Ijsn Vol7416 5
Ijsn Vol7416 5
Ijsn Vol7416 5
net/publication/320627741
CITATIONS READS
14 1,740
3 authors:
Komlavi Akpoti
International Water Management Institute
43 PUBLICATIONS 767 CITATIONS
SEE PROFILE
All content following this page was uploaded by Komlavi Akpoti on 20 October 2020.
CITATIONS READS
2 565
3 authors:
Komlavi Akpoti
University of Energy and Natural Resources
9 PUBLICATIONS 100 CITATIONS
SEE PROFILE
Some of the authors of this publication are also working on these related projects:
Effects of Large Scale mining on Land Use Land Cover Using Remotely Sensed Data View project
Novel Approaches for Efficient Targeting and Equitable Scaling of Rice Technologies in Togo and Benin (ETES-Rice) View project
All content following this page was uploaded by Amos Tiereyangn Kabo-bah on 27 October 2017.
ABSTRACT
Land cover is the natural or basic elements of the environment that link and impacts many parts of the local, regional and
global levels of the environment. The study was conducted in two capital districts of the Brong-Ahafo Region which lies
within the green belt of Ghana in the moist semi-deciduous forest zone. This research was to assess the effects of large-
scale mining on the Land Use Land Cover (LULC) using remotely sensed data. Also, this study tries to find out, total area
of the various land use categories, percentage change and annual rate of change of LULC changes as a result of the mining
activities from 2005 to 2015. Iterative Self-Organizing Data (ISODATA) under unsupervised classification showed an
overall accuracy and kappa coefficient of 80.8% and 0.754 for 2005, 92.8% and 0.908 for 2008, 89.2% and 0.861 for 2012
and 87.6% and 0.841 for 2015 respectively. The results from the LULC analysis showed that, Forest Evergreen was the
most dominant land cover type in 2005 with a total area of 1492.93 ha (44.94%), but decreased as the year’s increases with
increasing built-up areas. The built-up areas which consists of mining areas increased from 316.05 ha (9.51%) in 2005 to
1047.27 ha (31.53%) in 2015. We recommend effective management of degraded areas by incorporating tree planting as
this compiles with the 1998 Forest Policy. Concurrent reclamation should be adopted by the mining sectors to achieve a
sustainable and successful post-closure outcome. Also decision makers should adopt the use of remote sensing and GIS
tools as this would enhance identification of areas that are degraded, their rate and extent.
KEY WORDS: Mining, LULC, Remote Sensing, ANOVA, ISODATA, Landsat image, pattern.
724
The effects of large-scale mining on the Land use and Land Cover Changes using Remotely Sensed Data
population increase, depletion of natural resources and the natural resources and monitoring the environmental
pollution of the environment (Study & Zanjan, 2009). The changes (Mark & Kudakwashe, 2010). Detection of land
unplanned changes of the land use have become a major cover change has been found applicable in land use change
problem because of the absence of logical planning and analysis especially in the assessment of the extent of
consideration of environmental impacts (Study & Zanjan, deforestation in a particular area (Dale et al., 1993).
2009). For the past decades, Remote Sensing (RS) and Therefore viewing the Earth from space has now become a
Geographic Information System (GIS) technologies have necessity to understand the influence of man’s activities
been vital tools for mapping the Earth’s features, studying on the natural resources over a given period (Zubair,
the environmental changes in time and space, managing 2006). The focus of this research is to assess the effects of
the natural resources. This gives the most accurate means large-scale mining on the land use land cover changes of
of measuring the extent and pattern of the changes at a the Sunyani and Asutifi district in the Brong-Ahafo region
particular landscape over time (Kumar & Pandey, 2013). of Ghana for 2005, 2008, 2012 and 2015.
