Estimation of landslide processes and slope evolution from LiDAR ...

Estimation of landslide processes and slope evolution from LiDAR-derived DEMs
Mio Kasai
81-29-879-6787
81-29-879-6729
m-kasai@pwri.go.jp
Erosion and Sediment Control Research Group, Public Works Research Institute, 1-6
Minamihara, Tsukuba 305-0031, Japan
Manabu Ikeda
m-iked44@pwri.go.jp
Toshihiro Asahina
81-3-6412-2438
81-3-6412-2501
River and Erosion Control Technology Department, Pasco Corporation, 1-1-2 Higashiyama,
Meguro-ku, Tokyo, 153-0043, Japan
toshihiro\_\asahina@pasco.co.jp
Kazunori Fujisawa
fujisawa@pwri.go.jp
LiDAR-derived DEMs are making an increasing contribution towards landslide hazard
mitigation owing to their increasing availability. They not only produce detailed contour maps
and visualized images to create more accurate and complete landslide inventories, but also can
be analyzed to understand the surface process of landslides with the notion that their
activity relates to fine-scale surface roughness. At present, surface fabric filtering of these

DEMs has been used to examine individual landslides, although landslide processes can be
better understood when they are considered as a part of hillslope evolution of the area of
interest. This presentation introduces the work that the eigenvalue ratio, which represents the
3-dimensional surface roughness, can be used to analyze landslide processes and their
influence on hillslope evolution. A case study is presented from a 5 km 2 steep (average slope =
36 \%) area of Kii Peninsula, Japan, where a 1-m grid LiDAR-derived DEM was used to
derive eigenvalue ratios and related terrain attributes. The slopes of the v-shaped valleys are
particularly landslide-prone and there is little valley margin except for occasional terraces along
the river.
Based on a combination of the eigenvalue ratio and field survey, the slopes of the study area
were categorized into 4 principal types. Slopes with a filter value of less than 2.50 (roughest)
are bedrock-dominant, while those between 2.50 and 2.75 partly represent surface
collapses and loosened bedrock, and those between 2.75 and 3.00 are colluvial slopes with
large rock materials. Where the value exceeds 3.00 the slopes are inactive or covered with fine
colluvial materials. The results imply that more attention should be paid to landslide
blocks with a filter value of between 2.50 and 2.75, e.g. by continuous monitoring, as they are
at higher risk of entire slope collapse in the near future. The eigenvalue ratio tends to increase
and the terrain becomes smoother uphill, indicating that slopes evolve by losing stability from
the valley floor. The blocks with a value between 2.75 and 3.00 may in future gain similar slope
features to those with a value between 2.50 and 2.75, although how long this transition takes is
uncertain. The process has probably been driven by the river eroding the base of hillslopes and
striping away earth at high flows.

The results suggest that, not only can the current activity of landslides be evaluated, but also
their future behavior can be estimated by applying surface fabric filters to an area of interest
and considering their spatial distribution. Together with creating detailed landslide inventories,
utilizing LiDAR-derived DEMs in this manner will in future contribute to the planning
of landslide hazard mitigation programs.