Effects of Increasing Impervious Surface on Water Quality in Ile-Ife Urban Watershed, Southwestern Nigeria ()
1. Introduction
Globally, the urban environment has continued to experience land cover changes with the exchange of vegetation and water to impervious surfaces (Hua et al., 2020) and thereby increasing vegetation loss and highly impaired water bodies (Kaufmann et al., 2007; Xu et al., 2012; Ige-Olumide & Salami, 2018). The developing regions of the world most especially are extensively experiencing urbanization and each of its major regions will hold more cities than rural dwellers by the year 2030 as projected by the United Nations (UN) Population Division. It is also estimated that by 2050, two-thirds of the inhabitants in developing countries are probably going to live in metropolitan areas (Montgomery, 2008). Wu (2014) and United Nations (2013) emphasized that there has been a phenomenal population expansion in urban communities around the world (Farrell, 2017; Ritchie & Roser, 2018; Kuddus et al., 2020; Hua et al., 2020) which has prompted urban landscape changes. The population changes due to urbanization are increasing massively in developing regions and this exerts pressure on natural resources (Wang et al., 2018; Wang et al., 2020). This leads to the loss of forests, productive agricultural land, shrinking/loss of surface water bodies, other green open spaces, and perhaps irreversible damage to the earth’s ecological and environmental systems (Couch et al., 2005; Elmqvist et al., 2016).
Changes in these land use forms or initial cover are identified as some of the effects of urbanization (Mugiraneza et al., 2019a; Fernández-Götz et al., 2014). The erection of high-rise buildings, and the conversion of towns to urban areas and megacities are presently in operation, especially where there is a sporadic population increase. Urbanization has led to several ecological challenges which transcend from small to large scale (Zhao et al., 2006), incorporating various pollution types and reduced water availability (Shao et al., 2006), increased energy demands and nearby atmosphere modification (Zhou et al., 2004; González et al., 2005), inadequate infrastructures in urban areas (Jago-on et al., 2009), and a substantial loss of forest resources and carbon stock (Fang et al., 2003; Yuan, 2008). Increased urban growth causing negative changes to the ecological environment affects the quality of life which is most evident in the developing world without adaptative mechanisms (Rahman et al., 2011). United Nations (UN) (2013) established that the Sub-Saharan population increase will be roughly from 769 million to about 1.69 billion which is attributed to urban sprawl occurring in small and big cities. Given these projections, it is proposed that urban communities in developing nations should build impenetrable surface regions to sustain the present population expansion (United Nations (UN) (2013). Urban areas are becoming precipitous around the world, and the effect of urbanization will always bring along with it global, regional and local changes consistently. Urban expansion cannot be overlooked in LULC research (Rana, 2011). Developing countries are experiencing an unprecedented increase in population and urban expansion (United Nations (UN), 2013), which increases the percentage of impervious surfaces and probably elongates the severity of its impact on the copying capacity of the world (Raddad et al., 2010).
Impervious surface (IS) which is described as an urban area indicator from history is presently regarded as an environmental factor (Xu et al., 2012; Gillies et al., 2003; Arnold & Gibbons, 1996) used to address the various framework of complex environmental issues and predominately in urban and semi-urban areas. This ranges from its impact on water quality (Brabec et al., 2002; Arnold & Gibbons, 1996; Fu et al., 2019), rainfall run-off (Lohani et al., 2002; Aladejana, 2021; Brun & Band, 2000), and the urban heat island effect (Weng et al., 2004; Chen et al., 2006; Yuan & Bauer, 2007; Ige-Olumide & Garuba, 2018). Many land use planners and water resource stakeholders face extreme challenges of uncontrolled urban growth/sprawl and sometimes unmanaged construction development and their deleterious impacts on water quality (Kauffman & Brant, 2000). They emphasized that watershed impervious cover threshold or hotspot can be utilized to address growth into watershed zoning districts by identifying locations where the least impact on stream water quality will be recorded due to development. As the imperviousness of a watershed increases, the greater volume of stormwater increases the possibility of flooding (Fabiyi, 2006; Adetoro & Salami, 2018; Aderogba, 2012; Weng & Lu, 2008) and reduces the potential for pollutants to settle out; meaning that more pollution is delivered to drinking water streams and aquifers (Zhang, 2012). Too much paving and hardening of a watershed can reduce infiltration and groundwater levels which in turn can decrease the availability of aquifers, streams and rivers for drinking water supplies.
