The Drone-Willow-Bumblebee Project
In the rugged mountains of Western Norway, two species of willow (Salix lanata and Salix lapponum) play a quiet but vital role the life of bumblebees. As climate change reshapes mountain landscapes, understanding how these willow shrubs are spreading or retreating is becoming increasingly important. This project combines drone photography, field surveys, and artificial intelligence in a pilot study to attempt to automatically detect and map willow bushes across the Finse area.
By training a deep learning model to recognize willows from high-resolution aerial images, we should be able to survey vast and remote areas far faster than would ever be possible on foot. Although, gathering training data still involves a lot of manual identification and hiking in the gorgeous Finse mountain region. This research can open the door to large-scale monitoring of these ecologically important shrubs for years to come.
The project is a cross disciplinary pilot study, led by Ørjan Totland (Dept. of Biological Sciences) in collaboration with Ben Robson (Dept of Department of Earth Science) and the Mountains-in-Motion group by Suzette Flantua (Dept. of Biological Sciences). Last spring and autumn, two initial field campaigns took place. See this brief overview of some results to find out more.
To get oriented, have a look at the various topics we present here. Also, check out the interactive map of the Finse field work area below!
Finse research station is at the blue pin. Zones in pink are initial areas where the drone flies and gathers imaging data. The “leaf”-markers show where each willow bush was identified, gendered and geotagged in the field.