511 
Table l. Land use and land cover statistics for 10 districts in Kenya, as a percent of the total land 
surveyed. 
Land use and land cover 
District 
Class type 
Meru 
(6575) 
Kisii 
*(2375) 
South 
Kisumu Nyanza 
(2200) (5500) 
Bung- 
oma 
(2475) 
Mur- 
anga 
(1875) 
Nyand- 
arua 
(2400) 
Busia 
(1450) 
Ker- 
icho 
(4800) 
Trans- 
nzoia 
(2075) 
Cultivated (crops, ploughed) 
25.1 
43.9 
27.9 
15.1 
41.3 
52.6 
17.1 
30.3 
20.0 
39.9 
Pasture (improved, unimproved) 
20.1 
32.5 
36.2 
71.2 
36.0 
20.1 
58.0 
51.2 
45.5 
45.2 
Bush and Shrublands 
38.3 
0.0 
20.6 
5.6 
9.8 
6.5 
11.8 
6.3 
12.4 
3.1 
Forest/Woodlands 
8.1 
8.2 
0.0 
0.4 
9.0 
0.0 
0.0 
0.0 
15.4 
4.5 
Transportation (roads, paths, etc.) 
1.3 
3.1 
3.1 
1.5 
2.9 
4.7 
2.1 
1.9 
2.2 
1.4 
Structures (buildings) 
0.6 
2.4 
1.2 
0.5 
0.0 
1.8 
0.0 
0.7 
0.1 
0.0 
Woodlots 
1.0 
0.0 
0.0 
0.2 
0.0 
9.6 
0.9 
0.0 
0.0 
4.4 
Hedges 
0.6 
8.4 
6.6 
2.2 
0.7 
4.7 
1.8 
4.2 
3.5 
0.6 
Swamps/Rivers 
0.8 
0.6 
4.2 
0.7 
0.2 
0.0 
5.1 
2.8 
0.7 
0.1 
Barren (rocks, gulleys) 
2.3 
0.0 
0.4 
0.3 
0.0 
0.0 
0.9 
0.0 
0.0 
0.8 
Others 
1.9 
0.9 
0.0 
2.3 
0.0 
0.0 
2.3 
2.6 
0.1 
0.0 
Total 
100.1 
100.0 
100.2 
100.0 
99.9 
100.0 
100.0 
100.0 
99.9 
100.0 
Sources: Agatsiva 1984, 1985; Epp et al 1983; Muchoki 1985a,b; Mwendwa 1984, 1985; Ottichilo 1985; Peden et 
al 1984. 
* area surveyed in km^ - See text. 
Significant changes in future of any of these 
statistics in any district in area or amounts will 
be a reflection of a process of changes going on as 
a reaction to certain pressures or certain manage 
ment policies. A more detailed comparison between 
districts and detection of changes will be more 
useful when the data is disaggregated into differ 
ent ecological strata. 
The use of aerial sample photography for the 
mapping of resources has been advocated and dis 
cussed before by many authors for example Berry and 
Baker (1968) and Robertson and Stoner (1970). 
However, its application to a large scale mapping 
project has been lacking. The Kenyan project was 
adopted after a trial run in Kisii District (Epp, 
Killmayer and Peden 1983). In the Kisii study and 
those that followed the photo samples were strati 
fied and analyzed on the basis of sub-district 
administration units. Thus when the average per 
cent land covered by a class type was mapped, the 
distribution was depicted as evenly spread through 
out the administration unit through differential 
shading or pie charts. However for Meru District, 
class types' intensity and extents were first 
mapped independent of any factor. This method of 
mapping class type and intensities gives a picture 
close to real distribution within any type of 
stratification consequently overlayed. Stratifica 
tion can be on basis of administration units, 
soils, vegetation, climate or human population for 
various types of studies (e.g. factors affecting a 
class type distribution) in an effort to make 
rational development plans. Present generations of 
GIS systems with analytical capabilities can now 
perform these functions much faster. 
A major use of this type of mapping is the detec 
tion and measure of the movement and cultivation 
activities into the marginal areas. Epp and 
Killmayer (1982) suggested that areas with land 10% 
or more under cultivation be designated as 'agri 
cultural'. Jaetzold and Schmidt (1982) have how 
ever defined in details agroecological zones in 
Kenya in which agricultural areas are ecologically 
delineated. Using these two criteria, newly set 
tled areas as well as cultivation extending beyond 
ecologically acceptable boundaries can be detected 
using the photosample methodology described here 
and by map overlays. 
A recent experimentation using similar principles 
of methodology has been reported from Nigeria with 
success in providing fairly accurate statistics for 
planning (Bauchi 1984; Nigeria 1984). Mapping was 
done using pie charts and proportionate circles on 
each grid cell, to depict coverage or intensity of 
a class type in that cell. 
5 CONCLUSION 
The method described above has proved useful in 
crop hectarage estimation and distribution studies 
in the intensive agricultural areas of Kenya where 
a great variety of crops are grown in small land 
segments. Accuracy and consistency were found high 
in the second and third classification levels and 
in those finer class types which have extensive 
areal coverage like maize. However, some 
vegetation cover type classifications need more 
refined description and definition to reduce their 
class confusions, for example between unimproved 
pasture and bushlands, which was responsible for 
their relatively lower interpretation accuracy. 
It is anticipated that the infornmtion from all 
the districts will eventually be combined to give a 
national overview for every class or combination of 
class types. This is expected to be possible 
because the data is being collected using same 
method, measurement units (percent cover) as well 
as the UTM system base maps for every district, and 
thus produce a detailed land use and land cover map 
of Kenya. It is also anticipated that it will be 
possible to detect areas of land use conflicts and 
potential desertification frontiers. 
6 ACKNOWLEDGEMENT 
This study was funded by the Kenya Government 
through KREMU, while our studies, including the 
project time, were made possible jointly by the 
Kenya Government and the Canadian International 
Development Agency (CIDA). We are grateful to 
both. Professor Tom Henley of the Natural 
Resources Institute, University of Manitoba, and