Two Thousand Years of Engineering Genius on the Angkor Plain

By: Richard A. Engelhardt

Originally Published in 1995

View PDF

Commanding a strategic location on the upper­most tip of Cambodia’s great Tonle Dap lake, the ruins of the Angkor Empire expand north, east and west from the shores of the lake up to the sacred Kulen mountain plateau. This entire 5,000-square-kilometer site, once the location of one of the world’s largest met­ropolitan areas, is a relic cultural landscape—an envi­ronment which was intensively engineered by human activity over time to suit the Empire’s changing tempo­ral needs. During this process the land was continually reshaped into an ever more perfect image of the celes­tial city of which Angkor was the earthly reflection, ruled over by human incarnations of the very gods themselves.

The area we call Angkor today was, in fact, the site of at least seven capital cities built by a succession of Khmer rulers between the 9th and 13th centuries A.D. (Fig. 1). Before it became the capital of an empire, the region was already peppered with a mosaic of mound and moated villages inhabited by subsistence farmers and fishermen (Fig. 2; see box). As the wealth of the Khmer kingdoms grew, a succession of “kings of the kings” chose to live in and around Angkor and to build the metropolitan center of an empire. Even after the final breakup of the empire in the 15th century and the move of the capital elsewhere, the Angkor area contin­ued to be relatively densely populated. As we shall see, it was the very prosperity of the Khmer Empire which put eventual limits to its growth. This prosperity was based on the finely tuned and sustainable exploitation of the land and its resources—most importantly its water resources.

The climate of the heartland of the ancient Khmer Empire is controlled by the tropical monsoon system. Deasonal rainfall averages 1500 rnm/year and temperatures are constantly warm. During the rainy season, violent storms release large volumes of water within relatively short periods of time. The storms occur in a predictable cyclical pattern of frequency and intensity.

Each year, during the rains, the Tonle Dap River reverses its flow and quadruples the capacity of the Tonle Dap lake, flooding the forests and bringing a bounty of fish (Fig. 3). During the Angkorean period, hundreds of thousands of people from the farthest reaches of the Empire swarmed to the lake shores to harvest one of the world’s richest aquatic resources. This harvest supplied the Empire’s protein needs and assured the economic prosperity of Angkor as the cen­tral marketplace of the entire region (Fig. 4).

Large areas of the Angkor plain are also flooded by the Tonle Dap. The floods recede rapidly, however, and the subsequent dry season is characterized by drought and severe ecological stress. Intensive man­agement of water resources is necessary if the growing season is to be extended beyond the end of the rains.

To the north and east of Angkor, the Kulen plateau rises high above the ruined city on the plain, trapping monsoon water coming froni the southwest. The basic drainage pattern of the area consists of a series of radiating streams which rise in the hills and flow across alluvial fans to the Tonle Dap. The impor­tance of preserving this watershed as a source of water for the rice paddies and to fill the city’s transportation canals and municipal water system was clearly under­stood by the earliest inhabitants of the area. Phnom Kulen was perceived as a sacred site from which flowed

the very origin of Khmer civilization. Home of gods and retreat of kings, the mountain was for the ancient Khmer Mount Meru on Earth, and its revered primeval forests were left untouched over the centuries. The coronation rites for all of the kings of Angkor took place atop Phnom Kulen, vividly reinforcing the king’s para­mount role as protector of sacred ancestral lands as “king of the mountain.”

The Puok River is the only permanent stream draining the Angkor plain. It has a catchment area of 670 square kilometers. Rising in the hills at an altitude of 420 meters, the river cuts deeply into the plateau to discharge through a gorge onto the plains below. The main river is joined by a number of tributaries until it is diverted into an ancient artificial canal built to join the original Diem Reap River at Phum Khlat. The ancient bed of the Puok River continues southwest to the town of Puok, where it beconies lost in the marshes of the Tonle Dap.

During the reign of Rajendravarnian I in the 10th century A.D)., the Puok River was diverted eastward to join with the Diem Reap River which, for most of Angkor’s long history, was the capital’s principal water course. The diverted river, which has a total length of 80 kilometers, was canalized to supply the water needs of the Empire’s capital city, including the moats of Angkor Wat, Angkor Thom, as well as the Eastern, Western, and Northern barays (reservoirs). This half-natural, half-manmade river was the Ganges of the Khmer Empire, as important symbolically as it was eco­nomically and ecologically.

The second most important river system of the Angkor plain is the Roluos which rises from springs at the foot of the Kulen (Fig. 5) and eventually flows into the marshes of the Tonle Dap. This river was also canal­ized in ancient times to supply the Lolei Baray (the Indratataka), the moats of the city of Hariharalaya, and the monuments of the Preah Ko, Bakong, Trapeang Phong, and others.

