Web GIS
To
create an initial implementation of the Web GIS for the project, I used the free,
non-subscription version of ArcGIS Online, not only because my institution does
not have a subscription to ArcGIS Online but because I want to make this
project as useful as possible to people with limited resources. Anyone with a broadband
connection and a computer can now use the Web GIS published in the
second post to this blog as the basis for their own Web GIS by modifying
it, adding to it, deleting parts of it and saving their creation to their own free
account on ArcGIS Online.
The
first step in publishing the GIS on ArcGIS Online involved converting the
feature class files, both line and point, to a format that the free version of
that software accepts. In order to reduce the work involved in what was a
test, being my first experience with web mapping, I copied the GIS and renamed
it Slave_Voyages_WebGIS. Since I had used relative paths, the copy worked
perfectly. I then used a select by attribute query to find all the voyages that
carried enslaved Africans and deleted all the layers for other voyages as well
as the hurricane layer because the slave trade had ended by the time the
hurricane record begins in 1851. Then, to reduce the size of the files in the Web
GIS and ensure it ran smoothly, I used the Delete Field tool (in Data
Management Tools, then Fields) to delete all fields except VesselID, Flag,
DayOfWeek, Latitude, Longitude, TSTDURL, Cargo, and DATE from the attribute
tables of the point Feature Class files. Then I used the Feature Class to Shapefile tool (in Conversion Tools, then To Shape file) to export all the point and
line Feature Class files from the geodatabase to a new folder. Opening that
folder reveals that each feature class exports as several files, including the
essential file.shp, file.prj. file.dbf, and file.shx files. Each includes
essential information such as the projection, spatial reference, point or line locations,
and attributes. Selecting all the files associated with a point or line feature
class, right clicking, and selecting Send to Zipped Folder creates a compressed
folder that the free ArcGIS Online accepts as an upload, as long as it does not
contain more than 1,000 features.
Next, I
started a new map from within my free ArcGIS Online account by selecting the Map
link, then the New Map button, and then the Basemap button and adding the same Terrain
base map I used in ArcMap. I then saved the new map to My Content with the name
Dutch
and British Atlantic Slave Trade Voyages, 1751-1795. Once that GIS had been
saved to the ArcGIS cloud, I was able to add all the point and line layers for
the 48 slaving voyages by selecting the Add Content button, selecting Add Layer
from File, navigating to the folder of shape files on my computer, and
uploading each in turn.
Those shape
files do not contain the symbology and other features, such as the visibility range,
of the layers in the GIS from which they were exported, necessitating a lot of work
to reestablish appropriate symbols to indicate cargoes and so on within ArcGIS
Online. That is accomplished laboriously but straightforwardly through the
Contents panel of the Web Map and a menu
that appears when the arrow to the right of each layer is selected.
The
popup box associated with clicking on each feature, line or point, is
particularly useful in a Web GIS because it can include URL links to other
Websites. Most basically, I configured the popup boxes for the point symbols for
a particular voyage to include the URL for the search window in TSTD for that
same voyage, allowing users to immediately connect the data and representations
of one project to those of the other. I also ended up importing the
Ports_and_Landmarks shape file to identify places relevant to the slave trade,
from landmarks like Cape Mount, to ports like Paramaribo, to archives like the Zeeuws
Archief in Middelburg that hold the logbook data on which the GIS is
based. Each popup can contain a relevant image and “Get more info.” link to,
for example, the Website of the archive or a Wikipedia entry on the place. I
also added popup boxes to some of the lines that represent the routes of the
vessels, in this case linking to the shipping news sections of online newspaper
archives so that clicking on the line representing a specific voyage brings up
an image of the newspaper that contains a link to the reference in its shipping
news section for that specific voyage.
I also
wanted to represent one of the patterns that became clear in the ArcMap version
of the GIS, namely the pattern of deaths of enslaved Africans during the Middle
Passage and the disposal overboard of their corpses. The Dutch logbooks
recorded the number of deaths on each day of the crossing, making it possible
to visualize that horrific aspect of the slave trade. To do so, I used the
Merge tool in ArcMap (in Data Management Tools, then General) to generate a
single layer that combined all the individual voyages but eliminated all fields
except for VesselID, Cargo, and Occurrences, the last of which contained the
number of enslaved who died on each day at each position, from a low of 0 to a
high of 5. That created a new point Feature Class File with 8,889 records, or
rows in the attribute table. I used various tools (Add Field, Find &
Replace, Field Calculator, and Sort) available through the Table Options button
in the top-left corner of the attribute table window or through right clicking
the column heading to create a new short-integer field called SlaveDeaths and
populate it with the number of slaves who died at each latitude, longitude on a
particular day. I then eliminated all records with zero deaths and ended up
with a layer with only 389 records that I could export, compress, and upload to
ArcGIS Online in the same way as the other layers. Once added to the Web map, I
assigned different diameter point symbols to each number of deaths. The pattern
is striking enough with that representation, but I also wanted to visualize it
as a continuous density surface.
To
represent the pattern of deaths as a density surface, I first ran the Kernel
Density Tool (in Spatial Analyst Tools, then Density) in ArcMap. It calculates
the density of occurrences, deaths in this case, around each data point, the
vessel positions in this case, and creates a floating point raster layer as its
output, in other words, a layer that consists of square cells that each have a
value such as 0.349759 or 2.897347. The problem was that the free version of
ArcGIS Online does not allow you to upload raster layers. The raster layers must already exist within ArcGIS Online, for example, the base maps, which are raster layers. I used three additional tools to
work around that limitation. First I used the Raster Calculator Tool (in
Spatial Analyst Tools, then Map Algebra) to multiply all the raster values by
10,000, which changed the output from deaths per square kilometer to deaths per
10,000 square kilometers but would stop the next tool from truncating most of
the significant figures. That next tool was the Int Tool (in Spatial Analyst
Tools, then Math), used to convert a floating point raster file to an integer
raster file by truncating all of the cell values. And the last tool was the Raster
to Polygon Tool (in Conversion Tools, then From Raster), which converts integer
rasters, but not floating point rasters, to a series of lines that define polygons. Once
that polygon layer had been created, I exported it as a shape file, compressed
the results into a zipped folder, uploaded it to ArcGIS Online, and applied colors to each polygon to
represent different values of deaths per 10,000 km2.
The free
version of ArcGIS Online does not support uploading temporal layers, so the voyages
cannot be run as a temporal sequence in the Web GIS. But future posts will
explore solutions to that issue.
In the meantime, here is a prior temporal animation of the CLIWOC database for purposes other than historical climatology; here is another; and here is a nice static representation as well as one that shows seasonal patterns.
In the meantime, here is a prior temporal animation of the CLIWOC database for purposes other than historical climatology; here is another; and here is a nice static representation as well as one that shows seasonal patterns.
No comments:
Post a Comment