This technology affords a practical means of analysing the
changes in the land use pattern at the mine sites at METHODOLOGY
inaccessible places. It has also become possible to get a Description of the Study Area
synoptic coverage of a larger area, in a cost-effective and The study was conducted in two districts in the Brong-
in a repetitive way. Ahafo region of Ghana; Sunyani and Asutifi (Kenyasi) as
Assessing land-use and land-cover change has become a indicated in the figure 1 below:
central component in the current strategies for managing
The Sunyani Municipality surface area of 1500 km². The district has a total of 117
Sunyani Municipality is one of the twenty-two (22) settlements in the district and four paramountcy, thus
districts in the Brong-Ahafo Region of Ghana. Its capital Kenyasi No.1, Kenyasi No.2, Hwidiem and Acherensua
is Sunyani and this lies between latitudes 7019’57.43’’N (GoG, 2010).
and 7021’41.81”N and longitudes 2019’40.58’’W and
2020’51.47’’W (PHC, 2014). It shares boundaries with DATA COLLECTION
Sunyani West District to the Dormaa District to the West, Spatial Data Collection and Source
Asutifi District to the South and Tano North District to the In order to determine the effects of large-scale mining on
East the LULC changes of the study area, spatial data-sets were
The Asutifi District obtained from Landsat 7 and Landsat 8 archives from U.S
The Asutifi District is one of the twenty-two (22) Districts Geological Survey (USGS) and ground observations
in the Brong-Ahafo Region of Ghana and its capital is obtained from Google Earth. The Four data sets used for
Kenyasi. It is located between latitudes 6058’8.35’’N and the study, its source and date of acquisition are shown in
6058’7.55’’N and longitudes 2025’13.72’’W and table 1 below. The Landsat data were obtained from the
2026’43.56’’W (GoG, 2010).The district is one of the USGS and Earth Observation database. These imageries
smallest in the Brong-Ahafo Region with a total land were selected based on date of acquisition and its
725
I.J.S.N., VOL.7 (4) 2016: 724-733 ISSN 2229 – 6441
availability. To prevent bias in the data, the images were seasonal variation in the spectral reflectance of the land
of the same season free from cloud cover and have the cover data-sets (Nzunda, 2013). Also the data were
same identifiable features. This gives uniform radiometric georeferenced to the coordinate system of the study area
and spectral characteristics which helped reduce or prevent i.e. WGS84 projection; UTM zone 30N.
726
The effects of large-scale mining on the Land use and Land Cover Changes using Remotely Sensed Data
infrared, Red and Green) and this gave a false colour soils vary from dark to light brown, and water appears
composite. False colour composite was chosen because; very dark (Ned, 2004). Figure 3 and 4 shows the
vegetation mostly reflects in an infra-red colour, thus, they composite Landsat imagery of the study area in the false
appear in shades of red. Forest evergreen reflects in deep colour composite and they were carried out using ArcGIS
red, forest deciduous in light red, urban areas in cyan blue, version 10.2.
FIGURE 4: Landsat 7 TM & Landsat 8 of 2012 and 2015 scene of the Study Area
727
I.J.S.N., VOL.7 (4) 2016: 724-733 ISSN 2229 – 6441
the earth surface (Kashaigili & Majaliwa, 2010). The imageries were grouped into five (5) land cover class prior
unsupervised classified imageries also known as thematic to the knowledge of the area and this is as shown in table 2
map was generated using IsoCluster in ArcGIS version below;
10.2 and Google Earth as the reference. The classified
Accuracy Assessment Excel to determine the accuracy. The data exported were
The ideal knowledge for accuracy assessment was from used to determine the error matrix i.e. the kappa
