Digital Elevation Model
The image above is a map of elevation above sea level created in ArcView
and made into a gif image using XV. To create the input data the source
data was imported, projected into a 1DM geographic projection and ouput
as an ascii file using the following commands within ArcInfo:
Land One-KM Base Elevation Data version 0.1 (GLOBE).
Elevation and relief relate to and affect settlement patterns and accessibilty.
Although people are increasingly capable of building and supplying goods
and services to remote places and many people enjoy living in remote places,
the general pattern is that the higher more rugged and remote places have
a lower population. Lowland areas which are highly accessible, have
a less harsh climate and have plenty of adequate sites for development
provide lots more opportunity for it and also tend to have higher population
densities than highland areas. It is unlikely that very high population
densities will be observed in hilly remote places at a scale of roughly
Relief limits the availablility of naturally appropriate sites for modern
day residential and commercial developments and has obiously been a major
constraint to communication, transportation, development and the functionality
and size of settlements historically. There is a clear justification
for including height above sea level as a predictor for a population density
model and also many other obvious and subtle yet compelling ways of transforming
this and more detailed DEM's to estimate both population density and land
From the image above it is clear that Europe and the Mediterranean region
of the EU have a great range of different landscapes and reliefs more varied
than that found in a small part. It is likely that the most
general relationship between elevation, relief and population density in
Great Britain (described above) is similar to that in the rest of the EU.
However, for some regions or localities the relationship could be significantly
or even radically different because of the nature of the human activities
undertaken in those areas. Regional geographical knowledge about
how the general relationship between relief and population density differ
in one area compared to the next and compared to the more general broadbrush
patterns could be used to adjust and improve the population surface outputs
from the neural net modelling activities undertaken in Task1.
One way in which to do this is to create fuzzy rules and apply them to
(fuzzy) areas of the map. For example a fuzzy map modification rule
for southern Italy could adjust the neural net output to account for where
the historical old towns which have developed at the top of hills (originally
presumably predominantly for defensive purposes) have resulted in higher
population densities in general for that area (than might more generally
be expected) at the top of hills.