Waves generation due to a submarine earthquake
Initial deformation of bottom and surface
Submarine earthquakes generate tsunamis through the hydrodynamic response
to the sea floor deformation, or indirectly through triggering of submarine
slides. Most earthquakes can be associated with interaction of moving plates
at the earth's surface. When they make contact at the boundary the motion may
be obstructed and tension builds up until released through an earthquake.
The relative motion between the plates may be parallel to the
boundary (strike-slip), normal (dip-slip) or a combination.
It is the vertical, permanent displacement of the sea bottom that is important
for tsunami generation. Largest vertical motions are found with
earthquakes having a dominant dip-slip
component. In the figure we have depicted a cross-section of an inverse
trust fault, where two plates converge, one (dark green) subducting the other
(light green).
The bottom deformation display an asymmetry, with a large area of elevation
and a smaller area of marked suppression. The sketch is based on Okada's
formula applied to the parameters of the 1969 earthquake, magnitude 7.9,
southeast of Algarve,
Portugal, near the Azores-Gibraltar fracture zone. The epicenter was
located in a comparatively smooth
abyssal plain with a depth of roughly 5000 m.
A numerical simulation, taking into account the source-sink distribution at
the sea floor, is depicted in blue. We note the comparatively step gradient
above the fault, that stem from the discontinuity at the sea bottom.
Vertical scales are not in true proportions in the figure, the
total horizontal extent of the figure is 120 km, while the maximum elevation
of the sea bottom and surface equal 2.4 m and 1.8 m respectively.
Wave propagation and dispersion
After the initial rise of the sea surface waves propagates away from the
source region. To give an idea of the effect of wave dispersion we have
performed two dimensional calculations, solving the linear hydrostatic equations, the linear Boussinesq
equations and a linear, but otherwise fully inviscid, set of equations.
|
An mpeg animation of tsunami wave
propagation from an earthquake. All the results are linear and "Laplace"
correponds to full potential theory. |
Remarks:
- Dispersion is important and yields noticeable additional
crests following the leading one
- The first few crests are reproduced very
well by the Boussinesq equations
- The hydrostatic approximation soon becomes highly inaccurate
Comparison with asymptote for the
system propagating to the right
|
Starting from t=2 min we focus on the
part of the wave system going to the right. The key point is the approach to
the asymptotic solution. Click for more details on this solution. |
Remarks:
- Asymptote becomes close after less than 10 minutes for the first crest
- After 20 minutes the asymptote reproduces also second crest closely
Three dimensional effects
A three dimensional source reproduces many of the aspects from two dimensions,
but diplays a few additional ones. A three dimensional counterpart the
initial disturbance given above is used as initial condition for hydrostatic
and Boussinesq simulations.
|
Wave propagation as computed by a
hydroststatic finite difference model. The bottom deforamtion are indicated,
but is depicted with a smaller
scale than the surface. The numerical solution is carefully checked for
convergence. See
grid refinement for a cross section.
Note the strong numerical dispersion for coarse grid with 5 km increment.
3D animation in larger scale.
|
 |
Wave propagation as computed by a Boussinesq finite difference model. For better quality for click for an integrated
quicktime animation.
Grid refinement.
3D animation in larger scale.
|
|
An mpeg animation of a cross-section normal to the fault.
|
Remarks:
- Strong directional dependence
- Dispersive effects as important as in 2D
- Secondary depressions evolve also for hydrostatic solutions
Topographic effects
A complete idea of tsunami-propagation may be approached by studying 2D waves
in a realistic topography. Hence, the initial conditions, as given above is
specified within a 2D topography that is extracted from a digitized map of
the Iberian sea.
|
First phase: dispersion in deep water
The wave system is divided and spread out by dispersion. Note that the surface
elevation and the depth are in different scales and vertically translated relative to each other.
|
|
Second phase: Amplification in shoaling water
The wave system encounters the slope of the continental shelf and is
amplified, while the wavelength is reduced.
|
|
Second phase: Amplification in shoaling water
A close up of nearshore region. At the end we may observe a steepening of the
wavefronts.
|
|
Third phase: Bore formation
Nonlinear effects causes the wave front to break. In this case the amplitude
is then reduced due to a very slow rate of shoaling. For a steeper bottom
gradient the amplitude loss due to dissipation in breaking would compete with
amplification due to shoaling. Also the second crest breaks. In the present
long wave models the bore (breaking wave) is represented by a jump in the
surface. In reality we have a turbulent wave front of finite, but small, width.
|
A complete story, from generation to breaking, is available as a
quicktime animation. Note: The animation is
large (100 M) and includes speech.
Author of page: Geir Pedersen <geirkp@math.uio.no>