Ali Sakhaee successfully defended his dissertation, “Spatial Prediction of Organic Matter in German Agricultural Soil with Machine Learning”. Using the first German Agricultural Soil Inventory (3104 sites), he developed and compared state-of-the-art machine-learning approaches for nationwide prediction of soil organic carbon (SOC), C/N ratio, and SOM fractions.
Key innovations include a two-model approach (separate models for mineral and organic soils), an enhanced regressor-chain method, and the application of conformal prediction for spatial uncertainty quantification. The thesis provides harmonized, high-resolution maps for the whole of Germany and represents a major methodological advance in digital soil mapping.