Table of Contents
LIST OF TABLES
LIST OF FIGURES
RESULTS AND DISSCUSION
The study is the first attempt of mapping impervious surface that offers a very simple and visual understanding of the health status of the watersheds within Kaduna metropolitan area. This study examined the spatiotemporal growth of impervious surface areas in Kaduna metropolitan area, Nigeria in relation to the watersheds using the SAVI differencing approach. A supervised maximum likelihood algorithm was used to classify image derived from the SAVI differencing approach, for the time periods. A random sampling scheme was conducted to complete an accuracy assessment of the classification. As for the assessment, no overall accuracy was less than 0.7 and kappa coefficient below the range of 0.55-0.7 termed “Good” as suggested by Monserud and Leemans (1992). Results from the watershed categorization indicate it took 36 years to establish the first degraded watershed, this is from 1973 to 2009, going by the prediction in the analysis, it will take less years, i.e. 20 years to establish the next degraded watershed in the study area. This study demonstrates the feasibility of mapping and categorizing watersheds associated with impervious surface areas, through the combination of remote sensing data (LANDSAT and ASTER) with the help of a GIS software (ArcGIS).
LIST OF TABLES
Table 1. Data source
Table 2. Annual growth rate and impervious surface percentage for kd-shed
Table 3. Annual growth rate and impervious surface percentage for kd-shed
Table 4. Annual growth rate and impervious surface percentage for kd-shed
Table 5. Annual growth rate and impervious surface percentage for kd-shed
Table 6. Annual growth rate and impervious surface percentage for kd-shed
Table 7. Annual growth rate and impervious surface percentage for kd-shed
Table 8. Annual growth rate and impervious surface percentage for kd-shed
Table 9. Total impervious surface area for the two time periods
LIST OF FIGURES
Figure 1. Spatiotemporal analysis workflow
Figure 2. Kaduna Metropolitan area DEM
Figure 3. Kaduna metropolitan area delineated watersheds.
Figure 4. 1973 composite map
Figure 5. 1990 composite map
Figure 6. 2009 composite map
Figure 7. Impervious areas
Figure 8. Background colors corresponding to stream quality conditions.
Figure 9. Training areas / accuracy reference points
Figure 10. Impervious surface area coverage to watershed in 1973
Figure 11. Impervious surface area coverage to watershed in 1990
Figure 12. Impervious surface area coverage to watershed in 2009
Figure 13. Accuracy assessment result for impervious surface classification 1973
Figure 14. Accuracy assessment result for impervious surface classification 1990
Figure 15. Accuracy assessment result for impervious surface classification 2009
Figure 16. Watershed category in 1973
Figure 17. Watershed category in 1990
Figure 18. Watershed category in 2009
Figure 19. Watershed category in 2029 (predicted)
Figure 20. General overview of the percentage of impervious surface for the study area from 1993 to 2029 (predicted)
Figure 21. Impervious surface area in sq.km for kd-shed
Figure 22. Impervious surface area in sq.km for kd-shed
Figure 23. Impervious surface area in sq.km for kd-shed
Figure 24. Impervious surface area in sq.km for kd-shed
Figure 25. Impervious surface area in sq.km for kd-shed
Figure 26. Impervious surface area in sq.km for kd-shed
Figure 27. Impervious surface area in sq.km for kd-shed
Metropolitan areas are very large urban agglomerations with population in excess of a million. Countries all over the world have urbanized significantly since the 1950s and this trend is projected to continue through the 21st century (UN, 2008). This projection is expected to take place mainly in the developing countries, as the process of urbanization has virtually ended in Europe; whereas, in Africa, rapid urbanization is still aggravating the rural-urban dichotomy (Kreibich2003).
Cities in the developing countries are growing very rapidly due to distress migration from their rural hinterlands. Thus, cities in developing countries’ are expected to double their total population from 2 billion in 2000 to 4 billion by 2030, and triple their total built-up areas from about 200,000 sq. km. to more than 600,000 sq. km. within the same period, thereby constitute over half (54.5%) of total urban built-up areas in the world by2030 (Charles, 1989; Shlomo et al., 2005).
