Toward a Deep Map of Chang’an: Some Reflections of a (Beginning) Practitioner

Chang’an (modern Xi’an, Shaanxi province) is one of the most renowned cities in Chinese history. It was the capital of the Western Han (202 BCE–8 CE) and the Tang (618–907) dynasties: the twin peaks of imperial power in premodern China. In its heyday during the Tang dynasty, the area covered by the walled city (84.1 square kilometers) was even bigger than many modern metropoles, such as the island of Manhattan (59.1 square kilometers). When you visit the modern city Xi’an now, you can still see the remnants of the Han dynasty city of Chang’an to the northwest of the modern city center and, much more prominently, of the Tang city of Chang’an that overlaps with a large part of the modern city. As I witnessed firsthand walking the crowded streets of Xi’an during the sweltering summer of 2023, there is a tremendous amount of interest in the history of Chang’an among the general population, partly because of the remains of the ancient city, partly triggered by fictional works like the popular novel and TV drama The Longest Day in Chang’an.

Because of Chang’an’s historical significance, there have been efforts to map it since the Han dynasty.¹ The earliest extant maps date to 1080 in the Northern Song dynasty (960–1127), when a group of local elites created a map of the city itself and maps of the palaces.² These maps are also the earliest city maps in Chinese history. Subsequently, maps of Chang’an were made often as a part of a “local gazetteer,” a kind of geographically based compilation of local historical information.³ Such production continued into, and in many cases beyond, the Qing dynasty (1644–1911). In the twentieth century, historians and geographers began producing a kind of “historical map” (lishi ditu) that was intended to show the geographical contours of a place in a specific historical period. Chang’an, once again, was a center of scholarly interest in the production of such maps.⁴

During the building of an extensive subway system in 2006, many ancient tombs, old sites, and forgotten treasure hoards were excavated and rediscovered.

Yet, two trends that accelerated in the late twentieth century threaten to make these thoughtfully crafted historical maps obsolete, while offering new opportunities to reconceptualize how we represent Chang’an geographically. The first is new archaeological discoveries that have been appearing at a breathtaking pace. Because of the explosive economic development of Xi’an in the past half century and the accompanying construction of new infrastructure — notably during the building of an extensive subway system in 2006 — many ancient tombs, old sites, and forgotten treasure hoards were excavated and rediscovered. In the span of one year from 2021 to 2022, for instance, archaeologists found over 1,200 old tombs just from the west side of the city.⁵ These finds substantially changed our understanding of the shape of the city. When I visited an archaeological site in the summer of 2023, I observed how archaeologists uncovered a previously unknown section of the Zhuque boulevard — the main north-south axis of the Tang city — and found five unrecorded bridges over the boulevard. Such new discoveries far outpace the revision of traditional print books and atlases and cannot be properly and promptly reflected in these works.

At the same time, the development of GIS (Geographic Information System) technology, the use of portable GPS devises, app-based location services, and new developments in the field of digital humanities also made it possible to imagine a new kind of “deep map” that can incorporate and consolidate a huge amount of data, and to visualize them in ways that are easy to digest and manipulate. This deep map ideally will incorporate all the existing historical data about Chang’an, store and organize those textual, visual, and archaeological data in accessible and sensible ways, and visualize them in various forms for open-access use. This essay reflects some of my initial thinking as I begin to conceptualize this project, one that is certainly beyond the capacity of one person. To do so, I first examine existing maps of Chang’an produced by premodern cartographers and modern scholars, and discuss their uses and limits. Then I assess the nature of the data now at our disposal that are the basis of a “deep map” of Chang’an, and envision what kind of things this new map can do for us, as researchers, students, and even tourists of modern Xi’an.

Existing Historical Maps

To make new maps, we need to understand what “old” maps look like. Although maps were continuously made about Chang’an from the Han to the Tang dynasties, the earliest preserved map of Chang’an dates, as mentioned earlier, to the eleventh and twelfth century in the Northern Song dynasty (960–1127). Along with the map itself, a colophon was carved on a stone and preserved in later written sources, that allows us to understand the cartographic thinking of its creators. According to the colophon, the “principle” (fa 法) for the production of this map is as follows:⁶

The Sui capital city (which later became the Tang capital city) itself and the Daming Palace (a Tang palace located to the northeast of the capital city) are both converted on the scale of one li (about 460 meters) to two cun (each cun is about three centimeters). For [areas] outside of the city that are incorporated into the map, such principle of conversion does not apply. [This map] is largely based on old maps and the Record of the Western Capital by Wei Shu; we also consulted many other books and investigated traces of old sites. This conversion method does not allow us to show the many halls . . . within the three Palaces, the Taiji, the Daming, and the Xingqing, so we made additional maps [for these three Palaces].

