Abstract
Endemic pests are a fundamental part of forest ecosystems, they provide key ecosystem services such as nutrient cycling and support biodiversity. Still, massive outbreaks of these pests, triggered by events such as drought, windthrows, and snow breaks, can limit the provisioning of ecosystem services that are key for human populations such as water cycle regulation and, which can eventually trigger natural hazard events (e.g. landslides).
Unmanned Aerial Vehicles (UAVs) and miniaturized optical sensors can be used to support foresters in detecting, identifying, and quantifying pests and their diffusion by exploiting multispectral imagery at high resolution. Such platforms are especially suited for monitoring areas in mountain regions that are difficult to access.
This study focus on the pine processionary (Thaumetopoea pityocampa) and European bark beetle (Ips typographus) that affect many forests in the Province of South Tyrol, Italy. Here, we present an up-scale strategy that first identifies the presence of a pest at the centimeter level (ground and close-range scale) based on UAV-derived products on a plot level. We conducted three UAV-flight campaigns during the year corresponding to the insect-life cycle. Then, on the one hand, using simple RGB and NDVI mosaics the system delineates the trees, identifies nests (processionary) and quantifies their impact. On the other, using the NDVI time series collection the system classify healthy, infested or dead tree linked to the presence of bark beetle. The system classifies and quantifies its presence by presenting graduated symbol maps widely used by foresters. Then, we scale up to meter resolution (remote sensing scale) to detect changes due to certain conditions of stress that can link to the presence of the studied pests. The final aim is to create high-quality training datasets that will be exploited by remote sensing products (Sentinel) to study and cover wider areas.