The role of maps in a census is to provide the cartographic basis for enumeration and the platform for spatial census data products such as census atlases. This paper focuses mainly on mapping solutions for enumeration. The paper assesses the traditional methods of census mapping in African countries and aims to provide modern alternatives to the traditional methods that are realistic and achievable given the unique circumstances of the African continent. The paper is therefore of a practical rather than an academic nature. The author strives to provide hands-on and practical solutions based on experience gained in several countries over the last decade. The paper could therefore be of potential use to readers that are responsible for the planning and implementation of the 2010 round of population and housing censuses.
1. Introduction and problem statement
The aim of a population and housing census is to collect, process and publish data on every individual within a defined geographical area usually a country. The aim of census mapping is to provide the cartographic basis to be used during the actual process of counting. The census cartographer therefore needs to provide the census enumeration team with a set of unique maps covering the entire country that accurately defines the boundaries within which each interviewer (enumerator) has to work during the enumeration phase of the census.
In order to do this the census cartographer needs to have access to current maps of sufficient levels of scale and detail. In most African countries the required maps are either very old (often 30 years and older) or in some cases do not exist at all.
Furthermore the census cartographer usually relies on a very small permanent staff and a huge temporary workforce, funded directly from the census budget, to execute the census mapping plan. This is due to the fact that most statistical agencies in Africa are understaffed in general and seriously understaffed in terms of cartographers. It can therefore be stated that the capacity within statistical agencies in Africa to do census mapping is in most cases limited.
2. Conventional mapping
With the above being the status quo in most statistical agencies in Africa the mapping solutions in the past tended to focus on arriving at solutions within the existing framework rather than transforming the existing framework as part of the solution. Given the circumstances and the constraints that faced census cartographers most countries over the years adopted an approach of throwing extensive human resources at the problem once funding for the census become available. This approach implied that huge numbers of people (often referred to as field mappers) were deployed in the field to sketch enumeration area maps whilst sweeping the larger administrative area. The rationale behind this approach was to solve the lack of proper base maps by updating the maps in the field or by creating entirely new maps in the field. During the 2000 round of censuses this methodology was refined with the introduction of Global Positioning System (GPS) technology. GPS contributed to the increase of the absolute accuracy of map updates and sketch maps. The conventional approach provided and today still provides a workable solution in most African countries.
3. Why the need for change?
It can perhaps be argued that it is not necessary to revisit the conventional methods since it provided workable solutions in the past. There are however many reasons why there is a need for change:
3.1 User needs
The information age that we are current living in places more and more demand on the need for digital spatial data. This is also the case in Africa. Furthermore the need for quality and relevant statistics for developmental planning purposes is bigger than ever. A statistical agency therefore has to adjust in order to the respond to the spatial data needs of their users. Some statistical agencies are currently responding to user needs by using Geographical Information Systems (GIS) in post-enumeration activities (only). Although this is a step in the right direction it is argued that more emphasis should be placed on the use of GIS and geographical information technology (GIT) in pre-census activities. In doing so the platform for small-area (spatial) census data products are created as a direct deliverable of the methodology.
3.2 Advances in technology
GIT has advanced by leaps and bounds over the last few years. The processing power of computers and software packages are increasing by the day. Moreover, the integration and utilisation of GIS, GPS and remote sensing are also becoming easier and more cost effective. There can be no doubt that the right tools exist and it has never been easier to utilise these tools in geospatial applications. It is argued that, if implemented as part of a population and housing census project, the technology is now within the financial reach and technical capabilities of most statistical agencies in Africa.
3.3 The cost factor
Although the initial financial layout required for the establishment of a GIS Office at a statistical agency is still relatively high it is a much more affordable exercise than say ten years ago. Furthermore, if applied optimally, a statistical agency can recover the initial capital layout of the GIS infrastructure from the savings in the mapping sub-project of the first population and housing census that will benefit from the technology. This is made possible mainly by significant savings brought about by streamlining the pre-enumeration census demarcation fieldwork exercise.
3.4 Capacity building
The future of census mapping is digital. All developed countries are already using it. The sooner all statistical agencies starts building the required capacity the better since international standards and the overall demands driven by the information age will eventually demand statistical agencies to do so. It must be mentioned that building the required capacity is nowadays much more realistic than before since the technology (GIS, GPS, satellite imagery, digital aerial photography, etc.) is a lot easier to use and integrate than before.
3.5 The issue of quality
If digital census mapping is applied correctly and optimally it is guaranteed to provide better quality results when compared to conventional mapping. The normal errors associated with census mapping such as incomplete coverage or double coverage of the same area is to a large extent eradicated.
