![Hier klicken, um den Treffer aus der Auswahl zu entfernen](images/unchecked.gif) |
Titel |
Assessment of statistical uncertainty analysis of joint refraction and reflection travel-time tomographic models applied to real data off the Nicaraguan margin. |
VerfasserIn |
M. Prada, V. Sallarès, A. Meléndez, C. R. Ranero, K. McIntosh |
Konferenz |
EGU General Assembly 2010
|
Medientyp |
Artikel
|
Sprache |
Englisch
|
Digitales Dokument |
PDF |
Erschienen |
In: GRA - Volume 12 (2010) |
Datensatznummer |
250042424
|
|
|
|
Zusammenfassung |
The wide-angle seismic modeling of crust and mantle is key to understand the structure and
physical properties of the Earth’s interior. However, a reliable interpretation of the obtained
models requires having constraints on the model parameter uncertainties. For that reason, we
analyze here the potential and limits of most common, Monte-Carlo-based, statistical
uncertainty analysis schemes of velocity and reflector geometry models obtained by
travel-time tomography of wide-angle seismic data.
We have used a wide-angle seismic data set acquired off the Nicaragua convergent
margin. The data correspond to a profile that runs parallel to the trench axis along the
continental slope. It is 190 km long and includes 12 Ocean Bottom Hydrophones located
10-15 km apart. The data have been modeled following a joint refraction and reflection
travel-time tomographic inversion to obtain a 2-D velocity field of the overriding
plate as well as the location and geometry of the inter-plate boundary. The model
shows a ~5 km thick sedimentary cover with velocities varying from 2 km/s to 3.5
km/s, overlying a ~12 km thick basement with velocities between 4 km/s and 6.5
km/s.
Ray coverage shows that diving waves concentrate in the upper third of the model,
whereas the two bottom thirds are covered only by reflections from the inter-plate
boundary zone. This gives an idea of the areas where the model is likely to be better
constrained by the data, but the ray coverage alone does not provide a measure of the
accuracy of the results nor an idea of the significance of velocity-depth trade-off.
To analyze the accuracy of the model we have implemented a Monte Carlo-like
uncertainty analysis consisting of randomly varying an 1D velocity model, the depth of
the floating reflector, and the picked arrival times, to generate a total of 100 2D
reference models and perturbed data sets, to subsequently conduct an inversion for each
model-data set pair. Theory states that, under certain conditions, the mean and other
statistical measures such as mean deviation can be estimated from the velocity models.
By taking a range of variation that covers all the region of possible solutions, the
mean deviation of the Monte Carlo realizations give a measure of model parameters
uncertainty. It is therefore important to select the appropriate range of variation of
the different model parameters, so we have performed different tests changing the
range of variation of velocities, depth of reflector, and time perturbation to see its
influence in the obtained results. The objective is to assess the required conditions
to interpret the mean deviation of the model parameters as a reliable measure of
uncertainty.
Our results show that the improvement, or reduction, of mean deviation before and
after the inversion depends on different factors. The reduction of mean deviation
is much larger at the zones covered by diving waves (sediments and uppermost
basement) than at those controlled only by reflections (mid and lower basement). The
relationship between the mean deviation of the initial models (Ïi) and the mean
deviation of the final ones (Ïf) is linear below a critical Ïi value (Ï0) and converges
to a constant Ïf (Ïref) above Ï0, in areas covered by diving waves. This means
that we must select Ïi> Ï0 to be able to interpret Ïf (Ïref) as a measure of model
parameters uncertainty. The reduction of mean deviation of the floating reflector
depth is significant regardless of the significant velocity uncertainty. This means
that even if single velocity nodes are poorly resolved, the average of all velocity
values makes that the travel time to the floating reflector is not dramatically affected. |
|
|
|
|
|