247 
DEFORESTATION IN EQUATORIAL AFRICA: 
AN ASSESSMENT BASED ON ANALYSIS OF MULTIDATE LANDSAT MSS DATA 
Susan M. Berta, Paul W. Mausel, John A. Harrington, Jr. 
Indiana State University Remote Sensing Laboratory, Indiana State University 
Terre Haute, IN, USA 47809 
Robin Graham, Environmental Sciences Division, Oak Ridge National Laboratory 
P.O. Box 2008, Oak Ridge, TN, USA 37831 
Abstract 
Information on land cover, as interpreted from spectral signatures, is used in assessing rates of 
environmental change for three areas in equatorial Africa using LANDSAT MSS data. The three study areas 
are: the Berberati - Carnot, Central African Republic region and two sites in Zaire. Four different 
information extraction procedures were utilized: maximun likelihood and nonparametric classifiers and 
Normalized Difference Vegetation Index (NDVI) analysis and Principal Components Analysis (PCA). 
Trends in landscape change are identified with SITE ONE as the primary example. This study revealed that 
deforestation is not occurring in the large, contiguous patches of rain forest. The greatest 
deforestation pressures exist in close proximity to population centers, along roadways, and at the edges 
of the contiguous rain forest. Regional changes are toward less vegetative vigor and degraded nonforest 
environments. 
KEY WORDS: Central African Republic, parametric, NDVI, land cover change 
INTRODUCTION 
Global atmospheric CO2 concentrations are rising 
and portend a general global warming (Tangley, 
1988). Forests, which are an integral part of the 
global carbon cycle, represent 90 percent of the 
global carbon stored in vegetation. Estimates of 
the global carbon emissions from deforestation 
suggest that approximately a quarter are from 
Africa (Houghton et al., 1987). These estimates 
of carbon loss in Africa are uncertain since they 
are based on the estimated deforestation rates 
published by the 1980 FAO/UNEP Tropical Forest 
Resources Assessment Project (FAO/UNEP, 1981). 
Reliable, current, and quantitative data on forest 
area and deforestation rates are scarce for most 
tropical countries (Booth, 1989). Although 
satellite imagery has been used previously in 
South America, Southeast Asia, Madagascar, and the 
Ivory Coast of Africa (Nelson and Holben, 1986; 
Woodwell et al., 1987; Malingreau and Tucker, 
1988; Malingreau et al., 1985; Singh, 1986; Green 
and Sussman, 1990; Repetto, 1990) few analyses 
have been done for the rain forests of equatorial 
Africa. 
DESCRIPTION OF THE STUDY AREAS 
This investigation involves three locations in the 
Zaire/Central African Republic region (Figure 1). 
SITE ONE is located in the southwestern portion of 
the Central African Republic, with a small area of 
eastern Cameroon included (centered on 4°30’N 
latitude, 15°30'E longitude), while SITE TWO is 
located along the western edge of Zaire (2°47'S 
latitude, 18°17'E longitude) and SITE THREE is 
located in northeastern Zaire (2°49'N latitude, 
28°31’E longitude). 
SITE ONE lies at the northern margin of the 
Guineo-Congolian rain forest (Figure 2). 
Vegetation characteristics, described by White 
(1983), include drier Guineo-Congolian rain forest 
(rainfall averages between 1200 and 1600 nm per 
year) in the area south of Berberati and a mosaic 
of Guineo-Congolian rain forest and secondary 
grassland (where secondary grasslands refer to 
formerly forested areas destroyed by cultivation 
and fire) in the northern two-thirds of the study 
area. Within this latter cover type the 
grasslands occur in mosaic with small, usually 
severely degraded, patches of the original forest 
(found in lowland areas), and small patches of 
secondary thicket and secondary forest (White, 
1983, p. 84). Major waterways include the Mambere 
River near the town of Carnot (SITE 1-F) in the 
northeastern section of SITE ONE and the Baumbe 
River near Amada-Gaza (Site 1-D) which flows south 
and empties into the Kadei River. 
5*E 10*E 15*E 20*E 25*E 30*E 
Figure 1. Location of the study area.