GIS Research Brief · Khmer Empire · 10th Century CE
A history of the king known as the usurper, the capital he raised at Koh Ker, and a working catalogue of geospatial & LiDAR data sources for the site and Greater Angkor.
Jayavarman IV ruled the Khmer Empire from 928 to 941 CE. For most of the 20th century, French and Cambodian historians wrote him into the record as a usurper: a provincial strongman who broke from Yasodharapura (Angkor) and imposed himself on the throne by force. More recent work — particularly Chen Chanratana's monograph Koh Ker site and the reign of Jayavarman IV (2010s) and a body of inscription studies by Michael Vickery — has substantially complicated that picture.
He was the son of King Indravarman I's daughter Mahendradevi, and he married his own aunt Jayadevi, a half-sister of Yasovarman I. In a Khmer political tradition where succession traveled along matrilineal as well as patrilineal lines, this gave him a perfectly defensible claim — arguably stronger than that of Yasovarman's two sons, Harshavarman I and Ishanavarman II, who held Angkor between roughly 910 and 928.
The break came in 921. An inscription from that year records the moment with unusual directness:
"Jayavarman IV left the city of Yashodharapura to reign at Chok Gargyar, taking the Devaraja with him." Inscription, 921 CE — cited in Chen Chanratana & Wikipedia entry K.187
Taking the Devaraja — the god-king cult image — was the operative move. It transferred ritual sovereignty out of Angkor and to a new capital, ninety to one hundred twenty kilometres northeast, in what was probably Jayavarman's home district. For seven years the empire ran on two tracks: Yasovarman's sons clinging to the southern half from Angkor, Jayavarman holding the north from Chok Gargyar. When Ishanavarman II died in 928, the contest ended and Jayavarman became sole sovereign.
He reigned from Koh Ker for roughly thirteen more years, dying in 941 around the age of sixty. His son Harshavarman II survived him only briefly. By 944, Jayavarman's nephew Rajendravarman II had returned the capital to Yasodharapura — and Lingapura, the seat of an empire for two decades, was abandoned almost in its entirety. It would not be reoccupied at scale.
The traditional answer is political: he could not safely rule from a city dominated by his rivals' clients. Chen Chanratana adds an architectural one — Jayavarman wanted a blank slate on which to build at the colossal scale that defined his reign. The king's own inscriptions boast that his constructions surpassed those of every previous Khmer ruler, and the archaeological record bears that out. Where Harshavarman I had been content with modest brick shrines, Jayavarman built in megalithic sandstone, on a scale not seen again until the great period of Suryavarman II two centuries later.
The site sits in modern-day Srayang Commune, Preah Vihear Province, in the dry forest country northeast of the Tonlé Sap. It carries three names in the record: Chok Gargyar ("Island of Glory") in the contemporary inscriptions, Lingapura ("City of the Linga") in Jayavarman's own dedications, and Koh Ker in modern Khmer.
The walled urban core was roughly 1,200 m square, but the wider settlement — temples, satellite shrines, water-management features, road segments — extends across about 35 km². The site became a UNESCO World Heritage property in 2023 ("Koh Ker: Archaeological Site of Ancient Lingapura or Chok Gargyar").
The dominant water feature is the Baray Rahal, a reservoir of approximately 1,188 × 548 m. It is modest beside Yasovarman's East Baray at Angkor (~7,500 × 1,800 m), but its construction was harder: Koh Ker's terrain forced the engineers to cut into bedrock on one side rather than simply embanking on flat ground. Together with the natural steung (streams) descending from the surrounding hills, the baray defined the SSE–NNW alignment of the northern temple group around Prasat Thom.
There are two principal architectural groups, distinguished by alignment. The northern group — including Prasat Thom and the seven-tiered Prang state-temple — follows the baray's natural SSE–NNW axis. The southern group is laid out on a strict east–west line; none of its inscriptions predates 928, suggesting it represents the more "official" phase of construction after Jayavarman became sole sovereign and Lingapura became the formal imperial capital.
The map below plots the major monuments inside the protected core, with the Baray Rahal as the dominant hydraulic feature. Coordinates are derived from OpenStreetMap, the UNESCO nomination dossier, and the Hello Angkor archaeological gazetteer; they should be treated as indicative for routing or fieldwork — see the data sources section for the canonical layers.
Base layer: Esri World Imagery + OSM. Markers: 14 principal sites within the central Koh Ker complex. Elevation data from GLO-30 DEM.
| Monument | Type | Lat | Lon | Elev. |
|---|
Tip: drop this table into a CSV with columns name,type,lat,lon and you have the seed file for a Leaflet/Folium layer or PostGIS table. EPSG:4326. For metric work in-country, reproject to UTM Zone 48N (EPSG:32648).
For everything other than ground-penetrating LiDAR, the workhorse stack for Cambodia is OSM + Copernicus + the open archaeological gazetteers. None of these is gated and all of them clip cleanly to the Koh Ker bounding box (approx. 104.51, 13.76, 104.57, 13.81).
Country-level OSM shapefile bundle: roads, waterways, admin boundaries, places, POIs. Updates daily. Geofabrik is the cleanest extract source for Southeast Asia.
download.geofabrik.de/asia/cambodia.htmlTargeted queries for archaeological tags (historic=archaeological_site, historic=ruins, building=temple) within the Koh Ker bbox. Faster than full extracts when you only need monuments.
