452 
7 (180/61 of January 24th, 1976) was specifi 
cally used to delineate water bodies, urban- 
rural settlement and infrastructure. Band 5 
is used for the black and white reproduction 
in this paper, showing maximum detail. The 
land-cover map 1:1M based on visual interpre 
tation of LANDSAT imagery had been completed 
and was available for further exploration 
(Grootenhuis, Weeda and Kalambo 1986 a). 
Background material consulted were: 
Geological map, Degree sheet no: 51 
NE Quarter, 1971, 1:125.000 
Geological map, Degree sheet no: 52 
NW Quarter, 1963, 1:125.000 
Ministry of Natural Resources, Mines and 
Geology Department, Kenya 
Exploratory Soil map and Agroclimatic Zone 
map of Kenya, 1982, 1:1.000.000 
Ministry of Agriculture, Kenya 
Route map of Kenya, 1976, 1:1.000.000 
Survey of Kenya 
Nairobi and Environs, 1978, 1:100.000 
Survey of Kenya 
Topographical 
1 :50.000 
map, 
Limuru, 
Sheet 
148/1,1976 
Topographical 
1 :50.000 
map, 
Kiambu, 
Sheet 
148/2,1976 
Topographical 
1 :50.000 
map, 
Ngong , 
Sheet 
148/3,1976 
Topographical 
1:50.000 
map, 
Nairobi, 
Sheet 
148/4,1976 
Survey of Kenya 
The optical pantograph and distortion free 
overhead projector enabled comparison of the 
LANDSAT images and background information at 
the same 1:100.000 scale. The multi spectral 
additive viewer was used to study various 
combinations of bands 4, 5, 6 and 7. A print 
taken from the view screen showing a balance 
between bands 5, 6 and 7 at scale 1:250.000 
was used for crop distinction. The photo 
showed a discrimination between coffee, tea 
and pineapple plantation in different shades 
of orange and yellow. This appears to result 
from differences in band 5 and band 6 reflec 
tance of these crops. 
2.2 Method 
A tranparency of the Nairobi and Environs map 
1:100.00 was superimposed on the enlarged 
LANDSAT print. In consequence, the resulting 
geographical reference of the pixels simpli 
fied the visual interpretation of the images 
in relation to the ground truth. A land-cover 
map at 1:1M scale was already available as a 
result of previous work in the Nairobi area 
with LANDSAT images (Grootenhuis, Weeda and 
Kalambo 1986 a)„ This map was projected over 
the LANDSAT print, using the distortion free 
overhead projector, and the respective zones 
were transferred to the 1:100.000 scale map. 
By means of the geographical reference, the 
boundaries could be checked in the field sup 
ported by the topographical maps 1:50.000. 
The boundaries were redrawn where necessary. 
The identified zones showed homogeneous 
patterns, which enabled a representative 
choice of sampling points. About six field 
inspections were made per zone, depending on 
the size and complexity. In this way, sixty 
sampling points were located within the 60 km 
x 60 km study area. The field inspection 
points were located at random along accessible 
roads, using a specific land-use data sheet. 
The variables recorded, included details on 
altitude, landform, soil, drainage, vegetation 
land-use and visual characteristics. Cross- 
sections were sketched in the field to record 
the distribution of landscape elements. The 
land-cover/land-use classification system of 
the United States Geological Survey (Anderson 
1976) was applied. Integration of the visual 
interpretation of the enlarged LANDSAT image 
with the fieldwork and the consulted back 
ground data led to the production of a land- 
use map at 1:100.000 scale showing land-use 
units (Figure 2). 
3 RESULTS 
The land-cover map enlarged to scale 1:100.000 
and superimposed on the photographic print of 
the image at the same scale, provided the 
delineation of the land-use zones. The boun 
daries were checked using the 1:50.000 topo 
graphical maps, and only little adjustment of 
their location was necessary. Sharp boundaries 
were found around large scale plantations due 
to property boundaries. A similar situation 
is found at the forest edge where the boundary 
is secured by the Forest Department. In cases 
of transition between two zones, determination 
of the boundary is related to specific physi-r 
cal factors (as geomorphology and soil). 
Differentiation in texture and tone within 
one broad phototonal zone suggested sub-divi 
sions were necessary. Gathering additional 
knowledge about these areas in the field made 
it possible to distinguish the land-use units 
within the land-use zone. 
During the examination of the photo-tones, 
it was found that the same colour can repre 
sent different land-uses. For example, the 
reflection of coffee, pineapple, foodcrops 
and river bottoms in the near infra-red 
(band 7) is the same and all appear red in 
the FCC. Within the land-use zones, different 
tones and textures were identified. The ground 
truth provided confirmation of the different 
land-use units. Sets of land-use units are 
responsible for the mottled patterns within 
each zone. The cross-sections and the block- 
diagrams of the land-use zones (Figure 3-6) 
illustrate the sub-division into character 
istic land-use units and their relative dis 
tribution in percentages. 
The land-use units depict clearly the inter 
action between man and his environment. 
Agricultural practices relating to altitude, 
climate, type of soil, slope and landownership 
can now be identified.(Table 3). 
Table 1. Key to the land-use zones. 
Mapping unit Land—use zone 
(M) Mountain M1.M2 
(E) Escarpment E 
(S) Slopes si - S4 
(P) Plains P1 - P4