Target values of the sensor-based soil mapping are plant-available and fixed nutrients, acidity, organic substance, water content, texture, soil compaction, porosity and biological activity.
Methods of geoelectricity, VisNIR, Raman and XRF spectroscopy are used. A completely new technology is the terahertz-spectroscopy which is currently in the focus of basic research. The above-mentioned soil sensors are to be used on semi-automatic and (partly) automatic platforms. An autonomous sampler (robot) is being developed in cooperation with TU Berlin for fully automated applications. The sensory data are fused and evaluated using current methods of machine learning. Geostatistic methods are used to model the spatial prediction and estimation of sampling distances.
- How is soil fertility to be characterized economically in fields at high temporal and spatial resolution
- What are the relationships between relevant soil properties and the signals gained from different sensors?
- How can sensor data be integrated into decision support systems for soil management (fertilization, soil tillage, irrigation)?
- What is the variability of soil properties (contrast, frequency)?
- What are the impacts of management measures on soil properties, and which conclusions can be drawn for soil management?
- How can automatic sample feeding and conditioning be implemented in the mobile use of soil sensors in order to achieve good measuring results?
- How can the relationship of the sensor signal and ground characteristic be mathematically modeled and used for the prognosis?
Ongoing research projects