Nigeria Geographical Journal, Volume 8(1) June 2012: ©Association of Nigerian Geographers; All Rights Reserved
ISSN 1358-4319
ASSESSMENT OF URBAN GROWTH PATTERN IN SOKOTO METROPLIS, NIGERIA USING MULTITEMPORAL SATEELITE DATA
N. B. Eniolorunda and I. M. Dankani Geography Department, Usmanu Danfodiyo University Sokoto *Correspondence(
[email protected];
[email protected])
ABSTRACT Essential to town planning and management are estimates of the past, current and future sizes together with the growth rate of metropolitan area. This study assessed urban growth pattern in Sokoto Metropolis (Nigeria) using the Landsat data of 1986, 2002, 2005 and the quick-bird data of 2003. Principal Component Analysis (PCA) and Inverse PCA were run on the 2005 Landsat affected by striping. Unsupervised and supervised (by Maximum Likelihood method) image classifications were also performed to extract details on built-up areas for the various dates. Areal coverage for 2 2 2 1986, 2002 and 2005 were found to be 26.3 km , 32.2 km and 40.9 km respectively. Growth rates were put at 1.4% for 1986-2002, 8.9% for 2002-2005 and an overall value of 2.95% between 1986 and 2005. The size of the metropolis in 2020 was modeled using 2 exponential growth formula at 2.95% growth rate and this yielded 63 km (54% of its size in 2005). The study concluded that the study area would at least grow by half of its 2005 size by 2020 and that such growth would be spatially sporadic with monumental consequences for both the government and dwellers of the study area. Evaluation of the degree to which the master plan has been deviated from, harmonization of the current development with the master plan and enforcement of strict compliance to the master plan were recommended for effective and efficient city management. Keywords: Sokoto; Geomod; Simulation; PCA; Maximum Likelihood; Digital Image Classification
INTRODUCTION Sokoto metropolis has been the capital of Sokoto State since 1976. It is the administrative, socio-cultural, political and economic hub of the
state. According to Mamman (1999) and Shamaki (2006), the importance of Sokoto in the northwestern region of Nigeria derives from its being the seat of The Caliphate, dating back to 1804 when Shehu ___________________107
Nigeria Geographical Journal, Volume 8(1) June 2012: ©Association of Nigerian Geographers; All Rights Reserved
Usmanu Danfodiyo fought the Jihad war and subsequently established the throne for the Caliphate. Thenceforth till early 1960s when it became the capital of the then northwestern region, the growth of the city started increasing though at a slow rate. Between 1960s and late 1990s, coinciding with the military era in Nigeria, not much was achieved in the area of city infrastructural development across the country such as could have warranted rapid urban expansion as obtainable in most cities of Nigeria today. Beginning from 1999 when democracy was reintroduced into Nigeria, rural-urban migration and proliferation of economic, political, social and cultural events in the city took dramatic turns. Consequently, the growth of Sokoto metropolis became bolstered, and today, the master plan of the city is suspected to have grossly been altered. This will therefore pose serious challenges for government in allocating infrastructural facilities. The choice is usually to embark on urban renewal which is expensive for the government and unbearable for the city dwellers. According to John (2005), the explosive rates of growth in the Nigerian cities have not only progressively complicated interrelated problems of human settlements and the environment,
ISSN 1358-4319
but have also greatly accelerated poverty. Epstein et al. (2002) bring out the techniques for mapping suburban sprawl. They evaluate the traditional unsupervised classification and proposed GIS buffering approach for mapping the suburban sprawl. They also discuss the problems associated with the classification of urban classes (built-up) in comparison with rural and urban centers. Yeh and Li (2001) use Shannon’s entropy, which reflects the concentration of dispersion of spatial variable in a specified area, to measure and differentiate types of sprawl. This measure is based on the notion that landscape entropy or disorganization increases with sprawl. The urban land uses are viewed as interrupted and fragmented previously homogenous rural landscapes, thereby increasing landscape disorganization. Lata et al (2001) have also employed a similar approach of characterizing urban sprawl for Hyderabad City, India. Pontius et al. (2000) studied the scenarios of land use change in the Ipswich watershed, USA over a period of two decades. This study found that a conversion of forest into residential areas is a predominant land use change. Considering this type of land use change they predict the future land ___________________108
Nigeria Geographical Journal, Volume 8(1) June 2012: ©Association of Nigerian Geographers; All Rights Reserved
