GRAZEO - Grazing suitability indicators from Earth Observation to improve buffalo-cattle contact risk models

Context and objectives

Kruger National Park (KNP) and its surroundings are considered an area with a homogenous risk for foot-and-mouth disease (FMD) transmission between buffaloes and cattle. Stratification of this risk however, would allow prioritizing areas for vaccination and fence maintenance and a better allocation of resources. GRAZEO is a spin-off of the EPISTIS (SR/00/102) project where several models were developed for this purpose at different scales, using remote sensing and spatial modelling. In GRAZEO however, the emphasis will be put on the role of the natural vegetation as a key factor of habitat suitability (grazing suitability) for large herbivores. The abundance and quality of the forage source is known to be a key driver of feeding patterns and distribution of livestock and wild grazers in savannah rangelands. Therefore, the main aim of GRAZEO is to explore the potential of the new generation of very-high spatial and spectral resolution sensors, such as WorldView-2, for (i) developing methodologies for mapping grass patches, grass biomass and quality indicator (nitrogen concentration), as well as tree species communities as complementary forage quality indicator, at a scale compatible with savannah spatial heterogeneity and animal movements; (ii) Investigating how these improved inputs, along with adapted modelling processes and multiple scenarios testing can contribute to improving the buffalo-cattle contact modelling outputs.

Project outcome

  • Classification of tree cover, grass cover, and bare ground in a heterogenous savanna landscape by pixel-based and objectbased machine learning algorithms. The obtained accuracies were very high, especially when using an OBIA approach combined with innovative machine-learning algorithms such as RF or SVM;
  • Prediction of leaf N and biomass across season and at larger scales using random forests and WorldView-2 imagery with the red edge band capability;
  • Improvement of the modelling and mapping of the risks of contact between cattle and buffaloes which might support disease managers in their decision-making process by identifying the need for targeted surveillance, setting priorities and effective/efficient allocation of resources.