The colophon then continues to record the sizes of the Han dynasty Chang’an, which was included in the map, and of the Tang city, along with the number of wards in the city, the lengths and heights of walls, and the names of city gates. From this colophon, we learn two lessons about the principle of making this map of Chang’an.

This early map of Chang’an was created on the basis of textual evidence, earlier maps, ground surveys, and possibly even archaeological data.

First, the eleventh-century cartographers made use of a wide range of data, including old maps (presumably those made in the Tang dynasty), textual records of Chang’an (especially the Record of the Western Capital composed by a Tang author Wei Shu in 722),⁷ and personal investigations of extant sites. The leader of this project Lü Dafang (1027–1097) was from a local Chang’an family, and served as the governor of the Chang’an region when the map was made. His brother Lü Dalin (1044–1091), a famous antique collector, also participated in making corrections to drafts of the map. Therefore, Lü Dafang and his team must have been intimately familiar with the area and many of the ancient sites that were still visible.⁸ Given that the Lü family members have been engaged in archaeological investigations in the past, it is possible that the two brothers also excavated old sites in their making of the map of Chang’an.⁹ Therefore, this early map of Chang’an was created on the basis of textual evidence, earlier maps, ground surveys, and possibly even archaeological data.

Second, the method of incorporating these data on a two-dimensional surface is also revolutionary. For the core of the map — the Sui-Tang city and the three Tang palaces — the cartographers strived to represent them accurately by following a consistent scale of conversion, about 1:7,666. Within this core, information regarding specific places is included on the basis of textual information. For instance, for the Pingkang Ward to the northeast of the city, a ward renowned among the 110 or so wards in the city for its pleasure quarters, the map includes the name of the ward itself (Pingkang fang 平康坊) carved in the middle of the ward; the names of important sites are placed in different areas of the ward, including Buddhist monasteries and the residences of famous officials like Chu Suiliang (596–658). These sites are represented by a simple line drawing of a small house, reflecting the influence of Chinese painting at the time.¹⁰ In reality, however, each of these residences and monasteries must have consisted of many individual units of houses. Beyond the city core, areas were represented primarily for inclusion rather than precision.¹¹ The entire southern suburb of the city, which stretched for about thirty to fifty kilometers from the southern wall of the Sui-Tang city to the Zhongnan mountains, was compressed into a narrow strip at the bottom of the map.

From this brief introduction, we can see that, although remarkably accurate in its representation of the basic structures of the city (the walls, palaces, wards, and locations of gates), this unprecedented accomplishment in the early history of Chinese cartography nevertheless falls short in capturing the full richness of the city of Chang’an in a number of ways. Due to spatial limitations, the cartographers can only include a number of the most important sites, and in a much simplified way. This one map also cannot represent the changes that occurred within the space of the city, but rather combines parts of the city the developed from different times and places them on the same map. For instance, the Daming Palace was not part of the city’s original plan but was instead built several decades into the Tang dynasty. In this way, the map is not an accurate reflection of the city at any specific time, but an atemporal picture. Finally, because of the need to produce a map that is largely square for easy viewing and reproduction, the cartographers had to sacrifice accuracy for areas outside of the city, and squeeze in as best they could places they deemed important. The result was a rather distorted picture of the areas around Chang’an.

Lü’s eleventh-century map remained the standard Chang’an map until the late twentieth century.

These issues with the otherwise revolutionary map would not be addressed, or perhaps even recognized, for centuries. The maps of Chang’an produced after Lü Dazhong’s map in the Southern Song through the Qing dynasties, though interesting in their own ways, were as a rule even less accurate and less rich than the Northern Song map.¹² Therefore, Lü’s eleventh-century map remained the standard Chang’an map until the late twentieth century, when scholars began using newer technology to produce new kinds of maps of Chang’an. The culmination of such “modern” scholarship is the Historical Atlas of Xi’an (Xi’an lishi dituji) published by the historical geographer Shi Nianhai in 1996.¹³ Shi’s atlas contains eighty-nine maps of the Xi’an region from Neolithic times down to the late twentieth century. It is arranged chronologically, with an emphasis on the Han and Tang dynasties. In particular, there are twenty maps of the Sui-Tang city, which account for almost a quarter of the whole atlas.

