The Problem: Modern digital methods provide less information about data quality than traditional graphic products. A single 'accuracy' figure will not be enough to summarize the quality of a digital database.
Approaches to accuracy and data quality
· Map as mirror: exact replica of Earth
BUT a map is a purposeful simplification; represents chosen objects
and relationships.
Therefore, a map MUST lie. 'The map is not the territory.'
"Terrain nominale": testing depends on specifications
of how to see. (258 lecture)
· Responsibility (expressed or implied)
Specified procedures; accepted practice
Using a given method assures quality.
Examples: ISO 9000; US Federal Geodetic Control Standards
[Complicated when technology changes are rapid.]
Conformance to expectation (reducing error below a threshold)
testing? producer assumes liability; users passive.
Example: National Map Accuracy Standards (USA)
Truth in labelling
Producer documents procedures and results of tests.
[Role for distributor librarian; clearinghouse; facilitates
access]
User evaluates report => informed judgement about fitness
for specific use
· Goals: Yvan Bédard distinguishes:
error reduction: attempt to minimize error
error 'absorption': guarantee to compensate for losses; define
the error away
Example of absorption:
cadastral certification vs. US 'opinion of title', title insurance
Comparison of expectations (legalities):
nautical chart, cadastral survey, topographic map, thematic
map ...
· Database Management manages more than the data; access
to the data
Metadata: data quality/fitness for use describe contents of database.
· Evolution of standards
from paternalistic, technology-limited & assumptions concerning
users
Either: insiders set procedures OR set attainable thresholds.
towards fitness for use: SDTS, DIGEST (A & VPF), NTF etc.
Users decide.
Why there is not a single accuracy figure for modern GIS.
emphasis on integration of different sources
differences in measurement "framework"
differences in cost/accuracy/"up-to-date-ness"
What are all the concerns in Data Quality?
SDTS Data Quality Specification
[review of definitions- see SDTS text]