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

Precise knowledge about the soil organic carbon (SOC) content in cropland soils is one requirement to design and execute effective climate and food policies. In digital soil mapping (DSM), machine learning algorithms are used to predict soil properties from covariates derived from traditional soil mapping, digital elevation models, land use, and Earth observation (EO). However, such DSM models are trained for a specific dataset and region and have so far only allowed limited general statements to be made that would enable the models to be transferred to different regions. In this study, we test the transferability of SOC models for cropland soils using five different covariate groups: multispectral soil reflectance composites (satellite), soil legacy data (soil), digital elevation model derivatives (terrain), climate parameters (climate), and combined models (combined). The transferability was analyzed using data from two federal states in southern Germany: Bavaria and Baden-Wuerttemberg. First, baseline models were trained for each state with combined models performing best in both cases (R2 = 0.68/0.48). Next, the models were transferred and tested with soil samples from the other state whose data were not used during model calibration. Only satellite and combined models were transferable, but accuracy declined in both cases. In the final step, models were trained with samples from both states (mixed-data models) and applied to each state separately. This process significantly improved the accuracies of satellite, terrain, and combined models, while it showed no effect on climate models and decreased the models based on soil covariates. The experiment underlines the importance of EO for the transfer and extrapolation of DSM models.

No abstract

Reflectance composites that capture bare soil pixels from multispectral image data are increasingly being analysed to model soil constituents such as soil organic carbon. These temporal composites are used instead of single-date multispectral images to account for the frequent vegetation cover of soils and, thus, to get broader spatial coverage of bare soil pixels. Most soil compositing techniques require thresholds derived from spectral indices such as the Normalised Difference Vegetation Index (NDVI) and the Normalised Burn Ratio 2 (NBR2) to separate bare soils from all other land cover types. However, the threshold derivation is handled based on expert knowledge of a specific area, statistical percentile definitions or in situ data. For operational processors, such site-specific and partly manual strategies are not applicable. There is a need for a more generic solution to derive thresholds for large-scale processing without manual intervention. This study presents a novel HIstogram SEparation Threshold (HISET) methodology deriving spectral index thresholds and testing them for a Sentinel-2 temporal data stack. The technique is spectral index-independent, data-driven and can be evaluated based on a quality score. We tested HISET for building six soil reflectance composites (SRC) using NDVI, NBR2 and a new index combining the NDVI and a short-wave infrared (SWIR) band (PV+IR2). A comprehensive analysis of the spectral and spatial performance and accuracy of the resulting SRCs proves the flexibility and validity of HISET. Disturbance effects such as spectral confusion of bare soils with non-photosynthetic-active vegetation (NPV) could be reduced by choosing grassland and crops as input LC for HISET. The NBR2-based SRC spectra showed the highest similarity with LUCAS spectra, the broadest spatial coverage of bare soil pixels and the least number of valid observations per pixel. The spatial coverage of bare soil pixels is validated against the database of the Integrated Administration and Control System (IACS) of the European Commission. Validation results show that PV+IR2-based SRCs outperform the other two indices, especially in spectrally mixed areas of bare soil, photosynthetic-active vegetation and NPV. The NDVI-based SRCs showed the lowest confidence values (95%) in all bands. In the future, HISET shall be tested in other areas with different environmental conditions and LC characteristics to evaluate if the findings of this study are also valid.

There is a growing need for an area-wide knowledge of SOC contents in agricultural soils at the field scale for food security and monitoring long-term changes related to soil health and climate change. In Germany, SOC maps are mostly available with a spatial resolution of 250 m to 1 km2. The nationwide availability of both digital elevation models at various spatial resolutions and multi-temporal satellite imagery enables the derivation of multi-scale terrain attributes and (here: Landsat-based) multi-temporal soil reflectance composites (SRC) as explanatory variables. In the example of a Bavarian test of about 8000 km2, relations between 220 SOC content samples as well as different aggregation levels of the explanatory variables were analyzed for their scale-specific predictive power. The aggregation levels were generated by applying a region-growing segmentation procedure, and the SOC content prediction was realized by the Random Forest algorithm. In doing so, established approaches of (geographic) object-based image analysis (GEOBIA) and machine learning were combined. The modeling results revealed scale-specific differences. Compared to terrain attributes, the use of SRC parameters leads to a significant model improvement at field-related scale levels. The joint use of both terrain attributes and SRC parameters resulted in further model improvements. The best modeling variant is characterized by an accuracy of R2 = 0.84 and RMSE = 1.99.