For seven days in 2012 a helicopter flew a mowing pattern back and forth above the forest canopy of Phnom Kulen, firing a laser at the ground hundreds of thousands of times a second. Most of those pulses struck leaves and came straight back. A small fraction found the gaps, hit soil, and returned a few nanoseconds later than the rest. Sort the returns by timing, strip the vegetation out mathematically, and what remains is bare earth. When that image resolved on the screens, it showed streets. Ponds set out in rows. City blocks squared to the compass. Mahendraparvata - a capital that Khmer inscriptions had named for over a thousand years and that archaeologists had never once been able to draw whole - had been lying in a rectilinear grid under the trees the entire time.
Mahendraparvata means "Mountain of the Great Indra," from the Sanskrit mahendra, a title of the Hindu god Indra, joined to parvata, mountain. The name is attested in inscriptions at the temple of Ak Yum in the Angkor area, and it refers to the massif known today as Phnom Kulen. What happened on that mountain in 802 CE is the conventional starting point of Khmer imperial history: Jayavarman II was consecrated there as chakravartin, universal monarch, and declared his independence from the power that the inscriptions call Java. The city he founded on the plateau was one of three courts of his reign, alongside Amarendrapura and Hariharalaya. He did not stay. Later in his reign Jayavarman II moved down to Hariharalaya on the plain, where he died in 835. The mountain capital had been the empire's first address, and it lasted barely a lifetime.
The plateau was never truly lost, only illegible. In 1936 the French art historian Philippe Stern led an expedition into the Kulen highlands and came out with previously unknown temples, Vishnu statues, and the argument that this was the first true temple mountain of the Khmer. Later researchers added more dots to the map, one exhausting field season at a time, in country that was remote before the wars and mined after them. What nobody could do was connect them. "We didn't know how all the dots fitted, exactly how it all came together," Jean-Baptiste Chevance said of the state of knowledge before the survey. The ground phase of the modern expedition gives some sense of why: the team reached its starting point by motorbike and then worked forward along goat tracks and through waterlogged bog, with landmines among the hazards. Five new temples turned up that way. The lidar data eventually yielded thirty.
The airborne survey was co-led by Damian Evans, then of the University of Sydney, and Chevance of London's Archaeology and Development Foundation, working with the Khmer Archaeology Lidar Consortium. Initial findings were announced in June 2013; a second, far more extensive scan followed in 2015, and the peer-reviewed synthesis appeared in the journal Antiquity in 2019. What it described was not a scatter of shrines but a plan. A formally laid-out urban area of roughly 40 to 50 square kilometres, set on linear axes aligned close to the cardinal directions. Major thoroughfares carving a central zone into blocks, and a finer system of land parcelling subdividing the blocks in turn. Small shrines, house mounds and ponds distributed through them. A royal palace enclosure and a state pyramid-temple in a distinctive spatial relationship. It is the earliest large-scale grid city the Khmer are known to have built.
Two of the most telling features on the plateau are unfinished, and that is precisely what makes them eloquent. Prasat Rong Chen, the state temple, stands on the highest ground of the southern plateau: five tiers, the lower two raised from levelled earth and embankment, the upper three faced in laterite, the summit reached by ramps that were never replaced with stairs. Down in a nearby valley the lidar picked up a pattern of parallel striations where soil had been dug out systematically and heaped into dykes - the earthworks of a major reservoir. Completed, it would have held back the water running through the valley and flooded more than 35 hectares. It was never completed. The digging simply stopped, and the marks of the work are still legible in the ground twelve centuries later.
A city on a plateau has a hydrological problem that a city on a floodplain does not, and Evans has pointed to imagery showing the area was substantially deforested during its occupation. His argument is that clearance on that scale, combined with the difficulty of managing water on high ground, helped push the centre of Khmer power back down to the plain. It is a theory rather than a settled conclusion, and the unfinished reservoir sits awkwardly and interestingly inside it. What is not in doubt is where the empire went next. Hariharalaya, then Angkor - and Angkor, endlessly watered by the rivers running off this same mountain, became one of the largest pre-industrial urban complexes on earth. Mahendraparvata went back under the canopy for eleven hundred years and waited for someone to invent a laser.
Mahendraparvata occupies the Phnom Kulen plateau at approximately 13.613 N, 104.113 E, about 40 km north of Angkor Wat and 45 km north of Siem Reap. The plateau top holds a fairly even elevation near 400 m, rising to roughly 500 m, so terrain clearance is modest but the escarpment edges are abrupt. Nothing of the city is visible from the air - that is the whole point of the story - but the massif itself is unmistakable: an isolated 40 km chain of flat-topped plateaux running WNW-ESE across otherwise level country, with the Angkor barays glinting on the plain to the south. Suggested viewing altitude 3,000-5,000 ft AGL. Nearest major field is Siem Reap-Angkor International (VDSA), with Phnom Penh International (VDPP) as alternate. Expect orographic cloud and rapid afternoon buildups over the plateau in the May-October wet season.