Thesis subject

MSc thesis topic: Herb-rich grasslands: Exploring effects on nitrous oxide emissions and yield by classifying field plots with multispectral UAV data

From spring 2025-spring 2026 a field experiment is taking place with different mixtures of grasses (Lolium perenne and Festuca arundinacea), legumes (Trifolium pratense and T. repens) with biological nitrogen (N) fixation, and a herb Plantago lanceolata with demonstrated biological nitrification inhibiting capacity (i.e. limiting the conversion of NH4+ to NO3- in soil).

This field experiment will also be monitored with multispectral sensors underneath a UAV. This data can additionally tell something about the different plots and the changes in species composition in the plots throughout time. Five UAV flights will be performed to monitor the field experiment.

The main goal of this experiment is to investigate how different species can minimize nitrous oxide (N2O) emissions from N fertilization, whilst optimizing yield quality and quantity. The shifting in species distributions throughout the field plots need to be monitored and therefore a classification model need to be made which calculates the species coverages in the plots for different points in time. With both this data and vegetation indices, a relation between plant species and nitrogen deposition can be made.

The main goal of this experiment is to look at the effect of species (mixtures) on nitrous oxide (N2O) emissions resulting from manure and mineral N fertilizer sources, and exploring how N inputs can be reduced and N2O losses can be mitigated, whilst maintaining the productivity of the grassland.

Throughout 2025, N2O emissions from grassland plots will be measured, as well as microclimatic factors, (N) yield, fodder quality, soil mineral N pools, soil pH and the abundance of functional genes associated with the soil N cycle.

Objectives and Research questions

  • How can UAV images help understand the different plots where both fertilizers are applied as no fertilizers are used. What differences are visible in the different treatments?
  • With the data time series, does the composition of species change throughout time? Can we make a classification for the different plant species on the UAV images and see the changes in time and quantify this in percentages?
  • Which classification methods are most suited for detecting the percentages of plant species composition?

Requirements

  • Remote sensing knowledge
  • Image processing knowledge

Theme(s): Sensing & measuring; Modelling & visualisation