Based on current settlement practices, cities in the developing countries like Nigeria will most likely double their built up areas to accommodate their fast rising population. Like many developing countries Nigeria has experienced some sort of rapid development in terms urbanization and industrialization. These factors have considerably influenced growth in urban and suburban areas. All these have given rise to increase in impervious surface cover within urban and suburban areas in Nigeria.
Impervious surfaces are defined as surfaces that prohibit the movement of water from the land surface into the underlying soil (Civco and Hurd, 1997; Hurd and Civco, 2004). According to (Okeke, 2006) they are mainly constructed surfaces - rooftops, sidewalks, roads, and parking lots covered by impenetrable materials such as asphalt, concrete, brick, and stone. These materials seal surfaces, repel water and prevent precipitation from infiltrating soils.
Impervious surfaces are common features in every urban community and are easily recognized throughout the urban landscape, as roads and sidewalks, driveways and parking lots, homes and schools, and shopping centers and factories. Impervious surfaces are not as important for what they are but moreover for what they do. As the phrase descriptively implies, impervious surfaces do not allow other substances, water for instance, to pass through; instead it turns into runoff. In following this definition for this research, impervious surfaces specifically mean being impenetrable to the passage of water. (Johnson, 2004)
Irrespective of the global location, all urban communities are typically characterized by impervious surfaces. They are the cornerstones and foundations of a modern day society. As small communities grow and expand, the surrounding rural landscape is transformed into residential and commercial areas, which are largely comprised of impervious surfaces. Principally, impervious surfaces can be thought of as constructed surfaces created from materials such as asphalt or concrete, categorized as belonging to either transportation or rooftop land cover types (Johnson, 2004).
Infrastructures such as streets and drainages culverts are usually some of the features noticeable when a rural area is being transformed. These features create accessible paths for the frequent and reliable transit of motor vehicles in and out of the developing area. As buildings emerge, the amount of impervious surface increase. This transformation of land and its associated increases in impervious surfaces continue until the area is nearly all developed and has become a thriving center of residential and commercial activity.
From a progressive view of urban development, impervious surface development is both desirable and encouraged because of the economic and social benefits. Developed rural areas witness increase in property value, more jobs with skill specialization are created in order to construct and further develop the area. Therefore, as rural areas undergo growth and development, the increase in impervious surfaces facilitate society’s organization, consistency, and prosperity.
However, the negative effects of impervious surfaces are just now beginning to be recognized and understood. Studies have shown that impervious surfaces present on a landscape can act as an indicator for determining the overall environmental health of that area. For instance, as untarred roads and drainage areas become more impervious, there is less water infiltration into the soil to recharge the groundwater. With less groundwater, the loss of baseflows can trigger a chain reaction of negative impacts to various components of the landscape either direct or indirect leading to increasingly severe problems.
The measurability of impervious surfaces is the reason for its use as an environmental indicator. However, for its effective use as an environmental indicator, its measurement is usually done within the boundary constraints of a watershed. Measurement based on the boundary of a watershed not only identifies where to proceed, but it also facilitates a more effective assessment and course of action. (Johnson, 2004)
Kaduna metropolis is arguably the first urban center in northern Nigeria to have experienced development in terms of urbanization and industrialization in 1960s being the capital of defunct northern Nigerian government, it was therefore necessary to embark on constructions that will identify it as a regional capital. These constructions upon the landscape of Kaduna metropolis have gradually increased over space and time and are referred to as impervious surfaces Kaduna metropolis has in recent times witnessed increased cases of flooding during rainy seasons, which might not be unconnected with the increased amount of impervious surfaces within the metropolis. Increase in impervious surfaces results to more storm water runoff the increases the likelihood of flooding during rainy seasons, less groundwater recharge as such would translate to less ground water discharge to streams during dry season. Impervious surfaces also allow pollutants to accumulate upon them, and subsequently washed into water bodies by storm water runoff thereby degrading water quality.
Impervious surface mapping can serve as an indicator in ascertaining the health status of a watershed. In order to suppress subsequent environmental damages in the future development of the metropolis it becomes necessary to map impervious and categorize the watersheds within the study area. This study comes at the right time when the eastern part of the metropolis is undergoing rapid development, the study intends to serve as a reminder to the state policy makers on the need to monitor urban development in the metropolis with caution, in order to achieve an environmentally sustainable metropolis.
