SCECCME Metadata Definition

1
SCEC/CME Metadata Definition
Phil Maechling 9 June 2003 maechlin_at_usc.edu http/
/www.scec.org/cme
2
Metadata Handling

• Essential to Success of Project
• Initial Approach Defined
• Metadata files
• Data Dictionary
• Search For Standards

3
SCEC/CME Project Development

• Hazard Map Created on SCEC/CME System

Metadata for Map IMR Param List IMR
Abrahamson Silva (1997), Gaussian Truncation
None, Truncation Level 2.0, Std Dev Type
Total, Site Param List Min Longitude -120.0,
Max Longitude -116.0, Min Latitude 33.0, Max
Latitude 35.0, Grid Spacing 0.05, AS Site
Type Rock/Shallow-Soil IMT Param List IMT
PGA, SA Period 1.0 Forecast Param List Eqk Rup
Forecast Frankel96_EqkRupForecast, Fault Model
Frankel's, GR Fraction on B Faults 0.5,
Rupture Offset 10.0, Background Seismicity
Include
4
Metadata Handling

• Initial Approach Text files containing
  Name-Value Pairs
• Attribute_Name Value
• (e.g. Organization USC)
• Dotted notations ok
• Attribute_Categorey_Name.Atttribute Value
• (e.g. Location.Latitude 34.0)

5
Metadata Handling

• Advantages
• Common Approach (e.g. Java Properties Files)
• Easy to read and understand
• Can be used as the basis for other
  representations.
• (XML files and Relational Schemas).

6
Metadata Handling

• Disadvantages
• Frequent duplication of data elements between
  datasets.
• Not easily searchable.
• Keeping linkages between metadata files and data
  files can be a problem.
• Un-sure handling of Units in metadata files.
• Metadata files become data file (resources)
  themselves and may need metadata (infinite
  regression).

7
Metadata Handling

• AWM Metadata was selected as first data set
  because of the expense of reproducing the
  datasets.
• Data Attributes were defined in a Data
  Dictionary.
• Example AWM Metadata file was produced for Olsen
  AWM Wavefield data.

8
Metadata Handling

• Data Dictionary defines each Data Attribute and
  the following information about each data
  attribute
• Name of attribute
• Description
• Type (using IDL types)
• Value Values/Ranges
• Required
• Example
• Comments
• Given the data dictionary, you should be able to
  create a valid metadata file for the AWM data.

9
Metadata Handling

• Weve asked Pathway 1 to formally define their
  metadata for various datasets.
• These definitions will be added to SCEC/CME data
  dictionary.
• We expect some attributes (data elements) will be
  common (e.g. modeler name, Location, Coordinate
  system).
• Attributes common to more than one data type are
  defined once in a common section, and then
  reference in each data type attribute list.

10
Metadata Handling

• Metadata issues
• As a project, we dont have clear definition of
  what metadata we want to save.
• Scientist want to recreate the datasets.
• User want to intrepret the data correctly.
• System operators want to manage the datasets
  effectively
• It appears we need to save all three kinds of
  metadata.
• No easy way has been defined to verify we have
  appropriate metadata for a data type.
• We need ways to enforce (automate) saving of
  metadata before data is created. Proposed
  solution is to require entry of metadata before
  calculation will proceed.

11
Metadata Handling

• Metadata issues
• We are probably moving towards a Metadata schema,
  to reduce data duplication, and enhance
  searchability of metadata.

12
Metadata Handling

• What are appropriate Metadata Standards ?
• Our current definition lacks essential data such
  as name (URI) of dataset it refers to.
• Can we learn from Metadata Standardization
  efforts such as DublinCore (http//dublincore.org/
  )
• Content Intellectual Property Instantiation
  Coverage Contributor Date Description Creator
  Format Type Publisher Identifier Relation Rights
  Language Source Subject Title