Over the years, impenetrable surfaces are being used as one of the key indicators for city climate transformation (Yuan & Bauer, 2007; Morabito et al., 2018; Ige-Olumide & Garuba, 2018). Impervious surface is recognized as a key factor in the growth, monitoring and strategic planning of the watershed because of the effect on the health of the ecosystem (Arnold & Gibbons, 1996; Wu, 2009; Sugg et al., 2014). The advent of urban sprawl has resulted in the increase and number of impenetrable surfaces such as concretes, asphalts, etc. and a decrease in the extent of vegetated lands, wetlands, and other forms of open space that ingest and clean stormwater within a natural framework. Impervious surface (IS) coverage can be selected as a key and quantifiable factor/indicator of water quality (Kauffman & Brant, 2000). Increasing amounts of impervious cover contribute significantly to too many of the water quality and quantity issues facing cities today (Flinker, 2010; Wu, 2013; Wang et al., 2018). These surfaces prevent water from infiltrating into the ground and creating stormwater runoff (Millennium Ecosystem Assessment (M.E.A.), 2005; Raudsepp-Hearne et al., 2011). This runoff then carries organic matter, fertilizers, pesticides, oil and grease and other contaminants, directly into streams, water bodies, and local water supplies. Additionally, because water runs quickly off these surfaces, the quantity and velocity of runoff are increased, the physical structure of streams becomes altered and eroded, and there is a greater likelihood of more frequent and larger floods time (Daramola & Ibem, 2010; Wang et al., 2018; Aderogba, 2012). The degree of damage to watershed health depends on many factors: the nature of existing land uses or the topography, soils, and vegetation (Furniss et al., 2010; Ahn & Kim, 2017). For this reason, it can be difficult to attribute environmental impairment to just one source and therefore often hard to prevent or predict (Hassan & Lee, 2015). However, impervious cover, while perhaps not always the direct cause of environmental impairments, is a good representative substitute.
Numerous studies have demonstrated a correlation between impervious cover levels and the environmental integrity of streams, and their watersheds and this correlation can then be used to predict and manage both water quality and watershed health (Xian & Crane, 2006; Weng & Lu, 2008; Jacobson, 2011; Liu et al., 2012). Therefore, it is crucial to understand the specific percentage of IS extent linked with various urban and suburban land uses/cover. The main concerns in this study are those attributed to the health of water resources. The rate of change in the impervious surface in a settlement can be a reflector of the composition of the landscape (biotic, hydrologic and geomorphic) of the area (Paul & Meyer, 2001). The magnitude, location and spatial pattern of impervious surfaces induce many physical and biological changes affecting the urban watershed.
Geographic information science and remote sensing display higher data source potential adopted for the extraction of IS from global to local scales (Cai et al., 2016). Medium spatial resolution images such as Landsat, Enhanced Thematic Mapper Plus (ETM+) images, Operational Land Imager (OLI), Satellite Pour l’Observation de la Terre (SPOT), and China-Brazil Earth Resources Satellite (CBERS) images and ASTER images have been utilized globally in IM extraction (Yang et al., 2003; Pu et al., 2008; Powell et al., 2007; Zhang et al., 2009; Luo & Mountrakis, 2010; Lu et al., 2011; Zhang et al., 2014; Parece & Campbell, 2013; Cai et al., 2016). Low to medium resolution is considered very coarse for visible extraction analysis because of the heterogeneity in the urban landscape and the complexity of urban IS materials (Jensen & Cowen, 1999; Lu & Weng, 2004; Zhang et al., 2014). Due to challenges such as low accuracy (Wu & Murray, 2003; Lu & Weng, 2006) attributed to the mixed pixel classification due to low spatial resolution and the heterogeneity of urban features (Cai et al., 2016), spectral mixture analysis has been widely adopted for deriving IS. In recent times, high resolution (HR) images where mixed pixels are highly reduced such as those from WorldView-2 Ikonos, and GeoEye-1 satellites, adopted in the derivation IS have revealed more reliable results (Cai et al., 2016; Li et al., 2011). Various techniques are considered for the extraction of the impervious surface using very high-resolution images. Very high-resolution images, which include sub-pixel analysis (Ji & Jensen, 1999; Lu & Weng, 2009; Sun et al., 2020), the pixel-based method (Wu & Murray, 2003; Lu et al., 2011; Xu & Tang, 2013; Wei & Blaschke, 2018), the object-based method (Blaschke, 2010; Hu & Weng, 2011; Liu et al., 2020) and the pixel- and object-based hybrid method (Li et al., 2013; Zhang et al., 2014; Wei & Blaschke, 2018). Many studies have demonstrated the capability of object-based techniques to produce a better result than the pixel-based method or sub-pixel analysis (Blaschke, 2010; Lu et al., 2011; Wei & Blaschke, 2018). The object-based method in recent times is recognized to be effective for IS mapping from very high-resolution images (VHR) (Zhou & Troy, 2008; Blaschke, 2010; Wei & Blaschke, 2018). However, other problems are identified in IS derivation using very high resolution images. Most of this involves spectral challenges between IS and other classes. The reason for this is attributed to the low spectral resolution of very high resolution images and the high spectral variation of impervious surfaces; noise (shadows) from high objects, such as built ups and natural canopies (Hu & Weng, 2011). More classification of the grey portion is mostly considered as an option in other works to resolve this problem (Blaschke, 2010; Li et al., 2011; Wei & Blaschke, 2018; Liu et al., 2020). The recent use of Unmanned aerial vehicles/systems (UAV/S) to provide aerial images to support ground-truthing for accuracy assessment is of recent use (Feng et al., 2015; Popoola et al., 2016; Kalantar et al., 2017; Yao et al., 2019) to address some of these identified challenges and improve accuracy.
