European topsoil bulk density and organic carbon stock database (0–20 cm) using machine-learning-based pedotransfer functions
DOI: https://doi.org/10.5194/essd-16-2367-2024
IF: 11.4
2024-05-17
Earth System Science Data
Abstract:Soil bulk density (BD) serves as a fundamental indicator of soil health and quality, exerting a significant influence on critical factors such as plant growth, nutrient availability, and water retention. Due to its limited availability in soil databases, the application of pedotransfer functions (PTFs) has emerged as a potent tool for predicting BD using other easily measurable soil properties, while the impact of these PTFs' performance on soil organic carbon (SOC) stock calculation has been rarely explored. In this study, we proposed an innovative local modeling approach for predicting BD of fine earth ( BD fine ) across Europe using the recently released BD fine data from the LUCAS Soil (Land Use and Coverage Area Frame Survey Soil) 2018 (0–20 cm ) and relevant predictors. Our approach involved a combination of neighbor sample search, forward recursive feature selection (FRFS), and random forest (RF) models (local- RF FRFS ). The results showed that local- RF FRFS had a good performance in predicting BD fine ( R 2 of 0.58, root mean square error (RMSE) of 0.19 g cm −3 , relative error (RE) of 16.27 %), surpassing the earlier-published PTFs ( R 2 of 0.40–0.45, RMSE of 0.22 g cm −3 , RE of 19.11 %–21.18 %) and global PTFs using RF models with and without FRFS ( R 2 of 0.56–0.57, RMSE of 0.19 g cm −3 , RE of 16.47 %–16.74 %). Interestingly, we found that the best earlier-published PTF ( R 2 = 0.84, RMSE = 1.39 kg m −2 , RE of 17.57 %) performed close to the local- RF FRFS ( R 2 = 0.85, RMSE = 1.32 kg m −2 , RE of 15.01 %) in SOC stock calculation using BD fine predictions. However, the local- RF FRFS still performed better ( Δ R 2 > 0.2) for soil samples with low SOC stocks ( < 3 kg m −2 ). Therefore, we suggest that the local- RF FRFS is a promising method for BD fine prediction, while earlier-published PTFs would be more efficient when BD fine is subsequently utilized for calculating SOC stock. Finally, we produced two topsoil BD fine and SOC stock datasets (18 945 and 15 389 soil samples) at 0–20 cm for LUCAS Soil 2018 using the best earlier-published PTF and local- RF FRFS , respectively. This dataset is archived on the Zenodo platform at https://doi.org/10.5281/zenodo.10211884 (S. Chen et al., 2023). The outcomes of this study present a meaningful advancement in enhancing the predictive accuracy of BD fine , and the resultant BD fine and SOC stock datasets for topsoil across the Europe enable more precise soil hydrological and biological modeling.
meteorology & atmospheric sciences,geosciences, multidisciplinary