Temporal S2/S1 mosaics combined with environmental covariates for regional SOC mapping: lessons from la Beauce (France) and the Västra-Skaraborg (Sweden)

International audience

إنجليزي. Satellite-based soil organic carbon content (SOC) mapping over wide regions is generally hampered by the low soil sampling density and the discrepancy of soil sampling collection dates. Some unfavorable topsoil conditions, such as high moisture, roughness, the presence of crop residues, the limited amplitude of SOC values and the limited area of bare soil when a single image is used, are also among the influencing factors. For two contrasted wide agricultural areas in boreal and temperate zones, Veskra-Skaraborg (Sweden) and la Beauce (France), this study compares approaches relying on Sentinel-2 (S2) temporal mosaics of bare soil (S2Bsoil) over ≥5 years jointly with Sentinel 1, SOC measurements data and other environmental covariates derived from digital elevation models and/or lithology maps and/or airborne gamma-ray data. Prediction models relied on quartile random forest, with 10 fold cross-validation, according to several datasets: i) “Sentinel-2”, the Sentinel-2 bands of a given S2Bsoil; ii) “terrain”, the terrain covariates (Digital Elevation Model and its derivatives, plus oblique geographic coordinates); iii), “Sentinel-2” plus “terrain”; iv) “all”, i.e. “Sentinel-2”, “terrain” and a selection of relevant Sentinel-2 spectral indices.Lessons from la Beauce (Urbina-Salazar et al., 2023) deal with (i) the dates and periods that are preferable to construct temporal mosaics of bare soils while accounting for soil moisture and soil management; (ii) which set of covariates is more relevant to explain the SOC variability. The models using all the covariates had the best model performance. Airborne gamma-ray thorium, slope and S2 bands (e.g., bands 6, 7, 8, 8a) and indices (e.g., calcareous sedimentary rocks, “calcl”) from the “late winter–spring” time series were the most important covariates in this model. Our results also indicated the important role of neighboring topographic distances and oblique geographic coordinates between remote sensing data and parent material. These data contributed not only to optimizing SOC mapping performance but also provided information related to long-range gradients of SOC spatial variability, which makes sense from a pedological point of view.Lessons from the Veskra-Skaraborg deal with the impact of percentile thresholding for temporal mosaicking of bare soils in relationship with soil moisture and cloud frequency. Performance decreased from R90 to R25. Models were highly predictive over both Plain (RPIQ≤1.3) or Till, yet with a slight improvement for Till (best RPIQ 1.4). Results confirm the differences in performances according to soilscape and agricultural system, and the complex interactions due to soil moisture in satellite-based soil property mapping. For both study áreas, spectral models alone were not well performing, but covariates such as morphometric layers slightly improved the prediction from temporal mosaics of bare soils.Reference:Castaldi, F., Koparan, M.H., Wetterlind, J., Žydelis, R., Vinci, I., Savaş, A.Ö., Kıvrak, C., Tunçay, T., Volungevičius, J., Obber, S., Ragazzi, F., Malo, D., Vaudour, E. 2023. Assessing the capability of Sentinel-2 time-series to estimate soil organic carbon and clay content at local scale in croplands, ISPRS Journal of Photogrammetry and Remote Sensing, 199, 40-60, https://doi.org/10.1016/j.isprsjprs.2023.03.016Urbina-Salazar, D., Vaudour, E., Richer-de-Forges, A.C., Chen, S., Martelet, G., Baghdadi, N., Arrouays, D., 2023. Sentinel-2 and Sentinel-1 Bare Soil Temporal Mosaics of 6-year Periods for Soil Organic Carbon Content Mapping in Central France. Remote Sensing 15, 2410. https://doi.org/10.3390/rs15092410