3.6 International trends:
The United Nations (UN) [1] are promoting the use of GPS, Digital Imagery and GIS with Census Mapping. It can therefore be expected that more countries will be encouraged and empower by the UN to implement digital census cartographic methodologies in the 2010 round of censuses.
4. Modern (GIT) methods
Modern methods assume the optimal utilisation and integration of GIT tools. GIT tools include: GIS (including desktop GIS, mobile GIS and WEB GIS); GPS; remote sensing (satellite imagery and digital aerial photography).
It is important to understand the technical capabilities of current GIT tools, but it is also important to keep in mind that the crux of the matter is really the optimal integration of the available tools in order to arrive at the best possible solution for a given country. Experience has shown that no two countries have/will have the same solution. This is due to the fact that countries are unique in terms of the combination of variables that dictates the optimal solution that can be realistically implemented. These variables include population size, population distribution, settlement patterns and densities, geographical area, availability and specifications of base maps, agency capacity with regards to GIT, the census mapping budget, timeframe for implementation and commitment to the transition from conventional to modern census mapping methods at senior management level. If these factors are not taken into consideration and the available tools are not optimally used and integrated as part of a total census mapping solution it could easily become more of a hassle to the census cartographer than an aid.
4.1 Technical capabilities of GIT tools
GIT tools have developed tremendously over the last decade whilst at the same time becoming more and more affordable. GIS software and hardware is much more powerful, flexible and easier to use than was the case a few years ago. GPS receivers are more user-friendly and communicate elegantly with most GIS platforms. Mobile GIS with integrated GPS provides so-called “moving maps” that enable fieldworkers to navigate with relative ease to an area to collect spatial and attribute data. Electronic forms eliminate the need for paperwork. WEB-GIS is becoming more and more popular for census and survey operational management as well as dissemination of all statistical data with a geographic link.
Remote sensing in census and survey mapping is not Remote sensing (satellite imagery and digital aerial photography (DAP)) nowadays enables the census cartographer to have access to a GIS-ready base map for an entire country that meet all the needs of the census (i.e. currency, accuracy and level of detail). This technology enables the census cartographer to have a detailed base map for the entire country that is less than two years old thus eliminating the problems associated with dated base maps. This base map is suitable to be used as the base map for enumeration area and supervisor area maps. Because of differences in population characteristics (mainly geographic spread and densities) in urban and non-urban areas there are different base map solutions for urban and rural areas.
4.2 Rural base map: the SPOT5 solution
SPOT5 is the first commercial satellite that provides a feasible base map solution in terms of specifications and cost for census mapping purposes at a national (total coverage) level. The latest SPOT5 2,5m (pixel resolution) natural colour satellite imagery in particular provides the ideal base map for rural and even less dense urban areas. The technical specification of the imagery, as well as the ability of the satellite to do image acquisition of vast geographical areas such as national coverage’s linked to fast image processing capabilities ensuring relatively short delivery schedules, makes it an attractive option for census mapping purposes. Although still costly it is within the census mapping budget of most countries – especially if one takes into account the huge savings that can be made in reduced EA demarcation fieldwork as a result of the inherent benefits of having a detailed base map. SPOT5 5m natural colour imagery is lower resolution product from the same satellite. It is significantly cheaper than SPOT5 2,5m but there is an obvious compromise in terms of pixel resolution. A major advantage of SPOT5 5m is that the archive coverage on this product is, for most parts of the world, much more extensive than is the case with the SPOT5 2,5m archive. The archive imagery is cheaper than the programmed imagery and it takes the risk out of programmed image acquisition. Programmed image acquisition is determined by prevailing weather conditions as well as the priority of a given image acquisition project in relation to other projects at a given point in time. This is due to the limitations on the onboard storage capabilities of the satellite. SPOT Image is however currently the only satellite company that provides the client with a feasibility study prior to ordering the imagery.
The feasibility study tells the client how realistic it is to acquire the required imagery within the timeframe of the project. It takes into account previous experience based on the extensive historical image acquisition database of SPOT image as well as other important factors such as prevailing weather patterns and the effect of seasons on image acquisition1. With a detailed base map sketch mapping becomes unnecessary as well as extensive GPS fieldwork. SPOT5 2,5m natural colour supermode imagery is currently being used, for the first time ever in a population and housing census project, in the census mapping project of the 2006 Population and Housing Census of Lesotho. The project is currently still in progress but indications are that it will be a huge success. Ethiopia has also made provision for the acquisition of the same imagery for their 2007 census. This imagery is perfectly suited for rural areas as well as less dense formal urban settlements. The SPOT5 5m archive is furthermore quite extensive enabling statistical agencies to have access to relatively current imagery (<2 years) within a relatively short period of time.
