The point location table establishes spatial references that correspond to Hope Wright’s original designations while more accurately locating entities with reference to other digital historical gazetteers: the China Historical GIS, the Chinese Civilization in Time and Space Project, and the Robert Hartwell China Historical Studies GIS. Because of frequent changes in the names and ranks of Song jurisdictions, records in the DGSD and the other three digital gazetteers did not always correspond. This was, therefore, a rather painstaking process.
First we recorded the locations of prefectures based on Hope Wright’s designations. She provided latitude and longitude coordinates to only degree precision, yielding a possible spatial location anywhere within roughly 10,000 square kilometers. We also used print and digital reference works to identity the locations of prefectures for which Hope Wright neglected to provide any coordinates. Then, we compared the names, political ranks and approximate locations of prefectures with the corresponding and more precisely georeferenced entries from CHGIS, CCTS, and Hartwell.
From the prefecture point locations, we extrapolated the geolocation of counties based on Hope Wright’s information about the distance and bearing from their prefectural seat. We compared those approximated coordinates with those of the relevant entities in CHGIS, CCTS, and Hartwell, and used the best location from one of those sources wherever possible.
The small number of counties that could not be located within their parent prefecture based on the available data are located at a purely random point within a set radius from the prefectural seat based on the average distance from that prefecture seat of its other constituent counties. These point locations, and others that are not based on historical records, are labeled in the Notes section as “Extrapolated.”
When different coordinates for a single entity existed in CHGIS, CCTS, and Hartwell, we selected the best possible estimate based on our assessment of the precision of each gazetteer. We prioritized CHGIS first, followed by CCTS, and finally Hartwell. 18% of the prefecture coordinates could not be readily be identified with locations in any of the other gazetteers, and we have continued to use Hope Wright’s approximations or our own research. The source of each coordinate is identified in the Notes field.
3.a Unique ID
A four-digit numerical identifier unique to each point location entry.
3.b Entity ID
A six-digit numerical identifier that designates the entity associated with the geo-data.
3.c X/Y Coordinates
The longitude and latitude of the location. Data precision ranges from 0 decimal places for Hope Wright-sourced data to 4 decimal places for CHGIS-sourced data.
Information about the source of the geodata. Data created by the DGSD is identified with a particular reference or noted as extrapolated.
3.d Source ID
The original source of each piece of geodata.
The DGSD maintains the original Hope Wright point location data even when we have reference to more accurate point locations from other sources. This facilitates error-checking and review. However, where there is superior data, it will be listed as Priority 1, and the Hope Wright data as Priority 2. When creating shapefiles from this data, queries will need to account for data priority preferencing.
Changes in Point Location From Version 1.0 to Version 1.1
Version 1.1 DGSD geolocations are better aligned with the CHGIS. We matched DGSD data with information in the gis_info table of the CHGIS to identify shared traditional characters and feature types. We allowed a tolerance of 5 total degrees of difference from locations in the DGSD that were extrapolated from prefectural location and metropolitan distance recorded in Hope Wright’s Alphabetical List. Next, we matched DGSD shapefiles with CHGIS time series shapefiles for prefectures and counties. We required traditional characters and feature types to be shared in order declare a match for any entities, allowing 10 total degrees of the DGSD+gis_info locations. If entities shared characters, feature types, and change dates, we identified them as a match even if the locations differed by more than 10 total degrees. There were five entities in that category: Linjin county 臨津, Yongning county 永寧, Si’en county 思恩, Xiong 雄 zhou prefecture, and Anhua 安化 county. Since the CHGIS includes relatively few Song prefectures, we also identified DGSD prefectures that matched CHGIS county data. As a result there are prefectures in the DGSD with geolocations drawn from CHGIS data that are not direct matches, but that were counties with CHGIS matches at one time in their history.