Direct determination of soil carbon stocks by NIRS: an advanced approach.

FREITAS, V. S.; BENTO, L. R.; NAPOLITANO, V.; OLIVEIRA, P. P. A.; PEZZOPANE, J. R. M.; BERNARDI, A. C. de C.; MILORI, D. M. B. P.; MARTIN NETO, L.

Direct determination of soil carbon stocks by NIRS: an advanced approach.

Autores

BERNARDI, A. C. de C.

MILORI, D. M. B. P.

MARTIN NETO, L.

Descripción

Soil carbon stocks are obtained as the product of carbon content, soil bulk density, and specific layer thickness. The determination of soil bulk density requires field sampling procedures that can be labor-intensive and time- consuming, particularly when sampling deep soil layers or large areas. In this study, near-infrared spectroscopy (NIRS) was used to directly estimate volumetric soil carbon (Ps, kg C m 3 ), allowing easier estimation of soil carbon stocks. This approach aims to reduce the difficulties and high cost of traditional soil sampling, which involves opening trenches and collecting undisturbed soil samples using a volumetric ring for soil bulk density determination, as well as using conventional techniques as CHN elemental analyzer. A dataset of 576 soil samples (480 for calibration and 96 for validation) was measured by NIRS, collected from a long-term field experiment encompassing different tropical pasture management systems and a reference native vegetation area: recovered pasture (RP), intercropped pasture (CON), extensive pasture (EX), and native forest (FO). The best model, employing the spectral first derivative as a preprocessing step and Partial Least Squares (PLS) regression as the multivariate algorithm, achieved performance of R 2 2.85 in calibration, and R 2 = 0.79, RMSEC = 4.5 kg C m =0.85, RMSEP = 4.1 kg C m 3 3 , RPD = 2.16 and RPIQ = , RPD = 2.53 and RPIQ = 3.63 in validation. Carbon stocks calculated from NIRS-predicted values in the validation dataset showed no significant differences from reference measurements. These findings demonstrate that NIRS can provide reliable, rapid, and cost-effective estimates of soil carbon stocks, highlighting its potential application in carbon credit assessments.
Financiamento: FAPESP (Projeto 2023/02444-5 e 2023/18440-9). Agradecemos também ao Laboratório Nacional de Agrofotônica (Sisfoton/CNPq: 440226/2021–0). V.S.F. agradece a bolsa de pós-doutorado da FAPESP (número do processo: 2024/14266-7).