4.3 Urban base map:
Very high resolution satellite imagery and Digital Aerial Photography (DAP) Research has shown that colour imagery with a pixel resolution of about 0,5m or better is required for urban areas in general but specifically for high density informal urban settlements. High-quality examples of the 0,6m Quickbird satellite imagery2 or, in certain cases even 1,0m IKONOS imagery, can be considered for relatively large contiguous informal areas. In cases where there are however several small high-density settlements scattered all over a country (as is often the case) the fastest and most cost effective solution is the creation 0,5m colour digital aerial photography3. In this scenario satellite imagery usually also becomes relatively expensive (effective price per square km or hectare) because the user has to acquire minimum coverage
- See www.spotimage.fr for more information on SPOT imagery.
- Seewww.digitalglobe.com for more information on Quickbird imagery.
- See www.geospace.co.zafor more information on digital aerial photography.
imagery based on a pre-defined grid or footprint that is often much bigger than the area of interest or the area of interest might fall into two or more image blocks necessitating the acquisition of more than one image hence pushing the costs up even further.
A huge advantage of satellite imagery in general compared to aerial photography is that it does provide a workable solution in countries where it is difficult to obtain the required flight authorisations to do aerial photography (in a short period of time). Most satellite operators furthermore collect imagery on a continuous basis irrespective if the satellite is programmed for a specific customer or not. This implies that huge archives of data are constantly being collected. The archived data are usually cheaper than programmed data and will often suffice for census purposes – especially the data collected within two years before census day. For most satellites the archived data over Africa is however relatively limited compared to certain other parts of the world because the decision on where to collect archive imagery is mainly driven by potential future sales (of the archived imagery).
One of the biggest disadvantages of satellite imagery is that the image quality can vary significantly. This is ascribed to the variances in atmospheric properties that are caused and influenced by several factors including the effect of seasons, the effect of water masses and proximity to the equator. This is an important consideration given the fact that both the urban and rural satellite imagery solutions mentioned above pushes the technical capabilities of the respective satellites to their limits hence allowing virtually no room for inconsistencies in image quality. This is due to the fact that the satellite imagery available today is in essence only just suitable for census base map purposes. As image properties and resolution improves with newer satellites over the next few years this problem will however seize to exist.
DAP has proved to be the best solution for the creation of census base maps in high density urban settlements. DAP can be created at different pixel resolution usually starting at about 1,5m going down to as little as a few centimeters. This is important since the best pixel resolution for a given application can be determined on a case-by-case basis ensuring that the imagery will provide the required output. The biggest advantage of DAP is however the flexibility in the creation of the imagery. If say 100 urban areas within a country has been earmarked for digital aerial photography, it can be scheduled in such a way to optimise the weather conditions as well as the needs of the census mapping office. Experience has shown that it is by far the quickest and cost-effective method to create base maps for scattered small areas. Indicative of the quick turn-around time of digital photography creation is a sample survey mapping project that GeoSpace International [4] did in 2002 in South Africa. For this project digital aerial photography base maps were created for 970 areas scattered all over South Africa. This was done with one dedicated aircraft in less than three months. The significance of this is that census cartographers can now have access to a detailed image map of high density and often dynamic urban populations as close as possible to census day.
4.4 GPS
During the 2000 round of censuses GPS played an important role mainly to provide positional (absolute) accuracy to sketch maps. It was the first GIT tool that was used widely and with reasonable success in Africa. It was however used in isolation of other GIT tools mainly because other GIT tools were in most cases not used at all. For the 2010 round of censuses the role of GPS is likely to differ significantly from its role in the past. Since GIS implicitly creates coordinates as spatial data (vector or point) is captured it is no longer necessary to use GPS primarily to capture coordinates. A map (image or otherwise) created from a GIS provides the basis for the capturing of point and vector data in the field hence making it unnecessary to use GPS for this purpose. GPS is therefore used for navigational purposes to ensure that the right area on the ground is visited and as a quality control tool to ensure that the team was actually in the areas where they were supposed to have been. GPS is furthermore integrated into GIS making it completely unnecessary to transcribe any coordinates and thus eliminating one of the biggest sources of errors in the use of GPS. Data is simply downloaded from a computer into the GPS receiver or uploaded from the GPS receiver into the computer. GPS can furthermore be directly linked to mobile GIS enabling the use of so-called moving maps. This approach completely eliminates the need for any map annotations and the use of paper forms during mapping fieldwork since all data capturing is done directly onto a field computer linked to a GPS. Some of the latest model GPS receivers have built-in moving map functionality thus it further streamlines fieldwork by eliminating the use of a field computer.4.
