Two intensive field campaigns
Over two seasons, observations and measurements were carried out in 22 experimental trials and 103 sampling units in farmers' winter wheat fields in Wallonia. Data collected in the field include disease severity scoring, phenological stage and crop canopy reflectance measured with a portable spectrometer. Plants were also sampled to measure their biomass and estimate their nitrogen content. At the same time, EnMAP and PRISMA hyperspectral satellite images were acquired within the study area.
Biotic stress monitoring
Different modelling approaches were compared in order to detect the presence of biotic stress from spectral data, such as the use of spectral indices (traditional remote sensing approach), a multivariate linear approach (traditional chemometrics approach) and a machine learning approach. The results show fair detection performances from the phenological stage of heading.
Nitrogen status monitoring
The canopy nitrogen content was modelled using spectral bands and spectral indices for both satellites. The results are consistent with the literature (R²: 0.57-0.79 and RMSE: 20.86-18.65 kg.N/ha). Biomass was also modelled in order to estimate the Nitrogen Nutrition Index (NNI), which enables farmers to monitor nitrogen status.
Perspectives
The results of this project demonstrate that these hyperspectral satellites can be used to monitor biotic stress and the nitrogen status of cereals. However, the availability of these images is currently limited in Wallonia, due to the frequent dense cloud cover.
Many thanks to all those who gave us access to their wheat trials and fields, and to those who contributed to the measurement campaigns.
Cooperation partnership:
Private stakeholders (Constellr – project coordinator, AMOS, DELTATEC) and scientific/technical stakeholders (UCLouvain, CePiCop)
Funding:
"Skywin" competitiveness cluster, agreement no. 8614
Useful information
Team