the fact that it is essential for every classified imagery coefficient (k), overall accuracy, commission error (user’s
results, due to the fact that classified imageries are deemed accuracy) and omission error (producer’s accuracy), of the
inaccurate to be used for its intended purpose or as a images classified. Overall accuracy is the total accuracy of
decision tool. This helps in determining the feasibility of the classified images. Commission error (user’s accuracy)
the classified image depending on the acceptable level of is the probability of a specific class to be incorrectly
error in the imagery. The accuracy level of a map is classified on the map, whiles Omission error (producer’s
determined by selecting reference points identified in the accuracy) is the probability of a specific class is
imagery which are evenly distributed and by comparing it incorrectly classified on the ground. Kappa coefficient (K),
with the test pixel or corresponding reference location of a gives a discrete multivariate technique used in accuracy
ground observation. Equal number of text pixels selected assessment, thus K>0.80 gives a strong accuracy or
for the reference point is not advisable as some classes agreement of the class assessed, 0.40-0.80 is average and
may have larger number than the others, hence the larger <0.40 is poor (Maps & GIS Library, 2014). The formulae
the class, the more the test pixels. The reference points given below were used to determine the kappa coefficient,
were randomly distributed in the imagery and generated overall accuracy, user’s and producer’s accuracy
using ArcGIS version 10.2 and further exported to MS respectively;
r r
N x ii ( x i X x i )
K i 1
r
i 1
.........(4)
N ( xi X x i )
2
i 1
Change Detection and Analysis rate of change between the imageries i.e. 2005-2008,
Change detection measures the changes that have occurred 2008-2012 and 2012-2015. These were done by using Ms
in a particular area over a period of time. In this study, Excel and the rate of change for the different land covers
post classification method was used to determine the cover estimated based on the following formulae used by
change, annual rate of change and the percentage annual (Kashaigili & Majaliwa, 2010);
728
The effects of large-scale mining on the Land use and Land Cover Changes using Remotely Sensed Data
Where; Areai year x+1 = area of land cover (i) for the second date or the following date
Areai year x = area of land cover (i) for the first date/ initial date
n
Area
i 1
i year x = the sum of the land cover area for the first date
t years = the number of years between the first and second imagery date.
RESULTS & DISCUSSION matrix using the formulas in equation 1-9. The results
Accuracy Assessment obtained are summarised in table 3 below for the four
The accuracy of the classified imageries was determined periods of the study area.
by generating 250 reference points to obtain an error
As shown in table 3 above, the highest overall accuracy lowest accuracy was obtained in 2005 image classification
for the four period of the study was 92.8% in 2008 because of the low quality of the imagery as a result of
whereas the lowest was 80.5% in 2005. The highest kappa clouds and the scan lines in the imagery and this led to
coefficient was 0.908 and the lowest was 0.754. The misclassification of the LULC.
Land Use Land Cover Classification from 2005-2015 Five (5) land cover classes i.e. Forest Evergreen, Forest
The Landsat images for the study area were classified in Deciduous, Bare land, Built-up areas and Water Body
order to identify the changes in the cover between the four were identified. The land use land cover analysis is
periods, i.e. 2005, 2008, 2012 and 2015 respectively and presented in maps and tables as shown in figures 5, 6 and
this yielded four LULC maps from the satellite images. tables 4, 5 and 6 respectively.