The aim of the study is to identify impervious surface areas for three time periods, categorize the watersheds base on impervious surface cover of the watersheds and predict a future degraded watershed.
(i) To delineate watersheds of Kaduna metropolitan area
(ii) Study methods of impervious surface area mapping
(iii) Categorize Kaduna metropolitan area watersheds base on impervious surface area.
(iv) Predict a future degraded watershed in the study area
Watersheds within Kaduna metropolis are degrading fast due to increase in impervious surface cover.
(i) What are the methods of identifying impervious surfaces?
(ii) How many watersheds are there within the study area?
(iii) What is the growth rate of impervious surfaces area within the study area?
(iv) Has there been changes in the watershed category with regards to impervious surface area of Kaduna metropolitan watersheds over the time periods?
Outline of the Study
The dissertation is organized into five chapters, chapter one presents the research Introduction by highlighting the context of research, including the objectives and scope of the research, research questions and the hypothesis.
Literature review: Methods of impervious surface mapping, impacts of impervious surfaces on watersheds with particular reference to geomorphic impact and water quality impact were reviewed and presented in this chapter.
Methods: Describes how the impervious surface areas were mapped base on the theoretical framework of the chosen method in the previous chapter.
Results and analysis: Shows the results of the data analysis carried out for the study area, results of impervious surface area covered by each watershed, annual growth rate of impervious surfaces, watershed categorization base on impervious surface cover were compared and discussed to highlight the spatiotemporal differences over the time periods.
Conclusion: This chapter concludes on the spatiotemporal changes in watershed category of the study area, provides an overview of the limitation of the method used and recommendations for further development.
There are two major indices commonly used by Hydrologic researchers in measuring impervious surface cover total impervious area (TIA) or effective impervious area (EIA). TIA is a measure of the area of a watershed covered by all connected and disconnected surfaces that transport water over land to the drainage system. While EIA only measures the impervious surfaces that directly contribute to the drainage system (connected surfaces) (Young, 2010). Conceptually, EIA captures the hydrologic significance of impervious surfaces, thus making it the preferred parameter of urban development for use in hydrologic modeling (Booth and Jackson, 1997).
Studies have shown EIA and TIA to be equally accurate indices of urban induced hydrologic alterations (Sutherland, 1995; Booth et al., 2002; Lee and Heaney, 2003). However, EIA is rarely used in fluvial research focused on the geomorphic changes associated with urban development, in part, due to the difficulty of determining EIA, which requires substantial financial, human, and time obligations that detract from the main research priorities. Previous research concerned with urban-induced stream alterations have shown TIA to be a reliable indicator of urban development that yields consistent and accurate results (May et al., 1997; Paul and Meyer, 2001; Booth et al., 2002; Chin, 2006; Cianfrani et al., 2006). For these reasons TIA was selected as the index by which to represent the degree of urban development within drainage basins for this research.
Techniques of impervious surface mapping
Multi Endmember Spectral Mixture Analysis Method (MEMSA)
MEMSA is a more advance approach to spectral mixture analysis which is a method of modelling the sum of pure spectra called endmembers each weighted by the fraction of endmember required to produce the spectral mixture . While SMA is a powerful approach it fails to account for pixel scale variability in spectral dimensionality, spectral degeneracy between material and natural variation in spectra of most materials. MEMSA extends SMA by allowing the number and types of endmembers to vary on a per pixel basis (Robert et al, 1998).
MEMSA involves developing a spectral library, the unmixing an image using every possible combination of two, three, four endmembers applied to each pixel. With this approach, more materials can be mapped out from an image compared to SMA. In essence minimizing pixel scale fraction errors by selecting the best fit model for each pixel.
In order to map impervious surfaces (Weng, 2012) adopted the following procedures (1) endmember selection- by (purity pixel index (PPI) and manual selection. (2) Building a spectral library. (3) Impervious surface estimation Endmembers for constructing a spectral library in MESMA.
