Representations of Scale

I myself, have struggled with the issues of scale in health data whereby social and geographical scales interact but are very complex to model. The choice and use of scale naturally impacts the results of the analysis, and its limitations and complexities have been widely acknowledged in the human geography literature (Goodchild and Proctor, 1997, Longley and Batty, 1996, Sheppard and McMaster, 2003 and Atkinson and Tate, 2000). Indeed, even the terminology has different meanings, describing the same spatial data, a geographer may refer to it as small scale whereas an ecologist could refer to it as large scale. Therefore, representations of scale are something we should consider in our problem scenarios and discussed in the user interviews.

To use an example from health care: commonly web-mapping applications in health simply present data at one scale – so having a zoom function can be redundant functionality as the map never changes. What you often see is a thematic map showing the level of deprivation for a particular administrative boundary layered on top of a reference map but the detail in the thematic map never varies regardless of the level of zoom. If we were developing a health scenario around diabetes, what would be great to see is the layers change from household risk when the user is zoomed into the building detail on the map and then as you progressively zoom out the data changes. So with different zoom levels the data changes scale from individual to street to output area to lower super output area etc in line with the scaling of administrative boundaries/ policy decision making in the UK. This type of zooming would illustrate to new users of GIS, the issues of scale that experienced users of GIS take for granted.

Taking this one-step further, what would be nice is to have a zoom function that corresponds more to functional scale (similar to what Claire suggested in the post relating to the use of scale in biology). With functional scales of analysis, data are summarised and generalised according to the spheres of influence associated to the phenomena being mapped. So in the diabetes example risk data would be organised by the individual, neighbourhood community, GP catchments, (GP Constoria), hospital catchments etc.

Beyond the G3 Project

It is probably quite unusual to be thinking about things beyond this project when we've only just started work, but I've had a number of conversations over the last two years that highlight the potential of web mapping tools such as Google Maps in teaching in other disciplines that don't work at 'geographical' scale.

For example, biologists and nanotechnologists work at sub-mm scales - but more importantly they also have representations of cells from gene level to cell and tissue level - in other words, they work with what the GIS world calls 'Levels of Detail' and 'generalisation'. Apparently it is not yet possible in biology to automatically generalise from the detailed data upwards, but a web map that replaces the 'map' with cell-related images at the different scales would be a useful teaching tool (and is very easy to create with existing technology and web mapping tools).

Having such a 'map' would allow students to zoom in and out between the various scales and to click on various objects and identify them (linking to additional material). In other words, this could be a useful teaching tool. Animated data (what the GIS world calls time-series data) - showing the interaction between objects at a single levels - could also be included.

Perhaps something to think about as we are developing our tool kit?

I'd welcome some feedback from anyone out there who could provide more concrete, relevant details and terminology for what could be shown at the varying scales.

(With thanks to the people at the Crucible dinner at UCL last night for a very interesting conversation about scale)