The application of an Urban Sprawl Index: comparing towns and cities in the Western Cape Province, South Africa

The incidence and effects of urban sprawl have been the subject of a great many academic research mainly as a result of the challenges posed by continued urbanisation, especially in developing countries (see inter alia Jenks, Kozak and Takkanon 2008; Mander, Brebbia and Tiezzi 2006; Jenks and Burgess 2000; and Soja 2000). South Africa witnessed a proliferation of legislation and spatial policies to limit urban sprawl and contain the physical expansion and development of urban areas during the last two decades in response to exponential post-apartheid urbanization. In 2005, the Provincial Spatial Development Framework of the Western Cape Province, South Africa stated that “an Urban Edge shall be drawn around all villages, towns and cities in the province with the primary function to contain outward growth of urban settlements” (City of Cape Town, 2009) and in parallel the Provincial Department of Environmental Affairs and Development Planning published the Western Cape Urban Edge Guideline document assisting all municipalities in the province to delineate urban edges to be included in municipal Spatial Development Frameworks (SDF) thereby illustrating intent on maintaining urban footprints that are compact and limit developments that could be considered urban sprawl. Subsequently municipal urban edges have been delineated and are presently reflected in most municipal Spatial Development Frameworks in the province. This paper presents an Urban Sprawl Index as a tool to comparatively analyse the extent of urban sprawl between cities and towns of different sizes, making use of cadastre, land use and population data over time. The Urban Sprawl Index (USI) for the Western Cape put forward by this research will enable the comparative measurement of the extent of urban sprawl proportionately between the Metropolitan and local municipalities in the province and thereby aid in understanding the impact of planning instruments such as urban edges in the context of development dynamics and pressures experienced by individual cities.


Introduction
The incidence and effects of urban sprawl have been the subject of a great many academic research mainly as a result of the challenges posed by continued urbanisation, especially in developing countries (see inter alia Jenks, Kozak and Takkanon 2008;Mander, Brebbia and Tiezzi 2006;Jenks and Burgess 2000;and Soja 2000). South Africa witnessed a proliferation of legislation and spatial policies to limit urban sprawl and contain the physical expansion and development of urban areas during the last two decades in response to exponential post-apartheid urbanization (at a National legislative level see inter alia the Development Facilitation Act (1995), the National Development Plan (2013) and the Integrated Urban Development Framework (2016)), but the measurement of the successes or failures of these instruments has remained a largely unattainable goal. This can be attributed to the multifaceted nature of urban sprawl, the unavailability of comparable data over a reasonable time-period, and the absence of an appropriate measure of proportionate comparison. This argument is aided by authors such as Du Plessis et al (2015) who noted the surprising paucity of comprehensive cross city comparative empirical evidence to evaluate the changes of South African city structure since 1994 that would enable critical reflection on the impact of new generation spatial policy and planning instruments. In 2005, the Provincial Spatial Development Framework of the Western Cape Province, South Africa stated that "an Urban Edge shall be drawn around all villages, towns and cities in the province with the primary function to contain outward growth of urban settlements" (City of Cape Town, 2009) and in parallel the Provincial Department of Environmental Affairs and Development Planning published the Western Cape Urban Edge Guideline document assisting all municipalities in the province to delineate urban edges to be included in municipal Spatial Development Frameworks (SDF) thereby illustrating intent on maintaining urban footprints that are compact and limit developments that could be considered urban sprawl, in line with afore mentioned national policies. Subsequently municipal urban edges have been delineated and are presently reflected in most municipal Spatial Development Frameworks in the province. This paper presents an Urban Sprawl Index as a tool to comparatively analyse the extent of urban sprawl between cities and towns of different sizes, making use of cadastre, land use and population data over time. The Urban Sprawl Index (USI) for the Western Cape put forward by this research will enable the comparative measurement of the extent of urban sprawl proportionately between the Metropolitan and local municipalities in the province in an attempt to understand the impact of planning instruments such as urban edges in the context of development dynamics and pressures experienced by individual cities.

Background
In South Africa, as a result of the spatial political history of Apartheid, urban areas acquired a spatial form that contributed to the formation of single-core cities with far reaching and widespreading peripheral development, either in the form of suburban development or as a result of post-Apartheid programmes such as the Reconstruction and Development Programme (1994) which, in an attempt to assist the immediate housing shortage, favoured the establishment of single dwelling residential units for low income communities, on single plots on affordable land at the urban periphery (Horn, 2010). This invariably contributed to urban sprawl and even though this programme has since been abandoned, the trend towards low density residential development continues in South African urban areas, and is greatly exacerbated by the continuous accretion of informal dwellers of in-migrants at the urban fringe (Harrison et al 2008). The concepts of compaction and densification of urban form became entrenched in the South African spatial planning policy and approach since early 1990s and can be viewed as a key element of the overarching South African spatial planning doctrine (Du Plessis et al., 2015).
The research area for this paper is the Western Cape Province in South Africa, which in all aspects exhibit the spatial challenges described above (see Sinclair-Smith, 2014, Turok and Watson, 2001and Haferburg and Obenbrugge, 2003. The province consists of one Metropolitan Municipality (The City of Cape Town), and five district municipalities, which in turn accounts for twenty four local municipalities. The development economy of municipalities are seen as contributing to the perceived urban sprawl i.e. according to the Provincial Urban Edge Guideline (2005) "The income derived from high income market development, through rates and taxes, must pay for the cost of free services and relatively high service infrastructure maintenance cost in the subsidy and low income residential areas. Thus, the municipalities promote high-income market sector growth, regardless of the environmental and indirect cost thereof." The City of Cape Town has traditionally been the main receiver of in-migrants from other provinces, especially from the Eastern Cape, both from poorer families seeking proximity to employment markets and from more affluent pensioners seeking proximity to environmental assets (Eigelaar-Meets 2018). Both these trends have contributed to the expansion of population in other local municipalities in the province as well. A study by Jacobs et al (2016) evaluating the patterns and characteristics of substream migrants to the Western Cape found that more than 30% of migrants to the Western Cape settle either informally, and more than 21% settle informally in a backyard

