University of Twente Student Theses


Differentiating healthy and bark beetle infected spruce trees with Sentinel-1 SAR

Phuntsho, p (2020) Differentiating healthy and bark beetle infected spruce trees with Sentinel-1 SAR.

[img] PDF
Abstract:Forests are affected by insect pests globally resulting in tree mortality, and in Europe, the spruce bark beetle (Ips typographus) is known to have a large scale detrimental impact on Norway spruce (Picea abies) forests. Bark beetle infestations of the spruce trees have significant economic, social, and environmental impacts. The conventional method of field surveying is challenging in terms of resources to identify and differentiate healthy and infected spruce trees. There is a need for a cost-effective remote sensing technology to identify the sites of the bark beetle damaged spruce trees. The ability of remote sensing methods to identify bark beetle infected tree would help mitigate the further spread of the infestation and manage sustainable forest management. The study aims to determine the applicability of the Sentinel-1 SAR data and explore the method to differentiate between the healthy and bark beetle infected spruce trees. The study was conducted inside the Eifel National Park (ENP) in the North Rhine-Westphalia state of Western Germany. Google Earth imagery was used to obtain the samples for both healthy and infected spruce trees. A single time snapshot of the Sentinel-1 SAR image of the area was analyzed to compare the difference in Sentinel-1 backscatter response between the healthy and infected spruce trees statistically. Sentinel-1 SAR and Sentinel-2 image from 2014 to 2019 were used to extract SAR backscatter and NDRE spectral index, respectively. Temporal profile of healthy and bark beetle infected spruce trees was developed based on the time series of SAR backscatter and the time series of NDRE spectral index. An independent site outside the study area was used to validate the annual pattern of the SAR temporal to ascertain the robustness of the temporal profiling technique. The distribution of Sentinel-1 SAR backscatter response of a healthy and infected spruce tree overlap substantially, limiting the potential of SAR backscatter to differentiate between them, despite a statistically significant difference with their mean backscatter response. The Sentinel-1 SAR temporal profile of the healthy spruce tree shows an annual seasonal pattern. Adaptive Savitzky-Golay filter (ASAVGOL) helped to produce a pronounced annual profile pattern of the healthy spruce trees. On the other hand, the SAR temporal profile of a bark beetle infected spruce trees shows an irregular temporal profile pattern. The transition from an annual seasonal pattern to an irregular pattern on a temporal profile can be interpreted as the moment of a bark beetle infestation. There was no field data available at each stage of the infestation to relate with the moment of infestation. However, the ability of the temporal profiling method to graphically indicate a point of infestation was verified based on an independent site outside the study area using a discrete set of Sentinel-1 SAR temporal profile analysis. The temporal profiling method is suitable to exploit the Sentinel-1 SAR to approximately identify the moment of infestation on the time series profile. The moment of infestation shown by the SAR temporal profile is likely related to the annual phenological processes of healthy living trees and bark beetles infected dying trees. Dense weather independent time-series Sentinel-1 images facilitated to graphically visualize an annual pattern on a temporal profile, while the availability of fewer cloud-free Sentinel-2 observations impeded in showing the annual pattern explicitly. Understanding the moment of infestation would facilitate to identify the bark beetle affected sites for forest management applications. It will guide timely sanitation fellings to limit the further spread, though it may not alert forest managers to react immediately as not every deviation from the expected seasonal pattern is pointing at an infestation. Similarly, we can assess whether a timely bark beetle mitigation measures are being implemented or not, and make informed decisions on the sustainable status certification of sustainable forest management. It can be applied by relevant agencies to make an informed decision on the entitlement of the bark beetle damage compensation. Further investigation is necessary to understand at what stage of bark beetle infestation can be described by the SAR temporal profile using field-based samples at each stage of the bark beetle infestation in spruce forests.
Item Type:Essay (Master)
Faculty:ITC: Faculty of Geo-information Science and Earth Observation
Programme:Geoinformation Science and Earth Observation MSc (75014)
Link to this item:
Export this item as:BibTeX
HTML Citation
Reference Manager


Repository Staff Only: item control page