<scp>LIBS</scp> for Rapid Soil Bulk Density and Carbon Stock Estimations: Toward Scalable Soil Carbon Monitoring

ABSTRACT Measuring soil bulk density (as well as carbon content) is crucial for accurate soil carbon stock calculations. Given the growing interest in soil carbon sequestration on farmlands as a strategy for mitigating greenhouse gas emissions, effective large‐scale field monitoring has become more important than ever. However, traditional methods for measuring soil bulk density, such as the core (volumetric cylinder) and clod methods, require undisturbed samples, making them labour‐intensive, time‐consuming and costly—due to the high complexity of sample collection and preparation. To overcome these challenges, we developed a laser‐induced breakdown spectroscopy (LIBS)‐based method for efficient and cost‐effective bulk density estimation that does not require undisturbed samples. We trained and evaluated LIBS‐based models using a dataset of 880 diverse Brazilian soil samples, randomly split into 70% for training and 30% for testing. The LIBS‐based models, combining discrete wavelet transform (DWT), feature selection via F ‐test for regression, and Ridge regression, achieved an R 2 of 0.72 and a root mean square error (RMSE) of 0.12 g cm −3 on the test set for soil bulk density prediction. Furthermore, by combining LIBS‐predicted soil bulk density with measured soil carbon concentration, we estimated soil carbon stock, achieving an R 2 of 0.93 and an RMSE of 2.2 Mg C ha −1 on the test set, indicating that the uncertainty in bulk density predictions has a minor impact on soil carbon stock estimations. To further streamline soil carbon stock estimation, we developed a model to directly predict soil carbon density—the product of soil carbon concentration and bulk density—using LIBS‐derived spectral features, eliminating the need for separate measurements or estimations. Although this approach resulted in a lower R 2 of 0.78 and a higher RMSE of 4.1 Mg C ha −1 , its performance was adequate for carbon stock prediction while simplifying the estimation process. These findings highlight the potential of LIBS as a rapid and effective tool for assessing soil bulk and carbon densities, contributing to sustainable soil management and climate change mitigation and adaptation.