use changes in the Ipswich watershed based on the model calibrated for 1971 and 1985, and validated for 1991. With this model, the extent of deforestation in the watershed is predicted under different scenarios. The results of this are verified by Kappa index. In recent years, considerable interest has been focused on the use of GIS as a decision support system. The use of GIS as a direct extension of the human decision making process—most particularly in the context of resource allocation decisions is indeed a great challenge and an important milestone. With the incorporation of many software tools to GIS for multi-criteria and multi-objective decision-making — an area that can broadly be termed decision strategy analysis there seems to be no bounds for the application of GIS. The land use changes in the region under different scenarios are done using the multi-criteria evaluation through the decision support system. The decision support is based on a choice between alternatives arising under a given set of criterion for a given objective. A criterion is some basis for a decision that can be measured and evaluated. Criterion can be of two kinds: factors and constraints, and this can pertain either to attributes of the individual or to an entire decision set. In this case the objective being
ISSN 1358-4319
to urbanize; constraints include the already existing built-up area, roadrail network, water bodies, etc., where there is no scope for further sprawl; and factors include the components of population growth rate, population density and proximity to the highway and cities. The decision support system evaluates these sets of data using multi-criteria evaluation. This predicts the possibilities of sprawl in the subsequent years using the current and historical data giving the output images for the objective mentioned. Closely associated with the decision strategy analysis is the uncertainty management. Uncertainty is not considered as a problem with data, but else, it is an inherent characteristic of the decision making process. With the increasing pressures on the resource allocation process, the need to recognize uncertainty as a fact of the decision making process that needs to be understood and carefully assessed. Uncertainty management thus lies at the very heart of effective decision-making and constitutes a very special role in GIS (Eastman, 1999). This paper focuses on the urban sprawl pattern recognition and explores the causal factors. In recent years, considerable interest has been focused on the use of GIS as a decision support system. ___________________109
Nigeria Geographical Journal, Volume 8(1) June 2012: ©Association of Nigerian Geographers; All Rights Reserved
The use of GIS as a direct extension of the human decision making process—most particularly in the context of resource allocation decisions is indeed a great challenge and an important milestone. The use of Geographic Information Systems modeling has become quite prevalent within the field of urban sprawl research. Some research on urban sprawl uses GIS as a tool in understanding the effects of urban sprawl on the natural environment. GIS reveals spatial patterns of urban sprawl by measuring distances of new urban growth areas from town centers and roads for example (GarOn Yeh et al 2001). Because urban development is irreversible, GIS simulates future land development (Lee et al 1998). Because there is a lack of a universal definition of urban sprawl, a map of urban or built land is an adequate starting point in studying urbanization. A map provides the visual aspect from which studies on urban sprawl can begin in relation to urban growth. Similarly Before the introduction of Geographic Information Systems, mapping any phenomenon took an extremely long time. Maps produced through manual cartography for comparison were planned well in advance of a due date. Computer aided maps without GIS were very rudimentary and were not very aesthetically pleasing to say the least. The
ISSN 1358-4319
availability of different types of spatial data allows a GIS user to map virtually any phenomena with a geographic dimension applied to it. In addition, large amounts of data are processed before the creation of a map with much less work than with manual cartographic techniques. With a GIS, maps can be compared in a fraction of the time and can be done at variable scales with ease. Furthermore, there are no studies on assessing urban growth of Sokoto using GIS techniques. Similarly, the fact that the Master Plan of Sokoto has expired since 2003 and it was designed using the traditional cartographic methods, a GIS approach to assessing the growth of Sokoto will go a long way in bridging the gap between traditional approach to assessing urban growth and preparation of master plan to a more systematic approach integrating the use of latest GIS utility. Knowledge about what the current city size is, what it is expected to be in future, growth rate, among others are an essential component of strategic policies for infrastructural planning, development and management. Such knowledge can serve as a tool for checking uncontrolled urban sprawl and its attendant corollaries. Monitoring the pattern of urban growth has ___________________110
Nigeria Geographical Journal, Volume 8(1) June 2012: ©Association of Nigerian Geographers; All Rights Reserved
been demonstrated in many parts of the world using Remote Sensing (RS) in the Geographical Information System (GIS) environment. It is in the light of this that this study used the Landsat data of 1986, 2002 and 2005 to map the existing and simulate the future spatial growth of the metropolis. The aim of this study is to assess the growth of Sokoto metropolis using Landsat data. The specific objectives of the study were: i.
