Spatial Estimation of Surface Soil Texture Using Machine Learning and Google Earth Engine

Land management, modeling, and monitoring require knowledge about the distribution of soil properties across the landscape. The main objective of this study is to use a combination of remote sensing data and machine learning techniques to estimate surface soil texture properties (sand, silt, and clay) in the northeastern region of Libya. The results showed that the coefficient of determination (R 2 ) ranges between 0.39 and 0.48. The Root Mean Square Error (RMSE) varies from 5.04 to 8.28, while the Normalized Root Mean Square Error (NRMSE) ranges from 0.14 to 0.15. These values indicate that there is less residual variance. Therefore, the machine learning model was used correctly. These results demonstrate that the approach used for modeling and spatial mapping soil texture properties in this region is practical and valid. Furthermore, this methodology can be applied to other areas with similar characteristics.