Dince Angkorian times, dry-season flow in the rivers has been insufficient to meet potential demand for water, particularly for irrigation. Modification and canalization of the rivers ensured the urban settlements of the Khmer Empire with a year-round supply of water and enabled the control of water for irrigation of agri­cultural land. Retention and storage of surplus water during the rainy and flood seasons for use during the rest of the year was, along with the building of religious monuments, the major preoccupation of Khmer engi­neers throughout the long history of the empire.

All the rivers and streams draining the Angkor plain show entrenched meanders, indicating a slow low­ering of the base of the drainage system. As the chan­nels continued to cut down, the water level was lowered significantly. Waterwheels or other mechanisms to lift the water from the streams up into the city’s moats and canals were needed, and major maintenance of the water works would have been required to maintain water in the barays throughout the dry season (Fig. 6). The ever-increasing cost in terms of both labor and equipment needed to counteract the lowering of the base water system may have contributed to the eventual abandonment of the Angkor site as the metropolitan capital.

Ancient Field Systems: New Evidence from Remote Sensing

The stability of the food supply of the Khmer Empire depended on the modification and management of the hydrology of the area to ensure adequate rice production. A broad belt of land suitable for the cultiva­tion of rice was established across the Angkor plain at an early date. Old field patterns are still visible under some of the new land divisions: the old fields tend to be much smaller and run at a different angle to the present land holdings (Fig. 7).

In recent years, satellite and aerial photography have proven to be powerful tools in helping to identify and interpret many types of archaeological sites and, in particular, ancient field patterns (Fig. 8). Present-day botanical patterns remain extremely sensitive to past disturbance of the ground, especially if that disturbance depleted soil resources. In the upper reaches of the old Diem Reap-Puok river system, the early pattern is still visible in aerial photographs, although on the ground it is totally invisible and the area is both uncultivated and depopulated. One important implication of these dis­coveries is that the Khmer field pattern—and perhaps much else in Khmer culture—may have origins far ear­lier than the classic Angkor period.

Khmer engineers since well before the height of the imperial period seem to have been preoccupied with the problem of extending the growing season. To do so meant dealing with a surplus of water during the flood season and a shortage during the dry season (Groslier 1979). However, in the end, they did not suc­ceed. Observations from the separate disciplines of geology, hydrology, and botany all provide support for the same archaeological hypothesis: Angkor was aban­doned because of a collapse in agricultural productivity; the system could no longer sustain a large urban popu­lation. The cause was an overwhelmed ecosystem which was, in any case, deteriorating because of incremental changes in the natural hydrology of the region. To overcome these changes required a technological invest­ment which became increasingly less cost-effective to maintain.

The Baray: Key to Controlling Flood Water

In order to maximize the potential of their eco­logical setting, the ancient Khmer developed extensive hydrological systems. These retained and managed flood water for irrigation and ensured a continuous year-round supply of water for the urban centers and reli­gious complexes. Rivers were dredged and straightened into canals and vast water storage reservoirs called barays were created behind massive earth embankments (Fig. 9). Dikes were built across the flood plain to deflect and store flood waters to irrigate crops during the dry season. The annual rise and fall of Tonle Dap was exploited to grow first, floating rice on the rising flood and then, receding rice as the waters subsided.

To fill the barays, monsoon flood waters were trapped behind a system of dikes hundreds of kilometers long. In this way, the entire flood plain between the Kulen and the Tonle Dap was turned into a landscape of gradually sloping rice terraces. With triple-cropping, these terraces could supply the food needs of the popu­lation of between 200,000 and 1 million inhabitants.

Before the system collapsed, the farmers and engineers of Angkor had a remarkable record of success lasting over a thousand years. They gradually and pro­gressively prolonged the growing season with a simple but effective system of dikes. These trapped the early rain water as it flowed down toward the lake and then, at the other end of the wet season, retained the flood water retreating toward the lake. This system of diking appears to be part of the same simple technology as the building of a rice paddy bund and, when further elabo­rated, the same as the technology that went into build­ing the great baray water-storage reservoirs (Garami and Kertai 1993:29). This, more than any other single feature, is the diagnostic “technological marker” of the ancient Khmer Empire.

The Khmer word baray comes from a Danskrit word meaning “to transverse” or “to cross,” suggesting a local evolution from transverse dike to baray in the Angkor plain. Initially and in its simplest form, a baray consisted of a southeast angle built to trap the retreating water and to channel it, following the natural contour of the land. Later, structures were built to the north and west of earlier structures, forming a square reservoir. This enabled an additional amount of water to be trapped, and the level of the stored water to be raised above the surrounding plain. This water could then be fed into the city’s water and irrigation system. After the capital moved to Yashodarapura (the site we know today as Angkor), the old capital of Hariharalaya (Roluos) was apparently still linked into the greater metropolitan area’s water system through a long dike which chan­neled water down to the old city from the new.

