Soil organic carbon (SOC) plays a critical role in climate regulation, soil fertility, and ecosystem resilience, making its accurate spatial quantification essential for sustainable land management and greenhouse gas (GHG) reporting. However, mapping SOC in heterogeneous agroforestry systems remains challenging due to vegetation cover and landscape complexity. In this study, we develop and evaluate a hybrid bare soil modelling- Digital Soil Mapping supported by ML framework to generate high-resolution soil properties predictions in Mediterranean agroforestry systems (Extremadura, Spain). A dual modelling approach was implemented, combining (i) Bare Soil modelling using Sentinel-2 multi-temporal reflectance composites and (ii) Digital Soil Mapping (DSM) supported by environmental covariates (climate, terrain, vegetation) following the SCORPAN framework. Machine learning models, namely Quantile Regression Forests (QRF) and Extreme Gradient Boosting (XGBoost), were applied and optimised using automated hyperparameter tuning (FLAML). A total of 107 LUCAS topsoil samples and 36 complementary points from the Forest ICP Level I were used for calibration and validation, with a 70/30 train–test split. Results show that Sentinel-2-based modelling can effectively capture SOC spatial variability in bare soil conditions, while DSM improves predictions in vegetated areas. Model performance reached R2 values up to 0.76 (QRF, pH) and RMSE as low as 0.03 (XGBoost, N), with uncertainty quantified using the Prediction Interval Ratio (PIR) and performance further supported by RPIQ values up to 3.15. However, prediction accuracy remains sensitive to vegetation structure and sample density. The proposed framework provides a scalable and uncertainty-aware approach for SOC mapping, supporting Tier-3 GHG inventories and emerging Monitoring, Reporting, and Verification (MRV) systems. The results highlight the importance of integrating multi-source datasets and hybrid modelling strategies for reliable SOC estimation in complex landscapes.
Process-based soil organic carbon (SOC) models are widely used for carbon accounting and assessments. For applications beyond experimental sites, remote sensing–derived inputs are often used to complement or replace in situ boundary conditions. This substitution introduces additional sensitivity that has rarely been explicitly quantified. Here, we propose a general framework to quantify boundary-condition sensitivity in SOC modelling, distinguishing between dynamic sensitivity, which reflects deviations in the temporal trajectory of SOC stocks, and endpoint sensitivity, which captures differences in estimated SOC changes over defined reporting periods. Sensitivity is quantified by comparing substitution simulations, in which boundary conditions are replaced one by one with remote sensing–derived spatial datasets, against a baseline simulation driven by in situ observations. The framework is applied to 16 long-term experimental sites across Europe, covering diverse land-use types and environmental conditions. Boundary-condition substitution affected both SOC trajectories and endpoint estimates, although the magnitude and form of these effects varied among inputs. Initial SOC, vegetation-derived carbon inputs, and climate data were the dominant drivers of dynamic sensitivity. Fixed endpoint sensitivity was generally small relative to measured SOC changes, supporting the use of remote sensing–derived spatial datasets as scalable boundary-condition proxies for SOC modelling. Rolling endpoint sensitivity further revealed that SOC change assessments depend not only on SOC dynamics but also on the selected reporting window. Short reporting periods were more affected by short-term variability, whereas longer periods were more susceptible to accumulated deviations. A reporting window of approximately 15–30 years provided a practical balance between these effects. Overall, the framework enables a systematic quantification of sensitivity arising from boundary-condition substitution, enabling evaluation of both trajectory and reporting robustness. These findings highlight the importance of jointly considering critical boundary conditions and reporting-window design when applying remote sensing–driven SOC modelling for regional carbon accounting.
Process-based soil organic carbon (SOC) models are widely used for carbon accounting and assessments. For applications beyond experimental sites, remote sensing–derived inputs are often used to complement or replace in situ boundary conditions. This substitution introduces additional sensitivity that has rarely been explicitly quantified. Here, we propose a general framework to quantify boundary-condition sensitivity in SOC modelling, distinguishing between dynamic sensitivity, which reflects deviations in the temporal trajectory of SOC stocks, and endpoint sensitivity, which captures differences in estimated SOC changes over defined reporting periods. Sensitivity is quantified by comparing substitution simulations, in which boundary conditions are replaced one by one with remote sensing–derived spatial datasets, against a baseline simulation driven by in situ observations. The framework is applied to 16 long-term experimental sites across Europe, covering diverse land-use types and environmental conditions. Boundary-condition substitution affected both SOC trajectories and endpoint estimates, although the magnitude and form of these effects varied among inputs. Initial SOC, vegetation-derived carbon inputs, and climate data were the dominant drivers of dynamic sensitivity. Fixed endpoint sensitivity was generally small relative to measured SOC changes, supporting the use of remote sensing–derived spatial datasets as scalable boundary-condition proxies for SOC modelling. Rolling endpoint sensitivity further revealed that SOC change assessments depend not only on SOC dynamics but also on the selected reporting window. Short reporting periods were more affected by short-term variability, whereas longer periods were more susceptible to accumulated deviations. A reporting window of approximately 15–30 years provided a practical balance between these effects. Overall, the framework enables a systematic quantification of sensitivity arising from boundary-condition substitution, enabling evaluation of both trajectory and reporting robustness. These findings highlight the importance of jointly considering critical boundary conditions and reporting-window design when applying remote sensing–driven SOC modelling for regional carbon accounting.