Global vs Local variations

Deletions Deletions to commercial areas are generally low when compared to the amount of additions made between epochs. In Figure 7a Durban had significantly high deletions between late 2009 and early 2010 epoch 6-7. In figure 8a it can be seen that during this time period the 2008 data set was modified a lot, as if the entire data set had been shifting in position in 2009. This explains the increase in additions for this period in figure 6a. In Figure 7b deletions to Residential areas are lower than commercial areas, but this can be expected as the total contributions in residential areas are lower than commercial areas. The suburb Universitas also had high deletion values between early 2008 and late 2010 epoch 6-8. In figure 8b it shows that as with Durban, many modifications were made during this period, accounting for the high deletion values. Brackenfell had high deletion values in 2011 epoch 9. When examining the data sets, it was noted that an administrative boundary was removed from the early 2011 data set. a b Figure 8 Difference in position of line data between two datasets: a Durban – epoch 6-7 and b Universitas – epoch 6-8 a b Figure 9 Low urban density area: a additions and b deletions Figure 9 represents the additions and deletions for the low urban density category. The line colours represent the low urban density areas as follows: red-Gondeni, green-Makhwezini and purple-Stutterheim. Additions figure 9a for low urban density areas were very low when compared to commercial and residential areas. Gondeni had one small addition between late 2010 and early 2011 epoch 8-9. Makhwezini had two slightly larger additions between late 2009 and late 2010 epoch 6-8. Stutterheim had one significant increase, when compared to Godeni and Makhwezini, in 2011. Because additions for low urban density areas are low, the deletions are expected to be even lower figure 9b. Makhwezini and Stutterheim as with Durban figure 7a, Brackenfell and Universitas figure 7b, experienced significant modifications between late 2010 and early 2011 epoch 8-9 which accounts for relatively high deletion values.

5. DISCUSSION

5.1 Activity

Most line features represented in the test data are either roads or railway lines, Road classes range from national routes to footways. Single point features are used to represent amenities, e.g. banks, shops, hospitals etc., Quantity OpenStreetMap is first a repository for road data, thus the quantity of line features far exceeds the point features. But it can be expected that with the passage time the acquisition of point features will outstrip that of linear features as these point features are the most likely to change with time. Quality In earlier years, there were much more modifications to line features than in later years. This indicates that the base data is becoming more stable over time. a b Figure 10 aTwo datasets from 2011 display minor differences. b The same area compared in 2008 displays many more modifications. The structure of the OSM database allows for proper classification for both points and lines. This is however not enforced on the user and as a result most of the features are not classified, but only exists on a general field within the database. The result is variation in attribute information for the same features.

5.2 Densification

Rate of mapping From the results, it becomes clear that the rate of mapping is strongly correlated to the geographical location of an area. Highly populated urban and commercial areas experience greater contributions than towns. This is understandable as such places will contain more people with a culture of sharing information. Points The rate of mapping is very different for the three test area categories, from a steady increase for commercial areas, to a much lower rate for residential areas, to no data for low urban areas. The contributions to point features in commercial areas appear to still be increasing. Lines In cities and high urban density areas, the quantity of data contributed to OpenStreetMap increased dramatically since 2006. It does however appear that since 2010 very few contributions have been made in these areas. Low urban density areas continue to have a low contribution rate.

5.3 Global vs Local variations

Commercial areas had the highest mapping rate for the 2010- 2011 time interval. Residential areas have not had a steady increase in contributions, thus there is no common time period that can be said to have had the most mapping activity. Commercial areas have had the most additions of line features for the 2007-2009 time intervals. Residential and low urban density areas had the most contributions in 2010-2011. XXII ISPRS Congress, 25 August – 01 September 2012, Melbourne, Australia 557 Considering the time period of this study six years it would be expected that low urban density areas would have more contributions. The line features which have been mapped mostly represent the main roads running through the area. These contributions have most likely been made not by the residents but by people passing through. What should however be taken into consideration is the fact that there is generally not many features to map in these areas, thus high data volumes cannot be expected. Although the residents could provide valuable feature information, the likelihood of this is slim due to the lack of resources. The contributions to point features in commercial areas appear to still be increasing while the volume of line feature data have had a much higher mapping rate but have stabilised since 2010. One of the main drivers of a mapping initiative like this is the availability of resources. There exists a digital divide between urban and rural areas. Williams 2001 as cited in Genovese Roche 2009 describes this as the “gap between people with adequate access to digital information and technology versus those with very limited or no access at all”. User motivations play a big role in volunteered mapping. Some of these motivations include: “building professional networks”, “strengthening social relationships” Shekhar 2010 and benefiting others Coleman et al. 2009. The motivation of an interest group will vary with geographic location and therefore the type of data contributed will vary for different areas. This is seen in the comparison of amenity contributions between commercial and residential areas. The influx of tourists into an area does have an influence on the number of contributions as can be seen in figure 6, where Cape Town had a surge of contributions leading up to and during the 2010 FIFA Soccer World Cup period. The appreciation that tourists have for a location may have motivated South African citizens to contribute data. On the other hand the tourists themselves could be responsible for the increase in contributions made.

6. CONCLUSION