Low-cost drones help measure tree characteristics in the Sahelian savanna

Savanna is one of the main African ecosystems. The tree community in the savanna is a key element and provides many ecosystem services. Unmanned aerial vehicles (UAVs) combined with photogrammetric analysis are useful tools to produce accurate 3D maps, which, in turn, help describe the structure of the tree populations. The purpose of this study was to evaluate the use of commercial economical UAVs to assess tree characteristics in the savanna. A Dji Spark UAV was used to map 24 1-ha plots in northern Senegal. The images were processed using Pix4D software to produce a high-resolution canopy height model (CHM). A total of 239 trees were selected and their heights and crown areas were manually measured in the field. A strong correlation was found between UAV and field measurements with R-2 = 0.84 for height and R-2 = 0.93 for crown area. Based on tree canopy colours and tree morphologies measured by UAV, it was possible to predict tree species with an error rate of 20%, using a random forest classification. This study thus confirms the possibility of using low-cost UAVs to assess tree structures in the savanna not only for research on tree communities in savanna, but also by forestry agencies to inform stakeholders.

Automated summary

This study successfully evaluated tree characteristics in the savanna using commercial economical UAVs, finding a strong correlation between UAV and field measurements, and the potential to predict tree species with a 20% error rate through random forest classification.