Soil organic carbon and total nitrogen multivariate modelling from diverse FT-NIR spectral dataset DOI Creative Commons
Gbenga Daniel ADEJUMO, David Bulmer, Preston Sorenson

и другие.

Geoderma Regional, Год журнала: 2024, Номер 38, С. e00834 - e00834

Опубликована: Июль 2, 2024

This study linked soil FT-NIR spectroscopy with organic carbon (SOC) and total nitrogen (TN) content in Saskatchewan (SK) agricultural soils, using a multivariate approach. Soil spectra were acquired along laboratory measurements of SOC TN from 1965 samples. Spectral data transformed variety common pre-treatment approaches: Savitszky-Golay, first second derivative, standard normal variate, multiplicative scatter correction continuous wavelet transform. Models next built cubist regression tree (Cubist), support vector machine (SVM), partial least square (PLSR) to evaluate the performance different pre-treatment/modelling approaches. The wavelets transform was best performing spectral treatment method for SK TN. For predictive model an extensive dataset, performed (R2 = 0.80 0.85) followed by SVM 0.77 PLSR 0.63 0.73). However, all models demonstrated same correlation between predicted observed values (CCC 0.87 0.93). consistent accuracy dataset suggests model's ability generalize well beyond it trained on. varies if zones Sk sites, this suggest need careful selection specific site or soil-zone on which should be trained. Additionally, also underscores influence factors sample size variability, such as coefficient variation, predictions.

Язык: Английский

Drivers of Soil Carbon Variability in North America’s Prairie Pothole Wetlands: A Review DOI Creative Commons
Chantel J. Chizen, Angela Bedard‐Haughn

Wetlands, Год журнала: 2025, Номер 45(1)

Опубликована: Янв. 1, 2025

Язык: Английский

Процитировано

1
Soil organic carbon and total nitrogen multivariate modelling from diverse FT-NIR spectral dataset DOI Creative Commons
Gbenga Daniel ADEJUMO, David Bulmer, Preston Sorenson

и другие.

Geoderma Regional, Год журнала: 2024, Номер 38, С. e00834 - e00834

Опубликована: Июль 2, 2024

This study linked soil FT-NIR spectroscopy with organic carbon (SOC) and total nitrogen (TN) content in Saskatchewan (SK) agricultural soils, using a multivariate approach. Soil spectra were acquired along laboratory measurements of SOC TN from 1965 samples. Spectral data transformed variety common pre-treatment approaches: Savitszky-Golay, first second derivative, standard normal variate, multiplicative scatter correction continuous wavelet transform. Models next built cubist regression tree (Cubist), support vector machine (SVM), partial least square (PLSR) to evaluate the performance different pre-treatment/modelling approaches. The wavelets transform was best performing spectral treatment method for SK TN. For predictive model an extensive dataset, performed (R2 = 0.80 0.85) followed by SVM 0.77 PLSR 0.63 0.73). However, all models demonstrated same correlation between predicted observed values (CCC 0.87 0.93). consistent accuracy dataset suggests model's ability generalize well beyond it trained on. varies if zones Sk sites, this suggest need careful selection specific site or soil-zone on which should be trained. Additionally, also underscores influence factors sample size variability, such as coefficient variation, predictions.

Язык: Английский

Процитировано

2