The temporal evolution in Soil Organic Carbon (SOC) content is often used in estimations of greenhouse gas fluxes and is an important indicator of soil quality. Regional estimates of SOC changes can only be obtained by analyzing very large number of samples over large areas due to the strong spatial variability in SOC contents. Visible and Near Infrared Spectroscopy (VNIRS) provides an alternative to chemical analyses. The benefits of this technique include a reduction of the sampling processing time, an increase of the number of samples that can be analyzed within time and budget constraints and hence an improvement of the detection of small changes in SOC stocks for a given area. Carbon contents are predicted from spectra through Partial Least Square Regressions (PLSR). The performance of three different instrumental settings (laboratory, field and airborne spectroscopy) has been assessed and their relative advantages for soil monitoring studies have been outlined using the concept of Minimal Detectable Difference. It appears that ground-based spectrometers give Root Mean Square Errors of Cross-Validation similar to the limit of repeatability of a routine SOC analytical technique such as the Walkley and Black method (± 1 g C kg¿ 1). The airborne spectrometer, despite its greater potential to cover large areas during a single flight campaign, has some difficulties to reach such values due to a lower Signal-to-Noise Ratio. Because of its statistical nature, the method and its potential rely on the stability of the calibrations obtained. It appears that calibrations are currently site-specific due to variation in soil type and surface condition. However, it is shown that PLSR can take into account both soil and spectral variation caused by different measuring campaigns and study areas. Further research is needed to develop regional spectral libraries in order to be able to use VNIRS as a robust analytical technique for precisely determining the SOC content and its spatial variation.
- diffuse-reflectance spectroscopy
- meta analysis