Caves are defined as cavities large enough to admit a human being, typically formed in limestone1. The main origin of cave formation is the chemical action of water dissolving the limestone, which is very slightly soluble.
Caves can be formed in other types of rock by non-chemical action. For example, some sea caves may be formed by the action of waves on insoluble rocks, where they particularly exploit zones of mechanical weakness. However, these caves tend to be small in depth.
Lava caves, also known as lava tubes, are formed by the roofing-over of a flowing lava stream with a crust of air-cooled lava. The subsequent draining of molten rock from the now-covered channel can leave a tube-like cavity behind. Such caves can be several kilometres long, but are very linear and are close to the surface. Once formed, little change occurs, beyond the odd roof collapse.
The biological or physical concentration of calcium carbonate that forms the bulk of limestone rock builds up over geological timescales as sediment under marine or fresh water, usually shallow and clear. At times when significant carbonate deposition is taking place, relatively small amounts of other material washed into the area of sedimentation can become incorporated in the future rock, affecting the colour or texture.
During phases of minimal or no carbonate deposition, or when the influx of material is rapid, other materials may be washed in and deposited on the accumulating sediment to eventually form distinct layers of non-carbonate material (such as shale or clay) in the future rock. Distinguishable layers of rock that were deposited successively over time are referred to as beds, and the junction between two beds is known as a bedding-plane.
Eventually, if the sediment becomes buried sufficiently deeply, the resulting pressure and temperature results in the formation of solid limestone. The thickness of the individual beds within the limestone can vary from millimetres or centimetres up to several metres, and the resulting limestone is referred to as finely or massively bedded, respectively.
Subsequent geological processes can lift the rock to heights of hundreds or thousands of metres above sea-level. In the process, the rock may become stretched, thickened, folded, faulted, or tilted in comparison to the original horizontal orientation of the beds. In addition to the natural bed-structure, as a result of the upheaval procedure limestone often contains naturally occurring fractures, known as joints, which are broadly planar, and roughly at right angles to the beds2.
Carbon Dioxide and Limestone Solubility
The capacity of water to dissolve limestone depends on the acidity of the water. Carbon dioxide, when dissolved in water, undergoes a chemical reaction to form carbonic acid. As such, the greater the concentration of CO2 in water, the more acidic it is. Rainwater containing low levels of CO2 dissolved from the atmosphere is a very weak solvent for carbonate rocks, hence cave development is generally a slow process. Due to biological action, the level of CO2 in soil is very much higher than in the atmosphere, and therefore water which has drained through soil may contain much more dissolved CO2 than rainwater. This makes it more effective at dissolving limestone. In addition, other processes that occur in soil may add different acids to the water draining through it, so the nature of the landscape above caves can have great influence on the speed of dissolution.
The dependence of limestone solubility on CO2 levels also helps to explain the formation of the well known cave deposits called stalactites (which hang from the ceiling) and stalagmites (which grow on the ground), and other carbonate deposits inside caves. These deposits are commonly formed from water seeping through small fissures. The water moves slowly, allowing it to become saturated with dissolved limestone. When the saturated solution enters the cave passage, CO2 can leave solution, and if the humidity is below 100%, water can evaporate, leaving the solution supersaturated with limestone which is then deposited. Since little or no CO2 loss would occur if the concentration in the seepage water was in equilibrium with that in cave air, it is generally water which has collected an increased CO2 load by passing through soil that will result in the growth of cave formations. This helps explain why many cave systems formed in mountain areas with limited and very thin soil cover often contain few formations compared to lowland caves.
Cave Passage Development
Initial Channel Formation
For the sake of simplicity, in the descriptions below it is assumed that the beds of limestone are broadly level, and the joints are therefore relatively vertical.
The initial development process of a cave is extremely slow, lasting millions, or tens of millions of years. It involves the very gradual dissolution of limestone by water which either exists within the body of the rock from the time of its formation, or which percolates very slowly through the rock. This dissolution often occurs preferentially along joints, bedding planes, and faults if any are present. If sulphur-containing minerals, such as pyrite, are present within the rock3, any sulphuric acid formed by their oxidation can greatly assist this initial dissolution process.
