A place in history: a guide to using GIS in historical research

CHAPTER 9: PRESERVATION, DOCUMENTATION AND THE ROLE OF THE HISTORY DATA SERVICE

9.4 Documenting a GIS dataset

The maintenance of comprehensive documentation detailing the data creation process and the steps taken involves a significant but profitable investment of time and resources. It is more effective if documentation is generated during rather than after a data creation project. Such an approach is likely to result in a better quality data collection as well as better quality documentation, because the maintenance of proper documentation demands consistency and attention to detail. The process of documenting a data creation project can also have the benefit of helping to refine research questions and it can be a vital aid to communication in larger-scale projects.

Good documentation is crucial to a data collection's long-term vitality: without it, the resource will not be suitable for future use and its provenance will be lost. Proper documentation contributes substantially to a data collection's scholarly value. At a minimum, documentation should provide information about a data collection's contents, provenance and structure, and the terms and conditions that apply to its use. It needs to be sufficiently detailed to allow the data creator to use the resource in the future, when the data creation process has started to fade from memory. It also needs to be comprehensive enough to enable others to explore the resource fully; and detailed enough to allow someone who has not been involved in the data creation process to understand the data collection and the process by which it was created. Except in exceptional cases the documentation that accompanies a data collection deposited with the History Data Service must meet the standards set out in their Guidelines for Documenting Data.

Documenting any dataset is a resource-consuming task but GIS datasets are often more complicated. To start with, it is not always clear in a GIS what the documentation should refer to: should it be the whole dataset, each layer within the dataset, or even individual features within layers. The answer to this is often a combination of all three. While there is certainly a need for documentation about the dataset as a whole, individual layers or features may require further documentation to describe them.

The documentation required will vary between those captured from primary and secondary sources. It is impossible to produce definitive and exhaustive lists of the information required for both types of data, but primary source metadata is likely to need information on the following:

• the method of data capture
• the hardware and software used to capture the data (including version numbers)
• the resolution and accuracy of data capture
• an assessment of data quality
• a description of the purpose of the data capture
• who performed the data capture
• when the data were captured
• who owns the data
  

For secondary sources, this list is complicated by the need to include information on both the original source and on the data capture. In this case the list might include:

• the purpose for which the data were captured
• the hardware and software used to capture the data (including version numbers)
• the purpose for which the source was produced
• the scale of the original source
• the method of data capture
• the scale or resolution at which the data were captured
• the root mean square (RMS) error of the data capture
• who owns the copyright on the original source
• the date when the data were captured or purchased
• who captured the data
• who owns the captured data
  

Where additional work is done on an existing dataset, whether to derive a new dataset or merely update an existing one, this also needs to be documented. Here again the documentation would need to include:

• a description of what has been done to the data
• the hardware and software used to capture the data (including version numbers)
• the date of the work
• who did the work
• the relationship between this work and the dataset it was derived from