Adequate clarity on impervious surface changes is key to various ecological applications, for example, assessment of the impact of a watershed (Jennings & Jamagin, 2002), precipitation runoff volume, intensity, and length (Lohani et al., 2002), groundwater revive and base flow storm analysis, storm flow and flood recurrence (Brun & Band, 2000). Thus, producing a timely and accurate impervious surface distribution map is crucial to various challenges and matters related to worldwide ecological change. The overall objective of this study was to OBIA for the extraction of impervious surfaces from two VHR images of different years in an urban watershed and to investigate the effect of impervious surfaces on water quality using integrated data in the impervious surface analyst tool (ISAT).
2. Materials and Methods
2.1. Study Area
The study area is the urban settlement of Ile-Ife which is an ancient Yoruba town in southwestern Nigeria. The city lies between Latitudes 7˚28'N and 7˚45'N and Longitudes 4˚30'E and 4˚34'E as described in Figure 1. The city is located right at the center of the Yoruba-speaking states in West Africa and about 200 km NE of Lagos, which is Nigeria’s coastal capital city for over a century (Olupona, 2011). The rate of urbanization of Ile-Ife is dated back to approximately 500 ADS (Mabogunje, 1968, 1980) and has a social-cultural group known as the Yoruba ethnic group (one of the largest ethnic groups in Africa) (Levinson, 1998; Olupona, 2011). It is one of the popular towns of Osun state, extending over parts of Ife Central, East and North Local Government areas with a population of 644,373 (NPC, 2015).
![]()
Figure 1. Urban Area of Ile-Ife, Southwestern Nigeria, West Africa.
2.2. Data Sources and Sample Collection
To delineate the urban area into the watershed, the Advanced Spaceborne Thermal Emission and Reflection and Radiometer (ASTER) Global Digital Elevation Model (DEM) of 30 m spatial resolution from the archive of the U.S National Aeronautics and Space Administration (NASA). This is important for geospatial assessment and applications for catchment delineation, flood risk assessment, and the estimation of the transfer of pollutants and essential nutrients (Gaafar et al., 2020; Giannoni et al., 2005). The WorldView-2 image was acquired from the Digital Globe Foundation archive. The WorldView-2 sensor provides a high-resolution panchromatic band and eight multispectral bands (Nouri et al., 2014; Park et al., 2020): four standard colors (red, green, blue, and near-infrared 1) and four new bands (coastal, yellow, red edge, and near-infrared 2). The worldview provides 0.46 m panchromatic mono and stereo satellite image data (Nouri et al., 2014). The IKONOS image was also acquired from the archive of Space Application and Environmental Science Laboratory, Obafemi Awolowo University. It’s a commercial Earth observation satellite at 1 - 4-meter resolution (Fraser et al., 2002) that collects multispectral and panchromatic imagery. The population density was extracted from the ward map featuring the three local governments that fall within the watershed under study namely, Ife Central local government, Ife East local government and Ife North local government. Table 1 provides a summary of the data used and the relevance for the study.
