UNIVERSITY OF CAPE COAST DEPARTMENT OF GEOGRAPHY AND TOURISM LAND USE DYNAMICS IN BIEHA, SISSILI PROVINCE, SOUTHERN BURKINA FASO BY ISSA OUEDRAOGO A DISSERTATION SUBMITTED TO THE DEPARTMENT OF GEOGRAPHY AND TOURISM OF THE FACULTY OF SOCIAL SCIENCES, UNIVERSITY OF CAPE COAST, IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE AWARD OF MASTER OF ARTS DEGREE IN GEOGRAPHY. AUGUST 2007
147
Embed
UNIVERSITY OF CAPE COAST DEPARTMENT OF GEOGRAPHY …
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
UNIVERSITY OF CAPE COAST
DEPARTMENT OF GEOGRAPHY AND TOURISM
LAND USE DYNAMICS IN BIEHA,
SISSILI PROVINCE, SOUTHERN
BURKINA FASO
BY
ISSA OUEDRAOGO
A DISSERTATION SUBMITTED TO THE DEPARTMENT OF GEOGRAPHY
AND TOURISM OF THE FACULTY OF SOCIAL SCIENCES, UNIVERSITY
OF CAPE COAST, IN PARTIAL FULFILMENT OF THE REQUIREMENTS
FOR THE AWARD OF MASTER OF ARTS DEGREE IN GEOGRAPHY.
AUGUST 2007
DECLARATIONS
Candidate’s Declaration
I hereby declare that this dissertation is the result of my own original work and
that no part of it has been presented for another degree in this university or
of thermal characteristics as well as other environmental changes (Singh, 1989;
Zhongchao et al. 2002; You et al. 2004).
Macleod (1998, cited in Botomley, 1998) listed four aspects of change detection
which are important when monitoring natural resources: namely detecting the
occurrence of change, identifying the nature of the change, measuring the area
extent of change and assessing the spatial pattern of the change.
Scientific literature has revealed that digital change detection is a difficult
task to perform accurately and unfortunately many of the studies concerned with
comparative evaluation of these applications have not supported their conclusions
by quantitative analysis (Singh, 1989). All digital change detection is affected by
spatial, spectral, temporal, and thematic constraints. The type of method
implemented can profoundly affect the qualitative and quantitative estimates of the
change. Even in the same environment, different approaches may yield different
change maps. The selection of the appropriate method therefore takes on
considerable significance. Not all detectable changes, however, are equally
important to the resource manager. On the other hand, it is also probable that some
changes of interest will not be captured very well, or at all, by any given system.
Competing models for land use dynamics
Several models to be used for land use change exist depending on the
interest of each study. A summary of these models pointing out the variables used
30
31
for each model, including the strengths and weaknesses of each model are shown
in Table 5.
The weaknesses of most of the models concern their inability to take into
account one or two of the critical dimensions of time, space and human decision-
making, hence the choice of the model used by Agarwal et al (2002) to assess
human-environmental dynamics in this study. This model highly involves time
scale and complexity, spatial scale and complexity and human decision-making.
Table 5: Summary of land use dynamics models Model name
Model type Components/modules
Variables Strengths Weaknesses
1. General Ecosystem Model (GEM) (Fitz . et al1996)
Dynamic systems model
14 Sectors (modules), e.g. Hydrology Macrophytes Algae Nutrients Fire Dead organic matter Separate database for each secto
Captures feedback among abiotic and biotic ecosystem components
103 input parameters, in a set of linked databases, representing the modules, e.g., Hydrology Macrophytes Algae Nutrients Fire Dead organic matter
Spatially dependent model, with feedback between units and across time Includes many sectors Modular, can add or drop sectors Can adapt resolution, extent, and time step to match the process being modeled
Limited human decision making
2. Patuxent Landscape Model (PLM) (Voinov et al. 1999)
Dynamic systems model
Based on the GEM model (#1, above), includes the following modules, with some modification: 1) Hydrology 2) Nutrients 3) Macrophytes 4) Economic model
Predicts fundamental ecological processes and land-use patterns at the watershed level
In addition to the GEM variables, it -adds dynamics in carbon-to-nutrient ratios -introduces differences between evergreen and deciduous plant communities -introduces impact of land management through fertilizing, planting, and harvesting of crops and trees
In addition to the strengths of the GEM, the PLM incorporates several other variables that add to its applicability to assess the impacts of land management and best management practices
Limited consideration of institutional factors
3. CLUE Model (Conversion of Land Use and Its Effects) (Veldkamp
Biophysical drivers Land suitability for crops Temperature/Precipitation Effects of past land use (may explain both biophysical degradation and
Covers a wide range of biophysical and human drivers at differing temporal and spatial scales
Limited consideration of institutional and economic variables
34
Model name
Model type Components/modules
Variables Strengths Weaknesses Table 5 continued
and Fresco 1996a)
improvement of land, mainly for crops) Impact of pests, weeds, diseases Human Drivers Population size and density Technology level Level of affluence Political Structures (through command and control, or fiscal mechanisms) Economic conditions Attitudes and value
4. CLUE-CR (Conversion of Land Use and Its Effects – Costa Rica) (Veldkamp and Fresco 1996b)
Discrete finite state model
CLUE-CR an application of CLUE (#3, above) Same modules
Simulates top-down and bottom-up effects of land-use change in Costa Rica
Same as CLUE (#3, above)
Multiple scales - local, regional, and national Uses the outcome of a nested analysis, a set of 6x5 scaledependent land-use/landcover linear regressions as model input, which is reproducible, unlike a specific calibration exercise
Authors acknowledge limited consideration of institutional and economic factors
5. Chomitz et al. (1996)
Econometric (multinomial logit) model
Single module, with multiple equations
Predicts land use, aggregated in three classes: Natural vegetation Semi-subsistence agriculture Commercial farming
