Spatial analysis to prioritise spatial investment of social infrastructure

January 22nd, 2018, Published in Articles: PositionIT

This analysis formed part of a project for the Department of Rural Development and Land Reform where the core focus was on the development of differentiated provision standards for locating a range of social facility types. The ultimate aim is to improve access to, and the spatial distribution of, facilities in rural areas, and ensuring that facilities are not underutilised and unsustainable.

Delivery of social services in rural South Africa remains a major challenge 20 years after the demise of apartheid. Challenges include the poor planning and allocation of facilities, processes which can be vulnerable to politically-based decision-making. In a search for greater equity, spatial justice and efficiency of service delivery of social facility investment, the concepts of central place, agglomeration, and accessibility, together with GIS principles of catchment analysis, were applied to identify which towns and villages outside of the major metros are best located to serve the most people from the least number of service points.

Demarcating potential service regions around non-metro towns and settlements

In order to demarcate a potential service region that could be profiled with respect to population demand, a detail population accessibility analysis was conducted to identify the likely service region around each of 1328 identified town and settlement points in South Africa. Each part of the country was allocated to the closest town or settlement. The first step was to create a 1 x 1 km grid for the entire country, and then to allocate each grid cell to its nearest town or settlement using a detailed road network.

The result of this process defined the service catchment around all the towns and settlements in such a way that the potential service areas for each town, based on the cells that had been allocated to each point, are indicated. Then, making use of data disaggregation techniques based on dasymetric mapping and areal interpolation, demographic and economic data was disaggregated to the grid cells using high resolution ancillary datasets; and then was re-aggregated to the service catchments to be used to create profiles for these service regions [1, 2].

These defined service areas were ranked based on a combination of input variables, including the functional role of the town, the number of people for which it is the closest service point, and its administrative role – with order Level 1 being a metro with the highest rank, and order Level 10 being a settlement whose service region comprises less than 5000 people. The profile information included detail on potential linkages to other centres of a higher order.

Developing a hierarchy of social investment integration points

Local facilities, such as schools and parks, need to be provided in most areas and within five or less km from their users. On the other hand, middle to higher order facilities such as 24-hour health services (community health centres or hospitals), Home Affairs and Labour offices, regional sport facilities, libraries and police services offices have a greater access distance threshold. This presents a major opportunity to rationalise and spatially target the location of such middle to higher order facilities. These facilities have similar distance thresholds for services and their access reach can be tested using a detailed road network. The access distance threshold is typically 15 km in metro areas and large cities, but 25 to 30 km for regional/rural towns, and is extended to 50 or 100 km where the population density around towns is less dense and drops below 10 persons per km2.

The location criteria for middle order services thus presents an opportunity to test whether it is viable to use spatial optimisation analysis to prioritise towns with respect to optimal location and the size of middle order services, based on accessibility to the largest proportion of the population. In undertaking the analysis, the assumption was made to ensure that there was no potential overlap in service areas. The access distance around the town points was specified as 30 km road access distance in areas of medium to high population density, and 50 km in less dense areas.

Fig. 1: Prioritised towns by catchment orders showing service coverage for middle order services (travel distance bands within acceptable standards) [3].

Fig. 1: Prioritised towns by catchment orders showing service coverage for middle order services (travel distance bands within acceptable standards) [3].

An optimisation analysis routine was used to select the most optimal towns; and this was based on service catchment ranking, population density, centrality and access principles. The proposition is that these selected towns can serve as key points for integrated, spatially aligned inter-governmental service delivery investment, as these are where the most people could potentially access middle order services within government precincts from the least number of service points based on non-overlapping service areas around each town.

To enhance processing capability and to enable road-distance-based optimisation analysis of the whole country simultaneously, it was decided to use a larger analysis unit than the grid, but still one of equal areas and having a sufficiently fine grain. For this purpose it was decided to use the population data for 2011, which was aggregated to the Council for Scientific and Industrial Research’s (CSIR) Mesoframe (http://www.stepsa.org/socio_econ.html).

Using the catchment order ranking, the four highest ranks of settlement catchment points were included by default as priority investment locations for middle order services since these places include metros, cities, city regional and service towns. (For more information on the ranks, please see the technical report available under the “Documents” tab of the toolkit – www.socialfacilityprovisiontoolkit.co.za.) This resulted in 184 points being selected. These places were deemed to be critical focus points for their regions or to have sufficient population to automatically be required to provide middle order services; most are already established service points. A service area was defined around each place and the remaining area outside this service area was considered not yet served.

The next step in the optimisation analysis process was to select which of the remaining 1144 towns, outside the areas serviced by the already selected 184 points, have the greatest potential to act as service integration points for middle order services for the remainder of the country. Using an iterative selection process, the optimisation routine selected the next most optimal towns successively (from the remaining towns and settlement points), based on the population density within the specified cut-off distance, to select the next best towns not already in a serviced area. This means that the model iteratively selected from the Order Levels 5s, then 6s, and up to 10s to identify the best of the remaining towns to create a hierarchy of potential middle order service points for the country.

