Akira Ueda
What is Historical GIS?
Geographic Information System (GIS) is a category of computer software that visualises, integrates, compares, and calculates various data in a spatial context. Recently, many popular applications have integrated these functions. Online map services that include shopping and restaurant information or GPS-related pedometers in smartphones are a type of GIS. Notably, the COVID 19 crisis highlighted the utility of the GIS technique to visualise big data. ArcGIS and QGIS are common forms of GIS software used for business and academic purposes.[i]Both software packages have a wide range of spatial investigation functions. When GIS became popular in the 1980s, historians began using this new tool, which many scholars now use.[ii] In general, the information collected from various historical sources is organised as independent layers in a single GIS map. Comparing the layers allows the visualisation of the relationships between data and changes in regions over time.
GIS and Historical Sources
Historians face specific problems while using GIS, as historical data usually do not contain latitude and longitude information.[iii]Administrative workers, sociologists, and even archaeologists can obtain the correct latitude and longitude of objects using GPS. In many cases, historians have to rely entirely on place names written in narrative sources. If they do have access to historical maps, these are often without a geographic coordinate system. Place names change over time; political regime changes often cause a place name change. All post-Soviet countries experienced a series of place-name changes after the Russian revolution. We can trace the history of renaming famous cities, but it is difficult to reconstruct place-name changes in local and village areas. The georeferencing process in GIS is designed to convert paper maps without coordinate information into a digital format. The georeferencing process transforms the paper map into a digital map by connecting known coordinates (Figure 1). Known points consist of cities, roads, train stations, and capes that have not changed their locations. By using georeferencing, we can add temporary latitude and longitude to places recorded only on old paper maps. Even if the result of the first georeferencing is not perfectly correct, georeferencing is necessary for all GIS users because GIS cannot work without latitude and longitude data. We can correct incorrect locations in any phase after the first georeferencing.
Figure 1: How to integrate historical maps into GIS format Source: author
How To Use GIS in Historical Study
I would like to present three examples of historical GIS studies from my work: visualisation, time-series change, and spatial analysis.[iv]
1. Visualisation
Figure 2: Rainy and irrigated sowing in 1913 Source: author
The first example was chosen to demonstrate how GIS visualises historical data. I studied the characteristics of agriculture conducted by semi-nomadic Kyrgyz people and tried to clarify the difference in agriculture between sedentary and semi-nomadic people. Regarding historical sources, the Fergana region is suited for quantitative analysis because the Russian empire annexed this area by abolishing the Kokand Khanate in 1876 and conducted a series of statistical surveys on population and agriculture.
The Russian empire conducted a land survey of Central Asian nomadic and semi-nomadic regions to prepare for the colonisation of Russian peasants.[v] Though the purpose was problematic, the results of this survey remain a very important historical source that recorded the economic activity of nomads and semi-nomads before the Soviet reform of nomadic society. Figure 2 shows that the semi-nomads in the foothills mainly engaged in rainy sowing, unlike the sedentary population in the Fergana Valley. This difference cannot be explained only by natural conditions such as precipitation because semi-nomads far from the valley prefer irrigated sowing. If only the amount of precipitation had allowed the expansion of rainy sowing, semi-nomads far from the valley would have expanded their grain production with rainy sowing. This map raises the question of why semi-nomads near sedentary areas prefer simple but unstable rainy sowing rather than irrigated ones and, contrarily, why semi-nomads far from the valley did not expand rainy sowing, despite experiencing enough rainfall. A narrative source explains that the Kyrgyz semi-nomad engaged not only in agriculture and pasturage but also in small trade and seasonal labour in the valley. Kyrgyz semi-nomads multi-income, maintained by the economic connection with the sedentary population, enabled semi-nomads to expand ‘simple but unstable’ rainy sowing. This suggests that connections with sedentary people allowed agricultural development through the Kyrgyz semi-nomads.
2. Time series change
Figure 3: Ethnic data at the village level in the Fergana Valley Source: author
The second example shows how GIS reconstructs time-series changes. Figure 3 shows how ethnic names in official statistics changed dramatically after the Russian revolution and the establishment of Soviet Uzbekistan.[vi] These two maps were based on the list of settled points in Fergana province in 1909 and the All Soviet Census in 1926. The most evident change in the maps is the disappearance of the Sart and its change to Uzbek. Although this phenomenon has been well-researched in previous studies,[vii] the spatial characteristics of the change are informative. For example, some non-Uzbek settlements changed to Uzbek, while other settlements maintained the former ethnic name. This difference did not occur randomly, but with some spatial tendency. Figure 3 shows that non-Uzbek settlements isolated in the former Sart /Uzbek area more often changed their ethnic label in official statistics than in the non-Uzbek dominant area. Another finding from Figure 3 is related to the population having originated from East Turkestan/Xinjiang. They were recorded as Kashgar in 1909 and Uygur in 1926, but Figure 3 shows that the two categories did not spatially continue. Many Kashgar settlements in 1909 were recorded as Uzbek-dominant settlements in 1926, and most Uygur settlements in 1926 were recorded as Sart dominant in 1909. Although the cause of this situation remains unclear, it seems that mass refugees in Xinjiang in the 1916 revolt and their return and the Uygur national movement after the Russian revolution might have influenced this discontinuity.[viii]
3. Spatial analysis
Figure 4: Hotspot analysis on cotton planting in the Kokand oasis
Figure 4 shows how GIS can help researchers identify spatial tendencies from historical raw data. Hotspot analysis indicates a statistically significant area from the spatial dataset. This image suggests that hotspot analysis clarifies where the cotton ratio is high in an oasis. The processed image (right) shows the cotton monoculture district more clearly than the simple visualised law data (left). This result can be compared to Figure 3. This comparison shows that cotton monoculture in the Kokand oases expanded mainly in the peripheral area of the oases. In addition to the Sart population in the eastern wing, both the Karakalpaks in the north end and Uzbeks in the west wing participated in cotton cultivation. It seems that Sarts located in the central part surrounding the Kokand city were reluctant to participate in the cotton boom in the colonial era.
