historical geocoding and the city
TRANSCRIPT
Atlanta ExplorerAn Example in Historical Geocoding & the City
Michael Page, GeographerEmory Center for Digital Scholarship & Department of Environmental Science
Atlanta
Explorer
Project
(modified)
1. Construct a circa 1930 historical geodatabase of
Atlanta – 1878 now in production
2. Provide the community with a historical
geodatabase model and methodologies – We are
now partnered with University of Sao Paulo,
Brazil to help map their city
3. Create historical batch geocoders of Atlanta to build
new datasets and facilitate research in Atlanta
Studies - expanded the geocoder years from just
1928 to prior 1878 - 1930
4. Create an online search tool for the public –
developed a 3D/VR prototype and now retooling
our buildings database and models
5. Produce example data sets -- started with racial
segregation now includes historical epidemiology
and several other geospatial layers
6. Provide a base map data of historic Atlanta for
cartographers to use – will be expanding available
years to 1878 to 1930
Atlanta ~1930 Topo
Sao Paulo ~1930 Topo Mosaic and Historical Overlay
HYDROGRAPHY
ADMINISTRATIVE
BOUNDARIES
Data Extraction
Absolute (exact buildings are matched) vs. Relative (uses street network for address interpolation) Geocoders
Map showing new house numbering systems, courtesy University of Chicago
Student effort at Emory to digitize places and match with addresses (structures)
Students use machine learning in the Center for Digital Scholarship to fix errors in OCR’ed text
Opportunity: Speeds the process of creating address databases that can be matched to geography; working through years sequentially reduces the amount of manual corrections
Problems: Python code must be modified for each year because directories were structured differently and different typewriters were used; in some years there were major changes to street names or the numbering system
Atlanta Street Network Historical Morphology
Orange = streets added since 1928Red = streets that have been removed
Blue = streets that exist both now and in 1928Green = unpaved roads, alleyways, paths
RULES OF HISTORICAL GEOCODER UNIQUE IDENTIFIERS
• The unique identifier is meant to always equate to a specific structure as situated in a specific location.
• What might change year to year with a record is the box number and/or street name, and/or the person who is listed as owner or occupant.
• New “structures” are always given the next identifier in sequence. If a place no longer has a building or if a street is not there (this will occur frequently as we work backwards in time) then the identifier (and record) is removed from that specific year.
• It is possible that more than one unique identifier shares the same geographic coordinates across time if one structure has been removed and another erected.
• These unique identifiers reflect an record (entry) in the city directories from construction to removal of the structure.
Original prototype of a few city blocks centered on Atlanta’s Flatiron Building
Incorporating VR/AR in future prototypes
New questions arise : How do we reconstruct the historical topography of cities?
Collective Access AWS