Are differences between intercropping and monoculture practices reflected in satellite imagery data ? The Wallonia case

Details

Supervisors

Draye, Xavier ; Defourny, Pierre

Faculty

Faculté des bioingénieurs

Degree label

Master [120] : bioingénieur en sciences agronomiques, à finalité spécialisée

Abstract

In the context of sustainable intensification of agricultural production systems, there is a growing interest in diversified cropping systems such as intercropping, which involves cultivating two species simultaneously on the same field. However, implementing these practices raises the question of whether satellite data-based remote sensing tools, commonly used for agricultural monitoring, can be effectively adapted to these more complex systems. To effectively monitor intercropping systems using satellite data- based tools, it is crucial to identify differences between intercropping and monoculture systems that are detectable through satellite observations. The objective of this thesis is to assess the capacity of satellite-based remote sensing to detect differences between intercropping systems and monocultures in Wallonia, with a specific focus on the winter wheat-pea intercrop compared to pure winter wheat. The analyses relied on data from Sentinel-2 (NDVI, LAI, FCOVER, FAPAR), combined with agricultural declarations (SiGEC), which provide records of intercropping parcels. A significant part of the work involved processing SiGEC data to make it usable and associating temporal series of remote sensing variables with parcels in order to identify potential differences. Then, two approaches, univariate and multivariate, were employed, the former analyzing variables individually and the latter capturing combined effects to explore these distinctions. The results from both approaches revealed the satellite's inability to identify differences between fields of winter wheat and winter wheat-pea intercrop using the available data. The variability observed in the responses of key variable across individual parcels, likely stemming from plot-specific characteristics, introduced noise that these methods were unable to overcome. In the absence of information on these characteristics, these factors can only be treated as noise and blur the weak differentiating signal. This work underscores the challenges encountered and suggests directions for future research to improve the ability to distinguish between these systems using satellite imagery.