Methods in measuring urban sprawl
The measurement, quantification and modelling of urban sprawl have remained evasive objectives mainly as a result of the absence of a universal definition of what constitutes sprawl (see Bhatta et al., 2010 andWang et al, 2018). Soja (2000) in studying the landscape of the "postmetropolis" recognises five characteristics prominent to urban sprawl as 1) a scattered and discontinuous pattern of development 2) development of residential areas with low densities 3) commercial strip development 4) segregation of land uses and 5) low accessibility and high dependency on private transport. In summary sprawl is an urban growth pattern with the features of low density dispersion, auto-mobile dependence and other environmental and social effects. The direct consequence of sprawl is change in land-use and land-cover of the region as sprawl induces the increase in built-up and paved area (see Bhatta et al., 2010). Sutton (2003) charges the term urban sprawl as merely a politically correct means of complaining about the negative consequences of population growth or the changing scale of the total population of the city they live in. The fact is that the general known impacts of sprawl have been well defined however uncertainty remains regarding an unequivocal definition thereof and how to measure it. Empirical studies contribute to an understanding of the essence of sprawl as it adds quantitative knowledge into the discussion and suggests possible solutions without which ideological and practical discussion on urban sprawl and the effectiveness of a growth management policy remains only conceptual. Spatial metrics can thereby assist in quantifying indices to describe structures and patterns of a landscape (Frenkel et al., 2008).
Commonly used spatial indices include growth rates such as population growth rate measured against built-up growth rate (ibid) which calculates quite an "absolute" scale i.e. it creates a black and white distinction between a sprawled city and a compact city (Bhatta et al., 2010), assuming a predetermined threshold. Some theorists experimented with more relative measures that in contrast quantify several attributes of urban growth (see Feng et al, 2014 andGao et al., 2016) and can be compared among cities, however the complexity of the selection of indicators and data/resource availability renders such an approach expensive and time consuming.
Notwithstanding the discussion on measuring sprawl in literature, the absence of empirical evidence is still clear. Aside from the ambiguous definition of urban sprawl and the difficulties it generate in measuring its scope, another problem with most of the available sprawl measurement scales is the failure to define the threshold between what discerns a sprawling from a non sprawling urban area. Even so, relative comparisons between cities of different sizes and forms could already assist in providing some insight with regards to the occurrence of "more" or "less" sprawl per city.

Methodology
In a study by Donaldson et al (2012) ten local (non-metropolitan) municipalities in the Western Cape have been identified as having the greatest potential for future growth and development. The study presents an index of different aspects that contribute towards a town or city's development potential that relates inter alia to availability of infrastructure, social capital and economic growth.
In this regard the local municipalities of Drakenstein, Stellenbosch, George, Mosselbay, Overstrand, Knysna, Swartland, Saldanah Bay, Breede Valley, and Cape Agulhas were identified as having the highest composite development potential in the Western Cape. This study aimed to use the main urban area/town in the municipalities identified in the aforementioned study as having medium to high development potential as well as the City of Cape Town Metropolitan Municipality as subjects of analysis. Based on the immediate availability of data over a measurable time period, the municipalities that were investigated were Drakenstein local municipality, George, Overstrand local municipality, Plettenbergbay, Knysna local municipality and the City of Cape Town metropolitan municipality. The table below provides a summary of the municipalities that form part of this study and the spatial instruments referenced in their respective SDFs that would encourage the prevention of urban sprawl and contribute to a more compact urban form:  formal residential land use) and the 'township' land use category (i.e. established township areas as per land use data obtained from Mapable (ltd)) were used in calculating the urban areas for two reference years as these areas are the ones where increased population were assumed to be accommodated.
In the metropolitan area of Cape Town, the urban footprint covers close to the entire municipal area and therefore the total population was applied. In the case of the four other municipalities, the municipal area consist of a main urban centre as well as a number of smaller towns scattered across the area of jurisdiction of the municipality. In this instance, only the main central urban area's population figures were applied to the USI equation and not that of the isolated and scattered rural settlements. To compute the USI (Equation 1, a and b) the research makes use of South African census data for reference years 1996 and 2011 to ascertain the extent of population growth. The availability of land use data over a comparative time period was the biggest constraint to the research and the closest corresponding reference years for land cover changes were for 1990 and 2014. Given these parameters the USI was calculated using the formula described by Gao et al (2016): Where: 1 2 is the USI for a city measured over time period 1 T1 (1996) and time period 2 T2 (2009); 1 2 are the compound annual growth rate for urban formal residential expansion; 1 2 the compound annual growth rate for urban population ; Equation 1 was obtained by calculating the compound annual growth rate as follows: and Where: 2 and 1 refer to the urban residential area for a city in T1 and T2 and 2 and 1 are the urban population for each city in T1 and T2.
For the purpose of this research urban sprawl is defined as a land development pattern that occurs when the rate of residential expansion exceeds the urban population growth rate i.e. USI≥0. Table 2 presents the initial results following comparative analysis between population growth rate and existing residential footprints in all five municipalities under consideration. In all instances, the formal residential areas increased significantly less in the period 1990-2014 compared to the population growth during 1996-2011. This is illustrated by the last column in Table   2 (growth factor) indicating the rate by which population growth exceeded residential growth. In Overstrand municipality, for example this means that the population grew 19.3 times more than the formal residential footprint of the municipality. Noteworthy from the results is the USI in all instances below zero, mainly due to the disproportionate population growth compared to residential expansion.

