mixed cores. Data was analysed using a partial least squares regression (PLSR) model to identify the strength of linear relationship between reference chemistry data and predictions derived from spectral readings. Lime requirement maps using vis–NIR predictions were then theoretically compared against traditional aggregated sampling patterns while considering the trade-offs between accuracy, economics and agronomy associated with the identification of spatial variability. The vis–NIR measurements demonstrated moderate predictive capabilities in field for determining pH (R 2  = 0.3–0.5) and liming requirements (R 2  = 0.5–0.6) rapidly at high spatial resolution. Vis–NIR in field mapping techniques, which enable the use of site specific management of soil resources, were found to positively redirect lime resources from alkaline areas toward acidic areas of the paddock, resulting in minimal difference to overall expenditure on lime purchase and potential for increased agronomic benefits over the long-term. Further spectral library development, calibration, and research on in-field sampling methods is recommended.