Global 30 m DEM, derived from TanDEM-X. The best free baseline DEM globally; useful for regional context, watershed extraction, and viewshed studies before LiDAR is layered on. Available via OpenTopography's API.
portal.opentopography.org · GLO-30The original 1-arc-second SRTM and the reprocessed NASADEM. Older than Copernicus but the historical baseline against which the Evans et al. 2013 LiDAR work was published.
search.earthdata.nasa.govUNESCO Koh Ker dossier: property & buffer-zone polygons, inventory of monuments, comparative analysis. The official boundary file is the canonical reference for the core / buffer perimeter.
whc.unesco.org/en/list/1667Site-by-site gazetteer with ~50 named monuments at Koh Ker, each with descriptive notes. The richest open inventory for the southern and outlying groups; coordinates are approximate but useful as a seed.
helloangkor.com/attractiontag/koh-ker10 m multispectral (Sentinel-2) and 30 m (Landsat) for vegetation indices, dry-season change detection, and broader landscape context. April acquisitions match the LiDAR campaign window of minimum canopy.
browser.dataspace.copernicus.euWWF/USGS conditioned drainage network, watersheds, and flow accumulation derived from SRTM. The right starting point for situating the Rahal baray inside the regional hydrology.
hydrosheds.orgThis is the part where expectations need to be set. There is no public, click-to-download point-cloud LAS file for Koh Ker or central Angkor in the way US 3DEP coverage works. Cambodia has no national LiDAR program. What exists is a sequence of two extraordinary academic acquisitions, both led by Damian Evans (EFEO) — and an open-access derived product released in 2024 that is genuinely usable.
KALC, 2012 — The Khmer Archaeology LiDAR Consortium. ~370 km² across Angkor (the contiguous block), Phnom Kulen, and a 67 km² Koh Ker block. Leica ALS60 sensor on a Eurocopter AS350, ~2 ground returns/m². Published as Evans et al., PNAS 110(31), 2013. This is the dataset that put Koh Ker on the LiDAR map.
CALI, 2015 — The Cambodian Archaeological LiDAR Initiative, ERC-funded successor to KALC. ~1,910 km² total across all the major Khmer temple complexes (Sambor Prei Kuk, Banteay Chhmar, Preah Khan of Kompong Svay, Longvek, more). Leica ALS70-HP, higher density (10–95 points/m² depending on terrain).
Combined, KALC + CALI represent ~10 billion 3D points and ~13,000 aerial photos across 888 km² of curated coverage. The raw point clouds and orthophotos remain in the custody of the EFEO and the participating Cambodian authorities (APSARA, Ministry of Culture and Fine Arts), in line with longstanding ethical norms about archaeological LiDAR and looting risk — Cohen, Klassen & Evans, Ethics in Archaeological LiDAR (JCAA 2020), is worth reading on this point.
The first open-access ALS archaeology dataset for Southeast Asia. 888 km² of derived normalized Digital Terrain Models (nDTM) + orthophotos + 31,411 archaeologist-drawn polygon annotations, curated from KALC & CALI. Includes Koh Ker, Angkor, Phnom Kulen, and other complexes. Published with NeurIPS 2024 paper by Landrieu et al. Released as a benchmark for ML segmentation but fully usable as a GIS layer.
archaeoscape.aiThe KALC publication. Supplementary materials include figures derived from the LiDAR DTM at 0.5 m resolution and DSM at 1 m. Not a downloadable point cloud, but high-resolution hillshade tiles for Angkor and Koh Ker are in the SI.
pnas.org/doi/10.1073/pnas.1306539110The institutional custodian. For research access to the raw KALC/CALI point clouds beyond what Archaeoscape exposes, contact the École française d'Extrême-Orient research base in Siem Reap. Expect to provide a research proposal and to coordinate with APSARA for any Angkor-zone data.
efeo.fr · Siem Reap centreAuthority for the Protection and Management of Angkor and the Region of Siem Reap. The Cambodian state partner on KALC and CALI, and the gatekeeper for any in-country fieldwork or Angkor-zone derived products.
apsaraauthority.gov.khOngoing project (Sarah Klassen et al.) doing ground verification and excavation against the LiDAR base. Useful for derived feature inventories and contextual interpretation; some maps published in Antiquity and Journal of Field Archaeology.
sarahklassen.ca/researchprojectsIf you cannot get into KALC/CALI/Archaeoscape, OpenTopography's API serves Copernicus GLO-30, NASADEM, and SRTM clipped to any bbox. Not LiDAR, but the right baseline for hydrological and viewshed work at regional scale.
portal.opentopography.org/apidocsFor a portfolio piece or web-map deployment, the realistic path is: Archaeoscape's nDTM tiles for the LiDAR base layer, OSM + UNESCO polygons + Hello Angkor seed points for vector overlays, Copernicus GLO-30 for regional hillshade context, and Sentinel-2 dry-season composite for current land cover. That stack is fully open, redistributable, and renders well in Leaflet, Mapbox GL, or a tile server like Martin. Anything beyond — true point-cloud work on the 67 km² Koh Ker block or the central Angkor block — requires going through EFEO or APSARA.