The soil, a complex and major interface between the atmosphere, the lithosphere, the hydrosphere and the biosphere, is one of the elements at the source of life. Among other things, it provides food, regulates various environmental parameters (climate, natural disasters, water purification, etc.) and plays a cultural role (landscapes, leisure activities) for our societies. In addition, soil is able to sequester carbon from the biomass produced by cultivated plants and in particular from unharvested cover crops planted between two main crops. The need to monitor this biomass has been expressed by various actors in the food value chain, but today, despite the existence of remote sensing techniques allowing rapid and exhaustive measurements of biomass estimators, the lack of calibration data limits the widespread use of these techniques.
The use of drones offers a rapid, precise and spatially exhaustive assessment of various aspects of plant cover. Indeed, using photogrammetric processing of RGB or multi-spectral images, it has been demonstrated that it is possible to determine the height of plants and to estimate the biomass of vegetation cover. Unfortunately, these image processing techniques cannot be deployed by professionals in the agricultural sector because they require further engineering work and precise calibration at this stage.
To overcome this scientific obstacle, the project plans to generate a model for predicting the biomass of vegetation covers based on their height and cover rate. Then, after calibration on two series of experimental plots, we will test the sensitivity of their responses to plant growth during the intercrop period. Once their limits have been assessed on the basis of field truths, we will be able to transfer a user-friendly solution to agricultural advisory bodies and farmers.
Consequently, this Ra&D work will make it possible to further extend the use of drones in the agricultural sector to soil protection and management. The expected results will make it possible to respond to multiple challenges, including
Finally, such models adapted to the plot scale will have to be compared with regional predictive models obtained by satellite. These comparisons will make it possible to evaluate the respective advantages of each for better land and soil management.
Project leader - team
Ophélie Sauzet
(HEPIA),
Alain Dubois
(HEPIA)