Spectral compositing of Sentinel-2 data using SCMaP as input for soil parameter mapping

Kurzfassung

In the last couple of years, spectral-temporal reflectance composites have been used for spectral soil mapping as well as digital soil mapping. The compositing is beneficial since soils are barely uncovered in agricultural areas and often covered by different vital stages of vegetation. The quality of such composites depends on the input data quality, mainly the calibration of the data, atmospheric correction, available cloud and cloud shadow information and the geometry. Further, the bare soil selection technique is essential for the representation of undisturbed bare soils in the resulting data. In this contribution, we present the SCMaP (Soil Composite Mapping Processor) strategies and approaches behind the generation of the soil reflectance composites and other products used in the later soil information modelling and prediction. An essential step is to define spectral index-based thresholds for the bare and undisturbed soil pixel selection. We have developed a regionalized approach that is considering the environmental climatic and biogeographic conditions of smaller regions to produce European-wide threshold data set. The validation of the derived thresholds for the bare soil selection is an ongoing study for which we present first preliminary results. We also present the look and feel of the SCMaP data product suite in the European context produced by the WOSOMS (World Soils Monitoring System). Information such as the number of valid pixels (no cloud or cloud-shadow affected, no snow etc.), spectral reflectance variability of soil zones and confidence information are shown as well as statistical information such as soil frequency that describes the percentage of the visibility of base soils compared to the complete timeline. All data products are available via the WorldSoils web platform.