2.3. Field Survey and Accuracy Assessment
Ground truthing and field surveys using GPS receiver and recently archived UAV aerial images were utilized to facilitate more accurate interpretation of the satellite imageries, identify and verify important areas and other features as well as the classification accuracy assessment points on the ground and other salient features. Several points of interest were identified on the topographic map for easy navigation and identification of the area of interest on the ground. Areas of interest were identified on the field and their spatial and attribute information was recorded with a GPS and a field note. Most areas of interest on the map were accessible and this provided a better visual understanding and interpretation of the images during classification. After satisfactory visual interpretation of the images, the following land use and land cover types were identified and the layers created: Built-ups, roads, vegetation and water body. The Built-up comprises buildings and artificial structures. The vegetation represents areas covered by trees of different species and sizes, shrubs and woody plants, etc. water body refers to ground surface water. The visual interpretation was done employing the interpretation keys such as size, shape, texture, pattern, tone, color and association of features. Accuracy Assessment as reflected in most existing studies that used very high resolution (VHR) images for impervious surface extraction (e.g., Lu & Weng, 2009; Lu et al., 2011; Zhang et al., 2014; Sugg et al., 2014), reference data for both training and validation were mainly collected by visual interpretation of aerial photography acquired by low altitude UAV (unmanned aerial vehicle), VHR WorldView-2 images, and IKONOS images. Those samples of which the attributes are difficult to determine on remotely sensed data (e.g., the samples in shaded areas) were checked from the ground survey and acquired UAV aerial image. A ground survey was subjected to a probability sampling protocol involving stratified random sampling (Henrys & Jarvis, 2019) which was used to identify the locations at which validation points were obtained. This sampling method has been used widely and shown to improve sampling efficiency by reducing inconsistency within a stratum and increasing variability between strata (Jin et al., 2014).
2.4. Methodological Approach
![]()
Table 1. Datasets, characteristics, sources and relevance to the study.
2.5. Urban Watershed Delineation
The urban watershed was delineated from the Digital Elevation Model (DEM) using ArcSWAT toolbox. The DEM was set to projected coordinate system using the Universal Transverse Mercator (UTM) projection with World Geodetic System (WGS) 1984 datum and zone 31N using ArcGIS 10.3. The DEM was supplied to the Automatic Watershed Delineator in ArcSWAT (Luo et al., 2011) with a mask of the extent of Landsat images of 190/55 and 191/55 combined. Flow direction and Flow accumulation were generated. Stream network and outlet creation followed. Outlet points were selected along the stream network to delineate the watershed (Lindsay et al., 2008; Tarboton, 1997). The fill dem was carried out to remove the depression from an elevation raster and a common method for removing a depression was applied which involves increasing the cell value to the lowest overflow point out of the sink (Jenson & Domingue, 1998). The flat surface resulting from sink filling was interpreted to define the drainage flow. This was achieved by imposing two shallow gradients and forcing flow away from higher terrain surroundings the flat surface towards the edge bordering lower terrain (Garbrecht & Martz, 1997). The flow direction raster was carried out. Flow direction raster was carried out using a single D8. The cells’ flow direction was assigned to one of the eight surrounding cells that has the steepest distance-weighted gradient (O’Callaghan & Mark, 1984; Tarboton, 1997). The flow accumulation raster tabulates for each cell the number of cells that will flow to it. The tabulation is based on the flow direction raster. The flow accumulation raster records how many upstream cells will contribute drainage to each cell (Zhou et al., 2019).
2.6. Image Processing and Impervious Surface Extraction
This is an important phase in satellite image processing and analysis, which has an impact on all other subsequent procedures and results. The imageries were first pre-processed for geometric rectification. The image bands used for the study were geometrically rectified to the Geographic Coordinate System of the study area; WGS_84_UTM zone 31N. The shapefile of the study area was overlaid on the WorldView-2 satellite image and appropriate scenes of the image within the shapefile were selected. Each of the selected scenes was mosaicked to achieve a single image for further delineation. The shapefile of the study area was clipped from the mosaicked image (Blaschke et al., 2015) and gridlines were created over the scenes, numbering them into smaller scenes for easy use on the eCognition software.
2.6.1. Object-Oriented Classification
Image classification for the high-resolution satellite images was carried out on the eCognition developer software version 9.0. eCognition Developer is a development environment for object-based image analysis. It is used in earth science to develop rule sets (or applications) for the analysis of remote sensing data. The first step to image classification on eCognition was multiresolution segmentation. Image objects were extracted at modifiable scale parameters, single layer weights and the mixing of the homogeneity criterion concerning color and shape. The multi-Resolution segmentation (MRS) was adopted for this study which is a widely utilized algorithm embedded into ecognition (version 9.0, Trimble Germany GnbH, Munich, Germany, 2014). Bottom-up region (Fu et al., 2019; Blaschke, 2010) that uses a step wisely merges a small image into a larger image object until heterogeneity is reached. MRS was performed on the image using a scale parameter of 20, the domain was at the pixel level, while the composition of homogeneity criterion (Johnson & Xie, 2011) was applied also, with the shape at 0.1 and compactness at 0.5. This was done to ensure better segmentation thereby enhancing classification (Makinde et al., 2016; Russakovsky et al., 2015; Fu et al., 2019). The outcome of the segmentation is subject to several adjustable criteria—scale parameter, domain and homogeneity criterion (Blaschke, 2010; Fu et al., 2019). In particular, the scale parameter is a measure of the maximum change in heterogeneity that may occur when merging two image objects. A larger-scale parameter value leads to bigger objects and vice versa.
