Estimation of Soil Organic Carbon Contents in Croplands of Bavaria from SCMaP Soil Reflectance Composites

For food security issues or global climate change, there is a growing need for large-scale knowledge of soil organic carbon (SOC) contents in agricultural soils. To capture and quantify SOC at field scale, Earth Observation (EO) can be valuable data source area-wide mapping. The extraction exposed soils from EO challenging due to temporal permanent vegetation cover, the influence moisture condition surface. Compositing techniques multitemporal satellite images provide an alternative retrieve produce source. repeatable composites, containing averaged areas over several years, are relatively independent seasonal surface conditions new EO-based that used estimate large geographical with high spatial resolution. Here, we applied Soil Composite Mapping Processor (SCMaP) Landsat archive between 1984 2014 covering Bavaria, Germany. Compared existing modeling approaches based on single scenes, 30-year SCMaP reflectance composite (SRC) resolution 30 m used. SRC spectral information correlated point using different machine learning algorithms cropland topsoils Bavaria. We developed pre-processing technique address issue combining pixels purpose modeling. methods often studies choose best prediction model. Based model accuracies performances, Random Forest (RF) showed capabilities predict Bavaria (R² = 0.67, RMSE 1.24%, RPD 1.77, CCC 0.78). further validated results dataset. comparison measured predicted mean difference 0.11% RF promising approach distribution extents (30 m).