Results and discussion
What the results clearly indicate is the inability of the formal residential market (. i.e. the private sector and state-assisted housing schemes) to keep abreast the rapid population growth experienced by these municipalities. In response to rapid immigration, a large component of residential growth in South African happens in informal settlements, a development pattern that inadvertently contributes to urban sprawl. Based on this the analysis was performed again, this time including land use cover changes occurring as a result of as the growth of informal settlements. The results of the second analysis is presented in Table 3.  Table 3 reveals that even when including informal residential areas as a land use category in calculating urban expansion, the USI continues to suggest an absence of urban sprawl. One possible explanation for the reduction of the growth rate of residential area is that the population and built-up density increased over time (Sharifi et al, 2014  Thus far however the analysis failed to account for non-residential land use categories that may have resulted in land consumption contributing to urban sprawl. Consequently, the analysis was repeated by including "commercial" and "industrial" land use category data for 1990 and 2014. The results of this analysis is presented in Table 4.  Table 4 illustrates that the contribution of non-residential land categories to urban sprawl is fairly limited. With the exception of Drakenstein municipality, appearing to have experienced significant growth in commercial and industrial land uses, the remaining four municipalities see the addition of these land use categories leaving their USI's virtually unchanged.
The results of the preceding analyses suggest a noticeable reduction in land consumption and could be indicative of densification inside the urban edge. If the amount of sprawl is used as one indicator to measure the success of urban edges as planning instruments to curb urban sprawl, the results of this analysis suggest that the urban edge may have contributed to the containment of urban expansion.
Whilst the analysis clearly indicates the enormous difference experienced by all the municipalities in population growth measured against land development, it cannot account for the typology of urban development that did occur during this time period. Whilst land consumption may appear to be limited, it does not preclude that development associated with urban sprawl (i.e. low(er) density, cardependant, access controlled residential estates) actually occurred.

Limitations/Assumptions
A shortcoming of traditional density measures (see Du Plessis et al., 2015) that will also impact on urban sprawl analyses is its sensitivity to methodological issues around the delimitation of urban areas and population figures based on administrative boundaries.
The authors also recognise that this is not a compounded USI but that it focuses only on physical expansion of the formal residential footprint of urban areas compared to crude population growth, while there are many other factors such as leapfrogging, densification and infill that is important to take cognisance of when determining the extent of urban sprawl in an urban area.

Conclusion
One of the main reasons for the absence of cross city empirical comparisons is the absence of a common set of land-use or land-cover data at an appropriate scale for multiple time periods (Du Plessis et al., 2015). This research attempted to measure sprawl over comparable time periods in order to establish the extent of urban sprawl in the Metropolitan area of Cape Town measured against other municipalities in the Western Cape Province. The results indicated that whilst population have grown tremendously in all municipalities, the rate of land consumption have been limited resulting in an USI≤0 in all instances. Some of the smaller cities tended to experience more rapid population increases compared to the City of Cape Town. The negative values of the USI's calculated in this research points toward increased numbers of people not catered for by the formal residential development market and instead, and possibly contributing towards increased densities in formal and informal residential settlements.
In academia no-one has provided straight answers to questions like what should be the built-up growth rate in a non-sprawling city or what should be the per capita land consumption in a non sprawling city and therefore measuring the extent of urban sprawl by using an USI should not be used as an absolute indicator of the prevalence of urban sprawl but can at best be indicative of trends for comparative purposes. Presently the application of the USI to cities in the Western Cape yields results that suggest the success of the Western Cape Provincial urban edge policy by limiting urban sprawl and reducing land consumption per capita. However, the results cannot preclude that land development and consumption that did occur does not constitute in some instances resemble the typology of development associated with urban sprawl.