In this article we presented a frequency-domain kinematic finite fault
inversion that uses the fuzzy approximation system and reduces the
number of basis functions to expand the seismic source function.
The figure on the left show the general idea of using a fuzzy system to approximate an arbitrary function (black line) using four basis functions (dotted lines).
We employed this idea to reduce the number of fault discretizations in the kinematic finite-fault inversion.
By reducing the number of basis functions, we arrive at a more stable, less sensitive to the noise inversion.
As the following figure shows, using a low number of fuzzy basis functions, helps us to reduce the number of small singular values (according to GSVD: generalised singular value decomposition) have a more stable inversion, however with a reduced resolution.
Although this method works fine for inversion of low frequency ground motion, there are still room to improve for high-frequency components.
This article aimes to show the application of Neuro-fuzzy inversion method on the real-data recorded during the M6.2 August/2016 in central Italy. By inverting 3 datasets, including static GPS (~ 0Hz), high-rate GNSS(<0.09 Hz), and strong motion data (0.09<f<0.5 Hz) we obtained the slip-rate history for this earthquake.
Having abundance of observations, we're able use a part of data set for training the model, and leave the other part for validation of the model.
In the following figure, we use the model obtained from strong-motion data to simulate ground motion on the high-rate GNSS stations. we showed that our model is able to predict unseen data.
In this article, we presented a new probabilistic tsunami hazard (PTHA) model for the Iranina coasts of makran subduction zone.
Considering tsunamogenic events ranging between M7.8-M9.0, we numerically simulated run-up heights for return periods of 475, and 2475 years.
According to the PTHA results, Chabahar and Sirik towns are at the highest and lowest tsunami risk, respectively.
created with
Website Builder Software .