The data visualizations were created during a Spring 2013 class at Duke University and during fieldwork the following summer in San Carlos, Colombia.
The process of making the maps is outlined below.
1) Clean and re-structure data
The development plans were organized in spreadsheets according to village. The maps, however, needed to illustrate the quality of individual resources at the municipal level, rather than providing an overview of all of the resources at the village level, which was the structure of the development plans. This meant the spreadsheets needed to be re-organized according to resource rather than village.
Students re-structured this data as a class assignment, which complemented readings on the importance and difficulty of incorporating local-level data in national and international policy-making and project planning. This exercise helped illustrate the wide variety in the quality of socio-economic resources across the seventy-six villages in the municipality.
2) Making georeferenced files: AutoCAD to ArcGIS
The planning department of the municipal government provided us with mapping files. These were not georeferenced, however, but instead were AutoCAD (dwg) files, a format used for modeling and design.
Students converted these files using the ArcMap georeferencing tool. They overlaid the map with a geocoded satellite image of all of the municipalities in the department of Antioquia they found in public data available through ArcGIS Online. They chose the municipality of San Carlos from the map’s attribute table, zoomed in so the map of San Carlos’ boundaries filled the screen, and matched the boundaries of the AutoCad map with the georeferenced one.
They used the Select By Attribute tool to select only the village boundary lines from the Autocad file, defined a new layer with this information, and exported the layer to save it as a shapefile (.shp). After careful editing to close any gaps in the boundaries and delete any other unnecessary information, they used the Feature to Polygon tool to convert the closed spaces between the lines to polygons, which is the cartographic term to describe the area of a location. In this case, the polygons were villages.
This process was not as accurate as knowing exact coordinates and thus some parts of the map were not perfectly geo-referenced. If the maps were intended to show precise location information, such as the site of a well or a landmine, this would be a problem. Since the maps reflect information at the village level, however, exact geographic information is not necessary.
3) Making maps in ArcGIS and QGIS
The shapefile students made included an attribute table with the geographical information defining each village, the names of names, and corresponding identification numbers. Students added the re-formatted spreadsheets of the resources in the development plans to this attribute table, merging through the shared field of village names. During this step, there were some errors in matching, so verification was particularly important.
In order to display the indicators by color, they used the Symbology tab in the shapefile properties. All resources had been both numerically and color-coded in the development plans with green=3, yellow=2, and red=1. Villages with no data needed to be coded as 0 for the indicators to successfully display on the maps as they had been coded in the development plans. They used the Layout view to format the map.
Before traveling to San Carlos, we opened the maps in QGIS project files so the municipal government would be able to access and expand upon the maps using open-source tools. When we arrived, however, we discovered municipal officials were using ArcGIS. The use of QGIS turned out to be most useful for us in the field, since we were unable to remotely access Duke’s site license for ArcGIS. It was simple to share shapefiles between the two programs.
Here are several examples of the maps Saira, a student in the class, made and shared with the municipal government:
A word to the wise:
“A practical lesson I gleaned from working with ArcMap is that unexpected and untimely errors will often setback the process. Most of these errors will eventually resolve with no extra effort on my part except for some frustration. For instance, sometimes joining an Excel Spreadsheet with data in our ArcMap did not work, even though this is a simple ArcMap function. It took many tries to get Excel to cooperate. In other instance, the map was not displaying colors to project our indicators. It is important to allot enough time to create these maps with the expectation that one will run into an unforeseen errors.”