Soil nitrogen Available phosphorus Slope Ph Wetness Flood hazard Rainfall National land
Used spatially disaggregated information to calculate an integrated distance measure based on terrain and presence of roads Also, strong theoretical underpinning of Von Thünen’s model
Strong assumptions that can be relaxed by alternate specifications Does not explicitly incorporate prices
35
Model name
Model type Components/modules
Variables Strengths Weaknesses Table 5 continued
Forest reserve Distance to markets, based on impedance levels (relative costs of transport) Soil fertility
6. Wood et al. 1997
Spatial Markov model
Temporal and spatial land-use change Markov models
Land-use change
Models under development
Investigating Markov variations, which relax strict assumptions associated with the Markov approach Explicitly considers both spatial and temporal change
Not strictly a weakness, this is a work in progress and, hence, has not yet included HDM factors
Population growth submodel Spatial database, various layers merged to project Developable Land Units (DLUs) Spatial Allocation submodel Annexation-incorporation submodel
Explains land use in a metropolitan setting, in terms of demand (population growth) and supply of land (underdeveloped land available for redevelopment)
Population growth, DLUs, and intermediate map layers with: Housing prices Zoning Slope Wetlands Distance to city center Distance to freeway or BART station Distance to sphere-of-influence boundaries
Underlying theory of parcel allocation by population growth projections and price, and incorporation of incentives for intermediaries - developers, a great strength Large-scale GIS map layers with detailed information for each individual parcel in 14 counties provide high realism and precision
Compresses long period (20 years) in a single model run Has no feedback of mismatch between demand and supply on price of developable land/housing stock Does not incorporate impact of interest rates, economic growth rates, etc.
8. Swallow et al. 1997
Dynamic model
Three components: 1) Timber model 2) Forage production function
Simulates an optimal harvest sequence
Present values of alternative possible states of the forest, using the three model components
The long time horizon, and the annual checking of present values under alternate possible states of the forest makes it a useful forest management tool
Authors note that the optimal management pattern on any individual stand or set of stands
36
37
Model name
Model type Components/modules
Variables Strengths Weaknesses
3) Non-timber benefit function
for maximizing multiple-use values
requires specific analysis rather than dependence on rules of thum
b
9. Clarke et al. 1998, Kirtland et al. 2000
Cellular automata model
Simulation module consists of complex rules Digital dataset of biophysical and human factors
Change in urban areas over time
Extent of urban areas Elevation Slope Roads
Allows each cell to act independently according to rules, analogous to city expansion as a result of hundreds of small decisions Fine-scale data, registered to a 30 m UTM grid
Does not unpackhuman
e
map
not
decisions that lead to spread of built areas Does not yet include biological factors
10. CURBA (California Urban and Biodiversity Analysis Model) (Landis et al. 1998)
Overlay of GIS layers with statistical urban growth projections
1) Statistical model of urban growth 2) Policy simulation and evaluation model 3) Map and data layers of habitat types, biodiversity, and other natural factors
The interaction among the probabilities of urbanization, its interaction with habitat type and extent, and, impacts of policy changes on the two
Slope and elevation Location and types of roads Hydrographic features Jurisdictional boundaries Wetlands and flood zones Jurisdictional spheres of influence Various socioeconomic data Local growth policies Job growth Habitat type and extent maps
Increases understanding of factors behind recent urbanization patterns Allows projection of future urban growth patterns, and of the impact of projected urban growth on habitat integrity and quality
Human decision making not explicitly considered Further, errors arlikely from misclassification of data at grid level or misalignment offeature boundaries Errors also possible from limitations in explaining historical urban growth patterns
11. Gilruth et al. 1995
Spatial dynamic model
Several subroutines for different tasks
Predicts sites used for shifting cultivation in terms of topography and proximity to population centers
Site productivity (# of fallow years) Ease of clearing Erosion hazard Site proximity
Replicable Tries to mimic expansion of cultivation over time
Long gap between data collection; does include impact of landquality
Table 5 continued
38
Model name
Model type Components/modules
Variables Strengths Weaknesses
and socioeconomic variables
Population, as function of village size
Source: Literature review, 2006
Table 5 continued
Summary
The chapter has defined concepts and techniques related to land use
detection. These concepts are land use/land cover, land use change and its
consequences, land use evolution detection and competing models. Remote
sensing, GIS, GPS and Landsat TM imageries are tools to be used for land use
dynamics assessment. The local environmental issues concern the man who
contributes at a local level to the land use change.
The literature suggests that remote sensing and GIS are accurate tools for
land use change detection. Data from remote sensing platforms such as satellite
images provide information for GIS database. These data can be used for resource
monitoring, environmental analysis, forecasting and assessment. It was revealed
that several models for detecting land use change exist, but the one that takes into
account the dimensions Time, Space and Human decision-making ensures a better
understanding of the dynamics. Obviously, solutions must be found to resolve the
debate between the IBS and Howorth and O’Keefe’s findings on the state of the
environment in the province, and to generate results that best represent the reality
on the ground; hence the focus on land-use dynamics at a district scale (Bieha)
from 1986 to 2002 using Landsat TM images.