2.6.2. Arithmetic Feature
Arithmetic features are composed of existing features, variables, and constants, which are combined via arithmetic operations. Arithmetic features are composed of multiple features but apply only to a single object. After the multiresolution segmentation, an arithmetic feature was calculated, known as the Greenness Index. This is the ratio between the green band and the sum of the red band and the blue band.
According to the range of values gotten the boundary was 0.55, any object above 0.55 was vegetation and any object below 0.55 was not vegetation. The information derived from the greenness index aided in assigning classes. The features were assigned class names from the image object domain to a class array. The class was assigned based on the existing information, any object that has an arithmetic feature greater than 0.55 was classified as vegetation and any object that has an arithmetic feature lesser than 0.55 remained unclassified (Mugiraneza et al., 2019b; Makinde et al., 2016). This assisted in the separation of vegetation from other image objects. The shape index was used to determine the characteristics of other objects in the image. The Shape index describes the smoothness of an image object border. The smoother the border of an image object is, the lower its shape index (Lemenkova, 2015; Ikokou & Smit, 2013). From the value range of the shape index, objects greater than 3 were classified as roads and objects less than 3 belonged to other classes (Blaschke, 2010; Lemenkova, 2015). The relative border which describes the ratio of the shared border length of an image object (with a neighboring image object assigned to a defined class) to the total border length was also used (Flanders et al., 2003; Yang et al., 2019; Espriella et al., 2020). If the relative border of an image object to image objects of a certain class is 1, the image object is embedded in them.The relative border was calculated from the feature characteristics and objects with a relative border value of 1 were wrongly classified. The relative border was applied to misclassified objects to help recover them to the appropriate class. Furthermore, four classes were identified—Vegetation, Waterbody, Road, and Built-up. Two major classes (Impervious and Pervious) were further created from the classes above, with bare ground, roads and built up as Impervious and vegetation and water body as Pervious. All the feature objects in the same class were merged into a single multi-part polygon (impervious and pervious). The results were exported to a shapefile and opened on ArcGIS. On ArcGIS the study area (Ife urban shapefile) was clipped. Add geometry tool was used to calculate the total area of impervious cover and pervious covers in Square Meters which was eventually converted to Square Kilometers.
2.6.3. Percentage Imperviousness of the Urban Watershed
The percentage of the impervious surface area (PISA), which is defined as the ratio of impervious surface area to the total surface area of the watershed, has for many years been regarded as one of the useful indicators to measure the impacts of land development on the watershed environment.
2.6.4. Rate of Change in the Urban Watershed
The rate of change in the watershed between 2008-2016, was calculated using the formula;
2.6.5. Effect of Imperviousness in the Urban Watershed
The effect of imperviousness in the watershed was studied for an important phenomenon in the environment which is Water Quality.
2.7. Impervious Surface Analyst Tool (ISAT)
The effect of imperviousness on water quality was determined using the Impervious Surface Analysis Tool (ISAT), which is available as a geographic information system script. It was used to calculate the percentage of impervious surface area within the study area. Information derived from ISAT was used to predict the different impacts on water quality. ISAT was developed by the National Oceanic and Atmospheric Administration (NOAA) Coastal Services, it is an American scientific agency within the United States Department of Commerce. The specific data utilized for the successful running of the ISAT tool includes Population Density Map; Land Cover Grid- land cover data; Hydrological Unit and Impervious Surface Coefficient
2.7.1. Population Density Map
The population density map was derived by getting the ward map from three local governments in the study area.
● Ife Central local government
● Ife East local government
● Ife North local government
These maps respectively were subjected to further geoprocessing. The local government population values were resampled to the ward-level population using the number of buildings in each ward (simple relational association between building and population values). The number of buildings in each ward was related to the number of building in that same local government to derive the population for each ward, the population density was calculated by dividing by the area of each population value.
