Statistics South Africa made history in the recent census deciding to introduce the use of Geographical Information Systems (GIS) in the pre-census processes.

The census process begins a year before the actual count with a process that divides the country into smaller, more manageable units, called Enumerator Areas (EAs). For the first time in South Africa, this demarcation process was based on the availability of digital spatial information. The spatial database was populated with the latest digital data obtained from various government departments as well as private companies.

Once all the data had been accumulated it was loaded into the Oracle database at Statistics South Africa, using the GeoMedia suite provided by Intergraph Systems Southern Africa. Once all the data was loaded, data gaps were identified - these are areas with no available data or insufficient data. These gaps were then covered with digital aerial photography. Digital aerial photography now covers more than 80% of all Enumerator Areas. With a spatial database of a staggering 1,7 terabytes of live data, Stats SA manages one of the most comprehensive spatial databases in the country. This data was used to verify, correct and update EA boundaries in preparation for the actual count.

Overwhelming stats

Pali Lehohla, Statistician-General said: "When we decided to introduce GIS in the pre-census processes, none of us knew quite how big it would get. The team still cannot believe what they have achieved. The statistics are overwhelming - they verified 94 000 EAs, demarcated 81 000 EAs, quality controlled 81 0000 EAs, produced over 200 000 maps and printed, laminated, bound and distributed 81 000 enumerator summary books, all in a few months."

Due to the magnitude of the data, the backdrop data was extracted in smaller pieces, called delivery units, for distribution to the various sub processes. An Intergraph GeoMedia-based utility was developed to extract the delivery unit data with all the attributes from the Oracle database. Stats SA as well as the office and fieldwork contractors then updated this data, including both the spatial and attribute EA data, with the 2001 EAs. Each delivery unit was packaged in Microsoft Access along with customised software to enable contractors, subcontractors and any other parties involved to update the EA data according to standards specified by Stats SA. The integrity of the data was maintained at all times. The attribute data included the sectional title data (flats, town house complexes etc) and institutions such as old age homes, hotels, B&Bs; this data was used to ensure that the enumeration coverage would be as extensive as possible.

Lehohla continued: "Although time was a serious constraint there was no compromising on quality. The Quality Assurance process did a 100% check on all EAs ensuring the demarcation and attribute data were correct. Automatic attribute validation and report utilities were developed for this process. The EA geometry was also checked to ensure that it was 100% clean."

After passing QA, the cleaned EA delivery units were then imported back into Oracle. A GeoMedia application was developed to automatically renumber the EAs based on their spatial neighbourhood and to group them into supervisor units (a grouping of about five adjacent EAs based on their EA type. During the census enumeration a supervisor was responsible for about five enumerators. About 15 000 supervisor units cover the entire country).

Once the supervisor units had been defined the delivery units were re-extracted from Oracle and supplied to the contractor responsible for producing the census maps. Maps were printed for inclusion in the enumerators' summary book (09-books). Each book contained two maps, one forms the cover of the book and is used for orientation (to find the EA) and the other is a detail map showing the boundaries of the EA. The enumerators used these maps to find the EA and ensure that all the households contained in the EA received questionnaires and were counted.

Web map application

The census web map application was developed using Intergraph's GeoMedia Web Map to monitor progress and shortfalls during the actual counting phase of Census 2001. Data from various sources was presented on maps for progress reporting and identifying and fast-tracking problems in enumeration logistics, human resource related issues such as recruitment, training and payment and census mapping.

The spatial database has worked well for Statistics South Africa and they successfully demarcated the 81 000 new EAs and with the information in the spatial database, were able to ensure a good census coverage. In addition, this opens more opportunity for spatial information to be used by the government in general.

Intergraph Systems Southern Africa

(011) 313 1222