Interactive Large-Scale Seismic Noise Analysis in the Cloud Using Xarray, Dask and Zarr

Session: Applications and Technologies in Large-Scale Seismic Analysis
Type: Oral
Date: 4/28/2020
Time: 09:30 AM
Room: 120 + 130
Description: 

As research in seismology continues to identify valuable new signals in ever-growing data streams, it becomes more important to explore research tools and platforms that can scale from small exploratory analyses to large survey-style applications. The commercial cloud offers a diverse and powerful platform to quickly perform large-scale research, but it also comes with a number of practical challenges. Most existing research software in seismology is not compatible with a remote distributed system like the cloud. Additionally, there is a significant learning curve in using a new software ecosystem. Finally, standard seismic formats, such as miniSEED, SAC or PH5 may not be optimal for access on distributed systems, where the balance between compression, file size and network communication is different compared to local or HPC systems. In this work, we use the Xarray, Dask and Zarr libraries in the Python software ecosystem to address some of the challenges outlined above. We perform a continent-scale seismic noise analysis using the Amazon Web Services cloud platform to demonstrate an interactive and fully in-cloud research workflow that accelerates time-to-result.

Presenting Author: Jonathan K. MacCarthy

Authors