The Project GIS

 The Project GIS

To create the GIS, I used the ArcMap and ArcCatalog components of ArcGIS 10, for which my institution has a site license. I have been using GIS on and off since the late 1980s. Over the past two decades, GIS has gone from typing in commands to selecting them from pull-down menus, from relatively little functionality to high functionality, and from static dot-matrix printouts to high quality, dynamic output on the web. In parallel, the data sources have gone from having to convert them to digital format yourself, to a limited range becoming available on floppy disk, to many becoming available on the web in digital form. The latest innovation, and the one that prompted me to get into a relatively large GIS project for the first time, has been the addition of a temporal function to ArcGIS, finally making it quite easy to display changes in spatial patterns over time. Its productivity has not kept up with its impressive functionality, necessitating a lot a mouse clicks and keystrokes to accomplish many tasks that could be much better integrated, and its temporal functions remain limited compared to its spatial ones. But ArcMap has reached the level of utility that makes it essential to my research and instructional programs and, in fact, for anyone who does historical research.

The first stage of turning the prior work with the databases into a GIS involved setting it up and importing the data. I will be a bit more step-by-step about this stage because I assume those interested in the digital humanities have less experience with ArcGIS than with Excel.

For the initial setup of the GIS, I created a folder named ACLS_GIS to keep the project in, started ArcMap, used the File menu to start a New GIS, and saved it in the folder as ACLS_GIS.mxd. Then I went to the Customize menu and selected ArcMap Options to make sure that “Make relative paths the default” was selected, which makes it easier to move the project to another folder or computer. At the same time, you can check around the other tabs of the ArcMap Options dialog box to set other options to your personal preferences. Then I used the Windows menu to open the Catalog window, navigated to the folder ACLS_GIS, right clicked in it, selected New, selected File Geodatabase (not Personal Geodatabase, which is more limited), and named the resulting file ACLS_Geodatabase.gdb. That database, indicated by a special icon, has a specific format used by ArcGIS, will contain many of the files needed for the GIS, and will keep track of them and their relationships. The last step of the setup involved going to the File menu, selecting Add Data, selecting Add Basemap, and selecting one of the base maps that appear, everything from a basic gray outline to imagery from the Bing Maps Aerial view. The base map is the underlying map on top of which the data will be represented, and it can be changed for one of the other base maps at any time. I selected the Terrain base map, which includes four sub-layers named Reference, Borders and Places, Shaded Relief, and Terrain. For reference in the next step, the base maps use the Web Mercator projection of the WGS 1984 geoid, also known as the World Mercator; its projection reference file is named WGS_1984_Web_Mercator_Auxiliary_Sphere, and its datum reference is named D_WGS_1984 (more on this in the next step).

The next step involved adding the data. To add the worksheets in the Excel workbook named ACLS_Voyages_Data.xlsx to the GIS, I used the Folder Connection button (a folder icon with + on it) at the top of the Catalog Window to navigate to the folder that contained the workbook and establish a link to the ACLS_GIS folder and its ACLS_Geodatabase. I then opened each of the worksheets in turn (for example, Middelburgs_Welvaren_A$), right clicked it, and selected Create Feature Class From XY Table to open the dialog box that adds the data to the GIS. In the dialog box, I selected the longitude column as the X coordinates, the latitude column as the Y coordinates, and GCS_WGS_1984 as the Geographic Coordinate System of the Input XYs. Next, under Advanced Geometry Options, I selected Use Map Spatial Reference, which was the previously mentioned WGS 1984 Web Mercator. Those selections told the GIS the geographic reference of the data being added and the way in should project them onto a flat map. Without that information, the GIS has no idea if all those latitude and longitude numbers in the worksheet are in degrees, meters, kilometers, or something else; and it has no idea how to project those positions, which were taken from the curved surface of the Earth, onto a flat map. Then, still in the dialog box, I navigated to the ACLS_Geodatabase and saved the resulting Feature Class Files with the name of the vessel (for example, Middelburgs_Welvaren_A). Opening the ACLS_Geodatabase shows all of those files with an icon that indicates they are point features. After creating point feature class files for each of the voyages, I did the same for the Ports_and_Landmarks worksheet. (Note that if ArcMap does not recognize an Excel worksheet, saving it in an older version of Excel typically resolves the issue, in other words, as file.xls instead of file.xlsx.)

