Nitrogen (N) is one of the key plant nutrients and the most applied fertilizer in modern agri- and horticulture where high productivity is needed. Over-fertilization with N can have negative effects on the environment such as degraded groundwater quality or reduced biodiversity. Remote estimation of N status of agricultural production systems could assess the risk of N pollution of waterways and biodiversity, while also enabling smart farming techniques increasing N use efficiency.
Remote sensing of plant N status uses plant features such as biomass, canopy cover and chlorophyll content and is heavily dependent on the crop type and development stage. The instrument for remote sensing plant N status are spectrometers, which measure the light reflectance of plants across the electromagnetic spectrum of light. State of the art satellite systems, such as the ’Sentinel-2’ satellite constellation of the European Space Agency ESA, offer an orbital spectrometer to monitor vegetation with a high spatial and temporal resolution.
The project ‘DeepField’ is funded by the Swiss Federal Office for Agriculture (BLW) and includes one PhD position at the ETH group of Crop Science and two PhD positions at the ETH group of Photogrammetry and Remote Sensing. The project aims to combine satellite images and the latest methods of deep learning in order to record field calendar information (e.g. crop type, phenology) on agricultural land on a large scale in Switzerland. Within the project ‘DeepField’, ETH Crop Science aims to estimate the plant N status using the ’Sentinel-2’ satellites.
- Gather and combine available ground truth data sets into a large data base
- Apply current remote sensing methodology for estimating plant Nitrogen status
- Use machine and deep learning methods to analyze the combined data sets
- Estimate crop Nitrogen status on a regional level
- Crop Science Group of ETH Zürich https://kp.ethz.ch
- ETH Group of Photogrammetry and Remote Sensing (https://prs.igp.ethz.ch/)
Swiss Federal Office for Agriculture (BLW)