40
CHAPTER THREE
REVIEW OF ENVIRONMENTAL ISSUES
IN BURKINA FASO
Introduction
The Burkina Faso environment is characterised by great vulnerability; any
stability is only maintained by human management. However, several
unsustainable management practices are contributing to environmental
degradation, thus increasing people’s vulnerability through reduced productivity
and resilience to stress (Simonsson, 2005). According to MEE (1996) and MECV
(2004), from 1980 to 1992 the surface area of the forest of Burkina Faso reduced
from 15.42 million hectares to 14.16 million hectares. The annual loss of forest
was estimated at 105,000 ha (MEE, 1996). According to Kramer (2002), the
World Bank estimated the annual loss of wooded land surface in Burkina Faso at
80,000 to 100,000 ha while the FAO (2000) assumes an annual loss of 15.266 ha
equivalent to 0.2 % of wooded land surface, exclusively for those surfaces cleared
to make way for agriculture. According to Mongabay (2005), between 1990 and
2000, Burkina Faso lost an average of 24,000 ha of forest per year which
amounted to an average annual deforestation rate of 0.34 %. Between 2000 and
2005, the rate of forest change increased to 0.35 % per annum. In total, between
1990 and 2005, Burkina Faso lost 5.0 % of its forest cover, or around 360,000
hectares. Measuring the total rate of habitat conversion (defined as change in
41
forest area plus change in woodland area minus net plantation expansion) for the
1990-2005 intervals, Burkina Faso lost 2.8 % of its forest and woodland habitat.
Two studies related to land-cover using remote sensing and GIS tools have
been reported. The first was undertaken by IBS (1994) and aimed at examining
deforestation rates in Sissili since 1988 using Landsat and Spot images from 1988
and 1993. On the basis of maps produced from the study, extrapolations were
made on future deforestation which stated that the rate will increase from 21.6 %
in 1988 to 43.1 % by the year 2010. The second study was conducted by Howorth
and O’Keefe (1998) to investigate the new resource-use pattern that had developed
as a result of demographic changes. Based on maps of 1955 and 1983 and
interviews conducted in three villages of the province it was concluded that there
was a peaceful coexistence between the three ethnic groups living together in the
villages and there was no destruction of the environment at all in Sissili as IBS
forecasted. The study of Howorth and O’Keefe (1998) concluded that the
environment was improving in vegetation cover in Sissili.
The conclusions of the two studies were however contradictory. The time
interval used by IBS (1988 to 1993) may have been too limited to detect land use
changes, and worse of all, to forecast long-term changes in the environment. The
second study which used aerial photographs from 1955 and 1983 to map land use
changes in three villages of less than 40 km2 each, without detecting
environmental degradation, may be accurate at the village scale but, in terms of the
whole of Sissili province, it may be an exaggeration or overgeneralization.
As already reported in the literature the five broad and inter-related human
factors that lead to environmental degradation in Burkina Faso and specifically in
42
the study area have been identified as agricultural practices, migration,
overgrazing, fuel wood harvesting, timber logging and bushfires. Each of these is
briefly explained.
Agricultural practices and deforestation
The agricultural activities in Burkina Faso are undertaken in a rudimentary
and extensive way, with a low level of intensification (Bandre and Batta, 1998). In
this context, the only possible way to secure food security is to cultivate more land
(Reenberg and Lund, 2001). This is done by decreasing the ratio between fallow
and cultivated land within a village or by including new territory for cultivation,
thus contributing to the degradation of the environment (Howorth and O'Keefe,
1998; MEE, 1999). Under increasing population pressure, marginal lands are used
as farmers cultivate large area to maintain production but they do little to sustain
soil nutrient levels and the productive capacity of the soil. Short fallow periods and
inadequate use of fertilizers; coupled with overgrazing and deforestation through
fuelwood harvesting tend to cause loss of soil and vegetation cover, and water
degradation (MEF, 2000; Simonsson, 2005).
Elshout et al (2001) have argued that in the south and west zone of Burkina
Faso, the land clearance for extensive farming is the key contributor to vegetation
loss. In the study area, farming practices vary from one ethnic group to another.
Howorth and O'Keefe (1998) reported that the indigenous Nuni practice a gentle
form of agriculture which is exclusively manual with little inputs, relatively low
soil usage and use approximately 4.5 ha per family. This allows the retention of a
large number of trees and root systems, without causing great disturbance to the
43
agro-ecological system. The Fulani have settled extensively in Sissili, and tend to
concentrate their animal herding in the zones of low-intensive agriculture usually
in the periphery/wooded areas of villages. They cultivate about 1.5 ha per family
in old pasture zones containing high levels of cattle manure and, consequently,
have comparatively high yields. The Mossi, on the other hand, practise an
extensive form of agriculture with almost total field clearing, mainly for cereal
production. They tend so to exploit the lands in such a way as to degrade the
environment hence, there is an emergence of the relationship between migration
and deforestation.
Migration and environmental degradation
Migration is usually ignored in models of land use change (Veldkamp and
Fresco, 1996a; Shen, 2000; Stéphenne and Lambin, 2001), even though it is often
recognized to be the dominant demographic factor influencing land use (Lambin et
al, 2001). Many authors cite population growth as the single most important cause
of deforestation (Allen and Douglas, 1985; World Rainforest Movement, 1990;
World Bank, 1992). Population growth often leads to migration to the forest by
peasants seeking land to clear for subsistence farming.
