Early Detection of Bark Beetle Attack Using Remote Sensing and Machine Learning: A Review

Bark beetle outbreaks can result in a devastating impact on forest ecosystem processes, biodiversity, forest structure and function, and economies. Accurate and timely detection of bark beetle infestations is crucial to mitigate further damage, develop proactive forest management activities, and minimize economic losses. Incorporating remote sensing (RS) data with machine learning (ML) (or deep learning (DL)) can provide a great alternative to the current approaches that rely on aerial surveys and field surveys, which are impractical over vast geographical regions. This paper provides a comprehensive review of past and current advances in the early detection of bark beetle-induced tree mortality from three key perspectives: bark beetle & host interactions, RS, and ML/DL. We parse recent literature according to bark beetle species & attack phases, host trees, study regions, imagery platforms & sensors, spectral/spatial/temporal resolutions, spectral signatures, spectral vegetation indices (SVIs), ML approaches, learning schemes, task categories, models, algorithms, classes/clusters, features, and DL networks & architectures. This review focuses on challenging early detection, discussing current challenges and potential solutions. Our literature survey suggests that the performance of current ML methods is limited (less than 80%) and depends on various factors, including imagery sensors & resolutions, acquisition dates, and employed features & algorithms/networks. A more promising result from DL networks and then the random forest (RF) algorithm highlighted the potential to detect subtle changes in visible, thermal, and short-wave infrared (SWIR) spectral regions.