To extract the built-up area of Sokoto metropolis for each of the years; ii. To estimate the growth rate of the study area, iii. To simulate the future growth of the metropolis, and iv. To assess the spatial pattern of the simulated growth.
STUDY AREA The study area is the Sokoto metropolis, the capital of Sokoto State. It occupies an area between longitudes 5.136040 E to 5.302310 E and latitudes 12.956610 N to 13.083790 N (Figure 1). The area is drained by the westward flowing Sokoto-Rima River system which provides rich alluvial soil fit for variety of crops. The valley has an average altitude of 240 metres
ISSN 1358-4319
above sea level, representing the lowest level in the study area. The highest part of the study area has an altitude of about 321 metres above sea level, giving an altitudinal difference of 81 meters. The average annual rainfall is about 640mm, with the raining season lasting between May and October. The dry season spans between November and April. Temperature is as high as 430C around March/April (middle of the dry season) and as low as 230C in December/January (middle of the cold season).
MATERIALS AND METHODS Data Capture
For this study, the Landsat TM of 21st October 1986, Landsat ETM+ of 25th August 2002 and Landsat ETM+ of 15th September 2005 located in the path/row 191/051 were used. These data were downloaded from the Global Land Cover Facility (GLCF) of the University of Maryland, United State of America. Also, the ground control points (GCPs) of notable road junctions were collected from the study area for geometric assessment. ___________________111
Nigeria Geographical Journal, Volume 8(1) June 2012: ©Association of Nigerian Geographers; All Rights Reserved
5°12'0"E
ISSN 1358-4319
5°14'0"E
5°16'0"E
¯
13°6'0"N
So k o
u lt
tik u
l
ad
oad
od
R
iF
o
ad
o Ro
West
R
L a mi d
a in
d
ba
Ro a
lam
yo
i
13°4'0"N
n Bye a s -p s
an A
ul
ah
ster
S
Ab d
Ka
ern B ss ye-pa
d ll o R o a
a
Be
E
n lta
tan Abub a Sul k ar R o
Su
13°4'0"N
er
to
iv
University
Roa
d
R
ad
13°2'0"N
taria t
d Ro a
G
u
ad
Secre
13°2'0"N
Bi
ad
rn
Ro
i K eb
bi
u
Ro
sa
13°0'0"N
13°0'0"N
"
"
Sokoto Metropois
Legend "
12°58'0"N
NIGERIA
State Capital
Sokoto State
Major Roads
Nigeria Boundary
1
0.5
0
1 Kilometers 12°58'0"N
River 5°12'0"E
5°14'0"E
5°16'0"E
Figure 1: Map of Sokoto Metropolis (Digitized from Quick bird image of 2003)
Figure 2: Band 4 of Landsat ETM+ 2005 with Striping ___________________112
Nigeria Geographical Journal, Volume 8(1) June 2012: ©Association of Nigerian Geographers; All Rights Reserved
The Landsat data already georeferenced to Universal Transverse Mercator (UTM 31N) were imported into the Idrisi Andes RS and GIS software using the Tagged Image File Format (TIFF) module. The study area was submapped from each of the data and the data set subsequently explored for geometric and radiometric assessment. On assessment, the 2005 data was discovered to contain horizontal striping. Jensen (1996) and Eastman (2006) suggested the use of principal component analysis (PCA) so that the last few components which usually represent less that one percent of the total information are relevant to striping. When these are removed and the rest components reassembled, the improvement can be dramatic (Eastman, 2006). PCA was therefore run on the 7 bands of the 2005 data and the components so formed later reconverted into four refined bands with inverse PCA using the first component carrying 70% of the information content (see appendix 1). Also, all the data used were resampled to 28.5m resolution for overlay. Image Object Identification and Signature Development Several unsupervised classification attempts using CLUSTER module
ISSN 1358-4319
was performed in the Idrisi environment. All these yielded several confused classes (Figure 3a is a case in point). This prompted the formation of true and false colour composites of the study area using bands 1-2-3 and 2-3-4 respectively for object identification. Also, a true colour composite of mosaiced and georeferenced quick-bird satellite image (0.5m) of the study area for 2003 was formed in the Arcview3.3 environment to further aid object identification (as the number of pixels in the quick-bird image could not be handled in Idrisi). From a combination of the Landsat composites and that of the quickbird, built-up Area, Bare Surface, Water/Reflective surface, Moderate vegetation and Green vegetation were broadly identified. At least 100 pure pixels were digitized for each of these classes, the signatures of which were later extracted. Supervised Classification Having extracted the signatures of each class for each of the years, each of the images was submitted for hard supervised classification using maximum likelihood classifier. Figure 4a shows the classified 2005 image of the study area. The BREAKOUT module of Idrisi was later used to extract the built up ___________________113
