Metadata Tips
Defining your data set too finely or too broadly
It's easy to become overwhelmed trying to individually document every data table and resource. On the other hand, trying to cover all of your data resources with a single metadata record will drive both you and your data users crazy. A good rule of thumb is to consider how the data resource is used - as a component of a broader data set or as a stand-alone product that may be mixed and matched with a range of other data resources.
Metadata Description
Next to Title, Description is probably the most-read section of the metadata. It is composed of three narrative parts: Abstract, Purpose, and Supplemental Information. Potential users will consult these to understand your data.
- Abstract — a brief summary describing the data set. The abstract should address the following questions: What is the topic of the data set? What is the data set comprised of? Where were the data collected? What general area do the data represent?
- Purpose — the purpose should describe why the data were collected. Provide a summary of the intentions for the data set as it was originally developed. The purpose should address the following questions: Why were the data collected? How is this data set different from other well-known similar data? What role do the data serve? How can the data be used?
- Supplemental Information — if there is any other important information you feel should be included in the description of your data, this section is where you add it.
Confusing 'Currentness Reference' with 'Publication Date'
While the Currentness Reference (1.3.1) may refer to a publication date it is actually a qualifier to Time Period of Content (1.3).
- Does the time period refer to the date/time of data capture or ground condition as in photography or field data collection?
- Does it refer to the date the information was officially recorded as in a deed?
- Does it refer to a publication date as in a '1978 USGS Topo map'?
Basically, the idea is to let prospective users know how well you are able to 'nail' the actual time period of content.
Misunderstanding resolution
Who could blame us? The purpose of these fields is to indicate how coarsely or finely information was recorded. For example
For example:
| Value |
Resolution (4.1.1.1 or 2) |
Geographic coordinate units (4.1.1.3) |
| 30° 30’30” |
0.0028.(1° / 360”) |
degrees, minutes, seconds |
| 30° 30’ 30.01” |
0.000028 (1° / 36,000”) |
degrees, minutes, decimal seconds |
| 30.00001° |
0.00001 (1° / 100,000) |
decimal degrees |
NOTE: units of measures are provided under element Planar Distance Units (4.1.2.4.4) and would be ‘meters’ for the TM example provided and likely millimeters for the vector example.
Putting too much faith in metadata tools
Human review is the only thing that matters. The tools are there to help, remember:
‘garbage in - garbage out’.
Taking the minimalist approach
A common overreaction to the expansive nature of the CSDGM is to adopt ‘minimal compliance’ as an operational approach. Limiting your documentation to the ‘required’ portions of Sections 1 and 7, or even all ‘required’ fields, will limit the value of your effort and the metadata records you produce. Instead, identify those fields that apply to your organization and data, and create functional templates, or subsets, of the CSDGM.
Understanding assessments of consistency, accuracy, completeness, and precision
Section 2. Data Quality Information is intended to provide a general assessment of the quality of the data set. This represents the ‘Achilles heel’ for many RS/GIS professionals.
Consider it an ‘opportunity’ to get to know your data set. A brief summary:
- Attribute Accuracy Report (2.1.1)
Assessments as to how ‘true’ the attribute values may be. This may refer to field checks, cross-referencing, statistical analyses, parallel independent measures, etc. Note: this does NOT refer to the positional accuracy of the value (see 2.4).
- Logical Consistency Report (2.2)
Assessments relative to the fidelity of the line work, attributes and/or relationships. This would include topological checks, arc/node structures that do not easily translate, and database QA/QC routines such as:
- Are the values in column X always between ‘0’ and ‘100’?
- Are only text values provided in column Y?
- For any given record, does the value in column R equal the difference between the values provided in columns R and S?
- Completeness Report (2.3)
Identification of data omitted from the data set that might normally be expected, as well as the reason for the exclusion. This may include geographic exclusions, ‘data was not available for Smith County’; categorical exclusions, ‘municipalities with populations under 2,500 were not included in the study’; and definitions used ‘floating marsh was mapped as land’.
- Positional Accuracy (2.4)
Assessments of horizontal and/or vertical positional (coordinate) values. Commonly includes information about digitizing (RMS error), surveying techniques, GPS triangulations, image processing or photogrammetric methods.
An indication as to how ‘finely’ your data was recorded, such as digitizing using single or double precision. Note that the precision of the value in no way reflects its accuracy or truthfulness.
Glossing over Section 5. Entity and Attributes
Another of the GIS professional’s ‘Achilles tendons’, this section maps out data content and should be a product of your data design effort.
- Use the relational database format as a guide:
- Entity Label (5.1.1.1) – Table Title
- Attribute Label (5.1.2.1) – Column Titles
- Attribute Domain Values (5.1.2.4.X) – Recorded values within each column
- Domain Types – set of possible data values of an attribute Enumerated Domain (5.1.2.4.1)
- A defined pick list of values
- Typically categorical such as road types, departments, tree types, etc.
- Range Domain (5.1.2.4.2)
- A continuum of values with a fixed minimum and maximum value
- Typically a numeric measure or count, may be alphabetic (A–ZZZ)
- Codeset Domain (5.1.2.4.3)
- A defined set of representational values
- Coding schemes such as FIPS County Codes, or Course No. (GEOG 1101)
- Unrepresentable Domain (5.1.2.4.4)
- An undefined list of values or values that cannot be prescribed
- Typically text fields such as individual and place names
- Entity Attribute Overview (5.2.1)
- A summary overview of the entities/attributes as outlined in either Detailed Description (5.1) or an existing detailed description cited in Entity Attribute Detail Citation (5.2.2). Note that the field should not be used as a stand-alone general description.
Thinking of metadata as something you do at the end of the data development process
Metadata should be recorded throughout the life of a data set, from planning (entities and attributes), to digitizing (abscissa/ordinate resolution), to analysis (processing history), through publication (publication date). Organizations are encouraged to develop operational procedures that 1) institutionalize metadata production and maintenance, and 2) make metadata a key component of their data development and management process.
Not doing it!
If you think the cost of metadata production is too high – you haven’t compiled the costs of not creating metadata: loss of information with staff changes, data redundancy, data conflicts, liability, misapplications, and decisions based upon poorly documented data.
