I’ve previously referred to maps as representations of reality – cartography as model of geography -, and talked about how models are “fictional or idealized”. Our workshop on georeferencing served as evidence of this notion, given the discrepancies between the maps that we worked with. The construction of maps, though attempting to imitate reality, A) is incredibly complicated when high level technology is not used (drawing an accurate map is no easy task), and B) carries biases that somewhat distort reality.
Georeferencing means “aligning geographic data to a known coordinate system so it can be viewed, queried, and analyzed with other geographic data. Georeferencing may involve shifting, rotating, scaling, skewing, and in some cases warping, rubber sheeting, or orthorectifying the data.” This last part of the definition reveals the nature of maps as models, which differ from each other even if they’re representing the same location.
In the workshop, we used hand-drawn maps of the U.A.E. made by the Soviet Union in the 1970s (obtained from the NYU Spatial Data Repository), a polygon shapefile (downloaded from DIVA-GIS), Google satellite and street views of the region, and a map of Abu Dhabi city distributed by the First National City Bank (FNCB) to direct customers to their building, from the 1960s (before the unification of the United Arab Emirates). This last one was in physical format, found by the NYUAD library inside one of its books. It was scanned (digitized) before we could work with it.
The Soviet and the FNCB maps were georeferenced using ArcGIS, which I found to be very useful due to the versatility it gives its users. We had used CARTO before in class, and our professor had indicated that because it is more user-friendly than software like ArcGIS, it’s also more restrictive (which I agree with). The following are screenshots of what was done with the maps:
Figure 1. Scanned map and shapefile outline in red, before georeferencing.
Figure 2. Scanned map and shapefile outline in red, georeferenced.
Figures 1 and 2 show snapshots of the georeferencing process of 1 of 8 sheets that depict portions of the U.A.E. (these were tiled together over a different base map in Figure 3 below). This particular sheet shows a part of Sharjah, one of the seven emirates.
The process of adjusting the scanned map to the shapefile is evidenced in the change from Figure 1 to Figure 2. To do this, the shapefile was edited so that the polygons that represent the emirates had transparent fill and a red outline. This made it possible to view the two maps together, the scanned image in the layer below the shapefile. Next, coordinate points where “placed” in the shapefile (such as the green point in the corner of the Soviet map in Figure 2). The use of these points was essential. As can be seen in Figure 2, the maps are not exactly the same: their shapes don’t completely match, particularly along the coastline and inlets. Therefore, the attempt to georeference the scanned map solely by these lines was unfruitful. A great advantage of the Soviet maps is that they have coordinates: the latitude and longitude are indicated in the border of the image. This information allowed the corners of the map to be paired up (snapped) with coordinate points on ArcGIS; the parameters for the georeferencing were now quantitative, not qualitative.
The use of these coordinate points, however, revealed a “disagreement” between the scanned map and the shapefile. The last corner, with its latitude and longitude, couldn’t be positioned in the same place as the ArcGIS green point (though they were very close). This is normal in georeferencing, given inevitable discrepancies in the maps’ measurements.
Figure 3. Tiled map sheets of the U.A.E. The red arrow points to the map shown in Figures 1 and 2.
In Figure 2, it is clear that the shapefile and the Soviet map not only differ in the coordinates, but also in the shape of the land in general. In some parts, one of them shows a greater area of water, and in other a greater area of land. We talked about possible explanations for this obvious differences, which were most likely caused by a combination of factors, the main ones being A) natural transformations, B) human intervention, and C) cartographic design choices. Certainly, possible inacurracies due to measurement and drawing errors are part of the mix.
Factors A and B are very plausible, especially given the time that passed between the creation of the scanned map and the shapefile. The shapefile was created digitally, and is thus fairly recent. The scanned map, on the other hand, was created in 1978; however, the data used to draw it was collected in 1975 (a fact we were made aware of by a classmate who can read Cyrillic script). Figures 4 and 5 show where this information was written on the map:
Figure 4. Year when the map was created is circled in light blue.
Figure 5. Year when the data was collected is circled in light blue.
In the case of C, we knew that the Soviet Union had specific interests when creating maps of other countries. In preparation for the workshop, we read “Hyper Detailed Soviet Maps of Washington“,which explains how the Soviet maps of Washington D.C. are very exact in some aspects of the city (and incredibly detailed), such as military installations and water reservoirs, but have errors in others, such as certain building footprints and street names. Thus, the Soviet Union might have had reasons to worry more (and be more accurate) about certain features of the U.A.E.’s geography than others.
We knew from the start that it was going to be difficult to georeference the FNCB map; “simplified street map” is written on it, indicating that the map’s purpose is simple, to show customers how to navigate the city in order to reach the bank. It was likely that the shape of Abu Dhabi city wouldn’t match the base map for this reason, though the time in which the map was created also plays a role (just as with the Soviet map). Figure 6 shows the FNCB map in relation to a satellite view of Abu Dhabi, and Figure 7 shows the same map in relation to a street view.
Because this map has no coordinates, unlike the Soviet map, street intersections were used as points of reference. Figure 6 shows the attempt to georeference the map with some intersections. However, there are still many differences between the base map and the scan, meaning more points would be needed to adjust the FNCB map to the satellite image. Figure 7 shows the change from satellite view to street view, which might have been more convenient and easy to use in this case.
Figure 6. FNCB map and satellite view of Abu Dhabi.
Figure 7. FNCB map and street view of Abu Dhabi.
The differences between the Soviet and the FNCB maps demonstrate how essential information and details are in both the process of georeferencing and of analyzing its results. As models, maps tend to portray reality in a way that is useful/beneficial/convenient for the mapmakers. We ended our conversation with the notion of “the power of maps”, stating that people believe what cartography shows them, often unaware of the biases behind it. We mentioned how Google maps, created with satellite images (which don’t have the same geographical distortions as drawings do), alter the borders of disputed territories depending on where in the world the maps are being accessed (adminitrative distortions?). At a personal level, the issue is especially relevant to me in light of the relatively recent Costa Rica – Nicaragua conflict over Isla Portillos, where Google Maps played a central role.