The next step results in all those data actually appearing on top of the base map, a gratifying moment in any GIS project. I used ArcMap’s Windows menu to open the Table of Contents window, selected the List by Drawing Order button at the top of that window (on the far left), went to the open ACLS_Geodatabase in the Catalog Window, and dragged each of the Feature Class Files to the Table of Contents under layers, in alphabetical order and on top of the base map. The order of the layers in the Table of Contents represents the order of the layers on the map, with all the voyages therefore drawing on top of the base map. All the voyages then appeared as a series of linear point patterns, the overall pattern emphasizing the triangular routes of the slave trade in the North Atlantic. Checking or unchecking on any of the layers in the Table of Contents turns that layer on an off. Dragging them rearranges their order in the Table of Contents and on the map. Right clicking any of the layers and selecting Open Attribute Table opens a table that shows all the cargo, date, latitude, longitude, and other fields associated with each of the features, or daily positions, in that layer. Selecting a particular day or range of days in the attribute table, selects its point symbol on the map and vice versa.

Once all the data have been imported into the GIS and represented as point symbols, the next stage involves the use of various tools to alter the data layers and derive entirely new types of layers from them. Those tools are accessible in ArcMap by selecting the Geoprocessing menu and then ArcToolbox or by starting ArcCatalog.

The first tool I used was the Convert Time Field tool to get the dates of each position into a standard ArcMap format to use the temporal function. I opened the tool (in the ArcToolbox under Data Management, then Fields), navigated to and opened the ACLS_Geodatabase, selected each of the voyage Feature Class Files in turn, selected their date field, which was in a yyyymmdd format, and converted them to the ArcMap, standard time format. Opening the attribute table revealed that the tool has added a new field, DATE, to the attribute table and populated it with dates in the correct format while leaving the original yyyyddmm date field intact.

Next, I used the Points to Line tool to create lines to underlay the points that defined each voyage, which I believed would better define the route of each voyage by connecting its daily positions and increase the symbolization, labeling, and other visualization opportunities. I opened the tool (under Data Management, then Features), navigated to and opened the ACLS_Geodatabase, selected each of the voyage point Feature Class Files in turn, selected the field that contained the voyage name as the Line Field, and saved the resulting line Feature Class files in the ACLS_Geodatabase, naming them for the voyage with _line appended (for example, Middelburgs_Welvaren_A_line). I then dragged each of the line Feature Class Files to the Table of Contents, placing them immediately underneath their respective point layers. A check of each line layer’s attribute table reveals that it contains a field with the voyage name, which can later be used to label the line. Then I selected Start Editing from the Editor menu and used the split tool to cut each line at the beginning and end of the segments of the voyage, deleting the superfluous segments: for example, the segments that cut overland across West Africa from the coast lf Liberia to the coast of Ghana because of the missing days and positions while the vessel followed the coast trading for slaves. This changed the attribute table for each line by adding a record for each new segment. Finally, in the Table of Contents, I selected both the point and line layers for each voyage, right clicked, selected Group from the menu, and combined them as sub-layers of a single layer, naming it after the vessel and voyage (for example, Middelburgs_Welvaren_A).

