Fuzzy clustering of Vis-NIR spectra for the objective recognition of soil morphological horizons in soil profiles

- We used Vis-NIR and fuzzy clustering to group observations in air dried soil cores.
- We used a digital gradient to establish boundaries between homogeneous zones.
- This methodology can be used in conjunction with other proximal sensing devices.
- Spectrally derived horizons were more homogeneous than traditional horizons.

In the past decades the use of Vis-NIR spectra information applied to soil science studies has seen an exponential growth, specially in predicting commonly used soil properties. We used the ability of Vis-NIR for detecting physico-chemical characteristics along with fuzzy clustering techniques to discriminate spectrally homogeneous zones in soil cores and applied a DG to define its boundaries i.e., SPD hor. We tested this methodology in 59 air dried soil cores varying between 85 and 130 cm depth from the HWCPID, NSW, Australia. We observed that SPD hor had great similarity with traditional horizons. The SPD hor were more homogeneous in terms of Vis-NIR spectral variability and also offered more information about the relationship between the different spectral classes. Because of the intrinsic characteristics of the methodology it can be easily applicable with or in conjunction with other proximal sensing devices which can add further detail when recognizing morphological soil horizons.