Reducing Information Overload in Large Seismic Data Sets

Abstract

Event catalogs for seismic data can become very large. Furthermore, as researchers collect multiple catalogs and reconcile them into a single catalog that is stored in a relational database, the reconciled set becomes even larger. The sheer number of these events makes searching for relevant events to compare with events of interest problematic. Information overload in this form can lead to the data sets being under-utilized and/or used incorrectly or inconsistently. Thus, efforts have been initiated to research techniques and strategies for helping researchers to make better use of large data sets. In this paper, we present our efforts to do so in two ways: 1) the Event Search Engine, which is a waveform correlation tool and 2) some content analysis tools, which are a combination of custom-built and commercial off-the-shelf tools for accessing, managing, and querying seismic data stored in a relational database. The current Event Search Engine is based on a hierarchical clustering tool known as the dendrogram tool, which is written as a MatSeis graphical user interface. The dendrogram tool allows the user to build dendrogram diagrams for a set of waveforms by controlling phase windowing, down-sampling, filtering, enveloping, and the clustering method (e.g. single linkage, complete linkage, flexible method). It also allows the clustering to be based on two or more stations simultaneously, which is important to bridge gaps in the sparsely recorded event sets anticipated in such a large reconciled event set. Current efforts are focusing on tools to help the researcher winnow the clusters defined using the dendrogram tool down to the minimum optimal identification set. This will become critical as the number of reference events in the reconciled event set continually grows. The dendrogram tool is part of the MatSeis analysis package, which is available on the Nuclear Explosion Monitoring Research & Engineering Program Web Site.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Sep 01, 2000
Accession Number
ADA529887

Entities

People

  • Chris Young
  • Dorthe Carr
  • Jeff Hampton
  • John Merchant
  • Julio Aguilar-chang

Organizations

  • Sandia National Laboratories

Tags

Communities of Interest

  • Materials and Manufacturing Processes

DTIC Thesaurus Topics

  • Application Software
  • Clustering
  • Data Sets
  • Database Management Systems
  • Databases
  • Engineering
  • Graphical User Interface
  • Human Systems Integration
  • Information Overload
  • Internet
  • Overload
  • Relational Database Management Systems
  • Relational Databases
  • User Interface
  • User Interface Engineering
  • Waveforms
  • Web Browsers

Fields of Study

  • Computer science
  • Engineering

Readers

  • Database Systems and Applications
  • Neural Network Machine Learning.
  • Seismology