The carbon pool in soils consists of both organic and inorganic carbon. Determinations of their contents in soils are routine in this field, especially when there is high interest in the soil organic carbon pool and its stabilization behavior in the context of carbon dioxide emissions. Common methods for determining carbonate contents are gas-volumetric approaches, elemental analyses of carbon contents after combustion of organic material for 4 hr, and the pressure calcimeter method by Sherrod et al.1 A method that allows determinations of both parameters (organic carbon and carbonate) synchronously would save time and resources. In a previous study,2 the potential of mid-infrared spectroscopy was tested for carbonate and organic carbon determinations. A large and diverse data set of Austrian forest soil samples was investigated with this method to assess the applicability of mid-infrared spectroscopy for carbon determinations in soils. It was found that calculated models for organic carbon, which were based on areas of infrared bands, performed better in mineral soils than in all samples (organic layers and mineral soils). Partitioning into groups having less diverse chemical backgrounds would increase the performance of the models for organic carbon. For carbonate determinations, the correlations were high, even for the total sample pool having an R2 of 0.960.2 The aim of the present study was to evaluate the potential to enhance the performance of the models by group separations with respect to carbonate determinations. For the organic layers, in particular, these effects were expected to be quite high.