One of the features of the population of Burkina Faso is its mobility (Kress,
2006). In the period between 1985 and 1991, 10 % of the Burkina Faso’s
population of 7.5 million inhabitants migrated from one province to another or
abroad (Jeune Afrique Atlases, 1998). Within the country, people tend to migrate
to the agricultural zone and to Ouagadougou and Bobo Dioulasso (Simonsson,
2005; Kress, 2006). SIDA’s Poverty Profile of Burkina Faso concluded that
44
migration seems to be a strategy for households to reduce poverty, more important
in rural than in urban areas; among men more than women; and among the poor
more than the rich (Haberg, 2000). Due to the migration, the population growth
rate in the savannah region was reduced by 1.1 % while, in the forest regions, it
increased by 0.6 % (Zachariah and Conde, 1981). The regions of departure lose
labour force while the social and economic infrastructure in regions of arrival may
have problems coping with the rapidly growing population (Henry et al, 2002).
A study on the inter-provincial migration in Burkina in 2003 (Henry et al,
2002) showed the highest immigration rate of 4.88 % in the Sissili province. With
large number of immigrants, the province faces difficulties related to land conflicts
and access to social infrastructure. As a consequence, the population of some
villages in the Province has more than doubled in 20 years. An example is the
Bieha district whose population increased from 15,043 inhabitants in 1985 to
25,634 inhabitants in 2006. The effect of this migration driven population growth
on land use was quickly visible. The newcomers tended to reproduce the same
extensive farming practices followed in the centre-north and, encouraged by large
private companies, are keen to produce cotton and maize (Gray, 1999). The land
requirement for migrants was thus larger than that for the sedentary populations
and land supply became limited (Mathieu, 1998).
Authors such as Hardin (1968), Ehrlich (1968), Ehrlich and Ehrlich (1990)
and Meadows et al (1972) have a pessimist view of the relationships between
population pressure and environment. According to them, population control must
be a part of any development strategy, otherwise environment will collapse. On the
other hand, some optimists argue that population pressure does not necessarily
45
lead to environment degradation. It stimulates development rather than slowing it
down, and moreover, it leads to innovation in agricultural technology and
techniques which support the increased number of population (Boserup, 1972;
Simon, 1980; Tiffen and Mortimore, 1994; Fairhead and Leach, 1996; Bassett and
Bi Zueli, 2000).
Overgrazing
Chikamai and Kigomo (2003) reported that overgrazing is the most notable
factor in causing de-vegetation and hence degradation. The heaviest impact of
overgrazing takes place in the Sahel countries especially areas falling within arid
and semi-arid zones. Overgrazing is concentrated around settlements and is often
related to recent sedentarisations of nomadic herders.
In Sissili province, the number of animals has been increasing substantially
due to the continued in-migration of the pastoralist Fulani (Howorth and O'Keefe,
1998). In 1986 the total number of bovines, ovine and goats was estimated at
215,000, but these increased to 900,000 in 2003, and thus jeopardizing the
carrying capacity of the province (DPAHRH, 2006). During the 1980s, the
government created a pastoral zone in Yalle (Bieha District) with the purpose of
settling about 75 families of breeders (Fulani), to promote the best quality of the
livestock and to limit the conflicts linked to competition for space between farmers
and breeders. Unfortunately, the breeders of the area refused this offer and
preferred to walk along the forest to graze their animals by cutting palatable
species.
46
Firewood and timber request
Wood fuel is the principal source of domestic energy in developing
countries (Openshaw, 1974; Eckholm, 1975; Arnold and Jongma, 1978). Wood
fuel includes charcoal as well as firewood, brushwood, twigs and cut branches
(Openshaw, 1986). Bandré and Batta (1998) argued that wood is used in Burkina
Faso for two main purposes: source of energy and as building material. Wood
represented 96 % of the domestic energy consumed in 1993 and accounted for 64
% of the national requirement in primary energy. The average per capita
consumption is 300 kg/year in the north while in the southern and western areas it
is over 800 kg/year (Bandré and Batta, 1998). The annual commercial value is
estimated at 35 billion CFA (58,000,000 USD) for the firewood and 6 billion CFA
(10,000,000 USD) for the timber (Kessler and Greeling, 1994). According to
Kramer (2002), the consumption of wood in Burkina Faso is higher than the
production. This means that there is no ecological sustainability. At the same time,
because of the low economic productivity, import of necessary amount of energy
cannot be envisaged neither on the macro-economic level nor on the family budget
level.
The rate of deforestation in the eastern, southern and western zones of the
country to meet the needs for energy and timber consumption became so high that
the Ministry in charge of environment and water took a decision to suspend
charcoal production from July, 15th 2005 (Le Pays, 2005). The Ministry estimated
at 593,092 tons the quantity of charcoal consumed in 2004 in Burkina Faso while
the annual loss of forest to charcoal production was estimated at 370,000 hectares.
47
Bushfire
Fire has always played a major role in most of Sub-Sahara Africa in
clearing the field, hunting, improving visibility, accelerating the re-growth of
perennial grasses and in customary rituals as reported by Bandré and Batta (1998);
who also argued that frequent fire, and especially late fires, not only killed most of
the perennial plants, but also impoverished the soil and reduced its productivity.
Fire also causes loss of certain nutrients (nitrogen and sulphur), which are usually
dispersed in the atmosphere, and loss of organic matter.
Bushfire practices are ancient in Burkina Faso going back to the pre-
colonial period (Aubreville, 1949; Belloward, 1959; Ministère de l’Environment et
du Tourisme, 1991; Kambou and Poussi, 1997; Yameogo, 2005). Three types of
fires exist in Burkina: early fire, intermediate fire and late fire (Yameogo, 2005).