Over time, a connected network of small channels may be formed in the rock. If, as a result of gradual erosion or glacial action, the surface relief changes such that a point of this network becomes exposed, water from places in the network above that point can drain out. If the network extends upwards to one or more points where stream water or groundwater can enter the network at a higher altitude, a significant flow of water can start, and much more rapid solution can begin. The point where the water drains out onto the surface is known as a spring or resurgence.
The water table is the underground boundary dividing the saturated zone (where all pores and channels in the rock are filled with water) from the drier zone above. Initially, all channels are formed below the water table, and thus they are flooded. As their entire surface is in contact with the water, solution tends to result in a rounded channel cross-section.
In the case of joint-formed channels, it is common for the passage to become stretched vertically to some extent into a more oval form. This can be as a result of two processes. Either the original channel itself becomes flattened in the plane of the joint, or small channels formed earlier in the same joint become incorporated into the main channel as it expands. In some cases the 'ends' of the oval are simply rounded, but in others, there is a noticeable 'peaking' at the site of the joint itself.
The same effects can occur in the growth of bedding-plane channels, with the additional possibility that a band of physically weak material, such as a clay or shale layer, may exist between beds of limestone, and can easily be removed by the mechanical action of flowing water. This can cause a great sideways extension of the channel cross-section, leaving a conduit that could be centimetres high, but metres wide.
Another possibility in bedding-controlled channels is that the bed of rock on one side of the bedding plane may be significantly less soluble than the bed on the other side, so the developing circular or oval passageway may be relatively flattened above or below. In the extreme case where a passage develops along the junction between a bed of limestone and a bed of effectively insoluble rock, the developing passage can be completely un-eroded on the non-limestone side.
Subsequent Passage Development
Once a flow through the channel network has been established, the drainage of water from the rock can lower the water table such that channels above the level of the resurgence may become partially air filled. However, depending on the layout of channels, the level of the water table may be different at different parts of the network.
As a cave system develops, the water table can change further. Usually, this involves a lowering of water levels, as outside valleys deepen and uncover lower resurgences, or internal development of lower-level drainage networks.
However, it is possible, albeit much less likely, that passages may be blocked by material washed into the cave system, or that resurgences may be blocked by movements of surface material. This means that local or global rises in the internal water tables are possible.
Vadose and Phreatic Passages
The zones of rock above and below the water table are known as the vadose and phreatic zones respectively, and the passages formed there are named similarly. As mentioned above, all passages are initially phreatic, but changes in the water table over time may leave some of them partially air-filled channels, and vadose development can begin.
Since there is no flowing water in contact with the upper parts of vadose passages, solution only affects the lower portion. The passage develops by down-cutting, and usually develops a canyon profile. Traces of the phreatic origins may remain visible at the top of the passage. A roughly semi-circular cap above the canyon forms if down-cutting takes place over most of the width of the initial phreatic passage, whereas the bulk of the initial rounded passage may remain above a narrower canyon, giving a classic keyhole profile. In the case where the initial passage was a relatively flat bedding-plane, the resulting passage may either be a flat-roofed canyon, or a passage with a T-shaped cross-section.
In wide passageways, solution caused by water seeping through bedding planes and joints in the roof can cause weaknesses to develop to the point where collapse of boulders can occur. This sometimes leaves the roof rather irregular, but other times it can give a new flat roof in place of an old rounded or irregular one. This latter case can be explained by the post-collapse roof surface being part of a relatively thick, unfractured or otherwise mechanically competent bed of rock, which is separated from the underlying rock bed by a layer of particularly weak or soluble material.
In some cases, the fallen debris still remains as a boulder layer, often with distinctly flat-sided boulders. However, if the last collapse was a long time ago, and the passage carries a stream, it is possible for the boulders to be eradicated by slow solution or to be physically moved downstream in the case of flooding.