First, Purity Pixel Index (PPI) and manual selection were used. Second, The MESMA model (based on all combinations of final library endmembers) was applied to every pixel in the image, based on a set of criteria. Third, after satisfying the ideal criteria, the best-fit model were selected for each pixel. Then finally a Fraction cover derived from MESMA was produced.
Successful endmember selection For MESMA relies on identifying a series of spectra that represent the spectral variations for each material in the scene (Roberts et al., 1998; Okin et al., 2001). There are two sources of pure spectral signatures: (1) image endmember, selected from representative pixels from satellite sensor images (Elmore et al., 2000; Small, 2001; Song, 2005), and (2) reference endmember, collected from reference images, and measured in the laboratory or on the ground with a spectroradiometer (Smith et al., 1990; Adams et al., 1995; Roberts et al., 1999). In general, reference endmembers are of higher purity than image endmembers. Thus reference endmember spectra can be the optimal ones as the candidate endmembers for each group of interest, but they may not be appropriate for those interests within the image (Ballantine et al., 2005).
In this study, the endmembers were selected from images based on the PPI method and manual selection. Appropriate endmembers were effectively found by the PPI method, which identifies extreme pixels, or representative pure signatures, in data by searching for a set of vertices on a convex hull (Boardman et al., 1995; Chaudhry et al., 2006). Image endmembers can be collected by linking the PPI image with the image used to identify classes, avoiding the choice of edge pixels.
Building the spectral library
In building a spectral library for MESMA, the spectral library should contain enough spectra for each class of materials to represent spectral variation of materials on the ground; however, the number of endmembers in the library cannot be too great, because the total number of endmembers in the potential models is inversely proportional to the computational efficiency and accuracy (Powell et al., 2007).
In order to construct a comprehensive and representative spectral endmember library, Dennison and Roberts (2003) developed the following procedure (1) All possible endmembers should be acquired to augment the number of spectra in the library. (2) A series of representative spectra for each class of materials be selected to reduce computation errors and complexity by eliminating spectra from other material classes. (3) Endmembers that represent materials on the ground in the area of interest are selected for use with MESMA. (Timothy, 2011) identified a collection five different types of candidate image endmembers spectra, which included: Vegetation A, Vegetation B, High albedo, Low albedo, and Soil. The collection of endmember spectra were further reduced to five classes of optimal endmembers using advance image analysis software for the final spectral library of MEMSA.
The final spectral library for MESMA was used to map a model of Vegetation-Impervious-Surface fractions in the study area. Where shade-endmember in the model was converted to fractions that represent the physical composition of the materials in each pixel. After shade normalization, tree canopy, shrubs and grass were combined into one vegetation class, and then low albedo impervious surface and high albedo impervious surface were combined into one impervious surface class. The sum of the fraction values of vegetation, impervious surface, and soil were equal to 1. The fraction value of each class within a pixel was between 0 and 1. Water pixels, already masked out in a prior step, were assigned a fraction value representative of water. Finally, a map comprised by four classes: vegetation, impervious surface, soil, and water was finally produced.
Normalized Difference Impervious Surface Index
Various normalized difference indices have been developed and intensively used since the development of NDVI of rouse et al (1973). Present remote sensing methods used in estimating ISA are complicated with low degree of automation, most of these methods require some preprocessing.
(Xu, 2010) developed the normalized difference impervious index to enhance impervious estimation. The NDISI also adopts some basic ideas of previous research on normalized difference indices. In creating a normalized difference indices strong and weak bands of the land cover of interest are looked for in multispectral band to be enhanced, using the strongest reflectance band as the nominator and the weakest one as denominator. The normalized ratio of the two bands greatly enhances the strongest between the area of interest and background noise.
Most present impervious mapping methods are based on rids V-I-S model. However ignoring water and soil, one of the important components of urban ecosystem, in methods has brought problem to these designed to methods (Xu, 2007). Therefore his new normalized index is designed to take care of the water and soil feature problem associated with mapping impervious surfaces.