The three examples above indicate that GIS is suitable for demonstrating certain tendencies in big data and evaluating the statistical significance of spatial tendencies. In contrast, it is difficult to determine the cause and effect using GIS alone. Hence, it is necessary to use narrative sources to argue for the cause of any tendency in data. Traditional approaches to narrative sources do not lose any importance. However, historical GIS allows us to utilise fruitful data recorded in statistics. In other words, historians who adopt a traditional approach can use only the last line of hundreds of pages of a statistic, but GIS historians can use entire statistical tables for reconstructing past societies.
Akira Ueda is an area researcher at the Institute of Developing Economies JETRO in Chiba, Japan. He received his PhD from the University of Tokyo in 2017. A revised version of his dissertation titled 'Cotton and Nomad: the GIS-based economic history of the Ferghana region' was published by Hokkaido University Press in 2020 (in Japanese).
[i] ArcGIS (https: //www. esri. com/); QGIS (https: //qgis. org/) [ii] See. I. N., Gregory, and A. Geddes eds. Toward Spatial Humanities, Historical GIS & Spatial History (Bloomington: Indiana University Press, 2014). [iii] Even if a historic map contains latitude and longitude information, historians must conduct a georeferencing process on it. For example, see A. Kitamoto and Y. Nishimura’s works on Aurel Stein’s maps. Kitamoto and Nishimura’s other project on an old map of Beijing tells us that georeferencing on a city consisting of numerous straight roads requests a different solution (http://agora.ex.nii.ac.jp/~kitamoto/research/publications/cipa09a.pdf). [iv] See below for more specific arguments on these topics. A. Ueda “The demographic and agricultural development of the Kokand oasis during the Russian Imperial era: nomad immigration and cotton monoculture,” Central Asian Survey 38, no. 4 (July 2019): 510-530, https://www.tandfonline.com/doi/abs/10.1080/02634937.2019.1631754; A. Ueda, Cotton and Nomad: the GIS-based economic history of the Ferghana region, Hokkaido university press, 2020. (in Japanese); Jeanne Féaux de la Croix et al. “Roundtable studying the Anthropocene in Central Asia: the challenge of sources and scales in human–environment relations,” Central Asian Survey, Published online: 08 Sep 2021, https://www.tandfonline.com/doi/full/10.1080/02634937.2021.1960797. [v] The survey calculated the amount of excess (излишняя) land for the future colonisation of Russian peasants (Материалы покиргизскому землепользованию: Ферганская область: Наманганский уезд, 1913, С. IV). The background of this survey was Pyotr Stolypin's famous reform at the center of the empire (П.Н. Шарова, Переселенческая политика царизма в Средней Азии, <<Исторические записки>>, Т. 8, 1940. С. 9). [vi] Certainly, ethnic names in official statistics are only one aspect of the ethnic identity of Central Asian natives. See. J.S. Schoeberlein-Engel, Identity in Central Asia : construction and contention in the conceptions of "Ozbek," "Tajik," "Muslim," "Samarquandi" and other groups, Harvard University, 1994. [vii] For example, see. F. Hirsch, Empire of Nations: Ethnographic Knowledge & the Making of the Soviet Union, Ithaca: Cornell University Press, 2005; A. Khalid, Making Uzbekistan; Nation, Empire, and Revolution in the Early USSR, Ithaca and London: Cornell University Press, 2015. [viii] It is well-known that sedentary populations in Ferghana and Xinjiang are very similar in language, religion, and social habits, and migration among them was a common phenomenon. The ratio of the Ferghana population, which originated in East Turkestan, might be extremely underrated in colonial and modern statistics. See pages from 217 to 220 of С.Н. Абашин's “Чай в Средней Азии: История Напитка в XVIII–XIX веках,” (С.А. Арутюнов и Т.А. Воронина ред. Традиционная пища как выражение этнического самосознания, Москва: Наука, 2001, С. 204–231.)