|
Titel |
The 2014 Lake Askja rockslide tsunami – optimization of landslide parameters comparing numerical simulations with observed run-up |
VerfasserIn |
Sigríður Sif Gylfadóttir, Jihwan Kim, Jón Kristinn Helgason, Sveinn Brynjolfsson, Ármann Höskuldsson, Tomas Johannesson, Carl Bonnevie Harbitz, Finn Løvholt |
Konferenz |
EGU General Assembly 2016
|
Medientyp |
Artikel
|
Sprache |
en
|
Digitales Dokument |
PDF |
Erschienen |
In: GRA - Volume 18 (2016) |
Datensatznummer |
250128888
|
Publikation (Nr.) |
EGU/EGU2016-8927.pdf |
|
|
|
Zusammenfassung |
The Askja central volcano is located in the Northern Volcanic Zone of Iceland. Within the
main caldera an inner caldera was formed in an eruption in 1875 and over the next 40 years it
gradually subsided and filled up with water, forming Lake Askja. A large rockslide was
released from the Southeast margin of the inner caldera into Lake Askja on 21 July 2014. The
release zone was located from 150 m to 350 m above the water level and measured 800 m
across. The volume of the rockslide is estimated to have been 15–30 million m3, of which
10.5 million m3 was deposited in the lake, raising the water level by almost a meter. The
rockslide caused a large tsunami that traveled across the lake, and inundated the shores
around the entire lake after 1–2 minutes. The vertical run-up varied typically between
10–40 m, but in some locations close to the impact area it ranged up to 70 m. Lake
Askja is a popular destination visited by tens of thousands of tourists every year but
as luck would have it, the event occurred near midnight when no one was in the
area.
Field surveys conducted in the months following the event resulted in an extensive
dataset. The dataset contains e.g. maximum inundation, high-resolution digital elevation
model of the entire inner caldera, as well as a high resolution bathymetry of the lake
displaying the landslide deposits. Using these data, a numerical model of the Lake
Askja landslide and tsunami was developed using GeoClaw, a software package for
numerical analysis of geophysical flow problems. Both the shallow water version and an
extension of GeoClaw that includes dispersion, was employed to simulate the wave
generation, propagation, and run-up due to the rockslide plunging into the lake. The
rockslide was modeled as a block that was allowed to stretch during run-out after
entering the lake. An optimization approach was adopted to constrain the landslide
parameters through inverse modeling by comparing the calculated inundation with the
observed run-up. By taking the minimum mean squared error between simulations and
observations, a set of best-fit landslide parameters (friction parameters, initial speed
and block size) were determined. While we were able to obtain a close fit with
observations using the dispersive model, it proved impossible to constrain the landslide
parameters to fit the data using a shallow water model. As a consequence, we conclude
that in the present case, dispersive effects were crucial in obtaining the correct
inundation pattern, and that a shallow water model produced large artificial offsets. |
|
|
|
|
|