According to Ministère de l’Environment et du Tourisme (1991) and Zida (1993),
the surface annually touched by fires is estimated at 98,568 km², which is about 55
% of the forest surface of the country. In the provinces of the south of the country
(Sissili, Ziro and Nahouri), which possess 12,305 km² of forest, 9,844 km² (80 %)
are fired each year (Ministère de l’Environment et du Tourisme, 1991). Bushfires
cause a loss of 200 million Euros in animal resources, 10.7 million Euros in wood
production and more than one million in wildlife and cotton production (Zida,
1993).
48
CHAPTER FOUR
METHODS OF DATA COLLECTION AND
ISSUES FROM THE FIELD
Introduction
This chapter describes the data and sources as well as the methods and
tools employed in the data collection. It also covers the sampling techniques and
the problems encountered in the field and how they were solved.
Data and sources
The data collected were basically quantitative arising from primary and
secondary sources. The primary data were divided into two broad categories:
a) Data resulting from the satellite image processing that dealt with
quantitative variables and concerned the surface areas of the land use units
in time series;
b) Data from interviews of sample population in the study area which dealt
with quantitative variables such as the perception of density of trees and
the wild animals, the productivity and availability of food, etc.
The secondary data were collected from textbooks in documentation
centres (offices of agriculture and animal resources, etc.) and centres of primary
data storage such as the national meteorology, the national statistics and
demography office. These data concern the agricultural and pastoral practices and
yields, the rainfall and the quality of soils in the study area.
49
Images processing
Satellite images
Remote sensing and Geographical Information System (RS/GIS) tools
were used to carry out the different land-use units and their respective surfaces
based on satellite data. The main data used in the research included Landsat
Thematic Mapper satellite images of 1986 and 2002 (hereafter referred to as TM
images). Sissili province is covered entirely by the images number 195/52 of
Landsat TM (Figure 7). A brief description of the satellite images used is shown in
Table 6. Digital topographic data with contour interval of 10 m produced by the
Geographic Institute of Burkina Faso (IGB) were also used.
Figure 7: Landsat TM image mosaic of Burkina Faso
Source: Database of the Geographical Institute of Burkina (IGB), 2006
50
Table 6: Satellite images used for land use detection of Sissili
Satellite
Type
Sensor Image
number
Number of
bands
Pixel
spacing
Observation
date
Landsat TM 195/52 7 30 x 30 18 Nov. 1986
Landsat TM 195/52 7 30 x 30 21 Oct.2002
Source: Landsat database (2006)
The TM images were provided by the Institute for Environmental and
Agronomic Research (INERA) of Ouagadougou (Burkina Faso). They were
acquired within the same season (end of the rainy season) and are at the same
resolution: 30 meters resolution. The two dates have the same vegetation
conditions according to the climate of the study area. According to the farming
practices, October and November are the harvesting periods during which
precocious bush fires occur. As a consequence, there is a lot of haze in the images.
Those factors give effects on vegetation status causing reflectance values of land
use quite difficult to compare. However, possible similar nomenclatures were set
up based on physical characteristics of land use.
The ground-truth information required for the classification and accuracy
assessment of the images was collected from the field during January, 2006 using
a training sample protocol. In addition, a self-designed format was used to collect
vegetation level information on vegetation types, condition and history of land use
provided by the local people and direct observation in the field.
51
Geometric correction
Subsets of satellite images were rectified first for their inherent geometric
errors using digital topographic maps in Modified Universal Transverse Mercator
coordinate system obtained as the reference material. The image was registered to
the digital topographic maps using distinctive features such as road intersections
and stream confluences that are also clearly visible in the image. A first-degree
Rotation Scaling and Translation transformation function and the Nearest
Neighbour re-sampling method were applied. This re-sampling method uses the
nearest pixel without any interpolation to create the warped image. A total of 20
points were used for registration of TM image subset with the rectification error of
0.1083 pixels.
A very high level of accuracy in the geo-referencing of the images was
possible because of the use of digital source as the reference data that allowed
zooming to the nearest possible point location.
Classification
The supervised Maximum Likelihood Classification method was used for
the classification of all the images. Training areas corresponding to each
classification item (or, land use class), were chosen from among the training
samples collected from the field.
To produce land use maps of 1986 and 2002 and to investigate changes
that occurred between these periods, the following four land use classes were
considered in image classification: gallery forest, wooded savannah, shrubby
savannah and farm fields. The choice of these land use classes was guided by: i)
52
the objective of the research, ii) expected certain degree of accuracy in image
classification, and iii) the easiness of identifying classes on false composite of the
images and the ground. A brief description of each of the land use classes is
presented in Table 7.
Table 7: Land use classes considered in image classification and change detection Land use class General description
Gallery forest Forest areas mostly along the rivers with estimated 75
percent or more of the existing crown covered by
broadleaf trees. The predominant species are: Pterocarpus
erinaceus, Afzelia africana, Kaya senegalensis,
Anageissus leiocarpus, Parkia biglobosa, Cassia
sieberiana,Mitragina innermis, etc.
Wooded
savannah
Wooded areas with estimated 50 percent or more of the
existing crown covered by naturally growing trees. It
includes also old fallows. Common species are Vitelaria
(cotton), tubers (cassava, yam, potato) and plantation
(cashew, mangoes, orange, etc.).
Source: Landsat database (2006)
Detection of land changes
The Winships programme was used to convert the images from geotif
format to Idrisi raster (tfw). Bands 2, 3 and 4 were used for the image
classification because they are especially responsive to the amount of vegetation
biomass present in the images. Band 4 is put in the channel Red, band 3 in the
channel Green and band 2 in the channel Blue. A 432/RGB false colour
composites were produced (Figure 8). The specificity of each band of Landsat TM
images was described in Chapter Two.