According to (Xu, 2010) through inspection of spectral characteristics of major impervious surface types, he discovered that impervious surfaces generally have high emittance in the thermal band TR and low reflectance in near infra-red band (NIR), because impervious materials such as concrete and asphalt have a strong capability of emitting heat but do not allow vegetation’s to grow of them. Therefore a ratio of thermal band to NIR will enhance impervious features. However (Xu, 2010) realized that soil, sand and water possessed similar spectral feature, therefore adopting the thermal to NIR band ratio alone wasn’t going to effectively enhance impervious surface, the similarity in spectral signature could create a mix in impervious surface information.
In order to avoid a potential spectral signature mix up. He developed the normalized impervious surface equation as follows.
Abbildung in dieser Leseprobe nicht enthalten
Where NIR and MIR1 are near-infra-red and mid-infra-red bands respectively such as and 4 and 5 in ETM+, TIR (thermal-infra-red) as band 6 in ETM+ and WI represented as NDWI of Mc Feeters (1996) or the MNDWI of Xu (2006), where
Abbildung in dieser Leseprobe nicht enthalten
Abbildung in dieser Leseprobe nicht enthalten
Green is represented as the green band, such as band 2 in ETM+.
Dividing the index by 3 is to avoid small values for the index, in order to have an index result with values ranging between -1 to 1.
NDSI stands out as the most preferred choice of estimating impervious surfaces as it appears to solve the problem of spectral confusion associated with water feature ambiguity with impervious surface mapping. However NDISI is not designed to handle earlier landsat sensors such as the multispectral Scanner (MSS) that has four bands unlike the more recent thematic mapper (TM) and enhanced thematic mapper (ETM) with medium infrared (MIR) and (TIR) band as required by NDSI equation. For a spatiotemporal study that requires the use of earlier Landsat sensors, this method will not be suitable.
Estimating impervious surfaces employs the use of detailed maps. Investigators such as Martens (1968), Southard (1987), and Spencer and Alexander (1978) have all attempted to extract impervious surfaces estimations from detailed maps by overlaying them with grid frames from which to identify and quantify impervious surfaces. However, because maps are mere representations or models of reality, the identification and estimation of impervious surfaces is limited to the types of information and level of detail the maps display (Johnson, 2004). This level of detail, generally referred to as scale, can be defined as the proportional distance between what is represented on the map versus what is reality (Avery and Berlin 1992). A small-scale map covers a large geographic area with less detail as compared to a large-scale map covering a smaller geographic area with more detail. As a result, the ability to extract accurate impervious surfaces estimates from detailed maps is a process that is highly dependent on the information represented and the scale at which it is displayed.
Soil Adjusted Vegetation index (SAVI) Approach
The Soil-Adjusted Vegetation Index (SAVI) is an improvement on the normalized difference vegetation index (NDVI) which detects the presence of healthy or unhealthy vegetation. Unlike SAVI the effect of bare soil brightness on NDVI in areas with less vegetation cover has the potential of influencing the vegetation index values. Hence the creation of SAVI by (Huete, A. R., 1988) using a correction factor “L” to suppress the effect of soil brightness on vegetation index values.
The correction factor L varies depending on the amount of vegetation in the area. The general rule is in areas with no green vegetation cover L=1; in areas of moderate green vegetative cover, L=0.5; and in areas with very high vegetation cover, L=0. The results of the vegetation index for SAVI range between -1 to 1.
SAVI = ((NIR - Red) / (NIR + Red + L)) x (1 + L)
NIR and Red represent the near infra-red and visible electromagnetic spectrum respectively. These bands are used to detect whether a vegetation is healthy or not. In essence the bands are used to detect the level of greenness in a vegetation.
Like NDVI, SAVI could be used to detect impervious surfaces better by taking advantage of the soil correction factor introduced into SAVI. Unlike SAVI, NDVI has been widely used in impervious surface Gillies, R.R, et al 2003; Dougherty, M, et al 2004 ; Jantz, P et al 2005 Xian, G.; Crane, M. 2005 ; Shahtahmassebi, A.; et al 2012 .
