2024-03-29T10:23:50Zhttps://eprints.lib.hokudai.ac.jp/dspace-oai/requestoai:eprints.lib.hokudai.ac.jp:2115/872352022-11-17T17:14:37Zhdl_2115_54837hdl_2115_20129hdl_2115_54822hdl_2115_54823hdl_2115_20124Application of remote sensing for characterization of windthrow and landslides at multiple scales in forest landscapeマルチスケールでの風倒地・崩壊地の特性把握 : 森林景観におけるリモートセンシングの応用Furukawa, Flavio610Due to climate change, the frequency, size, and intensity of natural disturbances
are increasing, leading to increase damage to forest ecosystems. Proper management of
these disturbed areas is critical for the resilience of forest ecosystems. Remote sensing is
used for monitoring different stages of disaster management as it can provide essential
information about damaged areas, reducing the need for manual inspection of hazardous
sites. However, remote sensing has some challenges, such as the influence of elements
like atmosphere, clouds, topography, and sun position, generating noise in the output data.
The remote sensing community is addressing these challenges and developing new
techniques to improve scientific understanding. Still, there is a discrepancy between
technical development and application of remote sensing in the management of forest
landscapes, due to the need for interdisciplinary skills involved. Since quick and
accessible information on disturbed areas is critical, appropriate approaches according to
the scale are required to support forest managers.
In this thesis, three different approaches were proposed to facilitate the
implementation of remote sensing to characterize windthrow and landslides at three
different scales: regional scale, forest stand scale, and single tree scale. The first approach
(Chapter 2) compared three different classification methods using high temporal/ spatial
resolution satellite data: the normalized difference vegetation index (NDVI) filtering
method, spectral angle mapper (SAM) method, and support vector machine (SVM)
method. The results showed that the NDVI filtering method was better to identify
landslides, while the SAM method was better to identify windthrow; supporting forest
managers to choose appropriate methods to identify windthrow and landslides.
The second approach (Chapter 3) compared a Red Green Blue (RGB) unmanned
aerial vehicle (UAV) with a Multispectral UAV to characterize landslides throughout the
months at a forest stand scale. The results showed that the RGB UAV was able only to
monitor vegetation growth, while the Multispectral UAV, due to the higher spectral
resolution, could monitor vegetation, bare soil, and dead matter over the months. Both
systems, due to the high spatial and temporal resolution, were able to deliver an
understanding of the vegetation regeneration process in a landslide at a forest stand scale.
The third approach (Chapter 4) was based on full motion video (FMV) technology to
identify fallen and snapped trees, at a single tree scale. The results showed that FMV was
able to identify fallen and snapped trees in a windthrow area, even with the presence of
vegetation. The higher context-awareness provided by the video and simpler workflow
showed the potential to overcome the limitations of the UAV structure from motion (SfM)
photogrammetry process.
In conclusion, the study examined different approaches on three different scales
for the use of remote sensing in the management of disturbed forest areas. As climate
change advances, the need to take quick actions to mitigate and sustain resilience in forest
ecosystems is essential. Remote sensing will continue to develop and play an important
role in forest management after natural disturbances. However, more research needs to
be conducted on facilitating the implementation of remote sensing techniques and training
forest managers to take full advantage of remote sensing for forest management after
natural disturbances.118p北海道大学. 博士(農学)Hokkaido UniversityThesis or Dissertationapplication/pdfhttp://hdl.handle.net/2115/87235info:doi/10.14943/doctoral.k15153https://eprints.lib.hokudai.ac.jp/dspace/bitstream/2115/87235/1/Flavio_Furukawa.pdf2022-09-26engETD10101甲第15153号2022-09-26博士(農学)北海道大学