After selectively combining classes, classified images were filtered before
producing the final output (Figure 9). A 3x3 median filter was applied to smooth
the classified images. All activities related to image processing were performed
with IDRISI 32.
54
Classified images were converted into vector format, and then exported to
ArcView-GIS Version 3.2 from IDRISI. In ArcView environment, the vectors
were clipped with the real limit of the study area and intersected each other in
order to detect the land use change within these two dates. All these operations
were done with the module “geo-processing”. The land use polygon themes for
1986 and 2002 were converted into MapInfo format with the module “universal
translator”. Land use units computing and the finishing of the maps were done
with MapInfo. The data base was exported to EXCEL (dbf format) for further
analyses.
55
Figure 8: False colour composites of satellite imageries
Source: Landsat TM image and author’s design, 2005.
56
Figure 9: Supervised classification of Landsat image of Bieha district in 2002
Source: Landsat image processing, 2006.
57
Problems encountered during the images processing
It was difficult to separate fallow from the other units since they appeared
like farm fields, shrubby savannah or wooded savannah in accordance with their
duration. Presence of cloud in parts of the TM image was the second major
problem encountered during image classification. The clouds were classified as
separate classes and later combined with their respective classes with the help of
ground-truth information. The third and last problem concerned the bushfire
detection. It was quite difficult to map the bushfires due to the fact that the early-
fires did not destroy completely the grasses since they were still green.
Immediately after the fire, grasses and leaves re-grew and affected the detection of
the impact of the fire in the ground by the satellite.
Population interviews
Instruments used
For the primary data, a structured questionnaire was developed to collect
information. The questionnaire was mostly close-ended and was categorised into
sixteen (16) sections (appendix 1). Each section represented a specific sub-theme
from the set of information to be collected (Table 8).
58
Table 8: Structure of the questionnaire
Section Sub-theme
1 Vegetation dynamics
2 Wild animals dynamics
3 Crops productivity
4 Food security at household level
5 Type of crops produced
6 Farming practices
7 Soil dynamics
8 Arable land dynamics
9 Household size
10 Income level
11 Drinking water sources
12 Living condition
13 Permanent migration
14 Temporary migration
15 Permanence of water in the rivers after the rainy season
16 Availability of fishes in the rivers
Source: Author’s construct, 2005.
Method of sampling
The target population was the total population (male and female) of 25,
634 in Bieha district who were 40 years old or more and have been living in the
district for at least 20 years. The assumption was that people who satisfied these
two conditions were old and qualified enough to provide accurate information
related to the sixteen sub-themes of the questionnaire in 1986, 1996 and in the
recent time. The target population was multiplied by 0.16 to obtain sampling frame
59
of 4,101 as the population of those old enough to provide the right information.
This was based on the fact that the population aged of at least forty years old
formed about 16 % of the total population in 2006 of Sissili province (I.N.S.D,
1996). Based on the sample frame, a sample fraction of 0.03 was purposively
chosen and used to generate a sample size of 123. From the 22 villages that made-
up Bieha district, 11 villages were randomly chosen for the survey (Figure 10).
The number of respondents selected from each village was based on the population
size of that village (Table 9).
Figure 10: Selected villages for the survey in Bieha district
Source: Author’s construct, 2005.
60
Table 9: Selected villages and number of respondents for survey
Selected villages Population in 2006 Sample size
Bieha 2,193 13
Binou 1,240 10
Boala 700 10
Danfina 1646 11
Prata 1,098 10
Kumbo 1,839 11
Kumbogoro 2,713 12
Livara 972 10
Pissai 1,933 12
Yalle 3,437 14
Yelbouga 1,549 10
Total 19,320 123
Source Author’s construct, 2005
Pre-survey activity in the villages
Official permission was first sought from the Prefect of Bieha district and
the central Chief of Bieha; while, in each of the selected villages, permission was
sought from the chief, elders and representatives (RAV) before interviews
commenced. The intension was to gain the support and cooperation of members of
the communities through these opinion leaders. At least one literate person in the
village was employed to translate the questions and answers from French to the
local language and vice-versa.
The Offices of Environment and Earth (DPECV) of Sissili and Bieha were
also informed of the survey to be conducted and its purpose. As the officers
responsible for the local environment, the foresters were also involved in the
61
administration of the questionnaire. One copy of the questionnaire translated into
French was given to each forester to enable them better understand the work.
The fieldwork
The fieldwork began on March 18, 2006 and ended on 28th of the same
month. In each of the selected villages, the questionnaire was administered using
snowball for two reasons: on one hand it was not evident to know at the first view
those fulfilling the age condition and on the other hand those who have lived in the
village since the last 20 years.
Issues from the interview
Response rate
Out of the sample size of 123 proposed, a total of 113 respondents were
interviewed. This gives a response rate of 91.8 % (Table 10).
62
Table 10: Respondents and response rate by selected village.
Selected villages Proposed
sample size
Total
respondents
Response
rate (%)
Bieha 13 11 85
Binou 11 11 100
Boala 10 10 100
Danfina 11 11 100
Prata 10 9 90
Kumbo 11 10 91
Kumbogoro 11 8 73
Livara 10 7 70
Pissai 11 11 100
Yalle 14 14 100
Yelbouga 11 11 100
Total 123 113 92
Source: Field work, 2006
Problems encountered
In Bieha, Prata, Kumbo, Kumbogoro and Livara, the total number of
respondents expected was not reached. This was due to the high number of recent
migrants in these villages. Most of the adults who were present during the survey
did not fulfil the condition of having lived in the village for 20 years. Some of
them had, due to the dry season, left for distant markets or towns to look for jobs.
