Directing tropical forest management from the sky

Preventing the conversion of the world’s two billion hectares of degraded tropical forests to agriculture is a top conservation priority. This requires effective land allocations for conservation, restoration and the production of commodities to be made, which are dependent upon high quality forest survey data across extensive areas. Unmanned aerial vehicles (UAVs) make possible the acquisition of high resolution forest canopy images at relatively little cost but assessing forest quality from these images presents significant technical challenges. The project will develop the use of UAVs for forest survey work at the 100,000 hectare Harapan Rainforest ecosystem restoration concession in central Sumatra and at the Dakatcha Woodland in Kenya. These sites have been degraded through the extraction of forest products, especially large trees for timber, resulting in complex mosaics of forest at various stages of regeneration. Aerial imagery acquired from the sites will be analysed at the University of Cambridge, where recent advances in image recognition software will be refined. Mathematical techniques will be developed to enable automatic measurement of tree size and abundance, and identification of indicator species. Landscape scale maps of forest quality and restoration potential will then be produced. Land management decisions regarding where to carry out expensive restoration techniques, such as planting trees or removing invasive species, or where to establish income generating plantations will benefit enormously from this information. It is the aim of the project that by developing the software into a freely available open source package, it will become an indispensable tool for forest managers across the world, greatly aiding the conservation of degraded tropical forest.