Impervious surfaces and urban watershed characteristic
Geomorphic Impact of Impervious surfaces
Urban watersheds show greater surface runoff than comparable non-urban watersheds, and this is attributed to the differences in impervious surface cover in the two settings (Booth, 1991; Horner et al., 1999). Peak stream flows following storm events increase with urbanization while low flows between storms decrease because less water is able to infiltrate the ground (Leopold, 1968). Less infiltration means less ground-water discharge, that is, base flow, into any given stream channel. This subsurface flow of water to the channel is relatively slow. Urbanization disturbs the pre-urban equilibrium between hydrology and sediment yield and this results in geomorphic alterations of stream channels (Hammer, 1972; Morisawa and LaFlure, 1979; Pizzuto et al., 2000). The volume of sediment supplied to a stream changes as a function of the degree of watershed development, as suggested by Wolman (1967) and upheld by Chin (2006) and Grable and Harden (2006), occurring in three distinct stages: 1) a stable or equilibrium pre-urban stage; 2) a construction period exposing large areas of bare land; and 3) a final, new urban landscape dominated by impervious surfaces.
During the construction phase of development, large expanses of land are stripped and left bare for extended periods of time inducing widespread erosion of the surrounding landscape. This erosion supplies a greater volume of sediment to a stream than in the pre-urban stage, leading to short-term aggradation. Sediment yield during the construction phase has been shown to be as high as 40,000 times greater than pre-urban rates (Harbor, 1999). Deposition and bed aggradation result in overall channel capacity reduction as channel width and depth decrease. As land development continues, impervious surfaces begin to cover the majority of the landscape, through the construction of roads, buildings, parking lots, and other cultural features. Compared to pre-urban conditions, these impervious surfaces generate a larger amount of surface runoff, which is directed into the sediment choked stream channels as a greater discharge capable of eroding the recently deposited sediment.
When fully urbanized, a watershed is unable to supply much sediment to its stream channels as the majority of land is now covered by impervious surfaces. In response to the increased water input and decreased sediment input the stream erodes its channel bed and banks. This channel erosion may be gradual or rapid depending on the geologic setting and whether human manipulation of the stream takes place (Booth, 1991). Urbanization increases downstream flood activity, impervious surface cover greatly diminishes the infiltration of water and subsurface base flow, thereby increasing the amount of water transported as surface runoff.
With water input to the stream no longer slowed by the process of ground seepage, the time it takes for precipitation to enter any given stream channel (lag time) decreases, resulting in a flashier hydrology (Morisawa and LaFlure, 1979; Chin and Gregory, 2001). Akintola (1994) in a study of the infiltration process in Ibadan metropolis, Nigeria from 1965 to 1994 confirmed also varying rates of infiltration for the different type of urban land use surfaces.
(Russell, 2010) created a model to simulate surface runoffs on urbanized watershed of Chicago, Illinois. Comparing simulated results executed with land cover data of 1992 and 2001, the simulation indicated that runoff was faster in 2001 than in 1992, reason being that there are more urbanized areas 2001 than in 1992.
Greater surface runoff combined with the shorter lag time in urban watersheds increases flood frequency and flood magnitude. Research has shown that flood recurrence in urban watersheds can increase by as much as five times while flood volume can be six times higher than a nonurban channel of a similar drainage density (Leopold, 1968).
Seaburn’s (1965) research found that flood discharges were at least 250% higher in urban catchments when compared to similar non-urban catchments. Current research across geographic regions supports this concept; annual flood peaks are amplified by 22 to 84% in urban watersheds compared to watersheds covered by more natural land cover (Poff et al., 2006). Booth and Jackson (1997) found that the discharge of the 2-year flood in urban catchments equals the discharge of the 10-year flood in similar forested catchments.
The urban hydrologic regime is comprised of water entering the stream system at a faster rate than was the case for the pre-urban hydrology, thereby increasing mean and peak discharge. Changes to the hydrologic regime, whether through land use change or otherwise, force stream channels to adjust to accommodate the new flow. Increased flood activity is a stream’s natural response to high stream velocity and peak streamflows that exceed original, and construction phase, stream capacity. However, the amount of time needed for a stream to adjust properly to an urban hydrologic regime remains unclear.
Increased flood activity should not continue indefinitely as flooding will force the stream channel to adjust until it is capable of handling higher peak streamflows. At the streamreach level, channel slope should decrease as the increased elevation is no longer needed to create energy for the system. The modified channel form, then, is a function of the magnitude of urban development and the length of time a watershed has been subject to urban development.