Particularly in Kumbogoro, we were forced to leave before we reached the
proposed number for security reasons. In fact, a poacher from the village had been
apprehended carrying bush meat in the Safari Ranch of Bieha and had been
severely injured by the patrolmen of the ranch. According to the villagers, the
63
forester of Bieha district who was doing the survey with me that day was
responsible for the poacher being apprehended. One of the poacher’s parents
threatened the forester with a dagger. We therefore suspended immediately the
interview and drove quickly to Leo (the provincial city).
The second problem encountered was the lack of effective communication
with respondents. This was because our dialect i.e. the forester and researcher,
differs slightly from the Nuni dialect. Fortunately in each village, we could find
someone who could speak both Moore and Nuni or French and Nuni. However,
the translation of the names of plants and animals was a little bit difficult and we
were obliged to write these names in the local language and later find out the
corresponding names in French and English.
Diagram of the methodology
On the basis of the conceptual model previously discussed and on the
methods of data collection (image processing and interview of population), a
scheme in four steps that summarizes all the methodological processes is presented
in Figure 11.
Step 1: Use of Remote Sensing (RS)
This referred to the processing of Landsat TM images of 1986 and 2002 as
a modern method to assess land use change and also to estimate the extent to
which the natural resources have been depleted as a result of human activities.
This step involved the fieldwork, GIS processing and interpretation of data.
64
Step 2: Fieldwork activities
This combined checking and confirming real field situation through check
points and observations and the collection of data related to the various uses of the
natural resources by the local population. This step involved also RS, GIS and
interpretation of data.
Step 3: Use of Geographical Information System (GIS)
This integrated raster and vector information and traced the maps of the
land use dynamics. The GIS provided preliminary maps that facilitated fieldwork
and interpretation of data.
Step 4: Validation of the changes
This step validated the dynamics that have taken place from 1986 to 2002.
The extent of resources degradation in the area and its consequences on the long-
term sustainability were verified.
65
(1) REMOTE SENSING
(2) FIELD WORK
Figure 11: Methodological approaches for the land use dynamics
Source: Author’s construct, 2006
Processing of Landsat TM images: 2, 3, 4 RGB (1986 and 2002). Result: raster and vectors.
- Observation - Ground-truth - Questionnaires to assess resource use
(3) GIS
Use of vectors to map the dynamics of land use and to compute the surface areas of the land use units.
(4) INTERPRETATION
Maps interpretation will show the change in the area from 1986 to 2002 while the resource use assessment points out the role of the population in the degradation of the environment.
66
Limitation of the study
The multispectral mapping of the land associated with digital remote
sensing and GIS techniques is characterized by inherent limitations. No map
produced by digital manipulation of multispectral data is ever 100 % correct when
it is produced by a computer (Robinove, 1981). By nature, the process of
classifying such a broad range of the Earth’s features into specific and often
simplified land use and land cover classes introduces error by drawing boundaries
around geographically located classes that are ‘homogeneous’ or acceptably
heterogeneous (Bottomley, 1998). However, these limitations can often be
overcome by sound statistical analysis to produce acceptably accurate land use and
land cover maps as derived from multispectral satellite data.
Three main difficulties were encountered during the field work which may
constitute the limitations of this study. The first limitation focuses on the
separation between fallows and the other land use units. The geo-processing
module permitted to identify and quantify the fallows in 2002, but the
identification of those of 1986 requested training areas corresponding to fallows
unit for the supervised classification, which unfortunately was not possible
because of their confusion with the other units, namely farm fields, shrubby and
wooded savannas. The identification of the fallow in 1986 would have permitted
to know whether they are diminishing as the population reported.
The second limitation concerned the identification of the bushfires in the
images. Bushfires were not drawn on the land use maps because the dates of image
captures corresponded to the period of early fires which left very few marks in the
ground.
67
The third limitation is the ground truth data acquired for accuracy
assessment. By utilizing the process of obtaining the ground truth data by
extensive GPS field surveys, bias with respect to proximity to roads is
characteristic of the data. It should be noted that this is not critical to the overall
accuracy assessment of the land use map; however, it is important to mention.
The forth limitation dealt with the population interview. It was difficult to
reach the expected number of respondents who satisfied the selectivity conditions,
so that in some villages, the number of respondents interviewed was below the
needed number. The situation may affect the high representativeness of the results.
68
CHAPTER FIVE
LAND USE ASSESSMENT
Introduction
This chapter, in two broad sections, presents the output of images
processing and population interviews. The first section deals with the states of land
use detected in 1986 and 2002 and the changes which occurred during that period.
The second section discusses the population’s perception on the environmental
change since 1986.
Land use detection
State of the land use in 1986
In 1986, the farm fields’ area represented only 2 % of the entire Bieha
district. The area was mostly confined to just around the villages. The most
important unit was the shrubby savannah which occupied 38 % of the district,
followed by the wooded savannah (32 %). The gallery forest was located along the
rivers and represented 27 % of the district (Figure 12 and Table 11).
In all, 3,438 hectares were used for farming activities in Bieha district. The
rest of the district was occupied by natural vegetation: shrubby savannah, wooded
savannah or gallery forest. It was obvious that some portions of the shrubby and
wooded savannah included both recent and old fallows, even though the image
processing did not permit their detection. The relatively low cropping surface is
likely to be due to the fact that at that time the population was less dense in the
69
district and the cotton and maize cultivation as well as the cashew production were
not so much practised.