The GIS-based optimisation analysis made use of catchment area analysis to iteratively demarcate non-overlapping service catchment areas of 30 km (based on road distance) around each town/settlement point based on the population density, and then marked the area as serviced. Once done, it searched for the best remaining potential service points in the unserved area, all the while continuing to not allow service areas to overlap.

This analysis output proposes a geo-spatially targeted hierarchy of places that can be used to prioritise social investment of (regional) middle-order facilities in government precincts. The selection is a ranked set of points indicating the most optimal towns that can be used to best roll-out service points across the non-metropolitan areas of South Africa in a way that is equitable and spatially fair to potential users. The prioritised town locations and the respective travel distances within service reach of the priority towns are shown in Fig. 1. Fig. 2 shows the number of identified prioritised towns that can be prioritised versus the total number of catchment town points for each order level.

The implication of spatially targeting prioritised towns that optimally reach areas of 30 km (50 km in sparse areas) or less with no overlap, is that it is possible to achieve high service coverage whilst minimising the number of service points (as has been illustrated).

Fig. 2: Number of prioritised towns versus total towns [3].

Fig. 2: Number of prioritised towns versus total towns [3].

The analysis has implications for service provision throughout the country. The prioritised locations specifically identified for middle-order service location mean that service providers can achieve high service reach levels using fewer locations, rather than trying to roll out services where there is insufficient demand for services to be viable. The 378 identified priority towns can potentially provide middle-order services to 92% of the country’s population within 30 km, for denser rural areas, and within 50 km, for sparse rural areas like the Northern Cape and parts of the Free State, Eastern Cape, Western Cape, and North West.

If a spatially targeted investment strategy is used to locate middle-order services as described above, it reduces the number of potential points to be serviced by over 50%, while still being within an acceptable travel distance for more than 90% of citizens, including those in rural areas. Thus, it is in the provision of clustered middle-order services that the opportunity exists to direct investment optimally outside the metros and rationalise services. This could achieve a more efficient allocation of resources, especially given the pressure on South Africa’s fiscus within the medium term.

Social facility toolkit and calculator

The information on the priority places and the profile information on all the service catchment areas is accessible in a free web-based online toolkit which is spatially relevant. The toolkit can be accessed at www.socialfacilityprovisiontoolkit.co.za.

It supports the planning of spatially balanced and sustainable services and their efficient allocation to meet the needs of both users and service providers, and is in line with the principles of equity, government policy and fairness.

In addition to displaying profile information, the toolkit can be used to calculate the number of social facilities that each service area can viably support. The number of facilities and the typical service capacities that can be viably provided is informed and calculated based on the provision standards.

Using the information packaged in the toolkit (documents, the Facility Calculator and service catchment profiles), planners can gain a clear understanding of what facilities can be viably provided in each area, and an indication of which towns have the spatial potential to provide the best locations to optimally locate facilities to service the most people in South Africa in an efficient manner.

Fig. 2: The Social Facility Provision Toolkit.

Fig. 2: The Social Facility Provision Toolkit.

The toolkit is the result of an integration of skills and tools developed by the Spatial Planning Support Group of CSIR Built Environment. This includes GIS-based accessibility analysis, spatial planning support, and the development of social facility standards provision standards. The toolkit is unique in that it combines minimum norms and standards for the provision of social facilities in an electronic viewer and facility calculator, while the catchment dashboard graphically displays critical indicators for each virtual service catchment areas that can be used to inform planning. This includes indicators such as population growth, age profile and population distribution, dominant economic sectors, distances to higher order places, which were developed using a range of GIS tools and processes.

Given that internet access is still a problem in many places, especially in the rural areas of the country, and that many local municipalities do not have good internet capacity, the toolkit was developed with both an online and off-line capability. The off-line capability enables easy access to the data for decision support even outside the office or in the field.

This approach and its results were only possible given the strength of the GIS tools and the processing power available that provides usable decision support tools that can support future development. The data packaged in the toolkit can also be used to support a range of other location-based decisions that rely on demographic data.

Acknowledgement

The author would like to acknowledge the Rural Infrastructure Development Group of the Department of Rural Development and Land Reform for the initiation and funding the project, as well the core team members Mawande Ngidi, Gerbrand Mans, Zukisa Sogoni, David McKelly, and Tansy Argue for their dedication, team work and innovation.

References

[1] Mans, G. 2012. Developing a geo-data frame to facilitate data integration. PositionIT. Available at: http://www.ee.co.za/wp-content/uploads/legacy/positionit_2012/Developing_proc_aprl-may12.pdf
[2] Ngidi, M, Mans, G, McKelly, D and Sogoni, Z. 2017. Using a hybrid methodology of dasyametric mapping and data interpolation techniques to undertake population data (dis)aggregation in South Africa. South African Journal of Geomatics, Vol 6 (2), pp 232-244.
[3] Green, Chéri, Mans, Gerbrand, Ngidi, Mawande, Sogoni, Zukisa, Maritz, Johan. 2016. Using Catchment Areas Analysis and GIS based Spatial Analysis for Prioritising Spatial Investment in Non-Metro South Africa. ISOCARP Durban, 12-16 September 2016. Available at: https://isocarp.org/app/uploads/2016/09/Congress-Papers.pdf

Contact Chéri Green, CSIR Built Environment, cgreen@csir.co.za