Bedrock Geology

Hang Son Doong is formed in the Ke Bang massif, which exposes a section of Devonian (360-416 My) to Permo-Carboniferous (251-360 My) limestones that were originally deposited in the Paleotethys ocean.These limestones are now part of the Truong Son Fold Belt, a zone of intensely folded rocks that extends northwest across the northern half of Vietnam and into northern Laos.These rocks were folded during a period of mountain formation in the Early Triassic (240-250 My) known as the Indosinian orogeny that is associated with closure of the Paleotethys ocean and assembly of the Indochina tectonic block (Lepvrier et al. 2004).The limestones of the Ke Bang massif are surrounded by insoluble rocks to the east and south.These include Carboniferous to Triassic metamorphosed crystalline rocks that form mountains to the northeast, and much younger Cretaceous (145-65 My) red sandstone and mudstone to the south.Some Cretaceous redbeds are also preserved near the temple to the northwest of Hang Son Doong.These Cretaceous red rocks represent continental sedimentation.

Beginning at about 50 My ago, during the Paleogene, India collided into Asia.Associated with that collision, what is now the Indochina block was extruded southeastward out of Asia along massive strike-slip faults.At that time minor strike-slip faults also formed throughout the Indochina block.A series of strike-slip faults trending NW-SE are widely observed in north-central Vietnam, as well as regionally.These faults are associated with extrusion along the NW-SE trending Red River fault that passes through Hanoi into the Gulf of Tonkin (Rangin et al., 1995).A second, somewhat younger set of stike-slip faults trends N-S, and cuts the older faults.The N-S faults probably date to the Neogene (beginning 23 My ago), and may be associated with opening of the South China Sea (Rangin et al., 1995).

The bedrock geology has had a strong influence on the formation of Hang Son Doong.The Permo-Carboniferous rocks are generally massive to thickly bedded, and are formed of relatively pure calcium carbonate that is conducive to cave formation.At the cave, the rocks are dipping approximately 20-30 degrees to the north.At only one place in the cave (Watch Out for Dinosaurs) was a section of thinly-bedded limestone observed, and this was associated with passage collapse. Interbedded cherts are generally rare, except in localized bands.Both the Paleogene and Neogene faults control the passage geometry.Hang Son Doong formed primarily along a single N-S fault, with a truncated passage (The Alcove) extending SE along an older fault.


Cave description


1. Geology

Besides its size, one of the defining characteristics of Hang Son Doong is that it is profoundly straight and generally lacking side passages (with one exception at The Alcove). This is because the cave has formed along a N-S fault.The fault is well-exposed in the ceiling and walls, and is almost never out of view.It is subvertical and usually consists of 1-2 major strands.The walls of the cave also often expose a fault breccia developed in the limestone.This breccia consists of striking black and white calcium carbonate that forms angular, sharply-defined patterns.The patterns tend to be tooth-like near minor faults, merging into an almost kaleidoscopic background.Overall, the fault breccia seems to define a zone approximately 100 meters across, and may have helped to control the width of the cave.

The cave has been subject to two massive collapses to daylight.The southern collapse doline is named Watch Out for Dinosaurs (WOFD).It is 110 meters across at the base, and is colonized by thin forest.The walls of the WOFD collapse are the only place in the cave that expose thinly-bedded chert-bearing limestones.The doline collapse almost certainly occurred here due to the weakness of the rocks.The northern collapse doline at Garden of Edam (GOE) is considerably larger, 163 meters across at the base.A thicker forest covers the collapse pile.The rocks at GOE are thickly bedded, and so collapse here is not due solely to rock strength.Instead, it is apparent that GOE formed at the intersection of two major faults.The main cave is formed along a N-S fault, presumably Neogene in age.A tributary passage that is filled by an earlier collapse formed along a NW-SE trending fault, presumably of Paleogene age.GOE is found at the intersection of these two faults, where the ceiling would have been weakened both by its extra width at the passage intersection, as well as by the presence of two faults in the ceiling which would promote collapse of the unsupported blocks between them.

2. Hydrology

Hang Son Doong can be conveniently broken into three smaller units.The southern section, south of WOFD, the central section between the WOFD and GOE collapse dolines, and the northern section, north of GOE.

a. Southern section

The southern section of the cave actively takes water today.The passage has many indications of annual flooding, including actively dissolving breakdown, re-dissolving speleothem, and coarse fluvial sediment derived from outside the cave, as evidenced by abundant red sand and mud from the Cretaceous rocks and crystalline fragments from the Triassic rocks.

