Scientific Visualization for Earthquake Science and Simulation

natureplaygroundAI and Robotics

Nov 14, 2013 (3 years and 8 months ago)

72 views

Scientific Visualization for
Earthquake Science and Simulation

Louise Kellogg, Tony
Bernardin
, Eric
Cowgill
, Oliver
Kreylos
, Mike
Oskin
, John Rundle
, Donald
L.
Turcotte
, M.
Burak

Yikilmaz

UC Davis: Geology, Computer Science, &
KeckCAVES


Earthscope

data


Seismic Tomography model (
Obrebski
,
et al 2010)

Scientific visualization research for
natural hazards at the
KeckCAVES

Virtual Reality User Interface (VRUI)


A platform
-
independent foundation for
development of virtual reality
applications

Lidar

Viewer

Earth

Viewe
r

Crusta

3D

Visualizer

CAVES

3D TV

Desktop

Laptop

Haiti: January 12, 2010

Mw 7.0


200,000


300,000
fatalities.


Massive damage
from building
collapse including
houses, govt.
buildings, UN
headquarters,
airport.



Analysis of high
-
resolution airborne and
terrestrial LIDAR after recent events


Goal:


support rescue and recovery
first


and then to support science


~2.7 billion individual point
measurements in (3D) space;
66.8 GB on disk



January 21


27, 2010, an area of 850 km
2

surveyed using
airborne
LiDAR

at an average density of ~3.2 points/m
2


Funded by World Bank, coordinated by USGS, collected by
Rochester Institute of Technology

Working with LIDAR point cloud data

Mapping the fault system

Remote mapping


Guided field work


Gave consistent
results as found in
the field


Can improve
quality and
quantity of rapid
scientific response


We concluded that the 2010
earthquake

was a relatively small

event between
the
1751 and
1770 ruptures.

El Mayor
-
Cucapah

M 7.2 April 2010

El Mayor
-
Cucapah

M 7.2 April 2010

Credit: Mike
Oskin
, Ramon

Arrowsmith
, Alejandro
Hinojosa
,

and
Javier Gonzalez

Removing vegetation
from LIDAR data

Interactive scientific visualization for
rapid response


Interactive visualization in a VR environment has the
potential to completely change rapid scientific
response to events


Visualization of these very large datasets is
challenging, but feasible, using
octree

data
representation.


Human
-
in
-
the
-
loop is essential to interpretation
(combined with automated methods)


Underway: change detection (time series)


Future developments: Coupling data interpretations
with simulations