Detecting multiple insect disturbance types across the western US using very high-resolution satellite data
![Damage severity classification for two forest plots](https://s3.wp.wsu.edu/uploads/sites/2472/2024/05/damage_severity_abhi-792x624.jpg)
Project Description
Insects, such as bark beetles, can cause widespread and severe tree mortality events. In the western USA, the primary means of monitoring yearly insect-caused tree mortality on a wide spatial extent is through aerial detection surveys (ADS) conducted by the US Forest Service. Although analyses of ADS have previously led to impactful results (Meddens et al., 2012), ADS have various limitations and barriers to long-term time series analyses. By investigating the efficacy of high-resolution satellite and unmanned aerial vehicle (UAV; e.g., drone) imagery to detect insect disturbance, we aim to offer an alternative to ADS which allows for time series analyses over wide spatial extents.
UAV imagery allowed for a 3-dimensional point cloud for a stand of conifers impacted by insect disturbance (Shrestha et al. 2024). Using field observations of healthy vs. damaged trees, damage severity class, and % canopy mortality (or top-kill), we trained algorithms using multispectral indices and bands for the point cloud. The algorithms yielded predictions of insect disturbance throughout the study area. Furthermore, by incorporating 3-dimensional data and predicting % top-kill and damage severity class, this approach could potentially allow us to attribute disturbance to specific insect species (i.e., different species cause different patterns of top-kill and different damage severities). This novel approach may be limited in applicability to a wider study area, due to challenges gathering data that offers 3-dimensional information and/or sub-meter resolution multispectral data. However, it points towards a link between high-resolution multispectral imagery and insect disturbance. We further investigated this link by comparing field data of insect disturbance to satellite imagery; however, satellite data – even high-resolution satellite data – showcased mixed capabilities to detect healthy vs. damaged trees.
Based on promising results using UAV imagery, we plan on continuing to investigate the potential of both UAV and high-resolution satellite imagery to identify patterns of insect-caused mortality and damage on conifer trees.