UNIT 7: FINDING, CREATING, AND INTERPRETING METADATA
Written by Mary Lynette Larsgaard, Map and Imagery Laboratory, Davidson Library, University of California, Santa Barbara, mary@sdc.ucsb.edu
Context
The quick definition of "metadata" is "data about data." Like most circular definitions, this is not very helpful. More usefully, metadata is relatively brief descriptive information about an item, that is a surrogate for that item, like a record in an online catalog in a library. For our purposes, metadata is surrogate information about digital spatial data. If you could directly search data, you might not need metadata; when this is not possible, then metadata is essential, if you want to search quickly through your own data and especially if ever you want other people to be able to find your data. It is essential also to enable users of GIS data easily to find out enough about data in a GIS so that the data may be easily processed and interpreted. In the context of GIS, metadata may exist at several different levels: for a GIS database as a whole; for individual layers; for individual tiles; and for individual points, lines, and polygons.
The purpose of metadata is to enable a user to find needed information quickly and efficiently. One difficulty with metadata is that cataloging-or, if you prefer, metaloging-is a considerable amount of work. This is because of the many standards that the library world uses to generate metadata:
- a set of cataloging rules (the Anglo-American Cataloguing Rules, AACR);
- a set of field order and punctuation rules (the International Standard for Bibliographic Description, e.g., ISBD(CM), for Cartographic Materials);
- several sets of subject heading lists of which the most prominent is the Library of Congress Subject Headings (LCSH); a database format for communication in online systems (USMARC = United States Machine-Readable Cataloging format); and the most current, the U.S. Federal Geographic Data Committees Content Standard for Digital Geospatial Metadata (FGDC; first edition, 1994; second edition in draft, 1997; link to http://www.fgdc.gov/Metadata/metahome.html). On the other hand, given the type of fields that are available for cataloging GISs, the producer of the data is the best possible cataloger of the data.
The advent in 1994 of FGDC has been extremely influential in the world-wide application of metadata to digital geospatial data. An excellent Website providing links to the ongoing metadata efforts in this field is "Jan Smits' bookmarks concerning metadata for digital spatial data," http://www.konbib.nl/persons/jan-smits/metadata.htm. This provides links to international efforts such as ISO/TC211 and CEN/TC287, and to national-level efforts, such as the British DESIRE. For persons in the U.S., two useful pages are http://www.fgdc.gov/Metadata/ and http://www.alexandria.ucsb.edu/public-documents/metadata/. The former is the page of the U.S. government agency that leads metadata efforts within the Federal government; the latter is the homepage of the Alexandria Digital Library, whose goal is to provide online access to georeferenced information.
A portion of the FGDC page of especial interest is the "Metadata Tools Survey, July 1997" (http://www.fgdc.gov/metadata/toollist/metatools797.html), which lists and briefly annotates about fifteen different tools for generating metadata. A prominent example of automatic extraction of metadata is ESRI's document.aml. Document.aml (version 7.1.1. to be the last while ESRI works on new metadata documentation tools) is available from ESRI's Web site at http://www.esri.com; go to the site and do a search on the term "document.aml" for instructions on how to get a copy of the software and steps in using it. For an article with examples of using document.aml, read "Automated Capture of Metatada; Simple Procedures and Tools for Editing Coverages" by Dan Williams, go to http://www.esri.com/base/common/userconf/proc95//to300/p259.html.
An effort to make manual metaloging easy by using a relatively limited number of fields is the Dublin Core. The Dublin Core contains fifteen fields:
- subject
- title
- creator/originator
- publisher
- other agent/contributor
- date
- resource type
- format
- resource identifier
- source
- language
- coverage
- rights management
- relation
- description
These metadata fields are intended to be the minimum amount of information that a Webpage producer can enter in order to assist prospective users to find the page. The Dublin Core homepage is at http://www.purl.org/metadata/dublin_core. Several agencies have used either Dublin Core or some more limited subset of it, and in some cases have homepages where you may catalog your homepage by filling in a form. The Nordic Metadata Projects "Dublin Core Metadata Template" is an example of this. It is at http://www.ub2.lu.se/metadata/DC_creator.html . For examples of others, see the Dublin Core generator of the United Kingdom Office for Library and Information Networking (UKOLN), at http://www.ukoln.ac.uk/cgi-bin/dcdot.pl, or one from Australia at http://www.edna.edu.au/edna/owa/.
If you feel brave or if you want maximum documentation of your GIS data, use the FGDC document Instructions for Implementing the FGDC Metadata Standard, and Testing Conformance (text and word-processor versions are at: http://www.fgdc.gov/Metadata/metahome.html). The FGDC document has the following sections:
- Identification information;
- Data quality information;
- Spatial data organization information;
- Spatial reference information;
- Entity and attribute information;
- Distribution information;
- Metadata reference information;
- Citation information;
- Time period information;
- Contact information.
Citation information forms the basic bibliographic reference. Fields in the latter and in Time period information and Contact information may be used as often as they are needed. Section 5, Entity and attribute information, is specifically for documenting GIS data.
Encouragingly enough, there are occasionally workshops to teach metadata creation. See for example the national satellite videoconference, "A Practical Guide to Metadata Implementation for GIS/LIS Professionals," Wednesday, October 15, 1997, 1-3pm CDT. This conference--produced by the University of Wisconsin, Cooperative Extension--has the following sections: What is Metadata and Why Is It Important?; Get Acquainted with the Content Standards for Digital Geospatial Data [FGDC]; Getting Started; Metadata Tools; Practical Experience with Creating Metadata, a State Agency Perspective; Other Metadata Issues.
There is another Federal effort to document data, the Government Information Locator Service (GILS). The object of GILS is to enable potential users of Federal information to readily locate that information. For a report on GILS, see:
Moen, William E.; and McClure, Charles R. 1997. An evaluation of the Federal governments implementaiton of the Government Information Locator Service (GILS). http://www.unt.edu/slis/research/gilseval/
Example Application
- You need to find existing information in digital form for a GIS database that you are compiling, and you don't want to scan or digitize hardcopy data that is already available in digital form. Use Webbrowser searches, or, for a short cut, http://www.alexandria.ucsb.edu/other-sites/
- You are looking at metadata for a digital data set; what does all this gobbleydegook mean? If its in FGDC form, to find out the meaning of the fields, go to http://www.fgdc.gov. If it is from a library's online catalog, consult a cataloger, or read AACR (current version) or USMARC.
- You need to use an existing set of digital data. But before you use the data, you need to know about data quality - accuracy, lineage, source of information. Go to the FGDC standard (in Website given in 2. above), and read section 2, "Data Quality Information." Take a look at the metadata for AVHRR, http://edcwww.cr.usgs.gov/nsdi/html/avhrr/, or any other metadata you can find. Figure out what all that information really means, and what it tells you about the quality of the data - is this is a good dataset or not, and why or why not?