2.7.2. Water Quality Analysis
Water is a dynamic medium and its quality varies spatially and temporally. The change in the impervious-pervious stability has altered the stream water quality (Brabec et al., 2002; Kim et al., 2016). To accurately determine the effect of imperviousness on water quality it was important to carry out a water quality analysis in the study area (Burton & Pitt, 2002). Areas of sampling were carried out based on the result generated from the Impervious Surface Analyst Tool (ISAT). The parameters sampled were dissolved oxygen, biological oxygen demand, temperature, total dissolved solids, electrical conductivity and turbidity. The TDS meter, Model YL-TDS2-A which was first subjected to calibration (United States Environmental Protection Agency (USEPA), 2003; ISO, 2016) was used to test for total dissolved solids, electrical conductivity and temperature. The temperature measurement was carried out by using a portion of the water sample (1liter) and immersing the multi-parameter meter, Model YL-TDS2-A into it for a sufficient period until stabilization was reached (Kim et al., 2016) and reading was taken, expressed as ˚C. The electrical conductivity was achieved using the TDS meter, Model YL-TDS2-A which was dipped into the sample, and the readings were noted for the stable value shown as µs/cm. Winkler’s method was adopted for the measurement of dissolved oxygen (DO). DO bottles-300 ml capacity was used to collect the water sample. Manganous sulphate (2 ml) and 2 ml of potassium iodide were added and sealed. This is properly mixed and the precipitate was allowed to settle down. 2 ml of conc. Sulphuric acid was added and was adequately mixed until all the precipitate dissolved. Also, 203 ml of the sample is measured into the conical flask and titrated against 0.025N sodium thiosulphate using starch as an indicator. The endpoint is the change of color from blue to colorless. The biological oxygen demand was achieved by filling water samples into an airtight BOD bottle (Kim et al., 2016) and was incubated at room temperature for 5 days. The difference in the dissolved oxygen measured initially and after incubation gives the BOD of the sample. The total dissolved solid was derived by immersing the TDS meter into the sample for a sufficient period (till the reading stabilized) and the reading was taken and expressed as ppm (parts per million). Lastly for the sampling is the turbidity, which was determined using the instrumental method, the spectrophotometer at a wavelength of 430. The water samples were thoroughly shaken in the nephelometric tube and the values were recorded appropriately.
2.7.3. Hydrological Unit
The hydrological unit was derived from the delineation process carried out.
2.7.4. Land Cover Grid-Land Cover Data
The land use land cover map for the area was derived using a 2016 Planet Scope 3 meters multispectral satellite image in line with World view-2 image as recent images were required for accuracy purposes. The Planet scope image was classified using unsupervised classification, relating it to the Worldview image the following classes were identified: Built up, Vegetation, Waterbody, Road and Bare ground. This gave the land use land cover requirement for ISAT.
2.7.5. Impervious Surface Coefficients
The impervious coefficient was derived from the land use using the map algebra on ArcGIS. The impervious coefficient was determined based on the land use type as shown in Table 2 below.
These values were generated based on the nature of the land use. The map algebra tool was used to calculate the value for each land use type. An excel spreadsheet was generated for the land cover impervious surface values and with the available parameters, the ISAT analysis was carried out.
3. Results
3.1. Delineated Watershed
The urban watershed of Ile-Ife in Osun state was delineated and the result is shown in Figure 2 and Table 3 which identifies the different catchments and places located within the urban watershed. The urban watershed was delineated
![]()
Table 2. Land use classes generated for impervious coefficient.
![]()
Table 3. Delineated watershed in the urban watershed of Ile-Ife.
![]()
Figure 2. Watershed delineation of Ile-Ife Urban Watershed.
into 7 different watersheds, named alphabetically watersheds A to G. The area of each of the watersheds was determined and it was observed that watershed C was the largest watershed having an area of 2223.252 ha and the smallest watershed was E having an area of 142.641 ha. Other watersheds A, B, D, F and G covered an area of 487.788 ha, 226.812 ha, 1504.430 ha, 1093.871 ha and 2012.661 ha respectively, which makes watershed G the second largest in the study area. The urban watershed amounted to a total of 7691.455 ha.
3.2. Percentage Impervious Surface (%IS) of the Urban Watershed of Ile-Ife
The percentage of impervious surfaces in the watershed was analyzed using the object-oriented image classification and the result shows a drastic increase in an impervious layer over the years (2008-2016) in the respective watershed. Using object-oriented image classification (OBIA), the LULC was classified into four categories, which are; water, Built up, road and vegetation, which were classified with the seven urban watersheds that were delineated.
3.3. OBIA Classification for 2008 and 2016
The object image analysis classification for 2008 was obtained because of the classification carried out on each watershed. The study area was majorly made up of Built up, accounting for 4504.32 ha (59%), followed by vegetation 3076.65 ha (40%), water body 47.8 ha (0.8%) and road 62.8 (0.6%). The Total Area covered is 7691.5 ha (100%) as shown in Figure 3.
