Designed experiments for optimizing infrared spectroscopic characterizations and quantifications of soil organic matter at scales of increasing complexity: importance of band shifts, spectral variable selection and subtraction approaches

• Applicants: Dr. Ruth Ellerbrock; Professor Dr. Bernard Ludwig
• Subject Area: Soil Sciences
• Term: since 2022
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 501354812

The accuracy of visual and near infrared (Vis/NIR) and mid-infrared (MIR) spectroscopy for soil organic matter (SOM) characterization and quantification has been very variable in a large number of observational studies and there is insufficient knowledge on the underlying reasons for this variation in performance. The reasons are mainly related to IR-active properties of SOM and SOM fractions being affected by interactions of SOM with (mineral-bound) ions that result in shifts of bands in the Vis/NIR and MIR spectra relative to those without cations, minerals and soils. This proposal aims by means of designed experiments (i) to improve the mechanistic understanding of the interaction between organic matter (OM) and cations using Vis/NIR, MIR and micro-MIR; (ii) to subsequently develop an adjustment procedure to quantify effects of cations on shifts and intensities of OM bands; and (iii) to finally optimize linear and non-linear chemometric regression approaches for the prediction of contents of bulk soil organic carbon (SOC) and SOC fractions using the adjustment procedure and optimized spectral variable selections. Salts, minerals, and soils will be added to predefined organic components to obtain different OM-cation associations. The complexity of OM in the experiments will increase from low molecular weight organic substances to commercial humic acid, pre-incubated wheat straw and forest floor Oa material. The Vis/NIR and MIR spectra (including micro-MIR) of the mixtures will be compared with those of the OM to advance the mechanistic understanding of the relations between OM-cation interactions and band shifts. The adjustment procedure, which quantifies the effects of cations on band shifts and intensities, will be based on the Lambert-Beer law and will use the data of the OM-cation experiments to calculate wavenumber-dependent weighting factors that reflect the cation effects on the band intensities. The weighting coefficients will be used to recalculate the band intensities of the OM components. The accuracy of re-calculated OM spectra will be checked by comparing them with the spectra of the original OM and in case of native OM by a spectral subtraction procedure after chemical oxidation. Linear and non-linear chemometric regression approaches for the prediction of contents of bulk SOC and SOC fractions in soils at field and regional scales will be subsequently optimized. For the optimizations, the usefulness of band-shift corrections and improved chemometric regressions that consider advanced spectral variable selection approaches and cation-/mineral-related information will be explored. Overall, the spectral data from experiments designed at scales of increasing complexity are expected to advance the understanding of effects of OM-cation interactions on band shifts which can then be applied for improving the signal to noise ratios and optimizing linear and non-linear chemometric regression approaches.

• DFG Programme: Research Grants