36°
]
20 KM.
15
10
45
-— — Loose Surface Road
"C. M. $
1982
position of these changes were determined on a qualitative basis through visual
inspection and correlation with the aid of the available limited ground truth.
(Short, 1982) The percentages were calculated by dividing acreage per category
by acreage for the whole subscene for both data sets.
RESULTS AND CONCLUSION
Changes in the land cover categories between the two dates and particularly
forestation, are based on the acreage estimations as developed in the above
procedure. The results of the acreages and percentages for each of the categories
for the two dates are noted in Tables | and Il. As indicated in Table |, for
the Mau East site, there is a decrease of over five percent in the forest cover
category between the two dates. However, for the Mau West site, there is no
significant difference in the overall forested category between 1973 and 1978.
Visual inspection of the data and ground truth imply that there is notably more
agricultural encroachment in the Mau East study area as compared to the other.
The results clearly indicate that there is a considerable change of ground cover
for the Mau Forest. LANDSAT data provide a valid basis for monitoring forestation
changes and trends in areas where traditional data collection methods are at best
infrequently utilized to provide reliable information for the management for such
resources. A more detailed analysis of the study sites would require more reliable
ground truth. Additional data processing would make data sets such as the Mau
West and Mau East reveal more insight, but it would be essential that the infor-
mation be verified to validate the findings.
This study supports the premise that relatively simple data processing pro-
cedures and limited ground truth application can indeed serve as an effective
monitoring device of the forest changes in Kenya. More sophisticated techniques
or additional data bases would undoubtedly result in more refined findings but
not change the basic premise.