Nigeria Geographical Journal, Volume 8(1) June 2012: ©Association of Nigerian Geographers; All Rights Reserved
ISSN 1358-4319
areas from other classes. Figures 4b,
4c and 4d show the details.
a
b
c
d Figure 3: Unsupervised Classification (a) and Composite Images (b, c and d) of the Study Area
Future Growth Simulation GEOMOD module of Idrisi Andes was used to simulate the urban
sprawl. It is a grid-based land-use and land-cover change model, which simulates the spatial pattern of land change forwards or ___________________114
Nigeria Geographical Journal, Volume 8(1) June 2012: ©Association of Nigerian Geographers; All Rights Reserved
backwards in time. GEOMOD simulates the change between exactly two land categories denoted as 1 and 2 i.e. developed and non-developed areas (Robert and Hao, 2006). Modeling in this study was preceded by simulating first what the metropolis was in
ISSN 1358-4319
year 2002 to test the efficacy of the simulator. Figure 5 shows the detail. Also a comparison was made between the predicted and the actual 2002 maps by crosstabulation. Figure 6 and Tables 1 and 2 present the details.
a
b
c
d
Figure 4: Hard Classification Image of 2005 (a), Classified 1986 (b), 2002(c) and 2005 (d) Images of the Study Area
___________________115
Nigeria Geographical Journal, Volume 8(1) June 2012: ©Association of Nigerian Geographers; All Rights Reserved
Figure 5:
Simulated 2002 Growth Predicted and Actual 2002
ISSN 1358-4319
Figure 6: Crosstabulated Image between Development
Table 1: Cross-tabulation of 2002Forcast (columns) against Actual 2002 (rows) 1 2 Total -----------------------------1 | 30166 5647 | 35813 2| 5647 472252 | 477899 -----------------------------Total | 35813 477899 | 513712 Chi Square = 354325.68750 df = 1 P-Level = 0.0000 Cramer's V = 0.8305
Table 2:Proportional Cross-tabulation 1 2 Total -----------------------------1 | 0.0587 0.0110 | 0.0697 2 | 0.0110 0.9193 | 0.9303 -----------------------------Total | 0.0697 0.9303 | 1.0000
___________________116
Nigeria Geographical Journal, Volume 8(1) June 2012: ©Association of Nigerian Geographers; All Rights Reserved
ISSN 1358-4319
Kappa Index of Agreement (KIA) -----------------------------Using Reclas02 as the reference image... Category KIA -------- ------1 0.8305 2 0.8305
Using 2002Forcast_1 as the reference image... Category KIA -------- ------1 0.8305 2 0.8305
Overall Kappa
0.8305
The diagonal values from top left to bottom right in Table 1 are pixel numbers that did not change for both dates. These constitute 98% of the entire pixels. Kappa Index is a measure of category persistence (agreement of pixels in a class between two dates). For all of the Kappa statistics, 0% indicates that the level agreement is equal to the agreement due to chance and 100% indicates perfect agreement. For this study, the overall Kappa is 0.83. The simulation performed in this study roughly assumed that development is Omni-directional but inhibited in the north due to the presence of the agricultural valley that should not be developed. It also assumed that built up areas would not be traded off for the nondeveloped areas. In other words, the
suitability map shows developed as totally unsuitable while undeveloped is suitable. This is automatically generated by the Idrisi based on instruction. According to Hofstee and Brussel (2006), the growth pattern of any city follows an exponential rate. To calculate this, the annual growth rate of the metropolis was derived by subtracting the 1986 area coverage from that of the 2005, the result of which was divided by the number of years between them. This result is factored into the exponential formula for future size (area coverage) prediction. To use GEOMOD, knowledge of the number of the pixels in future is important. This was derived by using the growth formula:
___________________117
Nigeria Geographical Journal, Volume 8(1) June 2012: ©Association of Nigerian Geographers; All Rights Reserved Growth rate = (Developed Size for Latter Date – Developed size for Previous Date)/Developed size for Previous Date)*100)/ Years between both Dates
Pixel information that serves as an input into growth rate estimation and exponential growth formula
ISSN 1358-4319
was derived using the Area module of Idrisi. Table 1 presents the data. It should be noted that the area coverage of the developed expressed in km2 does not translate into its spatial spread but rather its proportion in space.