In important ways, this atlas followed the practice of Lü Dazhong’s map. It included maps of the Sui-Tang city itself, but also maps of the three Tang palaces.¹⁴ But it also addressed the issues of Lü Dazhong’s map in several ways. For the city itself, the 1996 atlas included several kinds of maps, such as a map of residential units, one on commercial establishments, one on religious institutions, and one on gardens, wells, canals, and springs.¹⁵ In this way, it contained a significantly richer array of information than the eleventh-century map. By presenting three maps (Sui dynasty Chang’an, early Tang Chang’an before 755, and late Tang Chang’an after 756), the atlas tries to capture the historical changes of the urban space in the three centuries of the Sui-Tang empire.¹⁶ Finally, instead of squeezing the suburban areas into the margins of the main maps, the atlas uses separate maps to represent the southern suburb, an area of villas and monasteries, and the northern suburb, where one finds all the Tang imperial mausoleums. These maps do not use distorted proportions, and are thus more accurate representations of the city and its relation to its environs.¹⁷

The most notable innovation of this atlas compared to traditional maps is its treatment of archaeological materials. In addition to maps and their accompanying, brief textual explanations, this atlas includes many photographs of archaeological sites — places of importance that were still visible to visitors — and archaeological finds. We find images of the ruins of the Han city,¹⁸ as well as treasures found in the Tang city and the surrounding Tang tombs.¹⁹ Uniquely, for the map of the Daming Palace, a site that has been thoroughly excavated, Shi Nianhai and his team include two maps: one based on textual information and the other on archaeological data.²⁰ While they conform to each other to a large extent, the archaeological map includes more concrete shapes of buildings and the textual map contains many more names of places. The mapmakers indicate, subtly perhaps, that the job of future scholars is to further correlate these two groups of sources, both deemed important to the reconstruction of the city of Chang’an.

The Song cartographers’ treatment of the relation between these two cities reveals a more intimate, and less rigid, understanding of the geographical dimensions of historical change.

In certain ways, one may argue that the Song dynasty map remains superior in recreating the city to the much more voluminous and accurate 1996 atlas. The Song map’s attempt at drawing out the shape of the buildings and gates was largely abandoned in the atlas, where most monasteries were represented by a black dot and residences by a small square. This latter approach might be more cautious, but it also renders the image of the map dull and introduces inaccuracies in its own way. (The Buddhist monasteries were of course not just dots.) More importantly, the Song map includes the Han city to the northwest as an integral part of the Tang city, while the 1996 atlas separates the Han and the Tang cities, representing them in different sections of book. But in the Tang dynasty, the Han city was still prominently visible and frequently visited by Tang emperors. So the Song cartographers’ treatment of the relation between these two cities reveals a more intimate, and less rigid, understanding of the geographical dimensions of historical change.

There are other efforts to improve upon both the Song and the Shi Nianhai maps. In particular, the works by Heng Chye Kiang and Seo Tatsuhiko contributed significantly to the ways we reimagine Chang’an. Heng published a digital reconstruction of the city that, for the first time in history, transformed our view of the city into a 3D one, thus revealing the scale of the city in ways that are impossible for traditional maps.²¹ Seo, on the other hand, followed a traditional format but, by using color coding, thematic arrangements, and geographical correlation — one of his maps connects the places of residence and places of burial for residents of Chang’an — produced the most aesthetically pleasing and scholarly useful maps to date.²² Their work represents the best of existing scholarship on mapping Chang’an.

Toward a Deep Map

The survey in the previous sections shows that, although there is an abundance of sophisticated scholarship on mapping historical Chang’an, the overwhelming majority of approaches follows a tradition of mapmaking that in some ways began with the eleventh-century Song dynasty map of Chang’an. These maps are, although extremely useful for certain purposes, static. Once a map is made, it becomes difficult to manipulate it and incorporate new data. On the other hand, unlike these traditional printed maps, “deep maps,” in the words of David J. Bodenhamer, are:

fluid cartographic representations that reveal the complex, contingent, and dynamic context of events within and across time and space. They express the results of research but the answers they provide are not final and the map itself is always open to revision. ²³

To achieve such a fluid cartographic representation, “within and across time and space,” one needs to build a geotemporal database that can be queried, visualized, and regularly updated. The raw materials for such a database include primarily textual sources and archaeological data. To explain how I envision building this database, I will briefly examine the nature of these two kinds of data.

