Vis-NIR spectroscopy for rapid soil carbon and nitrogen estimation in humid tropical landscape

Tropical Plant Biology · Springer Nature · 2026-09-12 · DOI: 10.1007/s10333-026-01095-2

Аннотация

Carbon is a key component of soil health, playing a crucial role in nutrient cycling, microbial activity, and soil fertility. Its distribution varies significantly across different landscapes due to variations in soil forming factors and land use practices. Conventional soil analysis methods are labour intensive, time consuming, and costly, stressing the need for rapid and cost effective alternatives. This study aimed to assess the spatial variability of soil organic carbon (SOC), total carbon (TC), total nitrogen (TN), and available nitrogen (Av. N) across diverse landscapes, and to evaluate the predictive potential of Visible-Near Infrared (Vis-NIR) reflectance spectroscopy for estimating soil carbon and nitrogen fractions. A total of 210 georeferenced surface soil samples (0–20 cm depth) were collected from three Agroecological (AEU) landscapes: Kole lands (AEU 6), North Central Laterite (AEU 10), and Palakkad Central Plain (AEU 22). Standard laboratory procedures were employed to determine SOC, TC, TN, and Av. N. and soil reflectance spectra were acquired using an ASD FieldSpec spectroradiometer. The spectral data were pre-processed using multiple techniques, including first and second-order derivatives (FOD and SOD), continuum removal (CR), and logarithmic transformation [log(1/R)]. The pre-processed spectral data, along with laboratory measured soil properties, were used to develop predictive models. Partial Least Squares Regression (PLSR), Support Vector Regression (SVR), and Random Forest (RF) models were applied to establish relationships between spectral signatures and soil properties. Model performance was evaluated using coefficient of determination (R²), root mean square error (RMSE), and residual prediction deviation (RPD). Among the models, SVR demonstrated superior predictive performance, with R² values of 0.75 for SOC, 0.75 for TC, 0.68 for TN, and 0.55 for Av. N. These results highlight the strong potential of Vis-NIR spectroscopy, particularly when combined with SVR, as a rapid and reliable alternative to conventional laboratory methods. This approach can significantly support large scale soil monitoring and precision agriculture applications across similar agro-ecological landscapes.

Читать у издателя →

Похожие статьи