Table 3: Area Values of Developed/Non-Developed Classes Class Developed NonDeveloped Total
1986 (Pixels) 29204
1986 2 (Km ) 26.3
2002 (Pixels) 35813
2002 2 (Km ) 32.2
2005 (Pixels) 45394
2005 2 (Km ) 40.9
484508
436.0
477899
430.1
468318
421.4
513712
462.3
513712
462.3
513712
462.3
Source: Digital Imagery Cliassifcation
Since GEOMOD works on the number of pixels rather than Km2, pixel numbers would be used. Therefore,
However, the overall growth rate of 2.95% was used to predict the future size to 2020. This value was factored into the formula:
Growth rate = (((2005 Developed - 1986 Developed) / 1986 Developed)*100)/ Years between 1986 and 2005 = (((45394-29204)/ 29204)*100)/19 = 2.92%
Af = Ab * (1 + %/100) (f-b),
This value is for the whole period. However, a vivid look at the area coverage of the years suggests inconsistency in growth rates. Using the same formula, the growth rates of 1986/2002 and 2002/2005 are put at 1.4% and 8.9% respectively.
Where: A is the area, f is the future year, b is the base year and % is the growth rate per year. 45394 * (1 + 15 = 2 2.92/100) 69,903 pixels (62.9 km ).
Figure 7 shows the 2005, 2020 Developed/Non-Developed and Growth Difference map. Also Figure 8 demonstrates the developed amidst other land types for 2005 and 2020. ___________________118
Nigeria Geographical Journal, Volume 8(1) June 2012: ©Association of Nigerian Geographers; All Rights Reserved
temporal inconsistency in growth rates within the period under consideration. Between 1986 and 2002, the metropolis grew by 1.4%. It also grew by 8.9% from 2002 to 2005. An overall growth rate estimate of 2.95% was arrived at between 1986 and 2005.