I then customized the way in which the GIS represented each voyage by right clicking each layer and selecting Properties. The dozen tabs of that dialog box allow you to set everything from how the temporal display works, to the style and color of the symbols, to the labels, to the scales (zoom levels) at which the layer is visible. To set the time function for the point layers, for example, I selected the Time tab, enabled time, selected the DATE field, and set the time interval at 1 day. To label each point or line segment, I used the Label tab, selected label features, selected the date field for the points layers and the voyage name layer for the line layers, and set the scale range at which the labels would come into and out of view. To connect to TSTD, I used the HTML Popup tab to select the TSTDURL field to open a search results window for a particular voyage in that database’s website whenever someone selected one of the point symbols for that same voyage in the GIS.

The most time intensive task turned out to be setting the symbols that would represent each daily position. To do that, I used the Symbology tab in the Properties dialog box. Since I wanted each daily position to reflect the cargo, I selected Categories, Unique Values, and the Cargoes field in the attribute table. I then assigned standard colors to each type of cargo: gray for unknown, brown for tasajo, and so on. In addition to the standard symbols included in ArcMap, users can create their own symbols and save groups of symbols under unique headings so the process becomes somewhat more productive.

By that point, the GIS had much of the data and functionality I had planned. With all layers visible, it displayed all the voyage segments as lines overlain with the daily position points, each with a unique symbol that indicated the cargo. At broad scales all the voyages blended together and large-scale patterns were evident, such as the triangular trade between Europe, Africa, the Americas, and back. Zooming in revealed individual voyages, their lines labels by name and their points labeled by date. Queries to search for particular attributes (Selection menu, then Select by Attributes) could be run to select all voyages that carried, for example, enslaved Africans. And, when time was enabled by selecting Open Time Slider Window (the clock icon on the main menu bar), and the GIS run through its temporal extent, the point symbols associated with each day appeared and disappeared in sequence. Selecting Time Slider Options from the Time Slider Window menu bar allowed changes to the length of the period represented, the speed at which it ran through that period, and so on. Such functions were enhanced by running the Add Attribute Index tool (under Data Management, then Indexes) for each of the Feature Class Files so that ArcMap could quickly locate records.

The final major task to ready the GIS for use involved correcting the positions added in Excel to represent missing days, latitudes, and/or longitudes as well as check the integrity of each of the other daily positions for each voyage. I accomplished that objective by advancing through each voyage, day by day, position by position, with its attributes table open and editing turned on. Positions that had been added for missing days with last known latitudes and longitudes, for example, where selected in the attribute table and dragged to interpolated positions between known ones. Position makers that appeared too close to shore were moved off shore on the bases of the attribute table information, which sometimes named a landmark, its bearing, and its distance. Position markers that had been given in the logbooks but appeared erroneous relative the preceding and following days were also corrected at this point, for example, when a logbook reported a positive latitude as a negative one or vice versa, as sometime happened near the equator.

Once all the voyage data were complete and correct as far as the available data allowed, the hurricane Excel worksheet was converted to a points Feature Class File and added as a layer in the Table of Contents. I used the ArcToolbox to create a DATE field, time enabled the layer, and applied symbols to differentiate among categories of storms.  Even reduced to 1851-1900, the attribute table had 9366 records.

By this point the GIS was highly functional. It made it possible to query intercepts in time and space between hurricanes and vessels. It revealed the pattern of deaths and disposals overboard on Dutch slaving voyages, the captains of which have left us a daily record of such deaths in the logbooks. And therefore the time had arrived to begin implementing it, or at least parts of it, as a Web GIS.

Selected References

1. Bodenhamer D. J., J. Corrigan, and T. M. Harris, eds., The Spatial Humanities: GIS and the Future of Humanities Scholarship (Bloomington: Indiana University Press, 2010).
2. Gregory, I. N., and P. S. Ell, Historical GIS: Technologies, Methodologies and Scholarship (Cambridge: Cambridge University Press, 2008).
3. Hunter, Richard, Methodologies for Reconstructing a Pastoral Landscape: Land Grants in Sixteenth-Century New Spain, Historical Methods 43 (2010): 1-13
4. Knowles, Anne Kelly, ed., Past Time, Past Place: GIS for History (Redlands, California: ESRI Press, 2002).