The object image analysis classification for 2016 was obtained because of the classification carried out on each watershed. The study area was mainly made up of Built up, accounting for 5334.9 ha (69%), followed by vegetation 2246.6 ha (29%), water body 36.72 ha (1%) and road 73.26 ha (1%). The Total Area covered is 7691.5 ha (100%) as shown in Table 4 and Figure 4.
3.4. OBIA Re-Classification for 2008 and 2016
A reclassification was carried out for both 2008 and 2016 land features (Table 5, Figure 4). For 2008 Built up and roads were reclassified to the impervious surface, covering a surface area of 4567.1 ha (59.4%) while vegetation and water body were reclassified as pervious surface, covering a surface area of 3124.4 ha (40.6%). Built up and roads (impervious) for 2016 amounted to 5408.2 ha (70.3%) while vegetation and water body amounted to 2283.3 ha (29.7%) giving a total of 7691.5 ha (100%). The result showed an increase in impervious covers in 2016 as compared to 2008.
3.5. Rate of Change in the Watershed of Ile-Ife within the Years Studied
The change detection was determined in line with the result of the object-based image classification carried out on each watershed, and the classified maps of
![]()
Table 4. Land use land cover classification for each watershed.
![]()
Table 5. Area covered for land use land cover in 2008 and 2016.
![]()
(A)![]()
(B)
Figure 3. Classified object-based images for 2008 and 2016.
![]()
(A)![]()
(B)
Figure 4. Reclassified object-based images for 2008 and 2016.
2008 and 2016. When the 2008 land use land cover classification is compared with the 2016 LULC classification, some changes show a reduction or increase in specific LULC. The rate of change in watersheds A, B, C, E, and F amounted to 94.46 ha, 70.24 ha, 511.77 ha, 23.79 ha and 208.08 ha respectively. Vegetation in these watersheds amounted to −93.05 ha, −69.69 ha, −526.68 ha, −24.42 ha and 199.96 ha respectively. The rate of change in another watershed is represented in Table 6.
Accuracy Assessment of the OBIA Classification
Table 7 describes the OBIA classification accuracy assessment carried out. The overall classification accuracy was 87.97% and the overall kappa statistic was 0.81. The producer’s accuracy for the combined Built up/water was 77.82%, for
![]()
Table 6. Rate of change of land use land cover classification for each watershed.
![]()
Table 7. Accuracy assessment result of the OBIA classification.
vegetation was 91.56%, and water was 100%. The user’s accuracy for the combined Built up/water was 94.86%, for vegetation it was 81.72%, and for the water class, it was 96.24%. The kappa by landcover class was 0.92 for combined Built up/water, 0.75 for vegetation, and 0.94 for water class. The vegetation landcover class with a user’s accuracy of 81.72% and the impervious landcover class with a producer’s accuracy of 77.82% explained the overall accuracy of 87.97%. The accuracy assessment result assisted in the further breakdown of the different watersheds into different classes.
3.6. Watershed Impervious Classification
Each of the watersheds was reclassified into impervious and pervious layers. Built-up and roads were classified under impervious while vegetation and water were classified under pervious as shown in Figure 5.
3.7. Water Quality Determination on Impervious Surface Analyst Tool (ISAT)
The effect of imperviousness on water quality was determined using the ISAT tool and an ISAT map was generated, this showed the level of impact (Table 8, Figure 6 and Figure 7). The ISAT map was derived through integration and analysis of processes. The result presented the high, medium and low levels of impact in the study area accounted for 25 - 100, 10 - 25 and 0 - 10 respectively. It was observed that the Built up (impervious) areas have a very high level of impact on the water body in that region, the regions of medium impact were areas with less dense impervious surfaces and some surrounding vegetation and the areas of low impact were observed to be at areas with very little or no human activity, areas with much dense vegetation. A water quality field survey was carried out to verify the various regions of impact as generated by the impervious surface analyst tool. A water quality field survey was carried out to verify the various regions of impact as generated by the impervious surface analyst tool. Water samples were collected from two different locations in the low region and water samples were collected at three different locations in both the medium and high regions of impact, as generated by ISAT.
![]()
Figure 6. Derived impervious surface of Ile-Ife watershed.
![]()
Table 8. Generated level of imperviousness for water quality assessment
![]()
Figure 7. Derived impervious surfaces and assessed water quality locations in Ile-Ife watershed.
