Retrieval of canopy chlorophyll content from hyperspectral sensing for bark beetle detection
The European spruce bark beetle (Ips typographus) is an essential component of spruce ecosystems. Yet, for the forest industry, infestation may present serious economic loss. Early detection of bark beetles (green attack detection) is difficult due to infested trees not showing visual symptoms and biochemical changes being subtle. Chlorophyll is the primary pigment driving photosynthesis, and chlorophyll content is known to decrease responding to stress, including insect infestation. This study aimed at investigating the potential of airborne hyperspectral sensing to detect bark beetle green attack. It compared different models (narrowband VIs, PLSR and RF) in their accuracy to predict canopy chlorophyll content (CCC) by relating hyperspectral data with quantitative field measurements. Subsequently, it determined whether predicted CCC may be used to discriminate between healthy and green attacked areas. The study yielded the highest retrieval accuracies with PLSR (RMSE=0.25g/m2, R2=0.66). It indicated specific regions within the visible, red edge, NIR and SWIR regions that were important for CCC retrieval. The difference in CCC between green attacked and healthy areas was not significant. However, the constraints of the employed research design were identified, and changes for future research suggested to ascertain whether green attack detection using remote sensing technology is possible.