Linking grassland yield and nitrogen nutrition status to management practices: Towards large-scale grassland use intensity assessment

This is the authors' abstract. We don't add key points for this paper.

Adapting grassland management to meet the rising demand in food production, while sustaining their ecological value is a challenge. This highlights the need to monitor grassland use intensity to support sustainable agricultural policies. The objectives of this study are to (1) retrieve grassland yield and nitrogen content from satellite imagery and (2) link these to management practices and pedo-climatic conditions. A stepwise linear regression performed best out of 7 different methods to retrieve grasslands dry matter yield (DM, t/ha), nitrogen concentration (NC, g/kg), canopy nitrogen content (CNC, kg/ha) from Sentinel-2 reflectances and indices. The models are calibrated using field measurements made during three growing seasons. They are validated on an independent field dataset with a normalized RMSE of 9.5%, 15.6%, and 6.7% for DM, NC, and CNC respectively. In addition, the nitrogen nutrition index (NNI) is computed from the estimated DM and NC to evaluate grasslands nitrogen nutrition status. The retrieval models are then combined with grassland classification and mowing detection methods developed in previous studies to evaluate grasslands vegetation status in spring and harvested forage yield at regional scale. These large-scale experiments consistently showed the impacts of management practices and pedo-climatic conditions on grassland yield and nitrogen content. • Grassland biophysical variables estimated from Sentinel-2 data. • Linear model performs best out of 7 methods. • EO-based methods can provide valuable knowledge on the relationship between inputs and outputs of Grassland Use Intensity. • Differences between agro-ecological regions are retrieved without ancillary information.