729
I.J.S.N., VOL.7 (4) 2016: 724-733 ISSN 2229 – 6441
LULC Change between 2005-2015 0.15% respectively whiles built-up areas continued to
The various land use land cover classes for 2005, 2008, increase by 5.70%. The results are also presented in annual
2012 and 2015 were analysed to determine the area rate of change (ha/yr) and percentage annual rate of
covered by each class quantitatively. The LULC area, area change (%/yr) in table 4, 5 and 6, these were determined
of change and the rate of change between 2005-2008, using the formula in equation 8, 9 and 10 in the previous
2008-2012, 2012-2015 and 2005-2015 that have occurred section. In 2005-2008, forest deciduous, bare land and
in the study area were analysed and presented in table 3, 4 built-up areas increased by 240.75ha/yr (7.25%/yr),
and 5 respectively for the four (4) classified maps as 38.86ha/yr (1.17%/yr) and 13.51ha/yr (0.41%/yr) and
shown in the previous section. From 2005-2008 as shown forest evergreen and water body decreasing by
in table 4, the forest evergreen and water body decreased 289.14ha/yr (8.70%/yr), 3.99ha/yr (0.12%/yr). For 2008-
by 26.11% and 0.36% respectively, whiles forest 2012, most of the land covers of the study area decreased
deciduous, bare land, built-up areas and increased by whiles land use (built-up areas) was increasing. Thus
21.74%, 3.51% and 1.22% respectively. For the second forest evergreen, forest deciduous, water body, bare land
period (2008-2012) shown in table 5 above in the previous were decreasing in rate by 61.41ha/yr (1.84%/yr),
section, forest evergreen and water body still by decreased 52.15ha/yr (1.57%/yr), 1.34ha/y (0.04%/yr), 10.72ha/yr
by 7.36%, 0.61% and built-up areas increasing by 15.09%. (0.32%/yr) respectively whiles built-up areas were
During the second period, forest deciduous and bare land increasing by 125.34ha/yr (3.77%/yr). From 2012-2015
which increased in the first period decrease tremendously result, there was a continuous reduction in the forest
by 6.28% and 1.29% respectively. For the third period evergreen, forest deciduous, bare land, water body by
(2012-2015) shown in table 6 above in the previous 14.19ha/yr (0.43%/yr), 40.56ha/yr (1.22%/yr), 1.64ha/yr
section, forest evergreen, forest deciduous, bare land and (0.05%/yr), 6.71ha/yr (0.20%/yr) whiles built-up areas
water body still decreased by 1.28%, 3.66%, 0.61% and continued to increase by 63.10ha/yr (1.90%/yr).
730
The effects of large-scale mining on the Land use and Land Cover Changes using Remotely Sensed Data
2500
2000
LULC,2005
Area (ha)
1500
LULC, 2008
1000 LULC, 2012
LULC,2015
500
0
FE FD WB BL BA
FIGURE 7: The trend of the Land Use Land Cover Changes
Considering the trend for Forest Evergreen over the years This shows a sharp increase from 2005-2008 and a gradual
between 2005 and 2015 with respect to the percentage area decrease from 2008-2015. The percentage area covered by
covered as shown in figure 8, the trend shows a strong water body over the years between 2005 and 2015 showed
negative correlation of r2=0.88 as there was a gradual a strong negative correlation of r2=0.967.
decrease in the percentage area coverage over the years. Generally, from the results, forest evergreen, forest
Therefore, as the years increase, the percentage area deciduous, and water bodies were decreasing as the built-
coverage of the Forest Evergreen decreases. However, the up areas increased. The increased depletion rate of the
Forest Deciduous shows a weak positive correlation i.e. general land cover indicated that, human population and
r2=0.385 over the years. The results also indicate a strong mining activities were immensely destroying the
positive correlation r2=0.969 over the years between 2005 vegetation cover. However, the mining activity is the main
and 2015 in relation to the percentage area covered by reason for the tremendous changes in the study area. This
built-up areas. The trend shows that as the year’s increase, is because, these activities are linked to both direct and
the percentage area covered by the built-up areas indirect changes of the land cover especially in areas
increases. Comparatively, the trend of Bare land was where the operations are being carried out. These have
weakly positive r2=0.285 in correlation over the years. substantial effects on the land cover land use of which the
731
I.J.S.N., VOL.7 (4) 2016: 724-733 ISSN 2229 – 6441
forest is the greatest victim because these mineral come to settle at the location of the mining operation for
resources are embedded within the soils in which they are work and trade purposes. Though, mining activities boosts
found. The loss of the forest cover was due to the random a country’s economy, it leaves a negative impact on land
and rampant clearing of the forest for mining activities cover because it is a function of time that results in direct
(surface mining) especially those mostly found on the and indirect effects on different land use and land covers.
mountainous areas in the Asutifi district. This indirectly This however, confirms the studies conducted by (Opoku-
has affected the water bodies as most of them had their Ware, 2010; Pandit, 2011; Kumi-Boateng et al., 2012 and
source from these mountainous and forested areas leading Nzunda, 2013) who concluded that, increase in
to their drying up as these vegetation covers are cleared. populations and mining activities and the changes in the
Built-up areas were increasing because of increase in land use/land cover were mutually related to each other.