Base-level flow at the time of our visit was estimated to be about 10 m3/s, although no direct measurements were taken.Peak flow was estimated from scallops widely distributed on the floor and walls at the first river crossing.The scallop dimension can be related to flow velocity, and probably represents conditions under peak annual flow.Scallop lengths were measured at 10 places each on the floor and walls, chosen at random over an area of several square meters, and are reported in Table 1.

The standard method for interpreting scallop dimensions comes from Curl (1974).He suggests taking a weighted mean of the scallop dimensions (the Sauter average).For the floor and walls the weighted mean dimensions are 46 and 35 mm, respectively.The scallop dimension is then related to the flow velocity through a scallop Reynolds number and the ‘law of the wall’, a widely-used relation in hydraulics.

Table 1. Scallop dimensions at river crossing, in mm.























For flow velocity calculations, I assumed that the flooded passage at the river crossing was 35 meters wide and 10 meters tall.I calculated flow velocity for parallel walls 10 meters apart.I assumed a water viscosity of 1 cp, consistent with a water temperature of 20°C.For 35 mm scallops, the calculated water velocity is 1.2 m/s, or 4.4 km/hr.This corresponds to a discharge of 400-450 m3/s.

The peak annual discharge of rivers in central Vietnam is not widely available in the English literature, or is not known to me.However, there is a report for discharge of the Tra Khuc river, approximately 400 km to the southeast, available online ( Tra Khuc river has a catchment area of 3240 km2, and annual rainfall ranging from 1.8-3.6 meters across the watershed.The mean maximum discharge for this catchment is approximately 7000 m3/s, corresponding to a specific discharge of about 2.1 m3/s per km2.If the Tra Khuc river is an appropriate analog for the Hang Son Doong watershed, then the discharge estimated in the cave is consistent with an upstream drainage area of about 200 km2.

Active flooding of the southern section of the cave is evident all the way to the WOFD doline collapse.During floods, water is apparently dammed behind the collapse pile, filling the passage approximately 100 meters deep, to the elevation of the Level Playing Fields.The Level Playing Fields is a flat surface on the top of a gour sequence, capped by a layer of sand.The sand is active during the annual flood.Footprints from 2009 were not visible in 2010.There are small channels that are apparently active on this surface, and the sand shows evidence of water flow.Floodwater apparently drains today through a bypass called the Oxbow, where it goes to a lower level of the cave.

There is evidence that the southern section of the cave has a long history of being filled with ponded water during the rainy season.The lower half of the cave contains a thick deposit of laminated muds and fine sands.These sediments are best exposed underneath the formation called the Hand of Dog (HOD), which is composed of a porous combination of gypsum and carbonate.The massive speleothems in the vicinity of HOD all show evidence of faulting, slipping, and wholesale collapse into the main passage as the formations are undercut and lubricated by the mud banks on which they formed.Two samples were taken to ascertain the age of the sediments at HOD: one for paleomagnetism, and another for cosmogenic nuclide burial dating.Dating results are reported in detail in a later section, but the sediments here are likely younger than about 300,000 years old.

b. Central section

The central section of the cave is fossil, with no evidence of river flow in the recent geologic past.The area near the WOFD collapse doline has a fantastic set of gours formed from ceiling dripwater.Elsewhere in this section there are speleothems indicative of dripwater, but no externally-derived sediment and no evidence of flooding.

c. Northern section

The northern section of the cave is mostly fossil passage.There is some evidence of flooding in the geologic past in the vicinity of The Alcove in the GOE collapse doline.An unconsolidated deposit of large cobbles and small rounded boulders indicate very rapid streamflow through a passage near floor level just to the south of The Alcove.However, all of the cobbles and boulders are composed of limestone.I observed no externally-derived sediment.This indicates that the water is either coming from a different source, or that the sediment has been filtered and is not transported to this part of the cave.This deposit does not represent simply a downstream extension of the same river that entered the southern section.The sediment is also clearly not associated with the time of speleogenesis, when a river would have traveled through the entire length of the cave.

At Camp II (north of The Cormorant) there are several meters of laminated silt indicative of flooding.The source of the water is likely a rising sump.There is no evidence of externally-derived sediment, so this water is either from a local source or it is filtered from the main river below.The silts are of unknown age, but the surface shows signs of active transport, so they are probably flooded episodically today.

