 |
6.4 Producing atlases from GIS Extending the above approach allows atlases to be produced using a small amount of spatial data coupled to a wide variety of attribute data. A good example of this is Woods and Shelton (1997). They use a single generalised layer of 19th century registration district boundaries in England and Wales to produce an atlas of mortality in Victorian times. By looking at the spatial pattern they are able to provide new insights into phenomena that have a pronounced geographical pattern. This atlas shows the importance of spatial detail and the power of maps to present it. There were over 600 registration districts in England and Wales in the Victorian era, and the maps shade all of them. This allows the authors not only to comment on the general patterns, but also on specific details and exceptions. When looking at infant mortality in the 1860s, for example, the authors note that the maps show a general pattern of high rates in urban districts and lower rates in rural ones. Looking in more detail reveals exceptions to this, for example, there were high rates found in rural areas around the Wash, in Lincolnshire and East Yorkshire, and in Cornwall. This kind of detail is readily apparent from maps but is difficult to spot in other ways. Kennedy et al.'s atlas of the Great Irish Famine follows a similar approach but rather than using a single layer of spatial data they use different layers for different dates (Kennedy et al. 1999; see also Chapter 5). By mapping changing housing conditions they show that on the eve of the famine the lowest quality housing, mud huts and similar constructions, were concentrated in the west of Ireland where they could form up to 50% of the housing stock. In the east, housing of this quality was less common. The poor who lived in this housing were the most seriously affected by the famine, so as the famine progressed this class of housing all but disappeared in the east, while in the west its importance also declined but to a lesser extent. In the above examples the atlases produced are cartographically simple products based around choropleth mapping. GIS can also contribute to atlases that are more sophisticated cartographically. Here, rather than linking a large number of attribute datasets to a limited number of spatial layers, the GIS becomes a database of the spatial features that will be used in the atlas and allows them to be combined to produce highly sophisticated cartographic products significantly more cheaply than through traditional methods. A good example of this is volume II of the Historical Atlas of Canada (Louis 1993), which was produced using ArcInfo. Pitternick explains the advantages that using this software gave to this volume, compared with volumes I and III where more traditional methods were used (Pitternick 1993). |