population and also immigrants from far and near have
732
The effects of large-scale mining on the Land use and Land Cover Changes using Remotely Sensed Data
Global demand for gold is another threat for tropical Environment and mining, A peep into deep. Agrotech Pub.
forests. Environmental Research Letters, 10(1), 014006. Academy.
http://doi.org/10.1088/1748-9326/10/1/014006
Maps & GIS Library (2014) Accuracy Assessment of an
Aryee, B.N. (2001) Ghana’s mining sector: its image classification in Arcmap. TEXAS. Retrieved from
contribution to the national economy. Resources Policy, http://library.tamu.edu/maps-gis
27(2), 61-75.
Mark, M. & Kudakwashe, M. (2010) Rate of Land-Use/
Bury, J. (2004) Livelihoods in transition: Transnational Land-Cover Changes in Shurugwi District, Zimbabwe:
gold mining operations and local change in Cajamarca, Drivers for Change. Journal of Sustainable Development
Peru. Geographical Journal, 170(1), 78–91. http://doi.org/ in Africa, 12(3), 107–121.
10.1111/j.0016-7398.2004.05042.x
Michael S. O’Neill (2006) Working with Landsat Data.
Dale, V.H., O’NEILL, R.V., Pedlowski, M. & NASA, USGS, Canadian Center for Remote Sensing, Del
Southworth, F. (1993) Causes and effects of Land Use Mar College.
Change in Central Rondonia. Brazil Photogrammetric-
Engineering and Remote Sensing, 59(6), 997 – 1005. Ned, H. (2004) Selecting the appropriate band
combination for an RGB image using Landsat imagery.
GoG. (2010) Government of Ghana 1998 Budget. Remote Sensing Resources (Vol. 3).
Retrieved January 28, 2016, from asutifi. ghanad
istricts.gov.gh Nzunda, H.P. (2013) Impacts of Mining Activities on
Land Cover and Forest Stock in Mbozi District, Mbeya
Jensen, J.R. (2014) Introductory Digital Image Processing Region, Tanzania.
(3rd editio). New Jersey, USA.: A Remote Sensing
Perspective. Opoku-Ware, J. (2010) the Social and Environmental
Impacts of Mining Activities on Indigenious
Kashaigili, J.J. & Majaliwa, A.M. (2010). Integrated Communities.
assessment of land use and cover changes in the
Malagarasi river catchment in Tanzania. Physics and Pandit, S. (2011) Forest Cover and Land Use Changes : A
Chemistry of the Earth, Parts A/B/C, 35(13), 730–741. Study of Laljhadi Forest ( Corridor ), Far-Western
http://doi.org/10.1016/j.pce.2010.07.030 Development Region , Nepal. Tribhuvan University.
Kumar, A. & Pandey, A. (2013) Evaluating Impact of PHC. (2014) Ghana Statistical Service. 2010 Population
Coal Mining Activity on Landuse / Landcover Using and Housing Census; Sunyani Municipality.
Temporal Satellite Images in South Karanpura Coalfields
and. International Journal of Advanced Remote Sensing Study, A.C. & Zanjan, O.N. (2009) Evaluation and
and GIS, 2(1), 183–197. forecast of human impacts based on land use changes
using multi-temporal satellite imagery and GIS: A case
Kumi-Boateng, B., Mireku-Gyimah, D. & Duker, A.A. study on Zanjan, Iran. The International Archives of the
(2012) A Spatio-Temporal Based Estimation of Photogrammetry, Remote Sensing and Spatial Information
Vegetation Changes in the Tarkwa Mining Area of Ghana, Sciences, 38(2), 548–552.
4(3), 215–229.
Zubair, A.O. (2006) Change detection In land use and
Lodha, R.M., Purohit, K.J. & Yadav, S.H. (2009) land cover using remote sensing data and GIS. University
Of Ibadan.
733