Continuing north, the Passchendaele section carries water even under base flow conditions.The presence of cave blindfish indicates persistent streamflow.The sediments here are several meters thick, and consist primarily of wet mud.A bathtub ring extends several meters above the sediments, illustrating the limit of annual flooding.Floodwater here is likely of two sources: a rising sump and dripwater.The Passchendaele section is beneath a doline on the surface that funnels dripwater into the cave, even during the dry season.The Great Wall of Vietnam is a muddy flowstone that has accumulated beneath the doline.The main source of water here is probably the same as at Camp II.There is no evidence of externally-derived sediment.


A dedicated effort was made to search for samples that could constrain the age of either the cave or the sediments within it.Suitable samples for dating caves are often found in sediments filling pockets on the cave wall, or high-level passages.Speleothem of pure calcite can be dated by U-series or U-Pb under suitable conditions.Despite a careful search, no such sediments were found.However, two sedimentary fills that postdate cave formation were found and samples were collected for dating.These were 1) the laminated silts behind WOFD at Hand of Dog, and 2) a passage-filling breccia at The Alcove.

1. Hand of Dog.

Laminated silt and mud beneath the speleothem at Hand of Dog were deposited in a quiet water environment, probably in ponded water behind a dam formed by the collapse at WOFD.The sediments thus probably date to the time of doline collapse.They are much younger than the cave itself, and have little bearing on the time of speleogenesis.

Two types of samples were collected.Three cubes of sediment were collected in oriented plastic boxes for paleomagnetic analysis.Each cube consists of a 1 inch (2.54 cm) square pressed into an oriented surface.The cube is 5/8 inch (1.6 cm) deep.Three samples were taken on a vertical face oriented N12°E.Each cube was marked, and oriented using a bubble level.The samples were sent to Josep Pares at CENIEH (Centro Nacional de Investigación sobre la Evolución Humana), Burgos, Spain for analysis.All three samples were determined to be magnetically normal, indicating that they were deposited less than 780,000 years ago.

A sample of silt was also collected for cosmogenic nuclide burial dating.This dating method is based on the decay of cosmogenic 26Al and 10Be in the mineral quartz after deposition in the cave (Granger and Muzikar, 2001).Successful dating requires that there be a measurable quantity of these nuclides, and that the aluminum concentration in the quartz be low(10-100 ppm).The silt sample was found to contain abundant quartz, however the aluminum content was moderately high (~250 ppm).The measurements were therefore not very precise. Cosmogenic nuclide data are reported in Table 2, and age calculations are reported in Table 3.The age of sediment at Hand of Dog was determined to be young, -0.27 +/- 0.32.Even though the age is negative (a physical impossibility), it overlaps within error with zero, and is thus statistically normal for a young sample.A sensible maximum age for the sediments is two standard errors older than the calculated age.The age of the sediments is therefore determined at 95% confidence to be younger than 0.37 My.

Both paleomagnetic and cosmogenic data indicate that the sediments at HOD are young.The best estimate for the timing of collapse at WOFD is within the past 370,000 years.I anticipate measuring this sample again as part of an experimental effort to improve the precision of 26Al measurements, so there is the possibility that the precision of this age will improve in the future.

2. The Alcove

The Alcove is a filled side passage that intersected Hang Son Doong at the Garden of Edam collapse doline.The Alcove represents a passage essentially as large as that of the main cave, and would have been a major tributary passage at the time of cave formation.The cave likely carried water sourced from a stream passing through the village along the trail to Hang En.At The Alcove, the passage is entirely filled with a breakdown breccia containing abundant externally-derived sediment.The passage collapse here therefore occurred after the time of speleogenesis, but while the river was still occupying the main passage.

A sample of sediment was collected here for cosmogenic nuclide burial dating.Results are reported in Table 2 and Table 3.This sample also contained a moderately high (~150 ppm) aluminum concentration, precluding very accurate dating.Moreover, the sample was surprising old, meaning that the cosmogenic 26Al concentration was very low.The 26Al measurement is therefore subject to very high uncertainty.The burial age of this sample is 3.0 +2.0/-1.0 My, bracketing the age of the sediment to 2-5 My.As with the sample from Hand of Dog, I anticipate re-measuring this sample in the near future in an experimental effort to improve 26Al precision.

Although the uncertainty in the burial age at The Alcove is disappointingly high, the result clearly points to the antiquity of the cave, placing speleogenesis in the Pliocene or possibly the latest Miocene.Effort should be made to re-date this sample if at all possible.

Table 2. Cosmogenic nuclide analysis


mass Qtz


Be spike









(x 10-15)

(x 10-15)

(106 atoms/g)

(106 atoms/g)





79.2 ± 11.5

89.2 ± 5

0.424 ± 0.065

0.055 ± 0.003





10.9 ± 6.4

36.3 ± 6.3

0.036 ± 0.021

0.023 ± 0.004

*Measured against standard KNSTD07.

