Abstract

Calibrations were developed to predict chemical and physical properties from near infrared spectra of an extensive spatial dataset of Scottish soils. For this purpose we used a spectral library of 1246 soil samples collected throughout Scotland in two campaigns: 546 samples collected in a 10 km grid between 1978 and 1988 (NSIS1); and 700 samples collected between 2007 and 2009 during a re-sample of the same sites but in a 20 km grid (NSIS2). The samples were split into validation (N = 250) and calibration (N = 996) sets, and global partial least squares regression (PLSR) was performed in combination with spectral pre-processing treatments, namely first or second derivative and, optionally, standard normal variate and de-trending or multiplicative scatter correction treatments. For the local model, the calibration set (N = 996) was split into test (N = 121) and library (N = 875). Local calibration was performed using PLSR in batches that iteratively selected from the library between 75 and 425 reference spectra, in increments of 50, in combination with spectral preprocessing treatments. Both global and local models were validated on the same validation set (N = 250). We succeeded in developing predictive calibrations with r² of validation greater than 0.60 for total elemental C and N, loss on ignition (450°C and 900°C), exchangeable H and Mg, moisture content, pH (in H₂O and CaCl₂) and dry bulk density. Promising results were also achieved for the prediction of total P, aqua regia-extracted Mg and P, and ammonium oxalate-extracted Al and Si, although these calibrations were highly biased. Predictive results for exchangeable Ca, sand, silt, clay, K (aqua regia), mineralisable N and δ¹⁵N were very informative, but not robust enough for predictive purposes. The lowest performance was observed for exchangeable Al and K, δ¹³C, Na (exchangeable and aqua regia), P (ammonium oxalate) and in particular for Fe and Mn (both exchangeable and extracted in ammonium oxalate). We found that local calibration was superior to global, in terms of accuracy and number of soil attributes successfully calibrated, and we observed better results when first derivative was applied to pre-process the spectra. Further work, including the expansion of the dataset or testing alternative calibration methods or spectral ranges, will be pursued.

© 2016 The Author(s)

Accurate and rapid detection of soil and fertilizer properties based on visible/near-infrared spectroscopy

Zhidan Lin, Rujing Wang, Yubing Wang, Liusan Wang, Cuiping Lu, Yang Liu, Zhengyong Zhang, and Likai Zhu
Appl. Opt. 57(18) D69-D73 (2018)

Evaluation of univariate and multivariate calibration strategies for the direct determination of total carbon in soils by laser-induced breakdown spectroscopy: tutorial

Wesley Nascimento Guedes, Diego Victor Babos, Vinícius Câmara Costa, Carla Pereira De Morais, Vitor da Silveira Freitas, Kleydson Stenio, Alfredo Augusto Pereira Xavier, Luís Carlos Leva Borduchi, Paulino Ribeiro Villas-Boas, and Débora Marcondes Bastos Pereira Milori
J. Opt. Soc. Am. B 40(5) 1319-1330 (2023)

Quantitative analysis of pH value in soil using laser-induced breakdown spectroscopy coupled with a multivariate regression method

Cuiping Lu, Gang Lv, Chaoyi Shi, Duoyang Qiu, Feixiang Jin, Man Gu, and Wen Sha
Appl. Opt. 59(28) 8582-8587 (2020)

Collinear double-pulse laser-induced breakdown spectroscopy based Cd profiling in the soil

Jie Ren, Zihan Yang, Yanru Zhao, and Keqiang Yu
Opt. Express 30(21) 37711-37726 (2022)

Spark-induced breakdown spectroscopy and multivariate analysis applied to the measurement of total carbon in soil

Morgan S. Schmidt, Kellen J. Sorauf, Keith E. Miller, David Sonnenfroh, Richard Wainner, and Amy J. R. Bauer
Appl. Opt. 51(7) B176-B182 (2012)