4.5 GIS
GIS is arguably the most important of all the GIT tools. GIS is the tool that is used to integrate all the other tools. It is used for the accessing and manipulation of all existing spatial data relevant to the census project. It is used for the accessing and processing of newly acquired image base maps. It is used for the integration of all the different spatial and attribute data sources and for the creation of fieldwork maps in electronic or hardcopy format depending on the fieldwork methodology. It can even be used to do preliminary EA demarcation in the office prior to fieldwork in order to limit the time in the field even further. It is therefore of critical importance during the preparations for EA demarcation fieldwork. All fieldwork maps in are created using GIS. After the fieldwork the updating of the spatial and attribute data base is done using the GIS. The creation and printing of all the final EA and SA maps are also done using bulk map creation and printing tools that are developed on a GIS platform. From the above it is clear that the tools exist to make the life of a census cartographer a lot easier. The challenge is the successful integration of the different tools in a total mapping solution. The author is therefore strongly of the opinion that the total solution is far more important than the using of the different tools in isolation. The real power of GIT is maximised when the different tools are properly integrated! This implies a complete rethinking of the census enumeration area demarcation methodology. Following is the process flow of a typical GIT-driven census mapping project:
Figure 1: Process flow of generic GIT-based census mapping project
The flow specified above is made possible by the use of all three GIT technologies and the integration of the data and tools provided by said technologies into one single system which links to one single data warehouse. Thus, linking a specific process flow to the various GIT options will typically look like this:
| Process Flow | Technology Used |
|---|---|
| Creation of spatial data warehouse | Database and GIS technology |
| Preparatory office demarcation | GIS and Remote Sensing |
| Fieldwork verification | GIS, GPS, Remote Sensing |
| Office demarcation updating and correction | GIS, GPS, Remote Sensing |
| Enumerator map creation and printing | GIS and database technology |
| Enumeration tracking and monitoring | GIS |
| Spatially enabled web application design and implementation | GIS, Web GIS, statistical analysis and database technology |
GIS is therefore a multi-disciplinary field integrating various other specialist fields into one single sustainable solution. The biggest mistake one can make is regarding the above mentioned fields as autonomous where census cartography is concerned.
5. The GeoSpace approach
GeoSpace International is a commercial company that has been involved in demographic projects of a spatial nature including five census mapping projects. The company was involved in the census mapping projects of South Africa (1996 and 2001), Namibia (2001), Tanzania (2002) and Lesotho (2006). Furthermore the company has been involved in technical consultancies on census mapping and GIS in Ethiopia, Swaziland and Botswana. As a result of the staff profile of the company we have census mapping expertise second to none on the African continent.
GeoSpace’s approach for the implementation of digital mapping for a population and housing census project consist of the following basic steps:
5.1 Needs assessment and user specification
During this step a thorough assessment is done of the status quo regarding GIS and the census preparations in general. The exact expectations with regards to the census mapping project are also identified and documented.
5.2 Project proposal
Based on the needs assessment and user specification and taking into account the status quo regarding GIT, a detailed project proposal is prepared in consultation with the relevant counterparts from the national statistical office (NSO). This project proposal documents the details of the most realistic census mapping solution for the particular country. The proposal would typically take into account the spatial data available for the country; the current census mapping methodology as well as the census mapping budget.
5.3 Project Implementation
Once the project proposal has been adopted by the census management/NSO the project is implemented. Since modern census mapping methods are relatively new to most census cartographers in Africa it is important and necessary for the CSO to have access to the required subject matter and GIS technical expertise. During project implementation GeoSpace usually provides the CSO with support with regards to training; base map data acquisition, project management and technical issues.
5.4 Capacity building
A key strategic focus of GeoSpace is to part with knowledge. This is important to the company not only because capacity building is important and necessary for the NSO but also because the more GIS users there are the more GIS industry as a whole will be stimulated. Moreover, the only way to really understand GIT is to do the work oneself. Geospace’s approach is therefore focused on training the local staff to do the work and then providing management support during the census cartography process itself, where possible.
5.5 Technical assistance
Since the technology as well as the modern methodology is usually new to the NSO it is important to provide technical support throughout the project to ensure the success of the project. GeoSpace provides the required GIS and subject matter technical support.
