Research
Time-reversal imaging of the Bonin earthquake, revealing a circular rupture on the surface projection.
Time reversal imaging using virtual seismogram
Supervisor- Dr. Pawan Bharadwaj
Collaborators- Isha Lohan, Aswini V.J., A P Singh
Summary- This work builds upon the foundations in the previous study titled 'Learning Earthquake Source using Symmetric Autoencoder'. In that study, we explored the process of generating uniformly scattered virtual seismograms (USVS) through redatuming techniques facilitated by a symmetric autoencoder. By engineering USVS we are using the same scattering effect across all the pixels due to this virtual seismograms are aligned with preserving directivity. In this current work, we utilize these virtual seismograms for time reversal imaging, enabling the creation of rupture images without the need for any prior knowledge of fault information.
Paper Link- Under Prep
Visualization of SymAE earthquake source codes using t-distributed stochastic neighbor embedding (t-SNE). Each point represents the source code of a focal-sphere pixel associated with a specific earthquake. The t-SNE analysis aids in exploring the complex relationships between seismic sources: the clusters formed by pixels from various earthquakes provide insights into the similarities and dissimilarities of source characteristics. Furthermore, instances where synthetic- and real-earthquake pairs cluster together, such as sum1 and sum1finite, indicate a teleseismic similarity between finite-fault models and actual seismic sources, as opposed to hnd and hndfinite.
Learning earthquake sources using symmetric autoencoders
Supervisor - Dr. Pawan Bharadwaj
Collaborators - Isha Lohan, A P Singh
Summary- We introduce Symmetric Autoencoder (SymAE), a neural-network architecture designed to automatically extract earthquake information from far-field seismic waves. SymAE represents the measured displacement field using a code partitioned into two interpretable components: source and path-scattering information. We achieve this source-path representation using the scale separation principle and stochastic regularization, which traditional autoencoding methods lack. According to the scale separation principle, the variations in far-field band-limited seismic measurements resulting from finite faulting occur across two spatial scales: a slower scale associated with the source processes and a faster scale corresponding to path effects. Once trained, SymAE enables the generation of virtual seismograms engineered to incorporate subsurface scattering effects from other seismograms. In this work, we first validate the accuracy of these virtual seismograms generated by SymAE using both P and S waveforms. We then demonstrate the use of virtual seismograms for estimating the similarity between collocated earthquakes. Additionally, we introduce the concept of uniformly scattered virtual seismograms, where seismograms associated with different focal-sphere pixels are engineered to contain identical path effects. SymAE is an unsupervised learning method that can efficiently scale with large amounts of seismic data and does not require labeled seismograms, making it the first framework that can learn from all available previous earthquakes to accurately characterize a given earthquake. This work presents the analysis of nearly forty complex earthquake events, revealing differences between earthquakes in energy rise times, stopping phases, and providing insights into their rupture complexity.
Paper Link- Learning earthquake sources using symmetric autoencoder
Based on the above figures, there seems to be a potential water flow within a distance range of 300 to 400 meters. However, the low values of natural current do not offer sufficient confidence to make definitive conclusions regarding the presence of water flow.
2D Inversion of Self-Potential Data in the Ambet Region
Aim- Exploring Potential Water Bodies
Superviosr- Dr. Anand Singh
M.Sc theises- Link
This report details the application of inversion techniques to self potential data collected in the Ambet region near Mumbai, with the aim of identifying potential water bodies in the subsurface. The report discusses the methods used for data acquisition and processing, as well as the specific inversion algorithms employed. The results of the inversion are presented and interpreted, highlighting potential locations of water bodies in the subsurface. The report concludes with a discussion of the significance of these findings and potential areas for future research.