There is a long tradition of writing about the city in China, and the writing about Chang’an is at the heart of this tradition. The Sanfu huangtu, a work of uncertain date that contained information about early Chang’an but was likely compiled much later, offers a textual dissection of the city, listing the palaces, gardens, gates, and streets of the city in the Western Zhou and the Han dynasties.²⁴ Under each heading, the book provides a short description of the place, including its location, its history, and famous anecdotes associated with it. This kind of thematic treatment is inherited by later geographies, such as the Leibian Chang’an zhi produced in the Yuan dynasty (1271–1368).²⁵

There is a long tradition of writing about the city in China, and the writing about Chang’an is at the heart of this tradition.

Another related kind of local writing began with the Song dynasty Chang’an zhi, produced by Song Minqiu (1019–1079) in 1076.²⁶ Song’s work includes a historical survey of the Chang’an area, including a summary of government institutions, palaces, and cities in the Zhou, the Qin, and the Han dynasties, as well as the city in the Tang dynasty and counties around Chang’an. This work is organized according to two different principles. First, the editor separates the materials between the city and the counties around it. Then, on the city, the book follows a chronological order, focusing on the Han and the Tang dynasty. Within the chapters on the city, it then includes textual information about the locations and history of notable sites. For example, the entry on the Pingkang Ward in the Tang city (represented above in the map excerpt, Figure 1) includes detailed information about several notable structures within the ward. An edition of the book printed in 1468 uses large characters to describe the name and location of these structures, while the smaller, double-line notes explain their history and incorporate stories about these places.

Therefore, unlike the thematically organized works mentioned above, the Chang’an zhi provides more concrete temporal and geographical information that can be associated with the texts it includes. For instance, we can “locate” the chunk of text about the residence of a notable historical figure geographically in a ward in the city, and temporally in the lifetime of this figure, which is often known or deducible from other sources. Because the Chang’an zhi provides the most complete list of Han and Tang place names in Chang’an and gives their location in a relatively concrete and reliable manner, it has long been the basis for the reconstruction of Chang’an in modern scholarship.²⁷ I also use the Chang’an zhi as a basis for my reconstruction of the city (Figure 2).

In addition to these geographical texts devoted to Chang’an, the vast ocean of traditional Chinese historical records includes many more sources for Chang’an: in official annals, political treatises, fictions, and poetry, among others. They are associable to different areas of the city because they often mention places in Chang’an as the location where the stories transpired. The Pingkang Ward, for instance, was renowned as the pleasure quarter of the city, and many encounters between courtesans and young officials and students are recorded.²⁸ Now that I have established a database of places according to Chang’an zhi, information from these other sources can then be added to each record, enriching the dataset as a whole. The result, hopefully, will be an extensive database of textual sources that are, to various degrees of preciseness, locatable in the city of Chang’an.

The second kind of data we possess in abundance now comes from archaeology. Unlike textual sources, whose geographical information can often be fuzzy, archaeological data generally contain concrete geographical information. For instance, when a hoard of Tang treasures was discovered in 1970, we know precisely where it was discovered: a modern village called Hejiacun (Figure 4).²⁹ Then, cross-referencing with textual sources from Chang’an zhi and others, we can determine that this hoard was located in the southwestern part of the Xinghua Ward on the west side of Chang’an.

Thousands of similar archaeological discoveries, from tombs and building foundations to ancient wells and gates, have been made in the past few decades. Almost all of them, when documented in official reports, are placed with a precise location that can be associated with a Tang or Han dynasty place. In this way, these archaeological data can be easily connected to the textual database. Our knowledge about their precise locations means that these archaeological data more easily lend themselves to geometric abstractions than does most textual information.

Furthermore, archaeology does not simply provide additional data, but can help revise and refine the textual data too. The archaeological work on the city itself — its walls, gates, bridges, and palaces — gives us a more precise picture of the shape of the city than textual sources and offers details unrecorded in texts. For instance, the excavation of what used to be the Western Market — the center of foreign trade in Tang Chang’an — revealed details about the internal structure of the market area, including the bridges and water system that must have been crucial to the function of the city (Figure 5). None of these details are recorded in textual sources. Such archaeological data complement, correct, and enrich textual data; a full picture of the city can only become possible if both types of sources are thoroughly incorporated.