5.6 Project management
GeoSpace has a proud record of project delivery. The Company has, up to date, not been involved with a census or survey mapping project that was not a success. One of the main reasons for this has been the fact that GeoSpace only exits from the project once the final deliverables have been met. Since GIT is a new concept GeoSpace normally prefers to be involved in the day-to-day project management of the mapping project. This enables the Company to identify possible problem areas early and to take remedial action before the problem gets out of hand to ensure that the project delivered as was outlined at the beginning.
6. The Challenges
Although the benefits of modern methods in census mapping are obvious there are several challenges associated with the migration from paper to digital mapping. Some of the more significant challenges are as follows:
6.1 Ignorance at the technical level
Statisticians and cartographers at the technical/operational level often feel threatened by the idea of introducing GIT into their organisations or projects because they either do not know what it can do or do not want to go through the perceived hassles of learning what it can do and how to do it. As a result it is often more convenient to remain ignorant. Other contributing factors are the fear of change and the fear amongst (certain) staff that they might become redundant once the technology is implemented.
6.2 Ignorance at the decision making level
Often decision makers within a NSO are not aware of the merits and implications associated with the modern approach to census mapping. There is also the perception that GIT is expensive and would bring unnecessary complexity to the (census mapping) project. The expertise required to provide decision makers with the right information to shape policy and decision making is often not available within the organisation and also not easily accessible from elsewhere and hence (negative/wrong) perceptions tend to drive decision making process.
6.3 Paradigm shift /
The implications of the migration from paper to digital are severe and hence it requires a paradigm shift within the cartographic unit of an NSO as well as the organisation as a whole. It cannot be implemented partially or half heartedly – failure is almost guaranteed if commitment and buy-in from the highest level within the NSO is absent. It is therefore important that the organisation needs to undergo a paradigm shift with regards to their thinking of spatial data and the role it plays in the collection, analysis and dissemination of statistics.
6.4 Fear for the risk of failure
A failed census mapping project will almost certainly guarantee a failed census or lead to a postponement of the census – both of which have serious negative implications for the NSO. As a result statistical agencies often prefer to follow the tried and tested conventional approach because they are familiar with the methodology and have better control over the delivery thereof. What makes matters even worse is that there are a number of examples where GIS/GIT implementation has ended in disaster.
6.5 Capacity
Most statistical agencies in Africa do not have the required capacity (human, financial, infrastructure, etc.) within the organisation to successfully implement a GIS. This is a serious constraint since GIS implies hardware, software, data and human resources and not only a subset thereof. It should however be noted that there are also often misconceptions about the financial implications of GIS. It is sometimes argued that GIS cannot or should not be implemented because it is prohibitively expensive. Case studies (Tanzania 2002 and Lesotho 2006) have however showed that, if applied correctly, GIT-based solutions can cut enumeration area demarcation fieldwork by up to 80% with the resultant saving often paying for the investment in GIS or a significant portion thereof. It is however unfortunately so that, after completion of a census project, newly established GIS offices often die a slow (or sudden) death because of a lack of resources to keep the momentum going. In many countries there is no provision for GIS staff in the formal structure of the organisation because it is a relatively new discipline. Furthermore trained GIS staff is in high demand. This leads to a situation where, once qualified or skilled, staff often leaves the statistical agencies for greener pastures.
6.6 The need for an expert group
According to the UN [1] there is a huge need for the establishment of an international expert group on GIT for census mapping. This is necessary to provide a forum to share experiences and discuss and brainstorm ideas. An expert group would, amongst other things, contribute to a more uniform approach on fundamental issues such as the role of (modern) base maps and the optimal use of GPS in census mapping. A more uniform approach would instill confidence amongst technical staff and decision makers within national statistical agencies. This is necessary to ensure more informed decision making and optimal utilisation of the technology.
7. In conclusion
The 2000 round of censuses saw the successful introduction of modern methods for census mapping in a few African countries. It is anticipated that the 2010 round of censuses will be the round where the revolution in census mapping will gain real momentum. Statistical agencies therefore have an important decision to take: they can either take the step now and become part of the international trend or they will probably have to wait another ten years for the next opportunity to catch up with the rest of the world.
8. References
- United Nations (2004), Integration of GPS, Digital Imagery and GIS with Census Mapping, Department of Economic and Social Affairs, Statistics Division, New York
- United Nations (2000), Handbook on geographic information systems and digital mapping, Department of Economic and Social Affairs, Statistics Division, New York
- Stern, M (1985) Census from Heaven?, Wallin & Dalholm Boktryckeri AB, Lund, Sweden
- GeoSpace International (Pty) Ltd (2006), GeoSpace International: List of Previous Projects, Pretoria, South Africa.