Table 3. Burial age calculations



Burial age*




7.704 ± 1.260

-0.28 ± 0.32


1.572 ± 0.966

3.05 +1.98/-0.99

*Burial age calculated using radioactive meanlives for 26Al and 10Be of 1.02 and 2.005 My.Production rates are taken as 4.1 and 27.9 atoms/g/yr for latitude 17.5°N and altitude 0.4 km.


There are several observations of local geomorphology and geology that lead to a general theory of speleogenesis for Hang Son Doong.Most tellingly, a high-level paleo-valley is sculpted across the surface, draining westward (Figure 1).This paleo-valley is carved into the limestone, but pre-dates cave formation.This implies that there was no subterranean outlet for the water from the basin, or else caves would surely have carried the water.The basin is surrounded to the east by metamorphic and crystalline rocks, and to the south by Cretaceous redbeds.A portion of the northern boundary of the watershed also abuts onto Cretaceous redbeds, preventing cave formation.The remainder of the watershed is surrounded by large faults.I suggest that these faults served as aquicludes, preventing water from passing through, and stymieing cave development.In particular, the fault along the ‘Road 20’ valley would have blocked water flow towards the Son River.It was only when this fault was breached, probably by incision of a surface stream in the valley to which Hang Son Doong exits to the north, that a hydrologic outlet was available for the watershed.At that time, the water draining the Hang Son Doong watershed would have been perched in a riverbed perhaps 200 meters above the new base level where the fault was breached.Water immediately began forming caves to this outlet.The fastest flow path to the outlet was along a pre-existing N-S (Paleogene) fault, and another NW-SE (Neogene) fault.The cave expanded rapidly, taking the entire discharge of the watershed by wholesale capture of the surface river.Cave formation was probably rapid, as suggested by the lack of tributary passages and the volume of water that must have been captured.

Summary of observations

Hang Son Doong formed as a result of stream capture along a pre-existing N-S trending fault.The capture likely occurred due to breaching of a different NE-trending fault that blocked water flow until sometime during the Pliocene to latest Miocene (2-5 My).The cave today takes an estimated peak annual discharge of 400-450 m3/s, consistent with a drainage area of 200 km2.The massive size of the cave is likely due to several factors.1) The limestone here is generally thickly bedded and able to support a wide ceiling.2) The cave has developed to a fairly uniform width, possibly influenced by the width of the fault breccia zone in which it formed. Cave enlargement has thus been primarily vertical rather than horizontal.3) The cave has few tributaries that would weaken the ceiling.In cases where either conditions (1) or (3) are violated, the cave has collapsed.The age of the collapses is probably variable, but Watch Out for Dinosaurs likely collapsed during the past 370,000 years.


Curl, R. L., 1974, Deducing flow velocity in cave conduits from scallops, National Speleological Society Bulletin, 36: 1-5.

Granger, D.E, and Muzikar, P.F., 2001, Dating sediment burial with in situ-produced cosmogenic nuclides: Theory, techniques, and limitations, Earth and Planetary Science Letters, 188:269-281.

Lepvrier, C., Maluski, H., Van Tich, Vu, Leyreloup, A., Phan Truong Thi, Nguyen Van Vuong, 2004, The Early Triassic Indosinian orogeny in Vietnam (Truong Son Belt and Kontum Massif); implications for the geodynamic evolution of Indochina, Tectonophysics, 393:87-118.

Rangin, C., Huchon, P., Le Pichon, X., Bellon, H., Lepvrier, C., Roques, D., Nguyên Dinh Hoe, Phan Van Quynh, 1995, Cenozoic deformation of central and south Vietnam, Tectonophysics, 251:179-196.


Figure 1.

Digital Elevation Model showing topography in the Ke Bang massif in the vicinity of Hang Son Doong.Major lineations (faults) are highlighted in yellow.Note the prevailing tendency of N-S and NW-SE trending faults, as well as the younger NE-trending major faults.A karst paleovalley is highlighted in blue.The valley does not take water today, but predates cave development.Hang Son Doong (HSD) formed near the intersection of a N-S fault with the paleovalley.At the time of speleogenesis the cave pirated water from the paleo-river and carried it along the N-S fault to discharge ultimately into the Son River.


Darryl Granger


2012 Report


report2009 d3d8182ca9deacbf240dc78f03bb63ba

2007report 7dfb2a4b00f807179d3d38fd3ebaa4fe

report2005 165c622e74bdbdb9191663324fefabeb