What can we do with a database that incorporates both textual and archaeological information? And how would this “deep map” differ from the existing maps surveyed above? At this point, in the beginning of my project, I can only offer a few conjectural remarks.

As I will be using GIS software like ArcGIS or QGIS to organize and visualize my data, the first step — by no means an easy one — is to produce a time-layered map of Chang’an, the scale of which can be easily manipulated through zooming-in and -out features. Unlike Shi Nianhai’s atlas, it is possible to create many more layers of different time periods, considering the changing political and cultural history of the city. The juxtaposition of maps from different eras could help readers better grasp the overlapping constructions in Chang’an across different dynasties, which formed a kind of imperial palimpsest. There will also be virtually no limit as to how much information one can include within a certain space. Going back to the Pingkang Ward example, unlike the Song dynasty map that can only include six building complexes, GIS technology allows us to put all the information contained in the Chang’an zhi and other sources under the entry of the “Pingkang Ward” in our database. This large amount of information can then be linked to specific locations on the map, showing either the names of specific buildings or even excerpts from the historical sources on these buildings. These annotations can be placed on a modern basemap like Google Earth’s satellite images, or a georeferenced old map like the fragments of the Song dynasty Chang’an tu. The result will be a Chang’an map of unprecedented detail.³⁰

Another aspect of the urban space of Chang’an that most existing maps fail to grasp is its three dimensionality. As scholars have shown, the vertical space of Chang’an was a central factor in the social life of its residents.³¹ Yet virtually all our existing maps use a bird’s-eye perspective, looking down on a two-dimensional city. As we have knowledge about the magnitude of Chang’an’s tall buildings — some of the pagodas in Buddhist monasteries rose to over one hundred meters — and the topographical features of the city — the Tang imperial palace in the north was the lowest elevational point in the city, and was about fifty to sixty meters lower than the southeastern corner of the city — it will be possible to incorporate these data into the “deep map” and offer a better approximation of residents’ experiences of the city in premodern times.

Fictional accounts like the love story between a courtesan and an aspiring examinee also include factual geographical information that allows us to place the flow of their plots in the space of Chang’an.

What makes a map “fluid” is its ability to illustrate not only buildings and walls, but also events. There is of course no doubt that many historical events played out in the city since the earliest days of its foundation. But it was not until the Tang dynasty that significant events were recorded in textual sources with (relatively) precise geotemporal information. We know, for instance, how the Xuanwu Gate incident (July 2, 626), in which Li Shimin (598–649) killed his two brothers and forced his father (the emperor) to designate him as the heir, played out, beat by beat, in different parts of the city. Neither is such knowledge limited to historical events. Fictional accounts like the love story between a courtesan and an aspiring examinee also include factual geographical information that allows us to place the flow of their plots in the space of Chang’an.³² The social lives of certain residents and visitors, such as the Japanese Buddhist monk Ennin who visited Chang’an in the mid-ninth century, can also be reconstructed with remarkable precision.³³ I plan to create subsets of data to document the flow of these events and show them in separate layers of maps. The data thus organized can then be animated, revealing the social lives of the city in ways impossible using printed maps.

Conclusion

This “deep map” of Chang’an will be useful for scholars and students to learn about the city and its history. The collection of a vast amount of textual and archaeological data with new ways to organize and visualize them will certainly lead to new insights into the history and culture of the city. I cannot foresee what these insights will be. In my own research, I have certainly already benefited from even thinking about this deep map: an article I am writing now discusses the continued presence, in Tang Chang’an, of physical structures and ghostly beings from the Han dynasty and earlier. The idea for this project came directly from when I tried to overlay the Han city map onto the Tang city map in QGIS. I hope this deep map will be similarly useful to other scholars in their research work.

As Chang’an was the capital of China in much of the early and medieval imperial period, a better understanding of the city also would lead to a more intimate knowledge about Chinese history in general. In this sense, a more dynamic and accessible deep map of Chang’an will no doubt be of pedagogical value and should be welcome in classrooms of Chinese history and culture, both in introductory lectures and in specialized seminars. A deep map of historical Chang’an, if combined with portable GPS devices, will give travelers an augmented experience that no print map can match. These multiple potential uses make this deep map a worthwhile endeavor, and I look forward to the collaboration with other students of China and Chang’an in the coming years on this project.