RESULT AND DISCUSSION This study has revealed that the area coverage of the Developed Sokoto metropolis for 1986, 2002 and 2005 are 26.3 km2, 32.2 km2 and 40.9 km2 respectively. The study also revealed
a
ISSN 1358-4319
b
c Figure 7: (a)2005 Map and (b) the Simulated 2020 Growth and Developed of the Study Area (c) Growth Difference between 2005 and 2020
___________________119
Nigeria Geographical Journal, Volume 8(1) June 2012: ©Association of Nigerian Geographers; All Rights Reserved
a
ISSN 1358-4319
b Figure 8: (a) Developed Area in 2005, (b) Developed Area in 2020
The 8.9% appears current for the study area, but for modesty of prediction, future growth was modeled using 2.95%. The period with the lowest growth rate coincided with the military era Mamman (1999) made reference to, while the period of the highest growth was when democracy had been well established. Also, the spatial simulation for the year 2002 the aim of which was to validate the efficacy of the model yielded 98% persistence of the developed and non-developed between the model and the reality of 2002. According to Robert and Hao (2006), the application of GEOMOD to the Atlanta Metropolitan Area renown for its land use change due to urban sprawl showed that 75% of the landscape persisted between 1973
and 1999, which are the years of rapid change. This was considered a success, though higher values could be achieved in other landscapes. For this study, 98% of persistence is acceptable with overall Kappa of 0.83. On this basis, the spatial growth of Sokoto for the year 2020 is considered valid. Furthermore, the built-up area of the metropolis is expected to cover 62.9 km2 at a growth rate of 2.95% by the year 2020. This is about 54% of its size in 2005 (Figure 7). According to Mamman (1999), the growth of every town involves the twin process of outward extension and internal reorganization. For internal reorganization, open spaces and greenbelts will be converted into developed area, thereby making it denser. The conversion of open ___________________120
Nigeria Geographical Journal, Volume 8(1) June 2012: ©Association of Nigerian Geographers; All Rights Reserved
spaces will result in loss of recreational grounds, while that of greenbelt will result in soil erosion which can develop into gullies that will destroy infrastructures. It will also lead to reduction in the absorption of CO2 gas in the atmosphere, thereby increasing warming. The outward extension of the metropolis will result in wetland encroachment which will reduce agricultural land, consequently leading to food (especially vegetables which are mainly consumed in the metropolis) production shortage. It will also result in loss of jobs by the farmers when the land is usurped for development. Although growth in 2020 is expected to maintain the normal spatial pattern, there will be dispersal of isolated built-up areas. This will affect infrastructural allocation such as roads and can lead to urban renewal, the cost and effects of which are monumental both on the government and the dwellers.
CONCLUSION This study has shown that Sokoto metropolis will at least grow by half of its size in 2005 by 2020 and that such growth will be spatially sporadic. Therefore, to forestall the likely negative consequences that
ISSN 1358-4319
might result from this exponential growth and ensure effective city management, there is a need for the government to evaluate the degree of deviation from the master plan, harmonize the current development with the master plan and enforce strict compliance to the master plan, as the situation is less expensive to manage now than later.
REFERENCES Eastman J. R. (2006): Remote Sensing Error and Haze Removal with PCA in Idrisi Andes Tutorial, Clark Labs, Clark University, 950 Main Street Worcester, M A, 01610-1477 USA Eastman J. R., J. E. McKendry, and A. F. Michele (2007): Change and Time Series Analysis Second Edition Clark Labs for Cartographic Technology and Geographic Analysis Clark University, Worcester, MA 01610 USA Gar-On Yeh, A., Xia, L. 2001. Measurement and Monitoring of Urban Sprawl in a Rapidly Growing Region Using Entropy. Photogrammetric Engineering & Remote Sensing 67 (1):83-90. Hofstee P. and I. M. Brussel (2006): Analysis of suitability for urban expansion ,Division of Urban Planning and Management, International Institute for GeoInformation Science and Earth Observation (ITC), P.O. Box 6, 7500 AA Enschede, The Netherlands. ___________________121
Nigeria Geographical Journal, Volume 8(1) June 2012: ©Association of Nigerian Geographers; All Rights Reserved
ISSN 1358-4319
Jensen J. R. (1996): Introductory Digital Image Processing: A Remote nd Sensing Perspective, 2 Edition, Clarke K. C. (Ed), Prentice Hall, Upper Saddle River, New Jersey 07458 John L.S. A. (2005): Planning Sustainable Urban Growth in Nigeria: Challenges and Strategies, Conference on Planning Sustainable Urban Growth and Sustainable Architecture, held at the ECOSOC Chambers, United Nations Headquarters, New York, on 6th June. Lee, J., Tian, L., Erickson, L.J., Kulikowski, T.D 1998. Analyzing growth- Management policies with geographical information systems. Environment and Planning B: Planning and Design, 25 (6): 865879. Mamman A.B. (1999): The Evolution and Growth of Sokoto City. Journal of Social Sciences and Administration, Vol.1 No.2. Robert G. P. and Hao C. (2006): GEOMOD Modeling in Land-Use & Cover Change Modeling. Clark University.. Shamaki M. A. (2006): The Incidence and Spatial Manifestation of Urban Poverty in Sokoto Metropolis: An Unpublished M.Sc. Dissertation, Dept. of Geography Usmanu Danfodiyo University Sokoto
___________________122
