Rivers.shp contains rivers of various sizes. It can
be displayed satisfactorily in two ways, either using a graduated symbol
in the legend, with Fc as the classification field (use increasingly
thick lines for codes 6223, 6224, 6225, then a thin line for 6255, which is
the code for lakes), or using a graduated colour, again with Fc as
the classification field, and the blue colour ramp. Roads.shp needs
a set of unique values, again based on the Fc field, which will distinguish
between A, B, and minor roads (codes 3001, 3002, 3004).
Looking at the sites on their own, it's immediately clear
that their distribution isn't smooth across the district. There's significant
grouping and a strong tendency towards linearity in some places. When the
rivers and roads are displayed as well, it's again clear that many of the
sites are placed close to features of one or another of these themes, and
this seems logical; both rivers and roads tend to follow river valleys, which
also have attractions for settlement of various kinds.
Displaying the archaeological sites on top of the elevation
data (which can usefully be shown as graduated colours, with min
elevation as the classification field, and one of the terrain elevation
colour ramps), there is an even more striking relationship between the locations
of sites and the nature of the terrain. The highest areas have few or no sites,
and there's an apparent division between prehistoric sites occurring on higher
ground and medieval ones at lower elevations.
These immediate impressions may be a valuable stimulus to
the formation of hypotheses, but one of the most useful functions of GIS software
is that it gives us the ability to quantify the relationships which we think
we see. Ultimately our hypotheses should be subjected to statistical testing,
a subject outside the scope of this document. Here we will concentrate on
investigating the basic spatial relationships between features of different
kinds in the landscape, and it must be borne in mind that it is the spatial
relationships which we are looking at. To take as an example the possibility
that there is a difference in distribution of prehistoric and medieval sites
with respect to elevation, if we decide in the end that there is a real relationship
between elevation and the age of a site, it could be caused by a number of
factors - perhaps there was a preference in the earlier periods for locating
activities at a higher level, or perhaps later settlement was more dense at
a lower level, or agricultural exploitation more intensive, so that prehistoric
sites which did exist have been destroyed or concealed by later activity.
GIS analysis may reveal spatial relationships, but it can't explain them.
We have already looked at simple analysis of spatial relationships,
using the Select by Theme function, but we will now investigate another
important GIS facility, the creation of buffers.
Buffers are polygons created around the selected features
of a Theme, which can be used to select features of other Themes which fall
within them. In this way they provide a facility similar to the Are Within
Distance Of option used with Theme - Select by Theme, but they
can be more complex than this. Buffers can be created as:
- Polygons with boundaries at a single specified distance
from the features of a theme (e.g. 500m)
- Multiple rings, forming a series of nested polygons where
the user controls the number of rings and the distance between them
- Polygons with boundaries at a variable distance from
the features of a theme, depending on the value in a field of the Attribute
Table.
Buffering is an operation which can place considerable demand
on your computer system. For larger datasets, as the operation starts you
will be given information about how many features and vertexes are being buffered,
along with a warning 'May take a long time'. Perhaps no two users of a GIS
will agree about the definition of 'a long time', but for complex buffering
operations you should certainly be prepared to wait for 10-15 minutes, and
perhaps much longer, depending on the capacity of your system. As the operation
proceeds a report will appear at the bottom right of the ArcView window, telling
you how much (in percentage terms) has been completed, along with a blue bar
which gradually lengthens, and there is a STOP button which allows you to
abort if necessary.
Before you can create buffers, the map and distance
units (both metres) must be set using View - Properties. If you
don't do this the buffer function will be unobtainable.
For the first exercise we'll buffer the river shapefile,
so add this to a View if you haven't already, then
- use Theme - Create Buffers. You can buffer either
graphics in the View, or the selected features of a Theme, so click in the
box next to the Theme option and select its name from the list. You will
see how many features are in the Theme, and how many (if any) are selected.
Click Next.
- Select the kind of buffer you want to create; in this
case choose At a specified distance, and leave the distance at the
default of 500 (the distance units will appear in a field towards the bottom
of the window). Click Next.
- Choose whether to dissolve the barriers between buffers
or allow them to remain separate; choose Yes (the default). You also
need to decide how to save the buffers, which may be as graphics, as features
in an existing Theme, or as a new Theme; choose the new theme option. On
this line is the default path and file name for the new theme (always buff1.shp
for the first buffer you create). If you click on the button at the end
of the line, which has an icon like an open folder, you can choose another
path and/or file name.
- Click Finish and the process will begin.
At the end you will have in the View a new and complex set
of polygons which outline those areas which are within 500m of any river in
the area. If you display the archaeological sites over this you will get some
impression of how may fall within the zone, but it's obviously much more useful
to use another spatial analysis tool to extract the sites so that they can
be examined more easily.
Activate the theme containing the archaeological information,
choose Theme - select by Theme, and opt to Select features of active
themes that Intersect the selected features of [name of your buffer theme].
The points representing sites within the buffer zone will turn yellow, and
if you open the theme's table the corresponding records will be highlighted
in yellow. Click on the name of the Period field, then use Field
- sort Ascending, which will sort all the records into the order of their
period, with the selected records still mixed up with the others. Now use
Table - Promote, and the selected records will move to the top, but
still sorted by period. This is one simple way to obtain a quantification
of sites which are within 500m of a river, and if you just count the number
of sites of each period, you should find that there are:
This operation could be repeated after sorting on the basis
of Type, and promoting the selected records again, which would tell
you, amongst other things, that in the buffer zone there are 8 chapels, 7
churches, 20 enclosures, 16 settlements, 6 shielings, and so on. It is, however,
a rather laborious process, and there is a better way to do it.
First of all, convert the selected features of the theme
containing the archaeological data to a new shapefile, then open its Attribute
Table. With the Table active, use File - Export, which allows the data
in the table to be saved in another format. In the export dialogue box choose
the delimited text option, which will create a file of plain text with
commas as the field delimiters. This file can then be read by almost any other
software package, and if you open it in Excel (or another spreadsheet) you
can use the Pivot Table facility to automatically generate a table showing
the counts of period or type without the time-consuming necessity of manual
counting.
Here are figures for the numbers of sites of different periods
which are within 500m of rivers, roads, and modern settlements, but converted
to percentages of the total number of sites of each period; there are very
few sites of the early medieval period, and only one modern one, so as ever
caution is required when looking at the figures.
