436 
Ltd in February 1972 at a nominal scale of 
1/25,00. The quality of the film and prints is 
good and features can be easily recognised on 
them. 
(d) Swamp forests 
(e) Cut over or burnt over forest areas 
(f) Forest with subsidiary cultivation. 
8. Swamps and marshes 
9. Unproductive land 
3.3 1950 photography 
This 1/25,000 photography was taken by the 
Royal Air Force. The flight lines are not 
East/West, as is usual, but are curved because the 
aircraft flew in a spiral around a radio beacon. 
The film is variable in quality and some of the 
prints made from it lack contrast which makes 
interpretation more difficult. 
Because of the different qualities and 
characteristics of the two sets of photography the 
question arises as to whether differences in land 
use are or are artefacts of the material. 
4. LAND USE CLASSIFICATION 
4.1 The choice of a suitable land use 
classification is usually a matter of compromise 
between the desirable and the possible. "There is 
no one ideal classification of land use (s) and 
land cover, and it is unlikely that one could ever 
be developed. Each classification is made to meet 
the needs of the user "(Anderson et al 
1976). 
4.2 "In principle a land use classification must 
consist of a number of classes which are mutually 
exclusive and which together cover all the 
categories of land use likely to be encountered" 
(Rhind & Hudson 1980) . Each class should be 
capable of clear definition and the criteria used 
should be of the same type throughout the 
classification - e.g. the vegetation cover, the 
nature of ownership, type of farming, but not a 
mixture of them all. It is desirable that 
each class should contain a reasonable percentage 
of the area to be viewed; empty or nearly empty 
classes are a nuisance and a waste of time. 
Each class must also be recognisable from the 
material available without any additional 
information required to back it up. Above all the 
interpretation must be repeatable within a 
reasonable margin of error. 
4.3 The starting point for the present 
classification was the World Land Use Survey 
classification which is cited by Bickmore et al 
(1970). This is a general classification with 9 
classes which does not fit very well into tropical 
systems of rural land use. Classes 4, 5, 6, are 
difficult to apply to the extensive farming 
methods with ill-defined boundaries typical of 
savannah areas. 
World land use survey land classification 
1. Settlements and associated non 
agricultural lands 
2. Horticulture 
3. Tree and other perennial crops 
4. Cropland 
(a) continual & rotation cropping 
(b) land rotation 
5. Improved permanent pasture (managed or 
enclosed) 
6. Unimproved grazing land 
(a) used 
(b) unused 
7. Woodlands 
(a) Dense 
(b) Open 
(c) Scrub 
Other classifications used by workers in 
tropical areas were also examined, including those 
by Sirdas (1966), Collins (1966) and Lebon 
(1965) . All these were useful but included 
sub-classes which were either not applicable to 
the area, or required additional information not 
interpretavle from sthe source material. 
Reference was made to local conditons, the work in 
particular Prothero (1957), Hopkins (1974)and the 
LRDC (1977-79) plus the w's won experience of the 
area. This resulted in an initial classification 
comprising 6 main classes and 23 sub-classes. 
However, there were considerable problems with 
this ideal classification was devised, comprising 
only those classes which could be unequicaocally 
recognised on the available material with the 
equipment used for interpretaion. The finally 
accepted classification was as follows: 
10. Settlements and Associated Features 
11. Settlements 
12. Communication 
13. Industry 
20. Cultivation 
21. Enclosed farmland enclosuresless than lOha 
22. Enclosed farmland enclosures greater 
than/equal to lOha 
23. Unenclosured farmland 
24. Tree plantation 
30. Uncultivated 
31. Fadama (floodplains) 
32. Grassland not in fadama 
33. Scrub and small trees 
34. Swamps and marshes 
35. Woodland 
36. Gallery forest and kutame (remnants of 
high forest 
40. Wasteland 
41. Inselbergs and bare rock outcrop 
42. River and lakebeds 
In practice, for purposes of data analysis, 
these classes are merged to form five larger 
categories: 
Group 
Containing 
1. 
Settlement 
11, 
12, 
13 
2. 
Farmland 
21, 
22, 
23 
3. 
Grass/scrub 
31, 
32, 
33 
4. 
Woodland 
24, 
35, 
36 
5. 
Barren 
41, 
42 
5. 
EQUIPMENT 
The aerial photography was available as 230mm x 
230mm paper prints. These were examined 
stereoscopically using a folding mirror 
stereoscope which, with its binocular high 
magnification attachment proved very suitable for 
single stereomodels. A CPI stereoplotter was 
used for a short time, but the better optics and 
high magnification were counterbalanced by the 
trouble of setting up the stereomodel. Land use 
information was collected in digital form and 
analysis performed using a 48k APPLE II 
microcomputer with floppy disk storage and a dot 
matrix printer.