Study of aerial photographs also shows that parts of the earlier system were incorporated into new town planning designs as the city of Angkor was reno­vated and remodeled by successive rulers. One example is the reuse of the old moat of the first city of Angkor (Yashodarapura) around the Phnom Bakheng as a spill­way. This spillway channeled excess water out of the Western Baray into the Diem Reap River/canal south of the new city of Angkor Thom, thus preventing the pos­sibility of a disastrous flood during periods of excess water.

Post-Angkorian Settlement and Land

Little is known about the land-use patterns in Diem Reap in the post-Angkor period. After the collapse of the waterworks, people apparently became more dependent upon the natural cycles of the Tonle Dap for agricultural production. There is also evidence that the population dispersed over the northern plains and used fire to clear land. This practice resulted in the degrada­tion of much of the forests and precipitated a further drop in productivity.

After the abandonment of Angkor as the capital and the dispersal of a major portion of its population, extensive areas which had been occupied or farmed were recolonized by evergreen forest species. These extended south towards the core monumental area at Angkor even in quite recent times. The presence of tall and moderately dense forest in the immediate vicinity of Angkor Thom indicates the extent of the woodland that developed after Angkor was abandoned. This forest remained relatively untouched into the 19th century.

Time, war, and illegal logging have all taken their toll on this, one of the most productive man-made landscapes of all time. Water continues to be a problem. A gradual, almost imperceptible geologic uplift tilting has occurred along a northwest-southeast axis following the orientation of the Kulen plateau. As a result, the rivers have cut 6 meters into the ground, putting them well below the level of the 10th-11th century reservoirs, canals, and moats (Fig. 10), and driving the upper reaches of the streams underground (Thung 1994:2). Various attempts to correct this problem have taken place over the past 1,200 years. Larger and ever more labor- and capital-intensive corrections have been tried, but in the long run the result has been an unprofitably expensive system which only functioned satisfactorily at the center. And as the system involved siphoning water from outlying farmland, agricultural productivity declined.

The monuments themselves have been affected by the disruptions in the hydrology, both natural and man-made. Isolated from the water system at large, many of the moats surrounding each monument have silted in and the water table underneath these monu­ments is no longer stable. The result is subtle seasonal shifting of the ground under the heavy stone monu­ments which leads to their breakup and potential col­lapse (Fig. 11). The situation is exacerbated by the breakdown of the drainage system, causing perpetual waterlogging of foundations and periodic flooding of the buildings themselves.

Water and the Khmer

It was probably the potential of the permanent rivers and seasonal floods between the Kulen plateau and the Tonle Dap that attracted settlers to the Angkor area long before it became the metropolitan capital. Early hydrological structures were extended and enlarged during the development of the sequence of city states. By the height of the Angkor period, these struc­tures were elaborated to create the complex water man­agement regime necessary to supply the barays, moats, and canals that lead water into the heart of the city and to the temple compounds. While the demand for water increased immensely over time, water resources remained constant. This put pressure on Khmer hydro­logical engineers to innovate and to continually adapt and modify the city’s public water works.

Each water-based feature fulfilled several func­tions. Barays provided agricultural and domestic water, and fish and plant foods. Canals channeled water for public sanitation, and were transport arteries. Embankments and dikes were usually oriented east-west following the contours and acted both as levees to con­trol floods and elevated causeways for roads. Moats sur­rounding temples, monuments, and inhabited areas also fulfilled several functions: they served as sacred bound­aries, they were a source of domestic water and food, and they provided fill for foundations to raise the level of the terrain for drainage and protection. Access to domestic water was provided by tanks or basins dug into the water table.

The unique character of Angkor’s water fea­tures indicates a highly sophisticated identification of ancestral spirits with water which prompted and sus­tained the repeated construction of water-related struc­tures. Religion and daily life were inseparable. Given the dependence of people living in the Angkor region on their environment, it is not surprising that a strong attachment to the land of the ancestors continues to this day. This is symbolized by an almost universal alle­giance to guardian spirits of the land and place which are called nak tha (Fig. 12). Over the years, these local ancestral spirits have been incorporated and subsumed into the structure of the elite gods of Hinduism and Buddhism who also guard the ancient lands of Angkor.

