Ordnance Survey – Great Britain's national mapping agency
![Ordnance Survey home page [logo]](/images/gui/logo-os.gif){kind=logo}
Modelling fuzzy and uncertain features
How Ordnance Survey can deal with fuzziness and uncertainty within data models.
The problem – multiple interpretations of boundaries
Traditional geographical information systems (GIS) are good at representing crisp boundaries but poor at representing uncertainty. Many features in the real world are uncertain. But GIS typically deal with empirical measurements – they store values that represent an interpretation of the world, but they have difficulty handling multiple possible interpretations. For example, topographic maps currently show features such as land cover types with crisp boundaries, but this method gives no indication of the transition between them. Place names are also represented on topographic maps, but places such as The Cotswolds do not have a specific boundary, so these named places are difficult to model and identify on maps.
Tackling the problem
This research aims to cater to the growing number of geographic information users who require this type of fuzzy information. This is done by exploring data modelling methods that support fuzzy objects and uncertainty and are already documented in the literature. The aim is to create a framework for the development of data models that support these concepts and can be used with Ordnance Survey data. Consequently, there are a number of research questions that we are tackling:
- What types of fuzziness and uncertainty occur in the real world?
- How can geographic uncertainty and fuzziness be stored in an information system?
- How can we represent fuzzy and uncertain objects meaningfully to data users?
- And can we develop a geographic data model to handle uncertainty?
Answering these questions will allow us to determine how best to incorporate these concepts into Ordnance Survey data in the future.
The impacts of uncertainty research
People often talk about objects that have uncertain boundaries or meanings, since we understand these in a more natural way than the crisp boundaries traditionally supported by GIS. Consequently, including these objects in the Ordnance Survey data model will open up a world of geography to many users who understand geographic information differently from its representation in GIS. This will create a broader appeal of such geographic information and the applications it will be able to support.