Digital mapping of soil organic carbon: advances in remote sensing, modeling, and environmental drivers

Soil organic carbon (SOC) plays a vital role in maintaining soil health, ensuring food security, and regulating the global carbon cycle. In his thesis, Xiande Ji proposes an advanced digital soil mapping (DSM) framework that integrates remote sensing and machine learning to enhance SOC predictions. Ji addresses key gaps in data selection, model performance, and SOC mapping under vegetation cover.

Ji describes recent advancements in remote sensing platforms, sensors, predictive models, and environmental drivers influencing SOC distribution. He also highlights ongoing challenges, such as data fusion and model interpretability, for future research.

Ji then presents a national-scale hybrid modeling approach that outperforms individual models for SOC prediction. It provides a high-resolution SOC map across various landscapes in Germany, emphasizing the value of hybrid techniques at multiple spatial scales.

Focussing on European forests, Ji quantifies current SOC stocks and projects future changes under different climate scenarios. He identifies key biogeochemical drivers and reveals biome-specific SOC patterns. Notably, Mediterranean young forests show high potential for carbon sequestration under high-emission scenarios by 2100.

In agricultural systems, selecting appropriate temporal satellite data, particularly images from bare soil periods, significantly improves cropland SOC monitoring. The findings also underscore the importance of non-photosynthetic vegetation signals for prediction.

By advancing SOC assessment across multiple landscapes and spatial scales, this thesis provides a robust foundation for improving digital SOC mapping, supporting smarter land management practices and more effective carbon sequestration strategies.

Dissertation: https://hdl.handle.net/11370/ab9708a9-c497-4a30-9059-4f8aca452f0c