Devising ways to control the flow and supply of water was at the heart of the genius of the ancient Angkoreans. The hydrological system of Angkor was complex and extensive, but it was based on the age-old principles of rice paddy construction begun in prehis­toric times. Recent archaeological survey and research have confirmed the long history and indigenous ingenu­ity of this work. As the Empire grew so did the water system, until engineering feats of truly awesome dimen­sions were accomplished. The result was a unique man­made environment which assured the productivity of the Empire and at the same time transposed onto the landscape the mental picture the people had of the per­fect universe of their gods, picture of old Angkor has emerged. Rather than existing in glorious isolation on the Angkor plain, the Khmer Empire’s capital city was situated in a heavily populated area, surrounded by settlements dating back before the founding of the Empire itself (Fig. 15, page 30). Analysis of aerial pho­tographs and satellite remote-sensing imagery revealed at least 69 elevated circular mounds scat­tered over the plain, some with associated earth­works and/or moats (Moore 1993). These mounds appear to be naturally occurring elevations which have been formed over time by down cutting and diversion of rivers, as well as by Aeolians action and the accumulation of debris from human habita­tion. Today their shape is defined by vegetation encouraged by the mounds’ slightly elevated posi­tion above the flood plain. Generally the mounds appear circular, but there is no standard form; each has a unique shape.

The density of the prehistoric inhabita­tion of the area is indicated by the fact that all mounds investigated to date have yielded stone tools, pottery, and other evidence of human occu­pation. The presence of stone tools in the Angkor plain links these mound sites to the few other “proto-historic” Cambodian sites that are known, such as Domrong Sen (Mansuy 1923) and Mlu Prei (Levy 1943) which also have polished stone tools.

The Puok River Basin

Thirty-four of the 69 newly identified prehistoric sites are located in the area of the old Diem Reap-Puok rivers drainage basin. This area was also the site of the earliest known urban set-dement in the area—a 7th century pre-Angkorian town called Ananditapura. The town or village was one of the spots to which Jayavarman II brought his nomadic capital during the Empire’s early days. The remains of the site are now mostly under the Western Baray which was built in the 11th century (Jacques 1990).

At least 8 pre-Angkorian settlement sites have been identified near Ananditapura. The remaining 26 sites are distributed continuously up the Puok valley. There is also evidence of a canal, with a concentration of ancient rice paddies nearby. In addition to diverting river water toward Angkor, the canal may have been part of an Angkorian period attempt to reanimate the rice fields of the upper Puok drainage basin.

The Diem Reap River Basin

In the Diem Reap River system, 14 habita­tion mounds were surveyed in the valley north of the main Angkor complex by Dr. Moore and the UNESCO team. An eastern tributary of the Diem Reap River flows into this valley from the princi­pal Kulen plateau springs. All the sites are located along or near a water course north of the Angkor-period dam where the artificial canalization of the Diem Reap River begins. As the diversion of the Siem Reap River dates from the 10th century reign of Rajendravarman I (Jacques 1990:165) while the present course of the river dates from the 16th century (Garami and Kertai 1993:27), and as no habitation mounds have been found south of the diversion, it seems a likely hypothesis that the sites north of the diversion are earlier in date and represent a pre-Angkorian settlement pattern (Moore 1993).

The Roluos River Basin

Sites in the valley of the Roluos River were ground surveyed by the initial UNESCO team. Subsequently, several additional sites have been identified near the ancient shoreline of the Tonle Dap. Like those in the Puok and Diem Reap river valleys, some of these sites have vestiges of moats. The mounds range in diameter from 150 to 350 meters, with elevations between 10 and 30 meters (Moore 1993).

When plotted, the distribution of sites appears relatively dense throughout the Angkor plain. However, when the size of each site is taken into account and “nearest-neighbor” statistical tests applied, the heaviest concentration of sites is evident in the three river valleys, with the inter­vening areas evenly but not so heavily settled.


I would especially like to thank the following consul­tants who worked with me from 1991 to 1994 on the UNESCO project to elaborate a Zoning and Envir­onmental Management Plan for the Angkor World Heritage Site: Dr. Elizabeth Moore of the School of Oriental and African Studies of the University of London; Dr. Ferenc Garami of Hydro-Ecosphere,

Budapest, and Dr. Heng L. Thung of the Mekong Secretariat. Their ideas and generously shared data con­tributed greatly to this paper, although I take complete responsibility for any mistaken interpretations. I would also like to acknowledge my intellectual debt to the late Bernard Philippe Groslier, and especially to his 1979 paper (see Bibliography).


Cite This Article

Engelhardt, Richard A.. "Two Thousand Years of Engineering Genius on the Angkor Plain." Expedition Magazine 37, no. 3 (November, 1995): -. Accessed May 19, 2024.

This digitized article is presented here as a historical reference and may not reflect the current views of the Penn Museum.

Report problems and issues to