4. Discussion and Conclusion
The LULC change documented in the research area affected mainly the conversion of vegetated areas to Built up in the watershed. Population growth applies pressure on land and hence influences land use change altogether, especially to supply new regions for improvement. Vegetation was at its maximum in the year 2008 with 40% of the land cover, then experienced a drastic reduction in the year 2016 to 29%. Furberg and Ban (2012) and Ige-Olumide and Salami (2018) studies agree with the research that the loss of these vegetated regions implies a broader cover by unnatural surfaces and few natural actions given by vegetation, for example, penetration, filtration and assurance against flood and erosion. An increase in impervious covers in the study area could prevent or upset the normal cycles that offer environmental types of services, for example, protection of wide life ecosystem, flood control and guidelines and water quality sustainability, to the district (Salzman et al., 2001). The decrease in vegetation cover has led to increased Built up and thus, impervious, exposing the urban watershed to incessant flooding over the years, including 2010 and 2014 (Liu et al., 2012; Akinwumiju, 2016). Orimoogunje et al. (2016), Rain et al. (2011) and Guyekye (2013) pointed out that decrease in vegetation cover results in increase in impervious surfaces, urban heat islands, and loss of biodiversity and incessant flooding.
The Built up that experienced an increase within the study area is suggestive of an increase in population and infrastructure. As urbanization continues to expand, there is increased pressure on the environment which could lead to environmental crises especially with a lack of proper management and this also agrees with the work carried out by Adetoro and Salami (2018) and Bennett and Saunders (2010). In a similar study, Mengistu and Salami (2007) attributed continued population expansion (Kuddus et al., 2020) as one of the key factors causing spatial growth of Built up and the related pressure on the available land resources and other land use types prompting land use class transition. Several studies have shown that impervious surfaces tend to grow due to urbanization in most parts of the country (Oyinloye & Kufoniyi, 2011; Obiakor et al., 2012; Jesuleye et al., 2013; Mahmoud et al., 2016; Usman et al., 2018; Wang et al., 2018; Wang et al., 2020). Haphazard development can be observed in the study area which is attributed to the uncontrolled increase in impervious surfaces coupled with the absence of well-defined policy implementation for impervious surfaces. Medium-scale enterprises such as shopping complexes, petrol stations and small-scale businesses are causing the study area to be densely urbanized. This increase in urban density takes a toll on the environmental components in the area, this can be seen in the shrinking water body in the study area, amounting to −0.3% (Elmqvist et al., 2016). The non-point source pollution rate related to impervious surface run-off could increase and result in lowering the water quality within the study area (Kim et al., 2016; Adetoro & Salami, 2018).
An increase in impervious surfaces increases runoff to surface water, thereby reducing the quality of the water. The water analysis carried out supports studies by (Osibanjo et al., 2011; Ajibade, 2004) in Ona and Alaro rivers in Ibadan, it agreeing that runoff harms the physical and chemical parameters of surface water. The results from the water sampling showed high turbidity, low dissolved oxygen, and high electrical conductivity in regions with high imperviousness as generated by the ISAT map. High total dissolved solids found in water reduce light infiltration prompting decreased photosynthesis with effects on both phytoplankton and zooplankton populations. Also, a striking water quality characteristic of surface water in urban areas is high turbidity (Adefemi et al., 2007). Turbidity results revealed that the value from most of the sample points in the high, medium and low areas was higher than the 5.0 NTU limit given by WHO (2008) which could be linked to runoff effects due to an increase in impervious covers (Wakawa et al., 2008). This study also revealed the dependence on UAVs and ground-truthing dependence for accuracy assessment carried out.
The study therefore concludes that as built up and other impervious man-made features increase due to rapid urbanization, Ile-Ife city becomes vulnerable to high runoff and thereby causes flooding. This has become a frequent and reoccurring event in Ile-Ife, such as reduced water quality, degraded and destroyed aquatic and terrestrial habitats. Important land use such as water and wetland were also adversely affected in terms of low quality and reduction in size. Based on the assessment of the LULC, the impervious surface development will continue to increase uncontrolled because of the high demand for land development by the various developers. However, it has been discovered that there is a strong connection between ambient temperature and impervious surfaces as both are critical in the study area. Finally, the increase in impervious surfaces if not checked, could affect other land use and cause an environmental disaster. Further research will focus on the use of artificial intelligence for the assessment of other water quality parameters such as the toxicity level of heavy metals in soil and water in fast-developing towns that are changing to urban centers.
Acknowledgements
The Federal Government of Nigeria and Obafemi Awolowo University are appreciated for the grant awarded under the Tertiary Education Trust Fund (TETFUND) Research Projects (RP) Intervention and DIGITAL GLOBE foundation for the Image award to support this research. Space Applications and Environmental Science Laboratory (SPAEL) appreciates the opportunity to utilize other facilities within the laboratory needed for the success of the research.