Figure 12: Land use units in Bieha in 1986
Source: Image processing and field work, 2006
Table 11: Surface area and proportion of land use units in 1986
Land use units Surface area in 1986 (ha) Proportion (%)
Farm fields 3,438.69 2.0
Shrubby savanna 67,427.46 38.5
Wooded savanna 56,967.57 32.5
Gallery forest 47,634.09 27.0
Total 175,467.81 100.0
Source: Image processing, 2006
70
State of land use in 2002
Land use units, surfaces and proportions in Bieha district in the year 2002
are presented in Figure 13 and Table 12. On the whole, the farming area reached
33,686 hectares, about 19 % of the whole district while the shrubby savannah
dropped to 20 % of the total surface area of Bieha. The wooded savannah and
gallery forest did not change in terms of surface area (Figure 13, Table 12). Apart
from the ranch and the forest reserve which are excluded from farming activities
and the extreme south-west of the district dominated by forests and perennial
rivers, the rest of the district was almost used for farming (Figure 13). The
remaining shrubby savannah was located in the centre-north of the district. The
cropping acreage increased at the detriment of the shrubby savannah; an indication
that most agricultural activities were practised in the shrubby savannah.
71
Figure 13: Land use units in Bieha in 2002
Source: Image processing and field work, 2006
Table 12: Area and proportion of land use units in 2002
Land use units Surface area in 2002 (ha) Proportion (%)
Farm fields 33,686.64 19.0
Shrubby savannah 35,818.88 20.5
Wooded savannah 58,714.6 33.5
Gallery forest 47,240.36 27.0
Total 175,460.48 100.0
Source: Image processing, 2006
72
Land use dynamics from 1986 to 2002
Two maps representing the period from 1986 to 2002 were superimposed
to enable significant evolution in land use dynamics to be determined. The four
land use units as previously defined were codified and used to present a better
understanding of the changes that were detected (see Table 13).
Table 13: Codification of land use units
Land use units Codes
Farm fields F
Shrubby savannah Ss
Wooded savannah Ws
Gallery forest Gf
Source: Author’s design, 2006
By using the Geo-processing module (ArcView) and the Cross Tabulation
module (Excel), the following combinations which present the dynamics of land
use in Bieha district was obtained (Table 14)
73
Table 14: Legend of the land use dynamics from 1986 to 2002
Combinations
(1986/2002)
significance
D y n a m i c s o f t h e f a r m f i e l d s
FF Farm fields in 1986, still Field in 2002
FGf Farm fields in 1986 changed into Gallery forest in 2002
FSs Farm fields in 1986 changed into Shrubby savannah in 2002
FWs Farm fields in 1986 changed into Wooded savannah in 2002
D y n a m i c s o f t h e g a l l e r y f o r e s t s
GfF Gallery forest in 1986 changed into Farm fields in 2002
GfGf Gallery forest in 1986, still Gallery forest in 2002
GfSs Gallery forest in 1986 changed into Shrubby savannah in
2002
GfWs Gallery forest changed in 1986 into Wooded savannah in
2002
D y n a m i c s o f t h e s h r u b b y s a v a n n a h
Shrubby savannah in 1986 changed into Farm fields in 2002
SsGf Shrubby savannah in 1986 changed into Gallery forest in
2002
SsSs Shrubby savannah in 1986, still Shrubby savannah in 2002
SsWs Shrubby savannah in 1986 changed into Wooded savannah in
2002
D y n a m i c s o f t h e w o o d e d s a v a n n a h
WsF Wooded savannah in 1986 changed into Farm fields in 2002
WsGf Wooded savannah in 1986 changed into gallery forest in 2002
WsSs Wooded savannah in 1986 changed into Shrubby savannah in
2002
WsWs Wooded savannah in 1986, still Wooded savannah in 2002
Source: author’s construct, 2006
74
75
The results of the combinations are presented in Figures 14, 15, 16, and 17.
Figures 14 and 15 illustrate the comparison between land use types in 1986 and
2002. The size of the total farm fields in 2002 was about nine times its original
size in 1986. The surface area of the shrubby savannah was reduced to half. The
sizes of the wooded savannah and the gallery forest had almost remained
unchanged.
Figure 16 presents the observations of land use change from 1986 to 2002.
The farm field and wooded savannah units increased in surface area (gain) while
the shrubby savannah and the gallery forest experienced losses in surface area.
Figure 17 shows the trend of land use in Bieha. The size of farm fields
increased within the period 1986 to 2002 with an expansion of 880 % in 16 years,
and annual expansion rate of 55 %. The surface area of the shrubby savannah was
reduced drastically from 65,427 hectares in 1986 to 35,818 hectares in 2002. The
reduction rate was 46.8 %; about 3 % annual average rate of reduction. The
tendency curves (Figure 17) of the wooded savannah and the gallery forest
remained horizontal, meaning that the change was not noticeable. The annual
expansion rate of the wooded savannah was 0.19 % and the reduction rate of the
gallery savannah was 0.05 % annually (Table 15).
76
Figure 14: Land use dynamics in Bieha district from 1986 to 2002
Source: Image processing, 2006
0
10000
20000
30000
40000
50000
60000
70000
Surface area (ha)
Farm field Shrubby savannah Wooded savannah Gallery forest
Land use units1986 2002
Figure 15: Comparison between the size of land use types in 1986 and 2002